A process capability convergence method and system based on cross-device bias aggregation

Abstract

The application discloses a process capability convergence method and system based on cross-device deviation collection. By identifying an uncompleted unstable compensation response in a performed compensation action, and collecting the quality characteristic deviation and the compensation response across devices, a deviation collection result with a reserved action attribution relationship is formed. According to the result, a device group process center, an inter-device dispersion state and a process convergence margin are determined, a compensation response is synthesized, and a compensation intervention relationship is established when the process convergence margin is exceeded. According to the intervention relationship, a compensation action, an execution authority and a compensation constraint are determined, compensation control data is formed, and parameter compensation is implemented. The deviation collection result and subsequent control data are updated with actual responses, and a process capability convergence result is generated when the device group meets a convergence condition, so that the cross-device quality consistency, the compensation control stability and the process capability convergence efficiency are improved.

Description

Technical Field

[0001] This application relates to the field of industrial production process control technology, and in particular to a process capability convergence method and system based on cross-equipment deviation aggregation. Background Technology

[0002] In discrete manufacturing, process manufacturing, and continuous production, multiple similar or process-related pieces of equipment typically share the production task of the same product, the same process, or adjacent processes. The quality characteristic deviations resulting from variations in processing load, material batches, equipment wear, environmental conditions, and control parameters often exhibit inter-equipment differences and continuity across production cycles. To maintain the process center position of the equipment group, minimize quality fluctuations between equipment, and ensure the overall process capability of multiple pieces of equipment remains stable, it is necessary to continuously adjust process parameters based on the quality inspection results of each piece of equipment, and to uniformly manage the quality characteristic responses resulting from parameter compensation.

[0003] Currently, multi-equipment production lines typically acquire the current quality characteristic deviation for each individual piece of equipment and generate corresponding process parameter compensation amounts based on preset thresholds, process experience, historical statistical relationships, or single-equipment feedback results. After the compensation action is issued, subsequent quality inspection results are used to determine whether the deviation of that equipment has decreased, and the compensation range is adjusted accordingly or the current compensation is terminated. For the control of multiple pieces of equipment, each piece of equipment is usually treated as an independent compensation object, with separate compensation records and quality inspection results saved, and the deviation of each piece of equipment in its current production cycle used as the basis for the next compensation judgment.

[0004] When using the above processing method, multiple devices perform deviation compensation separately. It is difficult to uniformly aggregate the quality characteristic responses of previously executed compensation actions that have not yet been completed with the newly proposed compensation responses. Compensation effects already in the response release phase are prone to superimposed with new compensation effects in subsequent production cycles. When multiple compensation actions act on different devices, their combined effect can also cause process center migration and changes in discrete states between devices. The compensation results of a single device are difficult to reflect the composite response at the device group level, and it is also difficult to identify and constrain compensation interference between different compensation actions. This can easily lead to duplicate compensation, overcompensation, and difficulty in achieving stable convergence of the device group's process capabilities. Summary of the Invention

[0005] In view of the above-mentioned actual situation, this application proposes a process capability convergence method and system based on cross-device deviation aggregation, in order to solve the problems in the prior art where multiple devices perform deviation compensation separately, the quality characteristic response of historically executed compensation actions that have not yet been completed and the newly added proposed compensation response are difficult to uniformly aggregate, and the process center migration, discrete state changes between devices and compensation interference between actions caused by the combined effect of multiple compensation actions are difficult to identify and constrain, thus causing repeated compensation, overcompensation and difficulty in stable convergence of process capability of device groups.

[0006] A process capability convergence method based on cross-device deviation aggregation includes:

[0007] Acquire historical process parameter compensation records and corresponding actual quality characteristic responses of multiple target devices; extract compensation actions executed before the current production cycle from the historical process parameter compensation records; identify compensation actions that have not yet met the preset response stability conditions based on the actual quality characteristic responses and preset response stability conditions; and determine the quality characteristic responses of the compensation actions that have not been completed since the current production cycle as unstable compensation responses.

[0008] Obtain the quality characteristic deviation and proposed compensation action of each target device in the current production cycle, and determine the proposed compensation response corresponding to each proposed compensation action; establish the association between the quality characteristic deviation of each target device and the corresponding non-stability compensation response and proposed compensation response and perform cross-device aggregation, retain the attribution relationship between the non-stability compensation response and the proposed compensation response and the corresponding compensation action, and form the deviation aggregation result;

[0009] The process center and discrete states between equipment in the equipment group are determined from the deviation aggregation results, and a process convergence margin is formed accordingly. The unsteady compensation response and the pseudo-compensation response that work together are synthesized to form a synthetic compensation response that characterizes the migration of the process center and the changes in discrete states between equipment in the equipment group. When the synthetic compensation response exceeds the process convergence margin, a compensation interference relationship is established between the corresponding compensation actions according to the attribution relationship.

[0010] Based on the compensation interference relationship, determine the current compensation action from the proposed compensation actions that participate in the compensation interference relationship and configure the corresponding compensation execution permission. At the same time, set compensation constraints for the proposed compensation actions to form compensation control data.

[0011] Process parameter compensation is performed based on the compensation control data, and the deviation aggregation result is updated based on the actual quality characteristic response of the compensation action already performed. After any current compensation action meets the preset response stability condition, the compensation interference relationship and subsequent compensation actions are re-determined based on the updated deviation aggregation result, and new compensation control data is formed. When the equipment group meets the process capability convergence condition, process capability convergence result is generated.

[0012] The process capability convergence method and system proposed in this application based on cross-device deviation aggregation achieve unified aggregation and action attribution maintenance of current quality characteristic deviations, historical unstable compensation responses, and newly added proposed compensation responses of multiple target devices. Through joint evaluation of the process center and discrete states between devices in the device group, identification of compensation interference relationships, configuration of compensation execution authority, and updating of actual quality characteristic responses, each compensation action is implemented in a controlled order, which promotes the continuous convergence of the process center and discrete states between devices in the device group to the corresponding target range, thereby improving cross-device quality consistency, compensation control stability, and process capability convergence efficiency. Attached Figure Description

[0013] Figure 1 This is a flowchart illustrating a process capability convergence method based on cross-device deviation aggregation.

[0014] Figure 2 This is a schematic diagram illustrating the formation of the compensation response reference continuation and the non-steady compensation response in this embodiment;

[0015] Figure 3 This is a schematic diagram of the cross-device deviation aggregation section in this embodiment;

[0016] Figure 4 This is a schematic diagram of the center-discrete state projection and boundary excess of the single-item-joint response in this embodiment;

[0017] Figure 5 This is a graph showing the boundary occupancy rate and out-of-bounds cycle of a single-item joint response in this embodiment;

[0018] Figure 6 This is a graph showing the full-cycle bearing capacity and execution range of the current compensation action in this embodiment;

[0019] Figure 7 This is a schematic diagram of the association closure of the compensation reconstruction domain and the domain division processing of control data in this embodiment;

[0020] Figure 8 This is a schematic diagram of a process capability convergence system based on cross-device deviation aggregation, provided as an embodiment of this application. Detailed Implementation

[0021] To make the objectives, technical solutions, and technical effects of this application clearer, the specific embodiments of this application will be further described below with reference to the accompanying drawings. The following embodiments focus on the quality characteristic deviations, process parameter compensation records, and compensation responses corresponding to multiple target devices, and are used to illustrate the specific processing procedures for cross-device deviation aggregation, compensation interference delimitation, and process capability convergence.

[0022] The equipment operation data, quality inspection data, and process parameter compensation records used in this embodiment are sourced from the manufacturing execution system, quality management system, process database, equipment controller, or compensation action log, and are acquired, retrieved, and written back within the data permission scope configured in the target production system. Industrial communication protocols, database read / write operations, timestamp synchronization, conventional unit of measurement conversion, and log storage are implemented using methods well-known to those skilled in the art, and will not be elaborated upon here.

[0023] This application applies to industrial production processes in which multiple target devices jointly undertake the same controlled quality characteristic processing task, or jointly constitute the same process capability evaluation domain. The target devices include production equipment of the same model, as well as process-equivalent equipment with different models but performing the same process function, and whose corresponding quality characteristic deviations and compensation responses can be converted to a unified evaluation scale.

[0024] Multiple target devices included in the same process capability evaluation domain correspond to the same controlled quality characteristics, or to controlled quality characteristics with the same engineering meaning after scale transformation; the quality inspection results of each device are converted into quality characteristic deviations relative to a unified quality target; the production data of each device are aligned according to a unified production cycle; each compensation response is associated with a specific target device, controlled quality characteristic, compensation action, and production cycle. The quality characteristic deviation, actual quality characteristic response, proposed compensation response, and non-steady-state compensation response corresponding to the same controlled quality characteristic use the same unit of measurement and sign direction.

[0025] For multiple controlled quality characteristics with different dimensions, deviation aggregation, equipment group status calculation, compensation interference judgment, and convergence verification are performed separately. When the same proposed compensation action applies to multiple controlled quality characteristics simultaneously, each controlled quality characteristic forms a corresponding constraint result, which is then uniformly converged to the proposed compensation action at the levels of compensation execution authority and compensation range. In other words, when the same process parameter adjustment simultaneously affects multiple quality indicators, the action is implemented according to the control range commonly allowed by all quality indicators.

[0026] The production cycle is a discrete time unit for performing quality characteristic deviation collection, compensation response projection, and compensation control update. One production cycle corresponds to a processing batch, a quality statistics window, a production interval formed by a preset quantity of products, an equipment control sampling cycle, or other production time periods that can form effective quality inspection results.

[0027] The current production cycle is the starting cycle for this cross-equipment deviation collection and compensation control calculation. The quality characteristic deviations formed by each target equipment in the current production cycle are used as the starting state for the forward projection of the unsteady compensation response and the proposed compensation response.

[0028] The response cycle is a relative periodic sequence formed by arranging the subsequent quality characteristic responses of a compensation action, starting from the execution time of that action. The response cycle and production cycle are correlated based on the execution time of the compensation action and a preset cycle division rule, ensuring that the production cycle in which the compensation action begins to produce a identifiable quality characteristic response corresponds to its first response cycle. When the compensation action is executed in the middle of a production cycle, the current production cycle or the next complete production cycle is determined as the first response cycle according to the preset division rule; the same response cycle division method is used for the same timing compensation response benchmark and the same compensation control calculation.

[0029] A production batch comprises one or more production cycles. The updating of compensation interference relationships and the reconstruction of compensation control data are performed according to the production cycle. After the process capability convergence result is formed, the actual compensation process corresponding to this production batch is written into the historical process parameter compensation record and used as historical data for determining the compensation response of the next production batch.

[0030] Historical process parameter compensation records are data records that are structured and saved from process parameter compensation processes that have already occurred. They include compensation action identifier, target equipment identifier, process parameter items, controlled quality characteristics, compensation direction, compensation magnitude, execution time, quality characteristic deviation before execution, actual quality characteristic response in each response cycle after execution, response stabilization cycle, action execution status, and data version.

[0031] The current quality characteristic deviation is the difference between the quality characteristic detection value of the target equipment in the current production cycle and the preset quality target. When the same target equipment generates multiple detection values ​​within a production cycle, the quality characteristic status of that production cycle is determined according to the preset cycle representative value formation rule; all target equipment corresponding to the same controlled quality characteristic adopts the same cycle representative value formation rule.

[0032] A compensation action is a control action that adjusts one or more process parameters of a target device to cause the corresponding controlled quality characteristic to change in the expected direction. An executed compensation action is one that has been sent to the target device and whose process parameters have been adjusted.

[0033] The proposed compensation action is a process parameter adjustment suggestion generated based on the current quality characteristic deviation and in an evaluation-pending state. It includes the target equipment, process parameter items, proposed compensation direction, proposed compensation range, and target controlled quality characteristic. The proposed compensation action can be generated by the single equipment compensation module, the quality deviation diagnosis module, or manual process adjustment suggestions, and will enter the equipment execution stage after compensation execution authority is granted.

[0034] The quality characteristic response is the change in the controlled quality characteristic relative to the deviation baseline before the compensation action, after the compensation action is performed. Let the quality characteristic deviation before the compensation action be... The quality characteristic deviation after execution in the t-th response cycle is The corresponding quality characteristic response is: The quality characteristic response is represented by a signed numerical value, with positive and negative values ​​indicating changes in the controlled quality characteristic along a preset positive and negative direction, respectively. The actual quality characteristic response, the pseudo-compensated response, and the unsteady compensation response all use the same sign and direction.

[0035] The actual response already experienced refers to the portion of the response for which detection results have been obtained after the compensation action was executed and before the current production cycle. The remaining response sequence is the quality characteristic response generated in subsequent production cycles, starting from the actual deviation state of the current production cycle. The actual response already experienced is reflected in the current quality characteristic deviation, while the remaining response sequence enters the subsequent deviation aggregation process.

[0036] When multiple compensation actions occur simultaneously for the same target equipment, the attribution of the current measured quality characteristic change to each compensation action is determined based on the action execution time, response cycle sequence, and already registered non-steady-state compensation responses. Historical records with clear response attribution and complete response processes are entered into the time-series compensation response baseline, and the response attribution status is saved along with historical process parameter compensation records.

[0037] Cross-device deviation aggregation is a process that, according to the target equipment, controlled quality characteristics, and production cycle, configures the current quality characteristic deviation of each target equipment, the non-stable compensation response corresponding to the historically executed compensation actions, and the proposed compensation response corresponding to the proposed compensation action into the same evaluation structure, while retaining the attribution relationship between each compensation response and the corresponding compensation action.

[0038] The unstable compensation response is the residual quality characteristic response that has been partially generated by compensation actions performed before the current production cycle and will continue to be generated from the current production cycle onwards.

[0039] The proposed compensation response is the quality characteristic response that the proposed compensation action will produce within the subsequent expected response period, determined based on the proposed compensation magnitude and the relationship between the compensation magnitude and the proposed compensation action.

[0040] Compensation constraints are restrictions on the execution status and execution amount of proposed compensation actions that interfere with each other. These include compensation execution authority, the range of executable compensation range, the effective production cycle, and the action acceptance restrictions corresponding to the current compensation action response status.

[0041] The preset response stability condition is used to determine the response state of the compensation action that has been performed. This includes ensuring that the actual quality characteristic response change within a preset number of production cycles is within a preset stability threshold range, and that the remaining response since the current production cycle is within a preset residual threshold range. When a compensation action is applied to multiple controlled quality characteristics, the compensation action enters a response stability state after all relevant controlled quality characteristics meet the corresponding conditions.

[0042] The deviation reference component is the forward response projection starting component formed by the quality characteristic deviation of each target device in the current production cycle according to the target device, controlled quality characteristics and subsequent production cycle configuration.

[0043] The compensation response component is a set of response components formed by aggregating the unsteady compensation response and the proposed compensation response according to the target equipment, controlled quality characteristics, and production cycle. The deviation baseline component, the compensation response component, and the attribution relationship between the response and the compensation action together constitute the deviation aggregation result.

[0044] The process center of a group of equipment is used to characterize the central position of the quality characteristic deviations of multiple target devices under the same controlled quality characteristic within the group. The discrete state between devices is used to characterize the degree of separation of the quality characteristic deviations of multiple target devices relative to the process center of the group of equipment.

[0045] Process convergence margin includes the directional center margin formed by the process center of the equipment group relative to the upper and lower bounds of the target center, and the discrete margin formed by the discrete states between equipment relative to the upper bound of the allowable discreteness.

[0046] Compensation interference relationship is the action correlation formed when different compensation actions work together in the same production cycle, jointly respond to break through the corresponding process convergence margin, and each compensation action makes an effective contribution to the same excess boundary.

[0047] Compensation control data is the control result formed based on the compensation interference relationship, process termination margin, and deviation improvement state of the proposed compensation action. It includes the compensation execution authority, the compensation range, and the response stability trigger conditions.

[0048] In this embodiment, "starting from the current production cycle" means taking the measured quality characteristic deviation already formed in the current production cycle as the forward starting state, and recording the remaining response sequence starting from the first response position after this forward starting state; the actual responses already carried before the current production cycle and in the current measured state are included in the deviation reference component.

[0049] Process capability convergence is achieved by constraining the process center and discrete states between equipment in a group of equipment, so that the controlled quality characteristics of multiple target equipment remain within a preset control range during continuous measured production cycles, and the state of the forward equipment group after the remaining unsteady compensation response continues to be maintained within the corresponding control range.

[0050] Figure 4 The deviation reference group state is a graphical combination term, representing the discrete state of the equipment group process center and between equipment formed based on the deviation reference components. Figure 4 The process convergence constraint domain in the diagram is a graphical region. Its horizontal range is defined by the lower and upper bounds of the target center, and its vertical range is defined by the upper bound of the allowable discreteness. It is used to represent the allowable state range corresponding to the process convergence margin.

[0051] Before implementing this method, standardize the target equipment identifier, controlled quality characteristic identifier, production cycle number, response cycle sequence, and unit of measurement, and ensure the time correspondence between compensation actions and quality inspection records. The same compensation control calculation should use a consistent target equipment scope, cycle division rules, historical response baseline version, parameter version, and data cutoff point.

[0052] In multi-equipment process compensation, executed compensation actions will continue to generate quality characteristic responses after the current production cycle. The combined effect of different equipment compensation responses simultaneously alters the process center and discrete states between equipment groups. Compensation actions within the allowable range are evaluated separately; when acting together within the overlapping response cycle, they will fill the same process convergence boundary. After the current compensation action enters a stable state, the deviation baseline, compensation interference relationship, and subsequent action succession status are updated accordingly. After the current measured state enters the control range, the remaining response continues to be released, further changing the state of the subsequent equipment group. The process of updating the compensation interference relationship, determining the current compensation action, configuring the compensation amplitude, and performing authorized joint verification is repeated until all process convergence margins in the authorized joint projection reach zero or higher, or until there are no pending compensation actions with open compensation execution permissions in this round.

[0053] The following is combined Figures 1 to 7 The document further explains the identification of non-steady-state compensation response, cross-device deviation aggregation, process convergence margin formation, compensation interference delimitation, compensation action inheritance, control data reconstruction, and process capability convergence process.

[0054] like Figure 1 As shown, this embodiment provides a process capability convergence method based on cross-device deviation aggregation, including steps S101 to S105.

[0055] S101, acquire historical process parameter compensation records and corresponding actual quality characteristic responses of multiple target devices; extract compensation actions executed before the current production cycle from the historical process parameter compensation records; identify compensation actions that have not yet met the preset response stability conditions based on the actual quality characteristic responses and preset response stability conditions; and determine the quality characteristic responses of the compensation actions that have not been completed since the current production cycle as unstable compensation responses.

[0056] S102, obtain the quality characteristic deviation and proposed compensation action of each target equipment in the current production cycle, determine the proposed compensation response corresponding to each proposed compensation action; establish the association between the quality characteristic deviation of each target equipment and the corresponding non-stable compensation response and proposed compensation response and perform cross-equipment aggregation, retain the attribution relationship between the non-stable compensation response and proposed compensation response and the corresponding compensation action, and form the deviation aggregation result.

[0057] S103, determine the process center and discrete state between equipment in the equipment group based on the deviation aggregation result, and form the process convergence margin accordingly; synthesize the unsteady compensation response and the quasi-compensation response that work together to form a synthetic compensation response that characterizes the migration of the process center and the change of discrete state between equipment in the equipment group; when the synthetic compensation response exceeds the process convergence margin, establish a compensation interference relationship between the corresponding compensation actions according to the attribution relationship.

[0058] S104. Based on the compensation interference relationship, determine the current compensation action from the mutually interfering proposed compensation actions and configure the corresponding compensation execution permissions. At the same time, set compensation constraints for the mutually interfering proposed compensation actions to form compensation control data.

[0059] S105, perform process parameter compensation based on compensation control data, and update the deviation aggregation result based on the actual quality characteristic response of the executed compensation action; after any current compensation action meets the preset response stability condition, redetermine the compensation interference relationship and subsequent compensation actions based on the updated deviation aggregation result, and form new compensation control data; when the equipment group meets the process capability convergence condition, generate the process capability convergence result.

[0060] In step S101, the historical process parameter compensation records and corresponding actual quality characteristic responses of multiple target devices are used as inputs to extract the compensation actions that have been completed in process parameter adjustment before the current production cycle, and obtain the corresponding actual quality characteristic responses according to the target device, process parameter item, controlled quality characteristic, compensation direction, compensation magnitude and execution time.

[0061] For each executed compensation action, a preset response stabilization condition is determined based on the changes in the actual quality characteristic response within the continuous production cycle and the remaining response state from the current production cycle. The actual quality characteristic response reflects the process of the compensation action already taking place, while the remaining response reflects the quality characteristic effect that the compensation action will continue to release after the current production cycle. Executed compensation actions that are still in the response release state are determined to be unstable compensation actions, and their quality characteristic responses that have not yet been completed since the current production cycle are determined to be unstable compensation responses.

[0062] The non-stability-reaching compensation response is associated with the corresponding executed compensation action, target equipment, controlled quality characteristic, and remaining production cycle position, and the response direction and response quantity for each remaining production cycle are recorded. When an executed compensation action acts on multiple controlled quality characteristics simultaneously, each controlled quality characteristic forms a corresponding non-stability-reaching compensation response, and together they maintain the attribution to the executed compensation action.

[0063] Figure 2 The reference continuation relationship is used to represent the non-steady-state compensation response. The historical compensation actions and their complete response profiles of multiple completed response stabilization processes constitute the time-series compensation response benchmark. After the action attributes of the compensation actions to be judged are matched, a compensation response reference sequence is formed. The current production cycle section is used to divide the actual response that has been experienced and the subsequent remaining response. The remaining response sequence after the section is used to determine the non-steady-state compensation response. Figure 2 The specific handling of reference continuation, residual response formation, and reaching the stable boundary will be discussed later.

[0064] The unsteady compensation response generated in S101, as a forward response that will continue to be generated outside the current equipment deviation state, enters the cross-equipment deviation aggregation performed in S102.

[0065] In step S102, the quality characteristic deviation and proposed compensation action of each target device in the current production cycle are obtained. The current quality characteristic deviation is used to characterize the equipment-level quality status at the starting point of this compensation control calculation, and the proposed compensation action is used to characterize the process parameter adjustment suggestions for the formation of the current deviation and awaiting control evaluation.

[0066] Based on the target equipment, process parameters, controlled quality characteristics, compensation direction, and proposed compensation magnitude of the proposed compensation action, the corresponding compensation magnitude response relationship is obtained from historical process parameter compensation records and corresponding actual quality characteristic responses. This is used to determine the proposed compensation response for each action within the subsequent expected response period. The proposed compensation response is recorded according to the target equipment, controlled quality characteristics, and expected response period, including the response direction and response amount, while maintaining the attribution to the corresponding proposed compensation action.

[0067] Based on the target equipment, controlled quality characteristics, and production cycle, the quality characteristic deviations of each target equipment in the current production cycle are correlated with the corresponding unsteady-state compensation responses and proposed compensation responses, and then aggregated across equipment. The current quality characteristic deviation forms the deviation baseline component, while the unsteady-state compensation responses and proposed compensation responses form the compensation response components. The attribution relationship between each compensation response and its corresponding compensation action is retained along with the compensation response components, thus forming the deviation aggregation result.

[0068] The deviation aggregation results use the equipment-level deviation state as the starting point for forward projection, and allocate the residual effects of historically executed compensation actions and the expected effects of current proposed compensation actions to the corresponding equipment, quality characteristics, and production cycles. To put it more intuitively, the current deviation is the underlying starting state, the unsteady compensation response and the proposed compensation response are the effects superimposed on a periodic basis, and the three maintain their respective sources in the same evaluation structure.

[0069] Figure 3 This is used to represent the hierarchical alignment of deviation reference components, unsteady compensation responses, and proposed compensation responses in the target equipment and production cycle dimensions, as well as the response attribution relationship between executed compensation actions and proposed compensation actions. Figure 3 The data structure, compensation action items, and multi-mass characteristic aggregation process of the deviation aggregation section shown will be elaborated later.

[0070] The deviation aggregation result formed in S102 enters S103, where the deviation reference component is used to determine the process center and discrete state between equipment in the equipment group, the compensation response component is used to form the compensation response projection, and the action attribution relationship is used to trace the joint out-of-bounds result back to the corresponding compensation action.

[0071] In step S103, according to the controlled quality characteristics and production cycle, the deviation reference component corresponding to each target equipment is obtained from the deviation aggregation result, and the process center of the equipment group and the discrete state between equipment are determined by the deviation reference component of each equipment.

[0072] The process center of the equipment group is used to characterize the group center position of the quality characteristic deviations of multiple target devices. The discrete state between devices is used to characterize the degree of separation of the deviations of each target device relative to the group center. The upward center margin and the downward center margin are formed according to the positional relationship between the process center of the equipment group and the preset target center interval. The discrete margin is formed according to the positional relationship between the discrete state between devices and the preset allowable discrete range. The process convergence margin is composed of the upward center margin, the downward center margin, and the discrete margin.

[0073] The unsteady-state compensation response and the proposed compensation response, which work together under the same controlled quality characteristic and within the same production cycle, are synthesized. Based on the equipment-level deviations after the combined effect of these compensation responses, the process center and discrete states between equipment are redefined. The change in the process center of the equipment group before and after the combined effect forms the center migration quantity, and the change in the discrete states between equipment before and after the combined effect forms the discrete change quantity. The synthesized compensation response is composed of the center migration quantity and the discrete change quantity.

[0074] Figure 4This is used to represent the projection relationship between individual compensation responses and joint compensation responses in the discrete state plane between the center and equipment in the process of the equipment group. The deviation reference group state constitutes the common starting state for all projections. The individual response projection reflects the equipment group state after each compensation action is taken, and the joint response projection reflects the equipment group state after multiple compensation responses are superimposed on the equipment-level deviation and recalculated. When the joint response projection exceeds the process convergence constraint domain, it forms a corresponding center boundary excess or discrete boundary excess.

[0075] Figure 5 This is used to represent the relationship between single-response projection, center migration occupancy, and discrete change occupancy in the production cycle dimension, and to determine the center out-of-bounds production cycle and discrete out-of-bounds production cycle accordingly. Figure 4 The focus is on representing the out-of-bounds state of the joint response in the device group state space. Figure 5 This focuses on indicating the production cycle position where the corresponding out-of-bounds state occurs. The specific formation process of individual response projection, joint response projection, and boundary occupancy will be discussed later.

[0076] When the synthetic compensation response exceeds the corresponding process convergence margin, the compensation action related to the corresponding out-of-bounds state is determined from the compensation actions that generate the quality characteristic response, based on the direction of the individual response corresponding to each compensation action and its contribution to the same out-of-bounds boundary. Then, according to the action attribution relationship retained in the deviation aggregation result, a compensation interference relationship is established between the corresponding compensation actions.

[0077] The compensation interference relationship is used to characterize the action-level correlation formed by the compensation actions on different target devices when they work together on the same process convergence boundary, and serves as the input for S104 to determine the order of compensation actions and the compensation control state.

[0078] In step S104, a follow-up evaluation is performed on the proposed compensation actions that interfere with each other, based on the compensation interference relationship, and the current compensation action is determined from the proposed compensation actions that interfere with each other. The follow-up evaluation is performed under the condition that the relevant unsteady compensation response continues to act, so that the selection of the current compensation action reflects both the residual effect of the previously executed compensation actions and the expected effect of the proposed compensation action itself.

[0079] For each proposed compensation action involved in the aforementioned compensation interference relationship, the changes in process convergence margin and target controlled quality characteristic deviation after the proposed compensation response are evaluated. The current compensation action is determined from the proposed compensation actions that satisfy the process convergence boundary constraints and reduce the target controlled quality characteristic deviation. For proposed compensation actions within the same compensation interference group that have not yet been determined as the current compensation action, corresponding compensation constraints are set according to the response state of the current compensation action.

[0080] For current compensation actions and proposed compensation actions not included in compensation interference relationships, based on the proposed compensation response, the unstable compensation response acting in conjunction with it, and the corresponding process convergence margin, the corresponding compensation magnitude constraints are determined, and compensation execution permissions, execution compensation magnitudes, and response stability trigger conditions are configured accordingly. Compensation execution permissions control whether the proposed compensation action enters the equipment execution phase; execution compensation magnitude limits the actual amount of process parameter adjustments issued to the target equipment; and response stability trigger conditions determine the trigger point for subsequent state updates after the compensation action has been executed.

[0081] Figure 6 This is used to represent the relationship between the directional center margin, discrete margin, and deviation reduction rate corresponding to the current compensation action and the compensation amplitude, and to form the action acceptance upper limit and execution compensation amplitude based on the margin constraints within the entire response cycle. The specific processing of compensation interference groups, compensation acceptance candidates, deviation reduction rate, compensation execution authority, and amplitude constraints will be discussed later.

[0082] The compensation control data is formed by the compensation execution authority, compensation range, and response stability triggering conditions corresponding to each proposed compensation action. The compensation control data is associated with the corresponding target equipment and the proposed compensation action, and then enters the process parameter compensation and actual response update process executed in S105.

[0083] In step S105, process parameter compensation is performed on the proposed compensation action with the compensation execution permission in the open state according to the compensation control data, and the corresponding process parameters of the target equipment are adjusted according to the execution compensation range. After the proposed compensation action is actually executed, it becomes an executed compensation action, and its corresponding proposed compensation response is updated according to the actual execution compensation range, and enters the subsequent actual response tracking process.

[0084] Before the stable trigger condition is triggered, the actual quality characteristic response corresponding to the executed compensation action is obtained according to the production cycle, and the deviation collection result is updated according to the actual response process, so that the non-stable compensation response used in subsequent control calculations is consistent with the actual action state of the executed compensation action.

[0085] Once any current compensation action satisfies the preset response stability condition, the deviation reference component is updated according to the actual quality characteristics of the executed compensation action, causing the corresponding non-stable compensation response to exit the compensation response component. The compensation interference relationship and subsequent compensation actions are then re-determined based on the updated deviation aggregation results, forming new compensation control data.

[0086] Figure 7This describes the associated reconfiguration process after the current compensation action reaches stability. The initial reconfiguration boundary is formed by the controlled quality characteristics affected by the stabilizing action. Unexecuted proposed compensation actions acting on the corresponding controlled quality characteristics are collected and extended along existing compensation interference relationships to related proposed compensation actions and newly added controlled quality characteristics, forming a compensation reconfiguration domain. The original compensation control data within the compensation reconfiguration domain is reconstructed based on the updated deviation collection results, while the original compensation control data maintaining consistency with external dependencies remains effective. The formation of the compensation reconfiguration domain, the reconfiguration of control within the domain, and the maintenance of control outside the domain will be discussed later.

[0087] During the continuous update of the compensation control state, the measured convergence state is determined based on the actual quality characteristic deviation of each target equipment in the continuous production cycle. The equipment-level quality characteristic deviation at the end of the continuous production cycle is used as the forward calculation state. The unfinished compensation response is then superimposed on the corresponding target equipment and controlled quality characteristics to determine the process center and discrete state between equipment in the subsequent equipment group.

[0088] When the discrete states of the process center and between equipment in the equipment group corresponding to a continuous measured production cycle meet the corresponding control range, and the discrete states of the process center and between equipment in the forward equipment group after the non-steady-state compensation response continues to act also meet the corresponding control range, it is determined that the equipment group meets the process capability convergence condition. The compensation execution permission for any unexecuted compensation actions is then disabled, the historical process parameter compensation record is updated, and the process capability convergence result is generated. The updated historical process parameter compensation record then enters the compensation response determination process for the next production batch.

[0089] In this embodiment, the non-stabilized compensation response is determined according to the following process: Compensation action description data is formed based on the target equipment, process parameters, and controlled quality characteristics affected by each executed compensation action, as well as the corresponding compensation direction, compensation magnitude, and execution time. Based on the compensation action description data, a time-series correspondence is established between historical compensation actions that have completed the response stabilization process and the actual quality characteristic responses of each production cycle after their execution, forming a time-series compensation response benchmark characterizing the response direction and response amount of the compensation action in each production cycle. For the executed compensation action to be judged, a matching compensation response reference sequence is determined from the time-series compensation response benchmark. Based on the response cycles already experienced by the executed compensation action to be judged, the remaining response sequence from the current production cycle is determined from the compensation response reference sequence. The preset response stabilization conditions include that the change in the actual quality characteristic response corresponding to the executed compensation action to be judged within a preset number of consecutive production cycles is less than or equal to a preset stabilization threshold, and that each remaining response amount in the remaining response sequence is less than or equal to a preset residual threshold. Executed compensation actions that have not yet met the preset response stabilization conditions are determined as unstable compensation actions, and the remaining response sequence corresponding to these compensation actions is determined as non-stabilized compensation responses.

[0090] like Figure 2 As shown, the multiple complete historical response contours on the left side of the figure collectively represent the time-series compensation response baseline. These historical response contours, after action attribute matching and response cycle alignment, form an action-level compensation response reference sequence. The actual responses already experienced in the figure represent the actual response process formed from the start of the quality characteristic effect of the compensation action to the current production cycle. The current production cycle section is used to divide the actual responses that have already occurred into the subsequent remaining responses. The response contours after the section represent the remaining response sequence formed after actual response state correction. The stability boundary in the figure corresponds to the cycle position where both the actual response change and the remaining response simultaneously enter the corresponding threshold range.

[0091] The compensation action description data is formed based on the target equipment, process parameters, controlled quality characteristics, compensation direction, compensation magnitude, and execution time of the executed compensation action, and is associated with the quality characteristic deviation before the execution of the compensation action, the quality characteristic test results of each production cycle after the execution, the actual quality characteristic response, the action execution status, and the data version.

[0092] When the same compensation action is applied to multiple controlled quality characteristics, the compensation action identifier remains consistent, and a separate quality characteristic response record is established for each controlled quality characteristic. Each response record is associated with the same target equipment, process parameter item, compensation direction, compensation magnitude, and execution time, and is saved according to its respective unit of measurement, sign direction, and response cycle sequence.

[0093] Historical response samples are categorized according to target equipment or equivalent equipment group, process parameter items, controlled quality characteristics, and compensation direction. Within the same historical response sample, the execution time of compensation actions, the pre-execution deviation state, the compensation magnitude, the response onset period, the actual quality characteristic response of each response period, the response peak period, and the response stabilization period form a complete correspondence.

[0094] The historical response samples used to form the time-series compensation response benchmark have complete action and response records. The corresponding quality inspection results have been validly verified, and there is a definite mapping relationship between the response cycle and the production cycle. The actual response process covers the continuous cycle from the start of the compensation action to the entry into the steady state of the response. The units of measurement and sign of the controlled quality characteristics have been converted to a unified evaluation scale.

[0095] When other executed compensation actions are present during the execution of historical compensation actions, the corresponding measured quality characteristic changes are assigned at the action level based on the execution time, target process parameters, quality characteristics, and registered response progress of each action. For historical compensation actions to be assigned, the response components already assigned to other executed compensation actions are separated from the measured quality characteristic changes, and a correspondence is established between the obtained action responses and the compensation actions to be assigned.

[0096] When the measured quality characteristic changes within the same production cycle are contributed by multiple executed compensation actions, the sum of the response components of each action corresponds to the measured changes in that production cycle. The action attribution result simultaneously records the attribution action identifier, the attribution production cycle, the attribution response quantity, and the attribution status, which are used for subsequent maintenance of historical response benchmarks.

[0097] Figure 2 Each complete historical response profile drawn on the left corresponds to a historical compensation action that has completed action-level response attribution and covers the entire stable process. Multiple historical response profiles retain their own characteristics in terms of compensation magnitude, response initiation location, and response evolution speed, thus forming a time-series compensation response benchmark with action attribute index and response period index.

[0098] Historical compensation actions that fall within the same aggregation range are assigned relative response cycles based on the execution time of the compensation action. When a compensation action begins to produce a change in quality characteristics that can be identified by quality inspection data, the corresponding production cycle is determined as the first response cycle. Subsequent production cycles are then assigned the second, third, and subsequent response cycles in chronological order.

[0099] Let the deviation of the quality characteristics before a certain historical compensation action is . The quality characteristic deviation corresponding to the j-th response cycle after execution is Then the cumulative actual quality characteristic response of the historical compensation action in the j-th response period is: ,in This represents the cumulative change in quality characteristics from the baseline deviation before the compensation action is executed until the j-th response cycle. Each historical response sample is periodically aligned using the cumulative response state, ensuring that different historical actions have a unified response meaning within the same response cycle sequence.

[0100] Historical response samples corresponding to the same target equipment or process equivalent equipment group, the same process parameter item, the same controlled quality characteristic, and the same compensation direction are configured to the corresponding compensation magnitude nodes in ascending order of compensation magnitude. Each compensation magnitude node records a complete cumulative response state sequence, response onset period, peak period, stabilization period, and response state after stabilization.

[0101] When a node with the same compensation magnitude has multiple historical response samples, the cumulative response states corresponding to the same response period position are statistically aggregated. The statistical aggregation rules are predetermined based on the number and dispersion of historical samples, and may use the median response state, the mean response state after processing outlier samples, or a weighted response state formed according to the similarity of operating conditions. The same statistical aggregation rules are used for the same time-series compensation response baseline version.

[0102] The time-series compensation response benchmark is formed by each compensation amplitude node and its corresponding cumulative response state sequence. The time-series compensation response benchmark uses the compensation action attribute and compensation amplitude node as the action index, the response cycle position as the time index, and the cumulative quality characteristic response as the response state value.

[0103] The time-series compensation response benchmark is also associated with its corresponding target equipment range, process equivalent equipment group, controlled quality characteristics, unit of measurement, sign direction, sample time range, sample quantity, and benchmark version. For the same non-steady-state compensation response identification, the same time-series compensation response benchmark version is used to complete action matching and residual response continuation.

[0104] For the executed compensation actions to be judged, historical response samples with consistent process parameters, controlled quality characteristics, and compensation directions are selected from the time-series compensation response benchmark. If the target equipment has corresponding historical response samples, the time-series compensation response benchmark corresponding to the target equipment is adopted; if the historical sample size of the target equipment is below the preset sample requirement, the time-series compensation response benchmark corresponding to the equivalent equipment group of the process is adopted.

[0105] Action matching also determines the corresponding amplitude position based on the compensation amplitude of the action to be compensated. When the compensation amplitude to be compensated is consistent with the historical amplitude node, the cumulative response state sequence corresponding to that amplitude node is read; when the compensation amplitude to be compensated is located between two adjacent historical amplitude nodes, the cumulative response states of the two amplitude nodes in the same response period position are interpolated according to the position of the compensation amplitude to be compensated between the two historical amplitude nodes, and the reference response states for each response period are formed in sequence.

[0106] When a difference in amplitude direction exists between the compensation amplitude and the historical amplitude nodes, the corresponding time-series compensation response benchmark is entered according to the compensation direction. Response sequences are established for both forward and reverse compensation actions, and the response direction is represented by the signed response state.

[0107] When the historical records also save product specifications, equipment operating modes, or major process status labels, the action matching, based on the matching of target equipment, process parameter items, controlled quality characteristics, compensation direction, and compensation magnitude, further uses the corresponding process status labels to limit the range of historical samples.

[0108] The reference response states corresponding to the position sequence of each response period after matching are arranged sequentially to form a compensation response reference sequence. The compensation response reference sequence records the cumulative reference response states from the first response period to the reference stable period, and associates them with the action identifier of the compensation action to be judged and the time-series compensation response benchmark version used in this case.

[0109] Figure 2Multiple complete historical response profiles converge into the main response profile, indicating that historical sample screening and response state collection are completed according to the action attributes, compensation magnitude, and response cycle position of the compensation action to be judged; the main response profile formed by the convergence corresponds to the compensation response reference sequence.

[0110] When the executed compensation action to be judged has gone through several response cycles, a sequence of actual responses is formed based on the measured quality characteristic deviations of each production cycle after its execution. Let the quality characteristic deviation before the execution of the compensation action to be judged, a, be... The measured quality characteristic deviation corresponding to the j-th already experienced response cycle is Then the cumulative actual response status corresponding to the j-th response cycle is: The actual response sequence and the compensation response reference sequence are phase-aligned according to the response activating position. At the start of the preset activation period, the first response period of the actual response corresponds to the first response period of the compensation response reference sequence.

[0111] When there is a periodic offset between the actual response initiation position and the reference initiation position, the reference response period position is moved within a preset phase correction range, and the fitting difference between the actual response state and the reference response state at each candidate phase position is calculated; the candidate position with the smaller fitting difference is determined as the response continuation position. The phase correction range is preset based on the historical response hysteresis range of the corresponding process parameters.

[0112] In one implementation, for each candidate phase position, the sum of the squared differences between the actual response state and the corresponding reference response state is used to characterize the fitting difference, and the position where the sum of the squared differences reaches its minimum value is determined as the current response phase. When the quality detection data have different confidence levels, each response period is weighted according to the corresponding detection confidence level.

[0113] After completing the response phase alignment, the response scaling coefficients are determined based on the experienced actual response states and the corresponding reference response states. Let the reference response state corresponding to the j-th experienced response cycle after response phase alignment be... Response scaling factor Determined according to the following formula: Where l represents the number of response cycles that have been completed. This represents the weight corresponding to the j-th response cycle that has already been experienced. When all response cycles are treated with equal weights, Take the same value.

[0114] The response scaling factor is used to characterize the overall ratio of the actual response intensity of the compensation action to be judged to the reference response intensity. When the number of experienced response cycles is small, the scaling factor is determined using the currently obtained effective response cycles; as new actual quality characteristic responses arrive, the response scaling factor is recalculated according to the newly added experienced response cycles.

[0115] Suppose that the compensation response reference sequence after phase alignment covers K response cycles, and l response cycles have already elapsed. This indicates the cumulative reference response status corresponding to the current response position. This indicates the cumulative reference response status corresponding to the s-th production cycle after the current response position.

[0116] The remaining response quantity for the s-th subsequent production cycle starting from the current production cycle is: ,in The remaining response quantity represents the new changes in quality characteristics that occur as the current actual response state evolves into the corresponding subsequent production cycle. The remaining response quantities corresponding to each subsequent production cycle are arranged in order of production cycle position, forming a remaining response sequence.

[0117] The current quality characteristic deviation already encompasses all actual responses generated prior to the current production cycle. Therefore, the remaining response sequence continues from the current response position into subsequent production cycles, recording only responses newly generated after the current state. This processing ensures that already experienced actual responses enter the deviation reference component through the current quality characteristic deviation, and that subsequent new responses enter the compensation response component through the unsteady compensation response.

[0118] Figure 2 The actual response already experienced corresponds to the measured response segment on the left side of the current production cycle section, which corresponds to the starting point of the deviation aggregation time. The response profile on the right side of the section corresponds to the remaining response sequence. The remaining response profile is jointly determined by the compensation response reference sequence, the actual response phase, and the response scaling coefficient.

[0119] When the actual response and the compensated response reference sequence maintain the same response phase, the phase offset is zero. When the actual response takes effect with hysteresis, the remaining response sequence continues to form from the corrected reference response position into subsequent cycles. The preset response stability condition simultaneously constrains the actual response change state and the forward remaining response state. The stability judgment window includes the most recent... For each consecutive production cycle, the preset stability threshold is [value]. If the compensation action to be judged satisfies the following conditions within the stability judgment window: , where W represents the set of response periods corresponding to the stability judgment window.

[0120] Let the preset residual threshold be The remaining response sequence satisfies: This indicates that the forward incremental effect of the compensation action since the current production cycle has entered the preset residual range.

[0121] When both the actual response change condition and the remaining response condition are met, the compensation action to be judged is determined to have entered a stable response state. Figure 2 The stability boundary shown corresponds to the response cycle position where both of the above conditions are met simultaneously. The curve flattens out near the stability boundary, indicating that the actual response change within the continuous production cycle enters the stable threshold range; the remaining response after the stability boundary enters the residual threshold range.

[0122] When a compensation action is applied to multiple controlled quality characteristics simultaneously, each controlled quality characteristic generates an actual response sequence, a compensation response reference sequence, and a remaining response sequence, and each undergoes a stability assessment. The compensation action enters an action-level response stability state when all relevant controlled quality characteristics have reached a stable response state.

[0123] Executed compensation actions that are still in the response release state are identified as unstable compensation actions, and the remaining response sequences corresponding to these compensation actions are identified as unstable compensation responses. Unstable compensation responses are associated with compensation action identifiers, target equipment, controlled quality characteristics, current response phase, remaining production cycle sequence, response direction and response quantity for each cycle, and the response baseline version used.

[0124] In this embodiment, the deviation aggregation result is formed according to the following process: Based on the target equipment, process parameter items, controlled quality characteristics, and compensation direction of the proposed compensation action, a compensation magnitude response relationship is formed from historical process parameter compensation records and corresponding actual quality characteristic responses, characterizing the mapping relationship between the compensation magnitude and the corresponding response direction and response quantity for each expected response cycle. The proposed compensation response is then determined based on the proposed compensation magnitude of the proposed compensation action. The controlled quality characteristic directly targeted by the proposed compensation action is identified as the target controlled quality characteristic. The executed compensation actions corresponding to the non-stable compensation response and the proposed compensation actions corresponding to the proposed compensation response are respectively constructed as compensation... Compensation action items; when the same compensation action item generates quality characteristic responses to at least two controlled quality characteristics, they are collected separately according to the controlled quality characteristics actually affected by each quality characteristic response, and the attribution relationship of each quality characteristic response to the same compensation action item is retained; according to the target equipment, controlled quality characteristics, and production cycle, the current quality characteristic deviation of each target equipment is configured as the deviation starting value of the production cycle covered by the unstable compensation response and the proposed compensation response, forming a deviation baseline component; the unstable compensation response and the proposed compensation response are collected into compensation response components, and the attribution relationship between each compensation response and the corresponding compensation action item is retained, forming a deviation collection result.

[0125] like Figure 3As shown, the deviation reference component, the unsteady compensation response, and the proposed compensation response form mutually aligned data layers. Each data layer is configured to the same deviation aggregation section according to the target equipment, controlled quality characteristics, and production cycle, and the attribution of executed compensation actions or proposed compensation actions is maintained through compensation action identifiers.

[0126] The compensation magnitude response relationship is established according to the target equipment or process equivalent equipment group, process parameter items, controlled quality characteristics, compensation direction, response cycle sequence, and compensation magnitude nodes respectively.

[0127] For the same target equipment or equivalent equipment group, the same process parameter item, the same controlled quality characteristic, and the same compensation direction, historical compensation actions that have completed the stable response process are divided into corresponding amplitude nodes according to the actual compensation magnitude. Each amplitude node records the cumulative quality characteristic response status for each response cycle, thereby forming a mapping relationship between the compensation magnitude and the corresponding response status for different response cycles.

[0128] For the proposed compensation action, the corresponding compensation magnitude response relationship is determined based on the target equipment, process parameters, controlled quality characteristics, and compensation direction. Then, based on the position of the proposed compensation magnitude in the historical magnitude nodes, the response state corresponding to each expected response cycle is determined one by one.

[0129] When the proposed compensation range is consistent with a certain historical range node, the response state of each cycle corresponding to that range node is adopted; when the proposed compensation range is located between two adjacent historical range nodes, the response state of the two nodes in the same expected response cycle is interpolated according to the proportion of the proposed compensation range between the two adjacent nodes; when the compensation range response relationship adopts the piecewise fitting result, the response state of each cycle is determined according to the position of the proposed compensation range in the corresponding segment.

[0130] The execution time of the proposed compensation action is mapped to the production cycle sequence, and the expected production cycle at which the proposed compensation action begins to generate a quality characteristic response is determined based on the response start cycle recorded according to the corresponding compensation magnitude response relationship. Before the response start cycle, the corresponding production cycle records the response waiting state; starting from the response start cycle, the response states corresponding to each expected response cycle are sequentially configured to subsequent production cycles to form the proposed compensation response.

[0131] The proposed compensation response record includes the proposed compensation action identifier, target equipment, controlled quality characteristic, expected response start period, expected response coverage period, and the response direction and response quantity corresponding to each expected response period. When the proposed compensation action acts on multiple controlled quality characteristics simultaneously, each controlled quality characteristic forms a proposed compensation response based on its corresponding compensation magnitude response relationship.

[0132] The compensation magnitude response relationship is mapped according to the expected response period. Therefore, the same proposed compensation magnitude will form its own corresponding response quantity in each expected response period. Thus, the proposed compensation response retains both the compensation magnitude effect and the response timing effect.

[0133] Figure 3 The proposed compensation response layer starts from the production cycle in which the expected response takes effect and extends along the direction of subsequent production cycles, indicating that the proposed compensation action has formed a forward response state at the equipment level, quality characteristic level, and production cycle level in the state to be evaluated.

[0134] The executed compensation actions corresponding to the non-steady-state compensation response are constructed as executed compensation action items, and the proposed compensation actions corresponding to the proposed compensation response are constructed as proposed compensation action items. Both types of compensation action items use the compensation action identifier as the action-level index and are associated with the target equipment, process parameter items, controlled quality characteristics, compensation direction, compensation magnitude, action status, and corresponding response sequence.

[0135] An executed compensation action associated with a non-steady-state compensation response indicates that the action has completed the adjustment of process parameters on the equipment side and will continue to generate residual quality characteristics in subsequent production cycles. A proposed compensation action associated with a proposed compensation response indicates that the action is in a control evaluation state, and its expected quality characteristics will be included in subsequent individual response projections and joint response projections.

[0136] The controlled quality characteristic directly targeted by the proposed compensation action is defined as the target controlled quality characteristic. When the proposed compensation action generates accompanying responses to other controlled quality characteristics, each accompanying response forms a response component according to the controlled quality characteristic it actually affects.

[0137] When the same compensation action generates quality characteristic responses for at least two controlled quality characteristics, each quality characteristic response is assigned to the corresponding controlled quality characteristic evaluation domain. Each controlled quality characteristic evaluation domain performs deviation aggregation, process center calculation for the equipment group, discrete state calculation between equipment, and process convergence boundary judgment.

[0138] Each quality characteristic response continues to be associated with the same compensation action identifier after being aggregated into sub-items. Thus, a compensation action item has multiple quality characteristic response components in different controlled quality characteristic evaluation domains, while maintaining a single action-level control object during the action execution permission and compensation magnitude configuration stages.

[0139] Figure 3 The executed compensation action attribution line connects the non-stable compensation response with the corresponding executed compensation action item, and the proposed compensation action attribution line connects the proposed compensation response with the corresponding proposed compensation action item. The action attribution line indicates that the compensation response retains its corresponding action source after entering the equipment, quality characteristic, and production cycle index, and is used for subsequent boundary excess contribution backtracking.

[0140] Based on the target equipment, controlled quality characteristics, and production cycle, the quality characteristic deviation of each target equipment in the current production cycle is configured as the starting value of the deviation for the production cycle covered by the unsteady compensation response and the proposed compensation response, forming a deviation reference component.

[0141] Assume the current production cycle is The current quality characteristic deviation of the target device i corresponding to the controlled quality characteristic q is: For the production cycle t that enters this forward evaluation, the deviation baseline component is expressed as: The deviation baseline component represents the unified forward calculation state of each target device before the superposition of the current non-steady-state compensation response and the proposed compensation response. Actual quality characteristic responses that occurred before the current production cycle are already included. middle.

[0142] Each non-steady-state compensation response is assigned to its corresponding index position according to its target equipment, controlled quality characteristics, and remaining production cycle, forming a non-steady-state compensation response component; each proposed compensation response is assigned to its corresponding index position according to its target equipment, controlled quality characteristics, and expected production cycle, forming a proposed compensation response component. The non-steady-state compensation response component and the proposed compensation response component together constitute the compensation response component.

[0143] In cases where there are multiple compensation responses for the same equipment, the same controlled quality characteristics, and the same production cycle, each compensation response is written as an independent response item into the corresponding index position, and the compensation action identifier, action status, response type, response direction, and response quantity are retained.

[0144] After each actual production cycle, the deviation baseline component is updated based on the newly acquired quality characteristic deviation; forward response items that have been converted into actual responses are written into the historical response process according to the actual response attribution, while response items that are still in subsequent cycles continue to be retained in the compensation response component.

[0145] The deviation aggregation results are in a structured data format with a common index of target equipment, controlled quality characteristics, and production cycle, and the action level is maintained by compensation action identifiers. Each equipment-quality characteristic-production cycle index position records the deviation baseline component, the non-steady-state compensation response item, the proposed compensation response item, and the corresponding action identifier and action status of each response item.

[0146] Figure 3 The deviation reference component layer represents the unified starting state formed by the measured quality characteristic deviation of each target equipment in the current production cycle; the unstable compensation response layer represents the residual effect of the compensation actions that have been performed before the current production cycle and continue to be released into subsequent production cycles; the proposed compensation response layer represents the expected effect of the proposed compensation action in the expected effective period and subsequent production cycles.

[0147] Figure 3 The current production cycle in the table represents the measured starting position of this deviation aggregation, while subsequent production cycles represent the forward configuration positions of the non-steady-state compensation response and the proposed compensation response. Each data layer uses the same target equipment position, controlled quality characteristic index, and production cycle direction, thus forming the deviation aggregation section.

[0148] For a specific target device, a controlled quality characteristic, and a subsequent production cycle, the forward deviation state corresponding to the index position is jointly determined by the deviation reference component and the compensation response component acting at that index position. The deviation reference component provides the current measured starting point, the unsteady compensation response component provides the remaining effect of historically executed actions, and the proposed compensation response component provides the expected effect of the action to be evaluated.

[0149] The actual response generated before the current production cycle is entered into the deviation baseline component through the current quality characteristic deviation; new responses generated since the current production cycle are entered into the compensation response component through the non-steady-state compensation response; the expected response corresponding to the proposed compensation action is entered into the compensation response component through the proposed compensation response. Each response forms only one forward action record at its corresponding production cycle position.

[0150] Figure 3 The executed compensation actions and proposed compensation actions shown constitute two types of action sources, respectively. After the non-stability compensation response and proposed compensation response are configured at the equipment layer, quality characteristic layer, and production cycle layer, they continue to be associated with the corresponding compensation action items through the compensation action identifier.

[0151] The deviation baseline component in the deviation aggregation result is used to determine the process center and discrete state between equipment in the equipment group according to the controlled quality characteristics and production cycle; the compensation response component is used to form the individual response projection of each compensation action item, as well as the joint response projection of multiple compensation action items working together in the same production cycle; the compensation action attribution relationship is used to trace back the center boundary excess and discrete boundary excess formed by the joint response to the corresponding compensation action item.

[0152] In this embodiment, based on the deviation reference component in the deviation aggregation result, the process center and discrete state between equipment in the equipment group are determined according to the controlled quality characteristics and production cycle; based on the upper and lower boundaries of the preset target center interval and the upper boundary of the preset allowable discrete range, the upward center margin, the downward center margin and the discrete margin are formed respectively, and the three constitute the process convergence margin.

[0153] Identify production cycles in which at least two compensation action items with effective compensation responses exist within the same production cycle for the same controlled quality characteristic, forming a response overlap cycle set; for each production cycle in the response overlap cycle set, superimpose the compensation responses corresponding to each compensation action item onto the deviation reference component to form a single response projection, and jointly superimpose the compensation responses corresponding to each compensation action item within the same production cycle onto the deviation reference component of the corresponding equipment to form a joint response projection.

[0154] Based on the changes in the discrete states of the equipment group process center and between equipment before and after the projection of the individual response, the central action and discrete action corresponding to each compensation action item are formed; based on the changes in the discrete states of the equipment group process center and between equipment before and after the projection of the joint response, the central migration amount and discrete change amount are formed, and the central migration amount and discrete change amount constitute the synthetic compensation response.

[0155] For the central boundary excess or discrete boundary excess formed by the synthetic compensation response, based on the central action, discrete action and action affiliation of each compensation action, the compensation action items that make the same contribution to the same process convergence boundary within the same out-of-bounds production cycle are determined, and compensation interference relationship is established between the corresponding compensation action items.

[0156] For any controlled quality characteristic q, the target equipment participating in this process convergence evaluation is obtained from the deviation aggregation results. All target equipment entering the same equipment group evaluation domain uses a unified unit of measurement, a unified deviation sign direction, and a unified production cycle number.

[0157] The same single-response projection, joint response projection, and compensated interferometry delimitation use the same target equipment range, equipment weight, preset target center interval, preset allowable dispersion range, and deviation aggregation result version. When the deviation aggregation result is updated, a new equipment group evaluation state is formed by the updated deviation benchmark component and compensated response component.

[0158] In this embodiment, each target device participates in the calculation of the process center and discrete states between devices using an equal-weighted approach. In some implementations, each target device is weighted based on effective output, effective quality inspection sample size, or quality inspection confidence level. In implementations using device weights, the deviation benchmark group state, individual response projection state, and joint response projection state always use the same set of device weights.

[0159] When a source process parameter compensation command applies to multiple target devices simultaneously, the command is broken down into multiple device-level compensation action items according to the target devices. Each device-level compensation action item is associated with its corresponding target device and quality characteristic response, and the source process parameter compensation command identifier is stored together. Subsequent device-level response aggregation, single response projection, and action contribution backtracking all use the device-level compensation action items as the processing objects.

[0160] Let the number of target devices correspond to the controlled quality characteristic q. The deviation reference component corresponding to target equipment i in production cycle t is Under equal-weighted evaluation conditions, the process center of the equipment group is represented as: The discrete states between devices are represented as follows: The process center of the equipment group represents the group center position of the quality characteristic deviation of each target equipment, and the discrete state between equipment represents the degree of separation of the deviation of each target equipment relative to the process center of the equipment group.

[0161] Let the preset target center interval corresponding to the controlled quality characteristic q be: ,in Indicates the lower bound of the target center. The upper bound of the target center is represented as: The upper boundary of the preset allowable discrete range is represented as: The upward center margin is expressed as: The downward center margin is expressed as: The discrete residual is expressed as: The upward centering allowance, the downward centering allowance, and the discrete allowance together constitute the process termination allowance corresponding to production cycle t.

[0162] like Figure 4 As shown, the horizontal axis represents the process center of the equipment group, and the vertical axis represents the discrete states between equipment. The lower bound of the target center, the upper bound of the target center, and the upper bound of the allowable discrete state jointly define the process convergence constraint domain. The deviation reference group state is jointly determined by the process center of the equipment group and the discrete states between equipment corresponding to the deviation reference components, and serves as the common starting point for both individual response projection and joint response projection.

[0163] When the deviation reference group state exceeds the target center boundary or the allowable discrete boundary, record the reference boundary crossing type, the reference boundary crossing production cycle, and the reference excess amount; when the deviation reference group state is within the process convergence constraint domain, proceed to the new boundary excess judgment formed by the combined effect of compensation response. The reference excess amount and the new boundary excess amount are saved separately.

[0164] For the same controlled quality characteristic, when there are at least two valid compensation responses corresponding to compensation action items within the same production cycle, the production cycle is determined as a response overlapping production cycle, and the entire response overlapping production cycle constitutes the response overlapping cycle set.

[0165] Effective compensation responses include non-stable compensation responses corresponding to executed compensation actions, and proposed compensation responses corresponding to proposed compensation actions. When the amount of compensation responses reaches the preset response inclusion threshold, the corresponding compensation action is included in the response overlap judgment for that production cycle.

[0166] When a proposed compensation action is in the response waiting phase, its proposed compensation response is configured cyclically according to the expected effective production cycle; from the expected effective production cycle, the corresponding proposed compensation response enters the effective compensation response phase. Unstable compensation responses enter the response overlap judgment according to the production cycle position in the remaining response sequence.

[0167] When multiple consecutive production cycles form a response overlap state, the corresponding consecutive production cycles are recorded as a response overlap interval; response overlap production cycles formed at intervals are recorded separately. Figure 5 The overlapping period brackets in the response indicate a continuous production cycle interval where multiple effective compensation responses work together.

[0168] Let the compensation action a produce the compensation response of the target equipment i to the controlled quality characteristic q during production cycle t as follows: .

[0169] For compensation actions that have already been performed Take the remaining response amount of the corresponding unstable compensation response in production cycle t; for the proposed compensation action item, Take the expected response amount of the corresponding proposed compensation response in production cycle t.

[0170] All effective compensation responses acting on target equipment i within production cycle t are aggregated according to their action affiliation to form the equipment-level compensation response aggregate: The summation range is all compensation actions that produce an effective compensation response to the controlled quality characteristic q of the target equipment i within the production cycle t.

[0171] The equipment-level compensation response aggregate stores the compensation action item identifier, action status, source process parameter compensation instruction identifier, response direction, and response amount. Joint response projection uses the equipment-level compensation response aggregate to recalculate the equipment status, and action contribution backtracking uses the action attribution information stored within it.

[0172] For any compensation action item 'a' within the overlapping production cycle, only the compensation response corresponding to that compensation action item is superimposed onto the deviation reference component of the corresponding target equipment to form a single projected deviation: Based on the individual projection deviations corresponding to all target devices, the process center and discrete states between devices in the device group are re-determined according to the aforementioned calculation rules, forming the individual response projection state corresponding to compensation action item a.

[0173] Let the center of the device group process after the projection of a single response be... The discrete states between devices are The central action corresponding to the compensating action term 'a' is expressed as: The discrete action quantity corresponding to the compensation action term 'a' is expressed as: .like Figure 4 As shown, Individual Response Projection 1 and Individual Response Projection 2 represent the discrete states between the center and equipment in the equipment group process after the individual action of two compensation actions. The Individual Response Projection is responsible for identifying the direction of action and the contribution of the action boundary.

[0174] For the overlapping production cycle t, the compensation responses corresponding to all effective compensation actions are superimposed onto the deviation reference component according to the target equipment, forming a joint projected deviation: Based on the joint projection deviations corresponding to all target devices, the center of the device group process after joint response projection is re-determined. Discrete states between devices .

[0175] Figure 4 The common superposition recalculation means that: each compensation action response forms a device-level compensation response sum according to the target device, the device-level compensation response sum is superimposed on the deviation reference component of each target device, and then the process center and discrete state between devices of the device group are recalculated based on the superimposed deviation of all devices.

[0176] The joint response projection result adopts the recalculated state after the equipment-level co-superposition. Individual discrete actions are included in the action contribution backtracking, and the joint discrete change is formed by recalculating the discrete state after the joint action of all equipment-level compensation responses.

[0177] For the overlapping production cycle t, the joint response projection uses the joint projection deviation corresponding to each target device to form the device group state.

[0178] Let the average value of the combined equipment-level compensation response corresponding to the controlled quality characteristic q be: ,in This represents the sum of the equipment-level compensation response of target equipment i corresponding to the controlled quality characteristic q during production cycle t. This indicates the number of target devices participating in the controlled quality characteristic q process convergence evaluation.

[0179] The device group process center after joint response projection is represented as follows: ,in This indicates the process center of the equipment group formed based on the deviation reference components. This represents the process center of the equipment group after the combined effect of all effective compensation responses.

[0180] The center migration before and after joint response projection is expressed as: The center migration amount represents the group average effect of the combined compensation response of each target device at the device level. When the center migration amount is positive, the center of the device group migrates along the upper bound of the target center; when the center migration amount is negative, the center of the device group migrates along the lower bound of the target center.

[0181] For target device i, the deviation of its deviation reference component relative to the process center of the device group is combined with the deviation of its device-level compensation response composite relative to the average response of the device group to form a device-level joint deviation component: ,in This represents the deviation of target device i from the process center of the equipment group before the compensation response is superimposed. This represents the response deviation of the combined device-level compensated response obtained by target device i relative to the average response of the device group.

[0182] The equipment-level joint deviation component aggregates the original deviation position of the equipment and the difference in equipment response to the same equipment-level state. When the target equipment, which was originally located on one side of the process center of the equipment group, continues to receive a response deviation pointing to the same side, the equipment-level joint deviation component increases accordingly; when the target equipment receives a response deviation pointing to the process center of the equipment group, the equipment-level joint deviation component decreases accordingly.

[0183] Based on the joint deviation components at the device level corresponding to each target device, determine the discrete state between devices after projection of the joint response: ,in This represents the discrete state between devices after the joint response projection.

[0184] The discrete changes in the joint response before and after projection are expressed as follows: When the discrete change is positive, the joint response increases the degree of deviation separation between the target devices; when the discrete change is negative, the joint response causes the deviation state of each target device to converge toward the process center of the device group.

[0185] The composite compensation response corresponding to production cycle t consists of the center migration and discrete changes. The composite compensation response is associated with controlled quality characteristics, production cycle, compensation action items participating in the joint response projection, the sum of equipment-level compensation responses corresponding to each target device, equipment-level joint deviation components, and the state of the equipment group after the joint response projection.

[0186] like Figure 4 As shown, the joint response projection result is formed by the migration of the deviation reference group state along the center migration direction and the discrete change direction. In the figure, the horizontal state change corresponds to the center migration amount, and the vertical state change corresponds to the discrete change amount.

[0187] When each target device obtains the same compensation response, the response deviation of each device is zero, and the joint response mainly forms the overall migration of the center of the device group process. When the compensation responses obtained by each target device differ in direction or amplitude, the corresponding response deviation enters the device-level joint deviation component and changes the discrete state between devices after the projection of the joint response.

[0188] In the implementation state using equipment weights, the average value of the combined equipment-level compensation response is formed according to the corresponding equipment weights. The discrete states between equipment after the joint response projection are aggregated according to the same equipment weights for the equipment-level joint deviation components. The same equipment weights are used for the same joint response projection as for the deviation reference group states.

[0189] In this embodiment, for each production cycle with overlapping response cycles, the corresponding upward center margin or downward center margin is selected according to the migration direction of the center migration amount. The boundary occupancy amount formed by the center migration amount along the corresponding direction is compared with the selected center margin, and the boundary occupancy amount formed by the discrete change amount along the discrete increase direction between devices is compared with the discrete margin.

[0190] Based on the comparison results, the central boundary excess state and discrete boundary excess state after joint response projection are respectively formed. The boundary excess already formed by the deviation reference group state is recorded as the reference excess state, and the boundary excess further formed by the joint response projection is recorded as the newly added excess state of the response.

[0191] Within the new excess production cycle of the central response, determine the compensation action items whose central action direction is consistent with the corresponding central migration direction from the compensation action items that generate effective compensation response; within the new excess production cycle of the discrete response, determine the compensation action items whose discrete action direction points to the discrete increase direction between equipment from the compensation action items that generate effective compensation response.

[0192] For the same controlled quality characteristic, the same additional production cycle in response, and the same process convergence boundary, compensation actions with the same boundary contribution are backtracked at the action level, and compensation interference relationships are established between the corresponding compensation actions. At least one compensation action corresponds to the action to be compensated, and at least two compensation actions act on different target devices.

[0193] The baseline excess state, the joint projection excess state, and the newly added excess state in the response are stored separately. The compensation interference relationship is established based on the newly added excess state in the response, so that the action interference boundary corresponds to the increased boundary occupancy caused by the combined effect of the compensation response.

[0194] In some implementations, a forward cutoff function is defined: The boundary occupancy formed by the joint response on the upper bound of the target center is expressed as: The boundary occupancy formed by the joint response on the lower bound of the target center is expressed as: The boundary occupancy of the discrete boundary formed by the joint response is expressed as: The baseline excess formed by the deviation baseline group state relative to the upper bound of the target center is expressed as: The baseline excess formed by the deviation baseline group state relative to the lower bound of the target center is expressed as: The excess amount of the deviation benchmark group state relative to the allowable discrete upper bound is expressed as: The excess of the upper bound of the target center after the joint response projection is expressed as: The lower bound excess of the target center after joint response projection is expressed as: The discrete boundary excess after projection of the joint response is expressed as: The increase in the boundary excess after the joint response projection relative to the baseline excess corresponding to the deviation baseline group state is determined as the new excess of the response.

[0195] The additional excess response corresponding to the upper bound of the target center is represented as follows: The additional excess response corresponding to the lower bound of the target center is represented as: The additional excess response corresponding to the discrete boundary is expressed as: When the deviation reference group state is within the process convergence constraint domain, the excess of each reference is zero, and the new excess of the response has the same value as the boundary excess after the joint response projection.

[0196] When the deviation benchmark group state has crossed the corresponding boundary, the benchmark excess amount record joint response projection of the excess state that was formed before, the response newly added excess amount record joint compensation response is the boundary excess that is further added above the benchmark state.

[0197] when When the value is greater than zero, the production cycle t is defined as the additional excess production cycle added to the upward center response; when When the value is greater than zero, the production cycle t is defined as the additional excess production cycle added by the downward center response; when When the value is greater than zero, the production cycle t is defined as the additional excess production cycle added to the discrete response.

[0198] like Figure 4 As shown, the distance between the deviation reference group state and the process convergence constraint domain boundary corresponds to the process convergence margin; the joint response projection result forms a joint projection excess after crossing the corresponding boundary; the part that exceeds the original reference excess state corresponds to the new excess amount of the response.

[0199] For compensation action item a, the individual contribution of the compensation action item to the formation of each process convergence boundary is determined based on the changes in the process center of the equipment group and the discrete state changes between equipment before and after the projection of its individual response.

[0200] The single contribution of compensation action term 'a' to the upper bound of the target center is expressed as follows: The single contribution of compensation action term 'a' to the lower bound of the target center is expressed as: The individual contribution of the compensation action term 'a' to the discrete boundary is expressed as: Within the additional production cycle of the upward-center response, compensation actions with a single contribution greater than zero at the upper bound of the target center form upward-center unidirectional contribution actions. Within the additional production cycle of the downward-center response, compensation actions with a single contribution greater than zero at the lower bound of the target center form downward-center unidirectional contribution actions. Within the additional production cycle of the discrete response, compensation actions with a single contribution greater than zero at the discrete boundary form discrete unidirectional contribution actions. The single contribution amount is used to characterize the intensity of the compensation action's effect on the corresponding out-of-bounds direction. When a compensation action simultaneously contributes unidirectionally to both the center boundary and the discrete boundary within a production cycle, the corresponding boundary type and single contribution amount are recorded respectively.

[0201] Figure 4 The same-direction boundary contribution identifier in the code indicates the action-level relationship between each individual response projection and the joint response projection as it crosses the corresponding boundary. The individual projection state is used to determine the direction of action contribution, and the joint response projection state is used to determine the excess state of the group boundary.

[0202] For the same controlled quality characteristics, the same additional production cycle for the same response, and the same boundary type, assume that compensation action a and compensation action b both form boundary contributions in the same direction.

[0203] The interference quantities formed by the two compensation action terms with respect to the upper bound of the target center are expressed as follows: The interference quantities formed by the two compensation action terms with respect to the lower bound of the target center are expressed as follows: The co-boundary interference quantities formed by the two compensation action terms on the discrete boundary are expressed as: The combined boundary interference quantities of compensation action a and compensation action b at the production cycle t are expressed as follows: It is worth mentioning that, for the same controlled quality characteristic, the same additional production cycle in response, and the same boundary type, two compensation action items that form the same boundary contribution are selected and denoted as compensation action item a and compensation action item b, respectively. Here, a and b are both compensation action item indices, indicating two compensation action items participating in the same action pair relationship. and These represent the individual contributions of compensation action a and compensation action b to the upper bound of the target center of the controlled quality characteristic q during production cycle t, respectively. and These represent the individual contributions of the two compensation actions to the lower bound of the target center; and These represent the individual contributions of the two compensation action terms to the discrete boundary. All of these individual contributions are calculated according to the aforementioned rules for calculating individual response projection and same-direction boundary contributions.

[0204] The boundary interference is jointly limited by the additional excess in the response and the co-directional contributions of the two compensation action items. The additional excess in the response is used to limit the actual increase in boundary excess range of the joint compensation response, and the co-directional contributions of the two action items are used to limit the scope of action of each action in occupying the corresponding boundary.

[0205] When the overall boundary interference is greater than zero, the two compensation action items generate effective compensation responses within the same response-added excess production cycle, the two compensation action items act on different target equipment, and at least one compensation action item corresponds to the proposed compensation action, a compensation interference relationship is established between the two compensation action items.

[0206] When there are three or more contributing actions in the same direction within the same production cycle, they are combined in pairs according to the compensation actions to form action pairs. Each action pair stores the controlled quality characteristics, production cycle, boundary type, the individual contribution of the two actions, the additional excess amount of the response, and the interference amount of the same boundary.

[0207] When two executed compensation actions together form a boundary contribution in the same direction, the corresponding action pair is written into the existing response boundary occupancy record; when the action pair contains a proposed compensation action, the corresponding action pair is written into the compensation interference relationship.

[0208] When the same action pair causes interference at the same boundary in multiple production cycles, the cycles are merged according to the action pair, and the interference quantity corresponding to each production cycle is retained. When the same action pair causes interference at both the central boundary and discrete boundaries, the corresponding boundary types and interference intensities are saved respectively.

[0209] To represent the degree to which center migration, discrete changes, and individual responses occupy the corresponding process convergence boundaries on the same vertical axis, the boundary occupancy and the corresponding process convergence margin are dimensionless.

[0210] When the center migration amount is greater than or equal to zero, the center migration occupancy rate is expressed as: When the center migration amount is less than zero, the center migration occupancy rate is expressed as: The discrete change occupancy rate is expressed as: The single-item boundary occupancy rate corresponding to compensation action item 'a' is expressed as: , Figure 5 The single-item response projection curve is fixed at the maximum value of the single-item boundary occupancy rate corresponding to all compensation action items within each production cycle: The aforementioned boundary occupancy rate is formed for production cycles where the corresponding process convergence margin is greater than zero. When the corresponding process convergence margin is equal to zero or takes a negative value, the production cycle is recorded separately according to the baseline boundary state, the joint projection excess state, and the response new excess state.

[0211] like Figure 5 As shown, the horizontal axis represents the production cycle, and the vertical axis represents the process termination boundary occupancy rate. The horizontal line with an occupancy rate of 1 represents the boundary full occupancy line. When the individual response projection curve is within the boundary full occupancy line, it indicates that the individual boundary occupancy rate formed by the individual compensation actions within the corresponding production cycle is within the corresponding process termination margin range.

[0212] Production cycles whose center migration occupancy curve exceeds the boundary full line enter the set of out-of-bounds production cycles of center joint projection; production cycles whose discrete change occupancy curve exceeds the boundary full line enter the set of out-of-bounds production cycles of discrete joint projection.

[0213] In joint projection over-limit production cycles, the newly added over-limit production cycle is further determined based on the change between the baseline over-limit and the joint projection over-limit. Production cycles where the joint projection over-limit is greater than the corresponding baseline over-limit are included in the corresponding set of newly added over-limit production cycles.

[0214] The earliest production cycle in each set marks the starting position of the corresponding boundary state, and subsequent production cycles that continuously meet the corresponding excess conditions are continued to be included in the corresponding production cycle set.

[0215] Figure 5 The overlapping period brackets in the response diagram correspond to multiple effective compensation responses entering the continuous production cycle interval of the joint response projection. The center migration occupancy curve and the discrete change occupancy curve represent the center boundary occupancy process and the discrete boundary occupancy process formed by the joint compensation response along the production cycle direction, respectively.

[0216] Figure 4 Used to represent the positional relationship between the deviation reference group state, the single-item response projection state, and the joint response projection state in the central-discrete state space; Figure 5 Used to indicate the timing of boundary overflows caused by changes in boundary occupancy rate during the production cycle. According to Figure 5 After determining the response to the newly added excess production cycle, the individual contribution of each compensation action item within the corresponding production cycle is read to form the same boundary interference quantity and compensation interference relationship.

[0217] For each compensation interference relationship, record the associated controlled quality characteristics, the response overlapping production cycle, the joint projection out-of-bounds production cycle, the response newly added excess production cycle, the out-of-bounds boundary type, the relevant compensation action item identifier, the source process parameter compensation instruction identifier, and the target equipment and action status corresponding to each compensation action item.

[0218] The compensation interference relationship also records the baseline excess corresponding to the deviation baseline group state, the joint projection excess corresponding to the joint response projection, the response addition excess corresponding to the joint compensation response, the individual boundary contribution of each compensation action item, and the same boundary interference of the corresponding action pair.

[0219] When a compensation interference relationship involves discrete boundaries, the system associates and stores the sum of equipment-level compensation responses, equipment-level joint deviation components, and discrete inter-equipment states after projection of the joint response for each target device within the corresponding production cycle. When the same action pair forms boundary interference in multiple production cycles, the action pairs are merged, and all newly added excess production cycles and the corresponding boundary interference quantities for each production cycle are retained. When the same action pair involves at least two types of boundaries, including the upper boundary of the target center, the lower boundary of the target center, or discrete boundaries, the corresponding newly added excess response quantity and boundary interference quantity are stored separately according to the boundary type. When a compensation action item participates in the compensation interference relationships of multiple controlled quality characteristics, each controlled quality characteristic forms an interference relationship record, and the action-level association is maintained through a unified compensation action identifier.

[0220] The compensation interference relationship is identified by the version of the deviation aggregation result and the version of the compensation interference relationship. When the deviation reference component, the non-steady compensation response, or the proposed compensation response is updated, the corresponding individual response projection and joint response projection are re-formed based on the updated deviation aggregation result, and the version of the compensation interference relationship is updated.

[0221] In this embodiment, the center-boundary production cycle includes production cycles in which the projection of the joint response of the center exceeds the target center boundary and results in an additional production capacity for the response; the discrete-boundary production cycle includes production cycles in which the projection of the joint response exceeds the allowable discrete upper limit and results in an additional production capacity for the response. The aforementioned upward center response additional production cycle and downward center response additional production cycle both belong to the center-boundary production cycle, and the aforementioned discrete response additional production cycle belongs to the discrete-boundary production cycle.

[0222] The resulting compensation interference relationship is processed in the compensation action acceptance process and serves as the relationship data for compensation interference group merging, candidate surplus data formation, current compensation action determination, and compensation execution permission configuration.

[0223] In this embodiment, compensation action items that are interconnected along the compensation interference relationship are grouped into compensation interference groups. Under the condition that the non-steady-state compensation response acting on the controlled quality characteristics associated with the compensation interference group and the evaluation of the production cycle continues to act, the proposed compensation response corresponding to each proposed compensation action item in the group is superimposed on the corresponding deviation reference component. Candidate margin data are formed based on the process center of the equipment group, the discrete state between equipment and the process convergence margin after the proposed compensation response has been applied.

[0224] The proposed compensation action items whose upward centering margin, downward centering margin, and discrete margin in the candidate margin data all reach zero or above, and whose absolute value of the deviation of the controlled quality characteristics of the target decreases after the proposed compensation response, are identified as compensation acceptance candidates. The deviation reduction rate of each compensation acceptance candidate is determined based on the change in the absolute value of the deviation of the controlled quality characteristics of the target before and after the proposed compensation response. The current compensation action is determined from the compensation acceptance candidate items whose deviation reduction rate reaches the maximum value. The compensation interference group whose compensation acceptance candidate set is empty is identified as the non-acceptance compensation interference group.

[0225] The compensation acceptance judgment determines the current acceptance order of the proposed compensation actions within the same compensation interference group. After the current compensation action is determined, the corresponding action acceptance limit and execution compensation range are determined according to the complete expected response cycle, all associated controlled quality characteristics, and the overall status of the equipment group. Based on the compensation interference relationship between compensation action items, compensation action items with direct compensation interference relationship or those forming continuous compensation interference relationship through other compensation action items are grouped into the same compensation interference group.

[0226] Each compensation intervention group includes compensation actions that have been executed and are still in the response release state, as well as proposed compensation actions awaiting compensation acceptance judgment. Based on the target equipment and controlled quality characteristics affected by each compensation action within the group, the equipment evaluation scope and quality characteristic evaluation scope of the compensation intervention group are determined; based on the overlapping production cycles and out-of-bounds production cycles corresponding to the compensation intervention relationships within the group, the action acceptance evaluation cycle range of the compensation intervention group is determined.

[0227] When a compensation action applies to multiple controlled quality characteristics simultaneously, all controlled quality characteristics are included in the evaluation scope of the compensation interference group. When a compensation action forms compensation interference relationships with other action items through different controlled quality characteristics, they are merged into the group according to the unified compensation action identifier.

[0228] For any compensation action to be evaluated within the compensation intervention group, read the deviation baseline component in the current deviation aggregation result, and retain all effective non-stable compensation responses that affect the associated controlled quality characteristics and the evaluation production cycle of the compensation intervention group.

[0229] Effective non-stability compensation responses include the residual responses corresponding to compensation actions already performed within the group, as well as the residual responses generated by compensation actions already performed outside the group on the corresponding equipment, controlled quality characteristics, and the evaluated production cycle. Each non-stability compensation response is superimposed onto the deviation baseline component according to equipment affiliation and production cycle order to form the candidate projection background.

[0230] The proposed compensation response corresponding to the current proposed compensation action to be evaluated is added to the candidate projection background, while other proposed compensation actions in the same group remain in the state of the actions to be evaluated. Based on the equipment-level deviation after the proposed compensation action to be evaluated takes effect, and in accordance with the aforementioned calculation rules for the process center and discrete state between equipment in the equipment group, the process center, discrete state between equipment, upward center margin, downward center margin, and discrete margin corresponding to each evaluated production cycle are re-determined.

[0231] The candidate margin data includes the identifier of the proposed compensation action to be evaluated, the corresponding target equipment, the associated controlled quality characteristics, the production cycle of the action acceptance evaluation, the upward center margin, the downward center margin, the discrete margin, and the pre- and post-deviation of the target controlled quality characteristics for each evaluation production cycle. The candidate action evaluation uses the proposed compensation response corresponding to the proposed compensation magnitude. When the proposed compensation response is in the response waiting period, the corresponding production cycle forms a candidate margin according to the candidate projection background; after the proposed compensation response enters the expected effective period, the corresponding response quantity enters the candidate projection state according to the expected response cycle position. The controlled quality characteristic directly targeted by the proposed compensation action is determined as the target controlled quality characteristic. When the proposed compensation action generates an accompanying quality characteristic response, each accompanying controlled quality characteristic is judged in the candidate center margin and candidate discrete margin.

[0232] In this embodiment, the production cycle in which the proposed compensation response reaches the expected steady state is used as the evaluation cycle for target deviation improvement. Within this production cycle, the target deviation corresponding to the candidate projected background is taken as the pre-action deviation, and the target deviation after incorporating the proposed compensation response is taken as the post-action deviation.

[0233] If the proposed compensation action maintains the upward center margin, downward center margin, and discrete margin corresponding to each associated controlled quality characteristic above zero during the entire evaluation production cycle, and the absolute value of the deviation of the target controlled quality characteristic decreases, then the proposed compensation action is identified as a compensation acceptance candidate.

[0234] The rate of reduction of deviation of the proposed compensation action a on the controlled mass characteristic q of the target is expressed as: ,in This indicates the deviation of the target controlled quality characteristics before the proposed compensation response. This indicates the deviation of the target controlled quality characteristics after the proposed compensation response. It represents a preset small positive number.

[0235] When the proposed compensation action directly targets the controlled mass characteristics of multiple targets, the action-level deviation reduction rate is expressed as: , where q traverses all target controlled quality characteristics directly targeted by the compensation action a.

[0236] For a compensation intervention group with compensation acceptance candidates, the current compensation action is determined from the compensation acceptance candidates whose action-level deviation reduction rate reaches the maximum value.

[0237] When multiple compensation candidates have the same action-level deviation reduction rate, in some implementations, the current compensation action is determined sequentially according to the expected response stabilization period, the candidate minimum margin, the completeness of historical response samples, and the preset process priority.

[0238] When the set of compensation acceptance candidates is empty, the corresponding compensation intervention group is designated as the non-acceptance compensation intervention group. The compensation execution authority of each proposed compensation action within the non-acceptance compensation intervention group remains closed, and the group will proceed to the next round of compensation acceptance evaluation after the deviation benchmark component, the non-steady-state compensation response, or the quality inspection results are updated.

[0239] In this embodiment, the current compensation action corresponding to each compensation interference group and the proposed compensation action not included in the compensation interference group are determined as compensation actions to be authorized; based on the proposed compensation response corresponding to the compensation action to be authorized, the unstable compensation response that works together with it, and the corresponding process termination margin, the range of executable compensation range that makes the process termination margin corresponding to each expected response production cycle reach a value above zero is determined.

[0240] The compensation execution authority of each proposed compensation action is set according to the action acceptance result of the compensation intervention group, and the proposed compensation range of the proposed compensation action with the compensation execution authority in the open state is limited to the corresponding executable compensation range to form the execution compensation range; the preset response stability condition is used as the response stability trigger condition after the proposed compensation action is executed, and the compensation control data is composed of the compensation execution authority, the execution compensation range, and the response stability trigger condition.

[0241] After each pending compensation action completes its amplitude constraint, the pending compensation actions for which compensation execution authority is to be opened in this round are collectively superimposed on the equipment-level deviation according to their corresponding execution compensation amplitudes, and together with all valid non-stable compensation responses, an authorized joint projection is formed; based on the authorized joint projection, the discrete states of each associated controlled quality characteristic in the equipment group process center and between equipment in each expected response production cycle are re-determined.

[0242] When the authorized joint projection is located within the corresponding process convergence constraint domain, the compensation control data of this round becomes effective; when the authorized joint projection results in an additional boundary excess, the compensation interference relationship is updated based on the additional boundary excess and the same-direction boundary contribution of each compensation action to be authorized, and the action acceptance and amplitude configuration of the relevant proposed compensation actions are re-executed.

[0243] The current compensation actions corresponding to each compensation intervention group, as well as the proposed compensation actions not included in the compensation intervention group, are identified as compensation actions to be authorized. For any compensation action to be authorized, the compensation magnitude is used as an evaluation variable. For each compensation magnitude to be evaluated, the proposed compensation response corresponding to each expected response production cycle and each associated controlled quality characteristic is determined according to the compensation magnitude response relationship.

[0244] The amplitude evaluation uses the current deviation baseline component, all effective unsteady compensation responses, and the proposed compensation responses of the compensation actions to be evaluated under the corresponding compensation amplitude to form the equipment-level amplitude projection state. Based on this equipment-level amplitude projection state, the process center of the equipment group, the discrete states between equipment, and the process convergence margins corresponding to all associated controlled quality characteristics and all expected response production cycles are redefined.

[0245] For any compensation range to be evaluated, in each expected response production cycle, the corresponding center margin is selected based on the migration direction of the process center of the equipment group caused by the compensation action to be authorized. When the process center of the equipment group migrates along the upper boundary of the target center, an upward center margin is selected; when it migrates along the lower boundary of the target center, a downward center margin is selected; when maintaining the original center position, the smaller value between the upward and downward center margins is selected.

[0246] Extract the minimum value from the directional center margins corresponding to all associated controlled quality characteristics and all expected response production cycles to form the minimum directional center margin for the entire cycle of the authorized compensation action a under the compensation magnitude α: ,in This represents the directional center margin of the controlled quality characteristic q in the production cycle t.

[0247] For the same compensation range to be evaluated, the discrete states between equipment corresponding to all associated controlled quality characteristics and all expected response production cycles are determined respectively, and the difference between the allowable upper limit of discreteness and the corresponding discrete state between equipment is determined as the discrete margin.

[0248] Extract the minimum value from all discrete margins to form the minimum discrete margin for the entire cycle of the authorized compensation action a under the compensation magnitude α: ,in This represents the discrete margin of the controlled quality characteristic q during the production cycle t.

[0249] Starting from zero compensation magnitude, continuously evaluate the minimum directional center margin and the minimum discrete margin of the entire cycle along the direction of increasing compensation magnitude.

[0250] Let the upper boundary of the engineering compensation range corresponding to the compensation action to be authorized be . Then the upper limit of action continuation for this action is expressed as: ,in, This indicates the upper limit of the compensation action to be authorized; 'a' represents the index of the compensation action to be authorized; 'α' represents the compensation range to be determined at present. β represents the upper boundary of the engineering compensation range corresponding to the compensation action to be authorized; β represents any intermediate compensation range between zero compensation range and the current compensation range to be determined. This represents the minimum directional center margin of the compensation action to be authorized (a) over the entire cycle under the intermediate compensation amplitude (β). The minimum discrete margin of the compensation action 'a' to be authorized under the intermediate compensation amplitude β is shown as the full-cycle minimum discrete margin; sup represents the supremum of the set of compensation amplitudes that satisfy the listed margin conditions; this process forms a continuous acceptance interval starting from the zero compensation amplitude, and the position where the central margin curve or discrete margin curve first reaches zero along the direction of increasing compensation amplitude is taken as the action acceptance boundary. When the compensation amplitude response relationship adopts segmented amplitude nodes, the zero compensation amplitude, each amplitude node, the proposed compensation amplitude, and the position where the central margin or discrete margin reaches zero are taken as amplitude segmentation points, and the equipment-level response projection and boundary judgment of each amplitude segment are executed sequentially.

[0251] When the equipment controller adopts discrete compensation step size, the amplitude to be evaluated is increased step by step from zero compensation amplitude according to the preset compensation step size. At each amplitude level, the minimum directional center margin and the minimum discrete margin of the whole cycle are re-determined, and the amplitude level that satisfies the two types of margin conditions along the amplitude increase direction is determined as the upper limit of action acceptance.

[0252] In some implementations, the compensation action to be authorized is applied to a single target device, and the compensation magnitude response relationship exhibits a local linear change within the interval between adjacent magnitude nodes.

[0253] Suppose that the unit compensation magnitude response of the compensation action to be authorized, 'a', to the controlled quality characteristic 'q' during production cycle 't' is: The number of target devices participating in the evaluation is .

[0254] When the unit compensation magnitude response points to the upper bound of the target center... Take the corresponding upward center margin; when the unit compensation amplitude response points to the lower bound of the target center... Take the corresponding downward center margin. The upper limit of the directional center amplitude corresponding to the production cycle t is expressed as: ,in This indicates the directional center margin of compensation action 'a' in production cycle 't' for the controlled quality characteristic 'q'. This represents the unit compensation amplitude response of compensation action a in production cycle t for the controlled quality characteristic q; when the unit compensation amplitude response is zero, the center of the equipment group in that production cycle remains in its original state, and the upper limit of the center amplitude in the corresponding direction is taken as the upper boundary of the engineering compensation amplitude.

[0255] The current deviation baseline component is aggregated with all effective unsteady compensation responses acting on the corresponding target equipment, controlled quality characteristics, and expected response production cycle to form the amplitude evaluation starting state for the compensation action to be authorized. Along the zero compensation amplitude to the upper boundary of the engineering compensation amplitude corresponding to the compensation action, the equipment-level projected deviation is formed amplitude by amplitude according to the adjacent amplitude node interval or the preset compensation amplitude step size.

[0256] For the controlled quality characteristic q and production cycle t associated with the compensation action to be authorized 'a', select the appropriate upward or downward centering margin based on the migration direction of the equipment group process center pointed to by the unit compensation magnitude response. When the unit compensation magnitude response points to the upper limit of the target center, select the upward centering margin; when the unit compensation magnitude response points to the lower limit of the target center, select the downward centering margin.

[0257] In the equal-weighted equipment group evaluation state, the selected directional center margin is converted into the corresponding compensation range based on the number of target equipment participating in the evaluation and the absolute amount of the unit compensation range response; in the equipment weighted evaluation state, the conversion is performed based on the equipment weight of the target equipment affected by the compensation action to be authorized and the absolute amount of the unit compensation range response.

[0258] The first compensation amplitude corresponding to the moment when the process center of the equipment group reaches the target center boundary along the corresponding migration direction is determined as the upper limit of the directional center amplitude of compensation action a in production cycle t for the controlled quality characteristic q, denoted as . .

[0259] When the unit compensation amplitude response is zero, the center of the equipment group remains in the amplitude evaluation starting state, and the upper limit of the center amplitude in the corresponding direction is taken as the upper boundary of the engineering compensation amplitude corresponding to the compensation action. After projection is completed along the entire engineering compensation amplitude evaluation range, if the center of the equipment group remains within the target center interval, the upper limit of the center amplitude in the corresponding direction is also taken as the upper boundary of the engineering compensation amplitude. When the amplitude evaluation starting state has reached the corresponding target center boundary, the upper limit of the center amplitude in the direction is set to zero, and the state of the corresponding center boundary being fully occupied is recorded.

[0260] For the same compensation action to be authorized, the projection deviation of the target equipment is determined on a per-amplitude basis, based on the deviation position of the target equipment relative to the process center of the equipment group, the direction of the compensation response, and the change in the compensation amplitude at the start of the amplitude evaluation. For each compensation amplitude to be evaluated, the process center of the equipment group and the discrete state between equipment in the corresponding production cycle are re-determined according to the projection deviations of all target equipment and the equipment weights used in this round of evaluation.

[0261] As the target device moves closer to the center of the equipment group as the compensation magnitude increases, the discrete state between the devices decreases as the deviation between the target device and the center of the equipment group decreases; when the target device continues to move along the compensation direction after reaching the center of the equipment group, the discrete state between the devices increases as the target device deviates to the other side of the center of the equipment group.

[0262] Along the direction of amplitude increase from zero compensation amplitude to the upper boundary of engineering compensation amplitude, the first compensation amplitude corresponding to when the discrete state between equipment reaches the allowable upper limit of discreteness is determined as the upper limit of the discrete amplitude of compensation action a for the controlled quality characteristic q in production cycle t, denoted as . .

[0263] When the unit compensation amplitude response is zero, the discrete state between equipment remains in the starting state of the amplitude evaluation, and the upper limit of the corresponding discrete amplitude is taken as the upper boundary of the engineering compensation amplitude corresponding to the compensation action. After projection is completed along the entire engineering compensation amplitude evaluation range, if the discrete state between equipment continues to be within the allowable discrete range, the upper limit of the corresponding discrete amplitude is taken as the upper boundary of the engineering compensation amplitude.

[0264] When the amplitude evaluation starts at the upper limit of the allowable discreteness and the authorized compensation action causes the discrete state between devices to change within the allowable discreteness range, the zero compensation amplitude is taken as the recovery starting point. Along the direction of increasing compensation amplitude, the first compensation amplitude corresponding to when the discrete state between devices reaches the upper limit of the allowable discreteness again is determined as the upper limit of the discrete amplitude. When the authorized compensation action causes the discrete state between devices to change outside the allowable discreteness range, the upper limit of the discrete amplitude is determined as zero, and the state of the discrete boundary being filled is recorded.

[0265] When the amplitude evaluation starting state exceeds the allowable discrete range, and the pending authorized compensation action gradually reduces the discrete excess, the first compensation amplitude corresponding to when the discrete state between devices enters the allowable discrete range is determined as the lower limit of the recovery amplitude, along the direction of increasing compensation amplitude. The first compensation amplitude corresponding to when the discrete state between devices enters the allowable discrete range and then reaches the upper limit of the allowable discrete range again is determined as the upper limit of the discrete amplitude. The corresponding executable compensation amplitude range is formed from the lower limit of the recovery amplitude to the upper limit of the discrete amplitude. When the pending authorized compensation action increases the discrete excess, or when the projected state corresponding to the upper boundary of the engineering compensation amplitude is still outside the allowable discrete range, the corresponding action acceptance state is recorded as no acceptance state, and the compensation execution permission remains closed.

[0266] When the discrete margin changes from increasing to decreasing with the compensation amplitude, the upper limit of the discrete amplitude is taken as the position where the allowable upper limit of discreteness is first reached along the direction of increasing zero compensation amplitude. The resulting discrete amplitude range is consistent with the aforementioned continuous action transition interval.

[0267] For any controlled quality characteristic q associated with compensation action a, obtain the upper limit of the directional center amplitude and the upper limit of the discrete amplitude of the controlled quality characteristic in each expected response production cycle.

[0268] For each expected response production cycle, the smaller of the upper limit of the central amplitude and the upper limit of the discrete amplitude in the corresponding direction is determined as the cycle acceptance limit for that production cycle. The minimum value among the cycle acceptance limits corresponding to all expected response production cycles is determined as the controllable quality characteristic level action acceptance limit of compensation action a for the controlled quality characteristic q.

[0269] When compensation action 'a' is associated with multiple controlled quality characteristics, a control quality characteristic-level action acceptance limit is formed for each controlled quality characteristic, and the minimum value among them is determined as the unified action-level acceptance limit for the compensation action. In the local linear amplitude response implementation state, the unified action-level acceptance limit is used as the specific calculation result of the aforementioned continuous action acceptance limit, and together with the proposed compensation amplitude of the compensation action, the execution compensation amplitude is determined.

[0270] like Figure 6 As shown, the horizontal axis represents the compensation magnitude, the left vertical axis represents the process convergence margin, and the right vertical axis represents the deviation reduction rate. The directional center margin curve is formed by the minimum directional center margin of the entire cycle corresponding to each compensation magnitude; the discrete margin curve is formed by the minimum discrete margin of the entire cycle corresponding to each compensation magnitude; the deviation reduction rate curve is formed by the change in the absolute value of the deviation before and after the action of the target controlled quality characteristics under different compensation magnitudes. When a small compensation moves the target equipment closer to the process center of the equipment group, the discrete state between the equipment decreases accordingly, and the discrete margin increases accordingly; when the compensation magnitude continues to increase and the target equipment crosses the process center of the equipment group, the discrete state between the equipment increases accordingly, and the discrete margin decreases accordingly. Figure 6 The upper limit of the action acceptance is formed by the position where the directional center margin curve or discrete margin curve first reaches zero along the direction of increasing compensation amplitude. The execution compensation amplitude is located at the smaller of the planned compensation amplitude and the upper limit of the action acceptance. The compensation execution permission is recorded as an action-level status, and the executable compensation amplitude range is recorded as a numerical range; both are entered into the compensation control data.

[0271] After determining the action acceptance limit and compensation range for each pending compensation action, compensation execution permissions are configured based on the action acceptance results corresponding to the compensation intervention group. For compensation intervention groups with a current compensation action, the compensation execution permission for the current compensation action is set to open, and the compensation execution permissions for other proposed compensation actions within the same compensation intervention group are set to closed. For compensation intervention groups without accepting compensation actions, the compensation execution permissions for each proposed compensation action within the group are set to closed. For proposed compensation actions not included in a compensation intervention group, if the action acceptance limit is greater than zero, the compensation execution permission is set to open; if the action acceptance limit is zero, the compensation execution permission is set to closed.

[0272] All pending compensation actions with the current compensation execution authority set to open will form corresponding proposed compensation responses according to their respective compensation ranges. These proposed compensation responses, along with all current valid non-stable compensation responses, will be superimposed on the deviation benchmark components of the corresponding target equipment, controlled quality characteristics, and expected response production cycle.

[0273] Based on the combined equipment-level deviations, the process center, discrete states between equipment, upward center margin, downward center margin, and discrete margin corresponding to all associated controlled quality characteristics and all expected response production cycles are redefined. When all process convergence margins in the authorized joint projection reach zero or above, the authorized joint verification result enters the pass state, and the compensation control data of this round takes effect as a whole.

[0274] When the authorized joint projection results in an excess of the newly added central boundary or the excess of the discrete boundary, the compensation interference relationship is updated based on the out-of-bounds production cycle, the out-of-bounds boundary type, and the same-direction boundary contribution of each compensation action to be authorized. The proposed compensation action associated with the newly added compensation interference relationship enters the re-evaluation state, and a new compensation interference group, current compensation action, action acceptance limit, execution compensation range, and compensation execution authority are formed.

[0275] For the same round of compensation action acceptance, action acceptance limit determination, compensation execution permission configuration, and authorization joint verification, the same deviation aggregation result version, the same non-stability compensation response version, the same compensation interference relationship version, and the same historical response benchmark version are used. New quality inspection results, equipment execution confirmations, or non-stability compensation response updates arriving during the compensation control data formation period are written to the next state version. Before the compensation control data takes effect, consistency verification is performed on its dependent versions and the current state version. When all dependent versions are consistent, the compensation control data enters the effective state in the scheduled effective production cycle; when dependent versions are updated, the corresponding compensation control data is re-formed based on the updated deviation aggregation results.

[0276] The compensation control data includes the proposed compensation action identifier, target equipment identifier, process parameter items, compensation direction, proposed compensation range, upper boundary of engineering compensation range, upper limit of action acceptance, range of executable compensation range, execution compensation range, compensation execution authority, response stability triggering conditions, corresponding compensation interference group, authorized joint verification results, effective production cycle, deviation collection result version, non-stability compensation response version, compensation interference relationship version, historical response benchmark version, and control data version.

[0277] In this embodiment, when a proposed compensation action with the compensation execution permission in an open state is implemented based on the compensation control data, the proposed compensation action is determined as an executed compensation action; the corresponding proposed compensation response is updated according to the actual executed compensation range and the compensation range response relationship, and the updated proposed compensation response is migrated to an unstable compensation response.

[0278] Before the response stability trigger condition is triggered, the unstable compensation response is corrected and the remaining response sequence is updated based on the difference between the actual quality characteristic response and the corresponding expected response that has gone through the production cycle. When the response stability trigger condition is triggered, the deviation reference component is updated so that the corresponding unstable compensation response is removed from the compensation response component.

[0279] Starting from the controlled quality characteristics affected by the already executed compensation actions, the relevant unexecuted compensation actions and compensation interference relationships are collected to form a compensation reconstruction domain. The original compensation control data within the compensation reconstruction domain ceases to take effect, and new compensation control data is formed based on the updated deviation collection results. The original compensation control data outside the compensation reconstruction domain that maintains the same dependency state continues to take effect.

[0280] The proposed compensation action, with its compensation execution permission open, sequentially goes through the authorized pending issuance state, the issued pending equipment confirmation state, and the confirmed execution state. After the compensation control data enters the effective state, the corresponding proposed compensation action transitions to the authorized pending issuance state. After the control system issues a process parameter compensation command to the target equipment, the action transitions to the issued pending equipment confirmation state. Once the target equipment returns confirmation of actual execution, the actual compensation magnitude, and the actual execution time, the proposed compensation action becomes an executed compensation action and enters the response release state. The actual execution time determines the start point of the response cycle, and the actual compensation magnitude determines the current effective response sequence. When there is a difference between the actual compensation magnitude and the proposed compensation magnitude, the response direction and response amount corresponding to each expected response production cycle are re-determined based on the actual compensation magnitude and the compensation magnitude response relationship. The response sequence formed according to the actual compensation magnitude is migrated to the unsteady compensation response.

[0281] Before the stable trigger condition is triggered, at the beginning of each new actual production cycle, the quality characteristic test results of each target device are acquired, and the measured quality characteristic deviation of the current production cycle is determined. The measured quality characteristic deviation of the current production cycle is written into the deviation aggregation result to form the deviation reference component for the next round of control calculation; the response that has occurred in the previous production cycle is entered into the actual quality characteristic response record; the non-stable compensation response is removed from the forward response position corresponding to the production cycle that has already been experienced, and the remaining response sequence continues from the new current production cycle.

[0282] The estimated production cycle index of the proposed compensation response is shifted forward synchronously with the current production cycle. Proposed compensation actions in the response waiting state are reconfigured in response position according to the updated estimated effective period, and executed compensation actions in the response release state are reconfigured in remaining response position according to the updated actual response phase.

[0283] The actual response formed before the current production cycle is carried by the latest measured quality characteristic deviation and actual response record, while the new effects after the current production cycle are carried by the updated unsteady compensation response.

[0284] When multiple non-stable compensation actions occur simultaneously for the same target equipment, the quality characteristic changes within the current production cycle are assigned to an action level based on the execution time of each executed compensation action, the order of the response cycle, the registered remaining responses, the controlled quality characteristics of the action, and the attribution relationship of the compensation actions.

[0285] The actual quality characteristic response belonging to the currently executed compensation action is written into the sequence of actual responses already experienced for that action. Based on the newly added actual response, the current response phase and response scale coefficient are re-determined according to the aforementioned response phase alignment, response scale correction, and residual response continuation rules, and the residual response sequence from the next production cycle is updated.

[0286] When a compensation action has been performed and belongs to a compensation intervention group, the compensation execution authority for other proposed compensation actions within the same compensation intervention group remains closed until the action reaches a stable response state.

[0287] When a compensation action is applied to multiple controlled quality characteristics, each controlled quality characteristic performs an actual response change judgment and a residual response judgment respectively. When all relevant controlled quality characteristics reach the preset response stability condition, the action-level response stability state is triggered.

[0288] When the stable trigger condition is triggered, the measured quality characteristic deviations of each target equipment that has reached a stable production cycle are written into a new deviation reference component; the corresponding compensation response that has not reached stability is removed from the compensation response component, and the compensation execution permission of the already executed compensation action is set to the off state.

[0289] The actual compensation magnitude, actual execution time, complete actual response sequence, response stabilization period, response phase, response scale coefficient, and action attribution results are written into the historical process parameter compensation record. The determination of the compensation reconstruction domain is jointly triggered by the update of the deviation benchmark component, the exit of the non-stabilized compensation response, and changes in the compensation action status.

[0290] like Figure 7 As shown, all controlled quality characteristics affected by the executed compensation actions that have reached a stable response state are defined as the stabilization-enhancing quality characteristics, and the stabilization-enhancing quality characteristics form the initial reconstruction boundary.

[0291] Figure 7 The first diamond-shaped object in the diagram represents the unexecuted proposed compensation action that directly affects the quality characteristics of the stabilizing effect. All unexecuted proposed compensation actions that affect the controlled quality characteristics within the initial reconstruction boundary are included in the reconstruction object, along with the compensation interference relationships associated with these proposed compensation actions.

[0292] Figure 7 The second diamond-shaped object in the diagram represents an unexecuted proposed compensation action that continues to be associated along the already included compensation interference relationship. When a newly included proposed compensation action also acts on other controlled quality characteristics, the corresponding controlled quality characteristic is identified as a new controlled quality characteristic, and the current reconstruction boundary is expanded.

[0293] The reconstruction boundary is repeatedly expanded in the following order: "controlled quality characteristics - unexecuted proposed compensation actions - compensation interference relationships - associated proposed compensation actions - newly added controlled quality characteristics". The reconstruction boundary reaches a closed state when the controlled quality characteristics, unexecuted proposed compensation actions, and compensation interference relationships obtained in two adjacent expansion rounds are consistent.

[0294] When a proposed compensation action is not executed and simultaneously applies to controlled quality characteristics within the reconstruction domain and other controlled quality characteristics, the proposed compensation action and its associated controlled quality characteristics are incorporated into the compensation reconstruction domain as a whole, so that the same proposed compensation action corresponds to a unified control data version.

[0295] Figure 7 The small dashed closed contour in the middle represents the initial reconstruction boundary, and the outer thick solid closed contour represents the compensation reconstruction domain formed after being expanded layer by layer by the unexecuted compensation action, compensation interference relationship and newly added controlled quality characteristics.

[0296] After the compensation and reconstruction domain is formed, any domain that has not yet executed the proposed compensation action will enter a reconstruction lock state, and the original compensation control data version within the domain will cease to be effective.

[0297] Based on the updated deviation baseline components, the currently released unsteady compensation responses, and the unexecuted proposed compensation actions within the domain, the deviation aggregation results within the domain are re-formed. Based on the updated deviation aggregation results, the process center, discrete states between equipment, and process convergence margins of the equipment group are re-determined, individual response projections and joint response projections are re-formed, and compensation interference relationships are re-established.

[0298] Based on the new compensation intervention relationship, a new compensation intervention group is formed, and the compensation acceptance candidates, current compensation actions, action acceptance limits, compensation execution range, and compensation execution permissions are redefined. Authorization joint verification is then performed to form a new version of compensation control data within the domain.

[0299] New compensation control data within the domain will take effect at the boundary of the next control-effective production cycle. Figure 7 The truncation of the old control line in the domain indicates the end of the original control data version within the domain, and the formation of the new control line indicates that the new compensation control data version within the domain has taken effect.

[0300] For the original compensation control data outside the compensation reconstruction domain, a dependency signature is generated. The dependency signature includes the deviation baseline component version, the non-steady-state compensation response identifier and version, the compensation interference relationship version, the associated controlled quality characteristics, the associated compensation actions, the effective production cycle, and the control data version.

[0301] If the current dependency signature is consistent with the dependency signature when the original compensation control data was generated, the original compensation control data from outside the domain will continue to be effective. Figure 7 The extra-domain continuous control line in the code indicates that the original compensation control data, which depends on the state, is executed along the original control version.

[0302] When external control data experiences changes in the version of the deviation reference component, changes in the non-stable compensation response, new compensation interference relationships, or changes in the status of associated actions during the consistency verification process, the corresponding control object is incorporated into the compensation reconstruction domain and the control data is reconstructed along with the objects within the domain.

[0303] In this embodiment, the quality characteristic deviation of each target device within a preset number of continuous production cycles, including the current production cycle, is obtained. Based on this, the process center and discrete state between devices corresponding to each controlled quality characteristic are determined. It is then determined whether the process center and discrete state between devices remain within the preset target center interval and the preset allowable discrete range, respectively, within the continuous production cycle, thus forming a measured convergence verification result.

[0304] Taking the quality characteristic deviation of each target device at the end of a continuous production cycle as the forward starting state, according to the production cycle covered by the non-stable compensation response, each non-stable compensation response is superimposed on the forward starting state of the corresponding target device and controlled quality characteristic, and the process center and discrete state between subsequent devices in the subsequent equipment group are re-determined to form the forward convergence verification result.

[0305] When both the measured convergence verification result and the forward convergence verification result indicate that the convergence verification has passed, the compensation execution permission for those who have not yet performed the proposed compensation action is set to the off state. The historical process parameter compensation record is updated according to the executed compensation action and its actual quality characteristic response to generate the process capability convergence result. The updated historical process parameter compensation record is then used to determine the proposed compensation response for the next production batch.

[0306] The actual measurement-forward dual convergence verification uses the same snapshot of the action state. Actions that have completed equipment execution confirmation and are still in the response release state are entered into the forward convergence verification with a non-stable compensation response; actions that have reached a stable response state have their actual effect borne by the measured quality characteristic deviation; actions that have not been executed to compensate remain in the to compensate state.

[0307] The dual convergence check is performed on a state snapshot where each issued action has a definite execution status. The proposed compensation response corresponding to the action that has not been executed is not included in the superposition of the forward residual response.

[0308] In some implementations, a range of continuous measured production cycles, including the current production cycle, is determined, and the target equipment, controlled quality characteristics, quality inspection data version, and action status version participating in this convergence calibration are identified. The number of continuous measured production cycles is pre-set based on the quality inspection frequency, the response cycle of the main compensation actions, and the production batch cycle time. A uniform number of production cycles and equipment group evaluation rules are used for the same convergence calibration.

[0309] The measured quality characteristic deviations of each target device are obtained according to the production cycle sequence. Based on the aforementioned rules for determining the process center and discrete states between devices in the device group, the measured state of the device group corresponding to each controlled quality characteristic is formed for each production cycle. The measured state of each device group includes the process center, discrete states between devices, upward center margin, downward center margin, and discrete margin for the corresponding production cycle.

[0310] The process center of the equipment group corresponding to each production cycle is compared with the target center interval of the corresponding controlled quality characteristic, and the discrete state between equipment corresponding to each production cycle is compared with the corresponding allowable discrete range. When the process center of the equipment group is continuously within the target center interval during the continuous measured production cycle, and the discrete state between equipment is continuously within the allowable discrete range during the continuous measured production cycle, the measured convergence verification result corresponding to the controlled quality characteristic is determined as a pass state.

[0311] When a controlled quality characteristic corresponds to multiple equipment evaluation subgroups, each equipment evaluation subgroup performs continuous measured state judgment, and the judgment results of all equipment evaluation subgroups are used as the measured convergence verification result of the controlled quality characteristic. When all measured convergence verification results corresponding to the controlled quality characteristic enter the pass state, the measured convergence pass result of the equipment group is formed.

[0312] When new quality inspection data appears within a continuous production cycle, the new quality inspection data is written into the measurement window according to the production cycle sequence, and the earlier production cycle data at the beginning of the window is removed, forming an updated continuous measurement window. The process center, discrete state between equipment, and process termination margin of the corresponding equipment group are re-determined according to the updated measurement window.

[0313] Production cycles in which the integrity of quality inspection data meets the preset verification requirements will enter the continuous actual measurement judgment; when there are production cycles in which quality inspection data needs to be supplemented, the corresponding continuous actual measurement window will be kept in the pending update state, and the actual measurement convergence verification result will continue to be generated after the corresponding quality inspection data arrives.

[0314] The measured convergence verification uses the currently confirmed actual state of the equipment. The effect of compensation actions that have reached a stable response state is reflected in the measured quality characteristic deviations of each target device; the effect of executed compensation actions that have already occurred in the response release state is also reflected in the current measured quality characteristic deviations. The resulting measured convergence verification results reflect the central and discrete states that the equipment group has formed within a continuous actual production cycle.

[0315] The measured quality characteristic deviations of each target device at the end of the continuous measured production cycle are determined as the forward calculation state, and the production cycle, deviation collection result version, non-stability compensation response version, and action state version corresponding to this forward calculation state are locked.

[0316] Forward convergence verification incorporates compensation actions that have completed device execution confirmation but are still in the response release state. Based on the target device, controlled quality characteristics, remaining response cycle sequence, and remaining response amount corresponding to each executed compensation action, a correspondence is established between each unsteady compensation response and the forward calculation state of the corresponding target device.

[0317] According to the remaining production cycle position of each unstable compensation response, the corresponding newly added remaining response is superimposed on the forward calculation state of the corresponding target equipment and controlled quality characteristics for each forward production cycle, forming the forward deviation state of each target equipment in the subsequent production cycle.

[0318] When multiple unsteady compensation responses correspond to the same target equipment and the same controlled quality characteristic, these responses are superimposed according to their action affiliation and production cycle sequence to form the cumulative forward deviation state of the target equipment in the corresponding forward production cycle. After the unsteady compensation responses are superimposed at the equipment level, the process center and discrete states between equipment are redefined based on the forward deviation states of all target equipment.

[0319] When each unsteady compensation response has a different remaining coverage period, the forward evaluation production cycle range is formed by the coverage periods of all remaining responses. After an unsteady compensation response reaches the end of its remaining response sequence, the cumulative response state corresponding to that action is carried over to the subsequent forward production cycle until the forward evaluation production cycle range ends.

[0320] For each forward production cycle, the redefined process center of the equipment group is compared with the corresponding target center interval, and the redefined discrete state between equipment is compared with the corresponding allowable discrete range. If the process center of the equipment group remains within the target center interval throughout all forward evaluation production cycles, and the discrete state between equipment remains within the allowable discrete range throughout all forward evaluation production cycles, the forward convergence verification result corresponding to the controlled quality characteristic is determined as a pass state.

[0321] When a non-steady-state compensation action applies to multiple controlled quality characteristics simultaneously, the residual responses of this action on each controlled quality characteristic are written into the corresponding forward states. Each controlled quality characteristic forms a discrete state between the forward device group process center and the forward devices. When all relevant controlled quality characteristics satisfy the forward convergence condition, the residual response corresponding to this action enters the forward convergence pass state. When the forward convergence verification results corresponding to all controlled quality characteristics enter the pass state, a forward convergence pass result for the device group is formed.

[0322] Forward convergence verification uses the same target equipment range, controlled quality characteristic range, and equipment group evaluation rules as the actual convergence verification. If the proposed compensation action is not executed, the proposed compensation state is maintained, and its proposed compensation response remains in the current control data. Forward convergence verification projects the subsequent state around the unsteady compensation response that has already entered the actual production process.

[0323] Upon the arrival of a new actual production cycle, the measured quality characteristic deviation corresponding to that production cycle is written into the deviation aggregation result, and a new forward calculation state is formed based on the updated end state of the continuous measured production cycles. Each unsteady compensation response synchronously removes the response position corresponding to the production cycle that has already been experienced, and the next round of forward convergence verification is formed based on the updated remaining response sequence.

[0324] The measured convergence pass results of the equipment group and the forward convergence pass results of the equipment group are written into the same convergence verification record, and a corresponding relationship is established through the target equipment, controlled quality characteristics, status version and verification production cycle.

[0325] The measured convergence verification results characterize the process center and discrete state between devices in the equipment group that have been formed in the continuous actual production cycle for each target device; the forward convergence verification results characterize the process center and discrete state between devices in the equipment group that have been formed in the subsequent production cycle for each target device after the current unsteady compensation response continues to be released.

[0326] When both the measured convergence verification and the forward convergence verification of the equipment group pass, the corresponding equipment group is determined to meet the process capability convergence conditions, and a double convergence closure state is generated. The double convergence closure state is associated with the continuous measured production cycle range, the forward evaluation production cycle range, the target equipment range, the controlled quality characteristic range, and the corresponding data version.

[0327] When the measured convergence verification result or the forward convergence verification result is in a state of pending observation, the current deviation aggregation and non-steady compensation response update processing are maintained, and the next round of dual convergence verification is formed with the new quality detection data, the remaining response sequence and the action status.

[0328] After the equipment group enters the dual convergence closure state, the compensation execution permission of the proposed compensation actions that are not yet executed will be updated to the closed state, and the effective production cycle, control data version, and closure reason identifier corresponding to the permission status update will be recorded. Compensation actions that have already received execution confirmation will continue to complete the actual response recording and historical data collection according to the current response status.

[0329] Write the target equipment, process parameter items, compensation direction, proposed compensation range, actual compensation range, actual execution time, response start production cycle, complete actual quality characteristic response, response stable production cycle, response phase, response scale coefficient, and action attribution result corresponding to each executed compensation action into the historical process parameter compensation record.

[0330] For executed compensation actions that simultaneously affect multiple controlled quality characteristics, the actual response sequence, steady state, and equipment group state changes corresponding to each controlled quality characteristic are saved, and action-level associations are established through a unified compensation action identifier. For multiple equipment-level compensation action items formed by splitting a source process parameter compensation instruction, the actual responses of each equipment level are saved, and a hierarchical relationship is established through the source process parameter compensation instruction identifier.

[0331] Based on the target equipment state at the time of double convergence closure, the discrete states of the end equipment group process center and end equipment corresponding to each controlled quality characteristic, the changes in the center state and discrete state within the continuous measurement cycle, and the changes in the equipment group state after the forward residual response are saved to form the process capability convergence result.

[0332] The process capability convergence results include the target equipment range, controlled quality characteristic range, continuous measured production cycle range, forward evaluation production cycle range, measured convergence verification results, forward convergence verification results, end-device group status, executed compensation actions and their actual responses, proposed compensation actions in closed permission state, deviation aggregation result version, compensation control data version, and result generation version.

[0333] Write the actual execution data and complete actual quality characteristic response of the compensation action already performed into the corresponding historical amplitude node; when the actual execution compensation amplitude is between the existing historical amplitude nodes, establish a new historical amplitude record based on the actual execution compensation amplitude, and save the corresponding response cycle sequence and actual response data.

[0334] The updated historical process parameter compensation records are compiled according to the target equipment, process parameter items, controlled quality characteristics, compensation direction, actual compensation magnitude, and response cycle sequence, and serve as the historical basis for determining the historical response, compensation magnitude response relationship, proposed compensation response, and expected response stability cycle for the next production batch.

[0335] The preset stability threshold, preset residual threshold, number of continuous production cycles with stable response, response inclusion threshold, number of continuous measured production cycles, target central interval, allowable discrete range, upper boundary of engineering compensation range, and number of abnormal continuous production cycles involved in this implementation method are configured according to the controlled quality characteristics, process parameter items, target equipment type, and production conditions.

[0336] The preset stability threshold is determined based on the actual response change in adjacent response cycles after historical compensation actions enter the stable phase. The preset residual threshold is determined based on the quality inspection resolution, the impact of the residual response on the process center and discrete states between equipment in the equipment group, and the allowable amount of residual response formed by process verification. The number of continuous production cycles with stable response is determined based on the historical response stability duration, production cycle time, quality inspection frequency, and compensation response hysteresis characteristics.

[0337] The target center range is determined based on product quality objectives, internal process center objectives, and equipment group capability verification results; the allowable discrete range is determined based on the discrete state between equipment and the consistency requirements of equipment groups in historical stable production batches; the upper boundary of the engineering compensation range is determined based on the equipment control range, process parameter safety boundaries, and historical effective compensation range.

[0338] Each parameter is associated with its source, applicable operating conditions, creation time, and parameter version. When a parameter is updated, a new parameter version is generated, and the effective production cycle or effective production batch is configured. The same parameter version is used for deviation collection, compensation interference delimitation, action succession, permission configuration, and convergence verification within the same round.

[0339] Historical response benchmarks are maintained according to the target equipment or process equivalent equipment group, process parameter items, controlled quality characteristics, compensation direction, actual compensation magnitude, and response cycle sequence.

[0340] After the compensation action has been performed and the response has reached a stable state, the actual compensation magnitude, the actual execution time, the actual quality characteristics of each response cycle, the peak production cycle, the stable production cycle, the final residual response, the response phase, the response scale coefficient, and the difference between the actual response and the original compensation response are saved.

[0341] The response samples entering the historical response baseline are associated with complete equipment execution confirmations, quality inspection data, action response attribution, parameter versions, and production conditions. When a compensation action involves multiple controlled quality characteristics, each controlled quality characteristic forms a response sequence, and the association is established through a unified compensation action identifier; when a source process parameter compensation instruction forms multiple equipment-level compensation action items, the equipment-level responses are aggregated through the source instruction identifier.

[0342] After the new stability action completes sample admission, it is written to the corresponding historical amplitude node. When the actual compensation amplitude is between the existing amplitude nodes, a new amplitude record is formed. After completing the collection of new samples and amplitude node maintenance, a new historical response benchmark version is generated and enters control processing at the next preset effective time.

[0343] If the historical samples of the current compensation action do not reach the sample size required for automatic response generation, the response reference is determined in the following order: samples of the same process as the current target equipment, samples of the same model of equipment, samples of process-equivalent equipment, and process calibration response curves. When using equipment-equivalent samples, the equipment equivalence relationship and response scale conversion data are saved simultaneously.

[0344] When the historical samples reach the tentative compensation conditions, the proposed compensation range will be adjusted to the preset tentative range, and the corresponding action will be transferred to the pending confirmation state; after the corresponding actual response is completed, the initial historical samples of the current target equipment will be formed.

[0345] When the reference cumulative response within the already completed response cycle is zero, the first production cycle with a definite cumulative response value is determined by moving forward along the reference response sequence, and this production cycle is used as the phase identification position. Before phase identification is completed, other proposed compensation actions within the same compensation interference group remain in the off state.

[0346] When the actual cumulative response direction is opposite to the reference response direction, an abnormal response direction record is generated, and the equipment status, process disturbance, action execution record, response attribution, and historical reference adaptation relationship are verified in sequence. After the response attribution correction or reference sequence switch is completed, the remaining response sequence is regenerated based on the updated actual response; when the equipment status or process status enters the abnormal range, the corresponding control object enters the compensation reconstruction and process review process.

[0347] When the target equipment is in a state of pending quality inspection data during the current production cycle, record the target equipment, controlled quality characteristics, production cycle, and inspection task identifier. New proposed compensation actions for the corresponding equipment remain disabled, while non-steady-state compensation responses corresponding to already executed compensation actions are retained. Once valid inspection results arrive, they are written to the corresponding production cycle, and a new deviation aggregation result is generated.

[0348] When the valid equipment data meets the equipment group evaluation conditions, a transient equipment group state with a data integrity identifier is formed; when the valid equipment data is below the evaluation conditions, the equipment group state calculation and convergence verification of the corresponding controlled quality characteristics enter a waiting state.

[0349] When the actual response exceeds the expected response coverage period and continues to change, the corresponding action is identified as a response hysteresis state. When the response hysteresis is within the allowable extension range, the response phase, response scale, and remaining response sequence are updated based on the newly added actual response, and the non-steady-state compensation response coverage period is extended.

[0350] When the response lag reaches the preset abnormal duration period, or when the actual response direction remains abnormal for a continuous preset number of production cycles, the associated actions and controlled quality characteristics are included in the compensation reconstruction domain, and the relevant automatic compensation control enters a pending review state. After completing equipment processing, process adjustment, or response attribution correction, compensation control data is regenerated based on the new state version.

[0351] When the unified acceptance limit of the action level for the pending compensation action is zero, the compensation execution permission of the corresponding action is set to the off state, and the controlled quality characteristics, production cycle, and boundary type that form the zero upper limit are recorded. The corresponding compensation intervention group enters a no-acceptance state, and the action acceptance evaluation is re-executed according to the release of the non-stable compensation response, the update of the deviation benchmark, or the result of the stable reconstruction.

[0352] If the same compensation interference group remains in a non-acceptance state after a preset number of reconstructions, a continuous non-acceptance record is formed. The continuous non-acceptance record includes candidate surplus data for each round, boundary occupancy position, associated action status, and reconstruction count, and triggers process parameter verification, re-formation of the proposed compensation action, and adaptation check of the target center interval and allowable discrete range.

[0353] While the existing compensation control data is in effect, newly generated proposed compensation actions are added to the set of actions to be reconstructed. Upon arrival of the next reconstruction node, the new proposed compensation actions, together with the current deviation benchmark component, the non-steady compensation response, and the existing proposed compensation actions, form a new deviation aggregation result. Then, compensation interference delimitation, action succession, amplitude limitation, and authorization joint verification are re-executed.

[0354] Each set of compensation control data includes the version of the associated deviation aggregation result, the version of the compensation interference relationship, the version of the parameters, the version of the historical response benchmark, the version of the compensation control data, the effective production cycle, the compensation reconstruction domain identifier, and the action execution status.

[0355] Before the target equipment executes a compensation action, a consistency check is performed on the target equipment identifier, compensation action identifier, process parameter items, compensation direction, compensation execution authority, compensation range, effective production cycle, and control data version. After the target equipment completes the execution, the control data version, actual compensation range, and actual execution time are returned, and the action state transition is completed accordingly.

[0356] New compensation control data within the compensation and reconstruction domain takes effect globally at the predetermined production cycle boundary, and the original control data within the domain is simultaneously transferred to the invalidated version. Original control data outside the compensation and reconstruction domain maintains version consistency through dependency signatures, which include the deviation baseline component version, the non-stable compensation response version, the compensation interference relationship version, the parameter version, the historical response baseline version, and the associated action status.

[0357] When the external dependency state is updated, the corresponding control object is included in the compensation refactoring domain, and new compensation control data is formed along with the objects in the domain. Each compensation refactoring uses the current effective deviation benchmark component, the current effective non-stable compensation response, and the currently unexecuted proposed compensation action as inputs, and historical failure versions are entered into the version tracking record.

[0358] When the measured convergence verification and forward convergence verification of all controlled quality characteristics enter the pass state, the permission to complete the unexecuted compensation action is closed, the historical process parameter compensation record is updated, and the process capability convergence result is generated, the automatic compensation control of this round enters the normal termination state.

[0359] When historical response samples are pending supplementation, quality inspection data is pending completion, the actual response direction remains abnormal, the action acceptance limit is zero, multiple consecutive rounds of reconstruction are in a state of no acceptance, control data version verification fails, or there is a discrepancy between the equipment execution status and the control data, the relevant controlled object enters an abnormal pause or pending review state. The abnormal state is associated with the abnormality type, starting production cycle, target equipment, controlled quality characteristics, compensation actions, and current data version.

[0360] After the cause of the anomaly is handled, the deviation aggregation result is re-formed based on the latest quality inspection data, action execution status, non-stability compensation response, parameter version, and historical response benchmark version, and then proceeds to the subsequent processing from the compensation reconstruction node.

[0361] After the process capability convergence results are generated, the actual execution data of the compensation actions and the complete actual quality characteristic responses are written into the historical process parameter compensation record. The response samples that meet the sample admission conditions are written into the new historical response baseline version, and the unexecuted compensation actions and their permission closed status are archived.

[0362] When the next production batch begins, the appropriate parameter version and historical response baseline version are called according to the target equipment, product specifications, process parameters and controlled quality characteristics. The deviation baseline component is formed by the initial quality inspection results of the new production batch, and the updated historical process parameter compensation records are used as the historical basis for accepting the non-stable compensation response and determining the proposed compensation response.

[0363] Based on the above method embodiments, this application also provides a process capability convergence system based on cross-device deviation aggregation. The system includes a processor and a memory. The memory stores program instructions. When the processor executes the program instructions, it calls a historical compensation response identification unit, a cross-device deviation aggregation unit, a compensation interference delimitation unit, a compensation authority control unit, and a process convergence closure unit to execute the aforementioned process capability convergence method. The data input, data output, and status update relationships between the units are consistent with the corresponding processing procedures in the aforementioned method. Figure 8 As shown, the process capability convergence system based on cross-device deviation aggregation includes the following units:

[0364] The historical compensation response identification unit 110 is used to acquire historical process parameter compensation records and corresponding actual quality characteristic responses of multiple target equipment; extract compensation actions that have been executed before the current production cycle from the historical process parameter compensation records; identify compensation actions that have not yet met the preset response stability conditions based on the actual quality characteristic responses and preset response stability conditions; and determine the quality characteristic responses that have not been completed since the current production cycle as unstable compensation responses.

[0365] The cross-device deviation collection unit 120 is used to acquire the quality characteristic deviation and proposed compensation action of each target device in the current production cycle, determine the proposed compensation response corresponding to each proposed compensation action, establish the association between the quality characteristic deviation of each target device and the corresponding non-stable compensation response and proposed compensation response and perform cross-device collection, retain the attribution relationship between the non-stable compensation response and the proposed compensation response and the corresponding compensation action, and form the deviation collection result.

[0366] The compensation interference delimitation unit 130 is used to determine the process center and discrete state between equipment in the equipment group based on the deviation aggregation result, and form a process convergence margin accordingly; it synthesizes the jointly acting unsteady compensation response and quasi-compensation response to form a synthetic compensation response characterizing the migration of the process center and the change of discrete state between equipment in the equipment group; when the synthetic compensation response exceeds the process convergence margin, it establishes a compensation interference relationship between the corresponding compensation actions according to the attribution relationship.

[0367] The compensation permission control unit 140 is used to determine the current compensation action from the proposed compensation actions participating in the compensation interference relationship according to the compensation interference relationship and configure the corresponding compensation execution permission, and at the same time set compensation constraints for the proposed compensation actions to form compensation control data;

[0368] The process convergence closure unit 150 is used to perform process parameter compensation based on the compensation control data, and update the deviation aggregation result based on the actual quality characteristic response of the compensation action already performed; after any current compensation action meets the preset response stability condition, the compensation interference relationship and subsequent compensation actions are re-determined based on the updated deviation aggregation result, and new compensation control data is formed; when the equipment group meets the process capability convergence condition, a process capability convergence result is generated.

[0369] The above description is merely a preferred embodiment of this application. It should be understood that this application is not limited to the forms disclosed herein and should not be construed as excluding other embodiments. It can be used in various other combinations, modifications, and environments, and can be altered within the scope of the concept described herein through the above teachings or the technology or knowledge in related fields. Modifications and variations made by those skilled in the art that do not depart from the spirit and scope of this application should be considered within the scope of this application.

Claims

1. A process capability convergence method based on cross-device deviation aggregation, characterized in that, include: Acquire historical process parameter compensation records and corresponding actual quality characteristic responses for multiple target devices; Extract the compensation actions that have been executed before the current production cycle from the historical process parameter compensation records, identify the compensation actions that have not yet met the preset response stability conditions based on the actual quality characteristic response and the preset response stability conditions, and determine the quality characteristic response that has not been completed since the current production cycle as the unstable compensation response. Obtain the quality characteristic deviation and proposed compensation action of each target device in the current production cycle, and determine the proposed compensation response corresponding to each proposed compensation action; establish the association between the quality characteristic deviation of each target device and the corresponding non-stability compensation response and proposed compensation response and perform cross-device aggregation, retain the attribution relationship between the non-stability compensation response and the proposed compensation response and the corresponding compensation action, and form the deviation aggregation result; The process center and discrete states between equipment in the equipment group are determined from the deviation aggregation results, and a process convergence margin is formed accordingly. The unsteady compensation response and the pseudo-compensation response that work together are synthesized to form a synthetic compensation response that characterizes the migration of the process center and the changes in discrete states between equipment in the equipment group. When the synthetic compensation response exceeds the process convergence margin, a compensation interference relationship is established between the corresponding compensation actions according to the attribution relationship. Based on the compensation interference relationship, determine the current compensation action from the proposed compensation actions that participate in the compensation interference relationship and configure the corresponding compensation execution permission. At the same time, set compensation constraints for the proposed compensation actions to form compensation control data. Process parameter compensation is performed based on the compensation control data, and the deviation aggregation result is updated based on the actual quality characteristic response of the compensation action already performed. After any current compensation action meets the preset response stability condition, the compensation interference relationship and subsequent compensation actions are re-determined based on the updated deviation aggregation result, and new compensation control data is formed. When the equipment group meets the process capability convergence condition, process capability convergence result is generated.

2. The process capability convergence method based on cross-device deviation aggregation according to claim 1, characterized in that, The determination of the non-steady compensation response includes: Based on the target equipment, process parameters, and controlled quality characteristics affected by each executed compensation action, as well as the corresponding compensation direction, compensation magnitude, and execution time, compensation action description data is generated. Based on the compensation action description data, a time-series correspondence is established between the historical compensation actions that have completed the response stabilization process and the actual quality characteristic responses of each production cycle after their execution, forming a time-series compensation response benchmark that characterizes the response direction and response amount of the compensation action in each production cycle. For the executed compensation action to be judged, a compensation response reference sequence matching it is determined from the timing compensation response benchmark. Based on the response cycle already experienced by the executed compensation action to be judged, the remaining response sequence from the current production cycle is determined from the compensation response reference sequence. The preset response stability conditions include: within a preset number of consecutive production cycles, the absolute value of the response change between adjacent production cycles does not exceed a preset stability threshold, and the absolute value of each remaining response in the remaining response sequence does not exceed a preset residual threshold. The executed compensation action that does not meet the preset response stability condition is determined as an unstable compensation action, and the remaining response sequence corresponding to the unstable compensation action is determined as an unstable compensation response.

3. The process capability convergence method based on cross-device deviation aggregation according to claim 1, characterized in that, The resulting bias aggregation includes: Based on the target equipment, process parameters, controlled quality characteristics, and compensation direction of the proposed compensation action, a compensation magnitude response relationship is formed from the historical process parameter compensation records and the corresponding actual quality characteristic responses. This relationship characterizes the mapping relationship between the compensation magnitude and the response direction and response quantity corresponding to each expected response cycle. The proposed compensation response is then determined based on the proposed compensation magnitude of the proposed compensation action. The controlled quality characteristic directly targeted by the proposed compensation action is determined as the target controlled quality characteristic. The executed compensation action corresponding to the non-stable compensation response and the proposed compensation action corresponding to the proposed compensation response are respectively constructed as compensation action items. When the same compensation action item generates quality characteristic responses to at least two controlled quality characteristics, they are collected according to the controlled quality characteristics actually affected by each quality characteristic response, and the attribution relationship of each quality characteristic response to the same compensation action item is retained. Based on the target equipment, controlled quality characteristics, and production cycle, the current quality characteristic deviation of each target equipment is configured as the starting value of the deviation for the production cycle covered by the non-stability compensation response and the proposed compensation response, forming a deviation baseline component; the non-stability compensation response and the proposed compensation response are aggregated into compensation response components, and the attribution relationship between each compensation response and the corresponding compensation action item is retained, forming the deviation aggregation result.

4. The process capability convergence method based on cross-device deviation aggregation according to claim 3, characterized in that, The determination of process termination margin and the formation of a synthetic compensation response include: Based on the deviation reference component, the process center and discrete state between equipment in the equipment group are determined according to the controlled quality characteristics and production cycle; the difference between the upper boundary of the preset target center interval and the process center of the equipment group is determined as the upward center margin, the difference between the process center of the equipment group and the lower boundary of the preset target center interval is determined as the downward center margin, and the difference between the upper boundary of the preset allowable discrete range and the discrete state between equipment is determined as the discrete margin. The upward center margin, the downward center margin, and the discrete margin constitute the process convergence margin. Identify production cycles in which at least two compensation action items and corresponding compensation responses exist for the same controlled quality characteristic within the same production cycle, forming a set of overlapping response cycles; For each production cycle in the overlapping response cycle set, the compensation response corresponding to each compensation action item is superimposed onto the deviation reference component to form a single response projection; the compensation responses corresponding to each compensation action item in the same production cycle are superimposed onto the deviation reference component to form a joint response projection. Based on the changes in the discrete states of the equipment group process center and between equipment before and after the projection of a single response, the central action and discrete action corresponding to each compensation action are formed; based on the changes in the discrete states of the equipment group process center and between equipment before and after the projection of a joint response, the central migration amount and discrete change amount corresponding to each production cycle are formed, and the central migration amount and the discrete change amount constitute the synthetic compensation response.

5. The process capability convergence method based on cross-device deviation aggregation according to claim 4, characterized in that, The establishment of the compensation-intervention relationship includes: For each production cycle in the overlapping response cycle set, the excess amount of the process center and discrete state between equipment in the equipment group before and after the joint response projection relative to the corresponding target center boundary or allowable discrete upper bound is determined. The increase of the excess amount after the joint response projection relative to the excess amount before the joint response projection is determined as the response new excess amount, and the production cycle in which the response new excess amount is greater than zero is determined as the out-of-bounds production cycle. Based on the central action and discrete action formed by the projection of each individual response, from the compensation action items that generate quality characteristic responses within the out-of-bounds production cycle, determine the compensation action items that make a unidirectional contribution to the boundary corresponding to the newly added excess amount of the response. A compensation interference relationship is established between any two compensation action items determined within the same out-of-bounds production cycle, wherein the two compensation action items that establish the compensation interference relationship correspond to different target devices, and at least one of the compensation action items corresponds to the proposed compensation action.

6. The process capability convergence method based on cross-device deviation aggregation according to claim 5, characterized in that, Determining the current compensation action from the proposed compensation actions involved in the compensation interference relationship includes: The compensation action items connected along the aforementioned compensation interference relationship are grouped into compensation interference groups; Under the condition that the unsteady compensation response involved in the compensation interference group continues to act, the proposed compensation response corresponding to each proposed compensation action item in the compensation interference group is superimposed on the corresponding deviation reference component, and candidate margin data is formed based on the process convergence margin after the proposed compensation response is acted. The proposed compensation action item is determined as the compensation acceptance candidate if all the remaining values ​​in the candidate remaining data are not less than zero and the absolute value of the deviation of the controlled quality characteristics of the target decreases after the proposed compensation response is implemented. The deviation reduction rate of each compensation candidate is determined by the ratio of the reduction in the absolute value of the deviation of the target controlled quality characteristics before and after the proposed compensation response to the absolute value of the deviation before the response. For compensation intervention groups with compensation acceptance candidates, the current compensation action of the corresponding compensation intervention group is determined from the compensation acceptance candidates whose deviation reduction rate reaches the maximum value; compensation intervention groups without compensation acceptance candidates are determined as non-acceptance compensation intervention groups.

7. The process capability convergence method based on cross-device deviation aggregation according to claim 6, characterized in that, The formation of compensation control data includes: The current compensation actions corresponding to each compensation intervention group and the proposed compensation actions not included in the compensation intervention group are identified as compensation actions to be authorized. Based on the proposed compensation response corresponding to the compensation action to be authorized, the unstable compensation response that works together with it, and the corresponding process convergence margin, the range of executable compensation range is determined, and the proposed compensation range of the compensation action to be authorized is limited to the corresponding range of executable compensation range to form the execution compensation range. The compensation execution permissions for the proposed compensation actions are configured according to the action acceptance results of each compensation intervention group. Each compensation intervention group opens the compensation execution permission for the corresponding current compensation action. The compensation execution permission for proposed compensation actions in no compensation intervention group remains closed. The compensation execution permissions for proposed compensation actions not included in the compensation intervention group are configured according to their executable compensation range. The compensation control data consists of the compensation execution authority, the compensation range, and the response stability triggering condition. The response stability triggering condition is that the executed compensation action, formed after the proposed compensation action is actually executed, meets the preset response stability condition.

8. The process capability convergence method based on cross-device deviation aggregation according to claim 7, characterized in that, The step of updating the deviation aggregation result, compensation interference relationship, and compensation control data based on the actual quality characteristics of the compensation actions already performed includes: When a proposed compensation action with the compensation execution permission in the open state is implemented, it is determined as an executed compensation action. The corresponding proposed compensation response is updated according to the actual compensation magnitude, and the updated proposed compensation response is migrated to a non-stable compensation response. Before the response stabilization trigger condition is triggered, the corresponding unstable compensation response and its remaining response sequence are corrected according to the actual quality characteristics of the compensation action that has been executed, and the compensation execution permission of the proposed compensation action that has not yet been executed in the same compensation interference group is kept in the closed state. When the response stability triggering condition is triggered, the deviation reference component is updated according to the actual quality characteristics, so that the corresponding non-stable compensation response exits the compensation response component, and the controlled quality characteristics affected by the executed compensation action and the associated unexecuted proposed compensation action and compensation interference relationship are used to form a compensation reconstruction domain. The original compensation control data within the compensation reconstruction domain is invalidated. Based on the updated deviation aggregation results, the compensation interference relationship, the current compensation action, and the execution compensation magnitude are re-determined to form new compensation control data. The original compensation control data outside the compensation reconstruction domain that maintains the same dependency state remains valid.

9. The process capability convergence method based on cross-device deviation aggregation according to claim 1, characterized in that, The process of determining whether the equipment group meets the process capability convergence condition and generating the process capability convergence result includes: The quality characteristic deviation of each target device is obtained within a preset number of consecutive production cycles, including the current production cycle. Based on this, the process center and discrete state between devices corresponding to each controlled quality characteristic are determined. It is then determined whether the process center and discrete state between devices remain within a preset target center interval and a preset allowable discrete range, respectively, within the consecutive production cycles, thus forming a measured convergence verification result. Using the quality characteristic deviations of each target device at the end of the continuous production cycle as the forward starting state, according to the production cycle covered by the unstable compensation response, each unstable compensation response is superimposed on the forward starting state of the corresponding target device and controlled quality characteristic, and the process center and discrete state between subsequent devices corresponding to each production cycle are re-determined; it is determined whether the process center and discrete state between subsequent devices are respectively within the preset target center interval and the preset allowable discrete range, forming a forward convergence verification result; When both the measured convergence verification result and the forward convergence verification result indicate that the convergence verification has passed, the compensation execution permission for those who have not yet performed the proposed compensation action is set to the off state. The historical process parameter compensation record is updated according to the executed compensation action and its actual quality characteristic response to generate the process capability convergence result. The updated historical process parameter compensation record is used as the historical basis for determining the proposed compensation response for the next production batch.

10. A process capability convergence system based on cross-device deviation aggregation, characterized in that, The system includes: The historical compensation response identification unit is used to acquire historical process parameter compensation records and corresponding actual quality characteristic responses of multiple target devices; extract compensation actions that have been executed before the current production cycle from the historical process parameter compensation records; identify compensation actions that have not yet met the preset response stability conditions based on the actual quality characteristic responses and preset response stability conditions; and determine the quality characteristic responses that have not been completed since the current production cycle as unstable compensation responses. The cross-device deviation collection unit is used to obtain the quality characteristic deviation and proposed compensation action of each target device in the current production cycle, determine the proposed compensation response corresponding to each proposed compensation action, establish the association between the quality characteristic deviation of each target device and the corresponding non-stability compensation response and proposed compensation response and perform cross-device collection, retain the attribution relationship between the non-stability compensation response and the proposed compensation response and the corresponding compensation action, and form the deviation collection result. The compensation interference delimitation unit is used to determine the process center and discrete state between equipment in the equipment group based on the deviation aggregation result, and to form a process convergence margin accordingly; it synthesizes the jointly acting unsteady compensation response and quasi-compensation response to form a synthetic compensation response characterizing the migration of the process center and the change of discrete state between equipment in the equipment group; when the synthetic compensation response exceeds the process convergence margin, it establishes a compensation interference relationship between the corresponding compensation actions according to the attribution relationship. The compensation permission control unit is used to determine the current compensation action from the proposed compensation actions participating in the compensation interference relationship according to the compensation interference relationship and configure the corresponding compensation execution permission, and at the same time set compensation constraints for the proposed compensation actions to form compensation control data; The process convergence closure unit is used to implement process parameter compensation based on the compensation control data, and update the deviation aggregation result based on the actual quality characteristic response of the compensation action already performed; after any current compensation action meets the preset response stability condition, the compensation interference relationship and subsequent compensation actions are re-determined based on the updated deviation aggregation result, and new compensation control data is formed; when the equipment group meets the process capability convergence condition, the process capability convergence result is generated.

Images