Automatic control system for manufacturing skim milk powder

Abstract

The application provides an automatic control system for manufacturing skimmed milk powder, which comprises a central cooperative controller, a plurality of field controllers, and pumping devices, valve actuators and buffer execution units controlled by the field controllers. The central cooperative controller determines a signature permission window according to a device running state sequence and a current delivery topology, generates a cumulative delivery amount deviation limited conservation type control signature, identifies a control action front and a production section front, generates production section identity data, a predicted device occupation interval and a stable occupation interval based on the front and rear boundaries of the production section, generates candidate control instructions according to the device capacity boundary and a withdrawal path, and adopts an ant colony algorithm to form a conditional control instruction set and a control execution sequence. By generating the conservation type control signature, the actual propagation path of the control action is identified under the condition that the cumulative delivery amount deviation is limited, and the influence of the path identification process on the normal production rhythm and the material delivery amount is reduced.

Description

Technical Field

[0001] This invention relates to the field of automation control technology for skim milk powder production process, and particularly to an automatic control system for skim milk powder manufacturing. Background Technology

[0002] Skim milk powder manufacturing typically involves multiple continuous or intermittent processes, including raw material transportation, temporary storage, concentration, drying, and finished product transfer. These manufacturing units are connected via pumping devices, valve actuators, pipelines, and buffer systems. Existing production lines usually start, stop, and adjust the load of each manufacturing unit based on equipment liquid level, pipeline pressure, pumping frequency, valve opening, drive load, and processing cycle time.

[0003] Due to transport delays, stagnation, and residual effects of materials within pipelines and equipment, pressure or load changes caused by actuator actions typically occur before the corresponding material reaches the downstream manufacturing unit. Existing control systems often determine control effectiveness based on actuator feedback or local pressure changes, making it difficult to distinguish the propagation location of the control action from the actual location of the target material section. Consequently, they cannot accurately determine whether the control action has been applied to the target material section.

[0004] Meanwhile, multiple production sections may exist consecutively within the same conveying path. After passing through branch pipelines, buffer devices, or periodic processing equipment, the leading edge, trailing edge, and equipment occupancy time of each production section will change. Existing systems typically estimate material positions based on planned time or cumulative conveying volume, making it difficult to stably bind control commands to specific production sections. This can easily lead to problems such as control actions arriving prematurely, only partially affecting the target production section, or extending to subsequent production sections.

[0005] Furthermore, when multiple manufacturing execution units adjust their capacity simultaneously, different control commands may compete for the same pumping device, valve actuator, or buffer device. Existing fixed priority or sequential queuing methods typically do not comprehensively consider current occupancy, occupancy of production sections in transit, committed occupancy, residual control effects, and abnormal withdrawal conditions, which can easily lead to control conflicts or equipment capacity exceeding limits.

[0006] Therefore, the present invention proposes an automatic control system for manufacturing skim milk powder. The information disclosed in the background section is only for enhancing understanding of the background of this disclosure and may therefore contain prior art information that is not common knowledge to those skilled in the art. Summary of the Invention

[0007] The purpose of this invention is to address the shortcomings of the prior art by providing an automatic control system for manufacturing skim milk powder, thereby solving the technical problems mentioned in the background section.

[0008] To achieve the above objectives, the present invention provides the following technical solution:

[0009] An automatic control system for manufacturing skim milk powder includes: a central coordinating controller, multiple field controllers, and a pumping device, a valve actuator, and a buffer actuator controlled by the field controllers; the central coordinating controller includes:

[0010] The control signature generation module is used to determine the signature allowable window and signature action boundary based on the equipment operating status sequence and the current conveying topology, control the target actuator to perform identification and compensation actions with opposite execution directions and limited cumulative conveying volume deviation, and generate a conservation control signature.

[0011] The dual-front identification module is used to identify the actual propagation path based on the actual action sequence and response of the conservation control signature, and to determine the control action front and the production segment front.

[0012] The segment action relationship construction module is used to generate target production segments based on the cumulative transport volume of the front boundary and the cumulative transport volume of the rear boundary, determine the stable occupied interval, and establish a set of prediction parameters for the control action shadow area.

[0013] The collaborative instruction generation module is used to generate candidate control instructions based on the equipment capability boundary and withdrawal path, calculate the control action shadow area and the allowed effective window, and use the ant colony algorithm to generate a set of conditional control instructions and a control execution sequence in the control instruction feasible connection graph;

[0014] The distributed execution and feedback correction module is used to execute conditional control commands, determine the control hit status based on the actual control action range and the actual occupied range of the target production section, and update the propagation time delay range, transmission time delay range, control action shadow area prediction parameter set, equipment capacity boundary, and control command feasible connection diagram pheromones based on effective feedback samples.

[0015] The beneficial effects of this invention are as follows:

[0016] This invention generates a conservation-type control signature to identify the actual propagation path of control actions when cumulative conveying deviation is limited, thus reducing the impact of path identification on normal production cycle time and material conveying volume. By separately identifying the leading edge of the control action and the leading edge of the production section, it is possible to distinguish the propagation position of the control signal from the actual position of the target production section, improving the accuracy of the correspondence between the control action and the target production section.

[0017] This invention establishes production segment identity data, projected equipment occupancy intervals, and stable occupancy intervals, enabling control commands to be bound to specific production segments, thus reducing premature action, partial hits, and cross-segment leakage. By comprehensively considering current actual occupancy, occupancy of production segments en route, committed occupancy, and occupancy of unfulfilled control actions, it constructs equipment capacity boundaries, reducing the risk of equipment exceeding capacity limits during control execution.

[0018] This invention establishes a feasible connection graph for control commands and uses an ant colony algorithm to generate control execution sequences, thereby coordinating the control needs of multiple manufacturing execution units and reducing competition and command conflicts among actuators. By verifying the actual control action range against the actual occupied range of the target production section, and updating propagation delay, transmission delay, predicted parameters of the control action shadow area, and equipment capacity boundaries based on valid feedback samples, closed-loop correction of control parameters is achieved, improving the system's continuous control accuracy and anomaly adaptability. Attached Figure Description

[0019] Figure 1 This is a framework diagram of an automatic control system for manufacturing skim milk powder according to the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Example: Figure 1 As shown, this embodiment provides an automatic control system for manufacturing skim milk powder, including: a central coordinating controller, multiple field controllers, and a pumping device, a valve actuator, and a buffer execution unit controlled by the field controllers;

[0022] The central coordination controller includes:

[0023] The control signature generation module is used to determine the signature allowable window and signature action boundary based on the equipment operating state sequence and the current conveying topology. It controls the target actuator to perform identification and compensation actions with opposite execution directions and limited cumulative conveying volume deviation, and generates a conservation-type control signature, which specifically includes the following:

[0024] S110. Constructing the equipment operating status sequence and current conveying topology: The industrial control system includes a central coordinating controller, field controllers located in multiple manufacturing execution units, and pumping devices, valve actuators, and buffer execution units controlled by the field controllers.

[0025] The central collaborative controller reads the equipment operation data uploaded by each field controller through the industrial communication bus. Among them, the start / stop status of the pumping device, the actual operating frequency, and the drive load are obtained from the corresponding frequency converter or motor controller; the valve command opening degree is obtained from the control output register of the field controller; the valve feedback position is obtained from the position feedback interface of the valve actuator; and the pipeline pressure is obtained from the pressure sensors installed at the inlet, outlet, or branch node of each conveying connection section.

[0026] The equipment liquid level is obtained from the liquid level sensor of the buffer execution unit and converted into the occupied volume according to the pre-calibrated liquid level-volume correspondence table; the cumulative delivery volume is obtained from the cumulative delivery metering device. If the cumulative delivery metering device is not configured, it is obtained by accumulating the calibrated instantaneous delivery volume according to the sampling period; the equipment processing cycle is obtained from the cycle start signal, cycle end signal and completion counter output by the field controller of the manufacturing execution unit.

[0027] The equipment's rated capacity, allowable continuous drive load, allowable maximum occupied volume, safety reserved volume, allowable range of motion of the actuator, and response time are derived from a pre-stored equipment basic configuration table. The equipment basic configuration table is determined based on the equipment's rated parameters, actuator technical data, production line debugging data, and background fluctuation amplitude during historical steady-state operation.

[0028] The sampling period for pressure, pumping frequency, valve feedback position, and drive load is preferably 200ms, while the sampling period for equipment level and cumulative delivery is preferably 1s. Each field controller writes a local timestamp and scan number when collecting data, and the central coordination controller performs time alignment according to a unified clock, with an allowable time synchronization error of no more than 200ms.

[0029] For data with a continuous missing time of no more than 2 sampling periods, linear interpolation is performed using the valid sampled values ​​on both sides of the missing interval; when the continuous missing time exceeds 2 sampling periods, the corresponding data is marked as invalid data, and the transmission connection segment that depends on the data is marked as pending verification.

[0030] The central collaborative controller uses manufacturing execution units, pumping devices, valve actuators, and buffer execution units as topology nodes, and physical transport connection sections as topology edges. The corresponding topology edge is marked as available only when the corresponding pumping device is in an operable state, the valve feedback position meets the connectivity conditions, the connection section pressure is within the allowable range, and the relevant field controller communication is effective. Based on this, the current transport topology and topology version number are generated.

[0031] Upon initial startup, an initial in-transit occupied range is generated based on the cumulative delivery volume increment, pumping device operating status, valve feedback position, and equipment liquid level. Subsequent control cycles update the in-transit production section status using the production section leading edge, production section trailing edge, occupied segment, and section reliability status feedback from S530.

[0032] The central collaborative controller combines the equipment operation data, data validity identifier, current transport topology, topology version number, and status of the production section in transit into an equipment operation status sequence according to the collection time sequence. This sequence is used by S120 to determine the signature permission window and by S210, S220, S310, and S510.

[0033] In one specific implementation, the manufacturing execution unit includes a concentration manufacturing execution unit and a spray drying manufacturing execution unit, each equipped with a field controller, which is implemented using a PLC. The concentration manufacturing execution unit and the spray drying manufacturing execution unit are equipped with online monitoring instruments for temperature, vacuum, pressure or flow. The detection data of each online monitoring instrument is uploaded to the central collaborative controller after being written with a collection timestamp by the field controller, and is written into the equipment operation status sequence as the equipment operation data of the corresponding manufacturing execution unit.

[0034] S120. Determine the signature allowable window and signature action boundary: The signature allowable window refers to the continuous time interval during which a conservative control signature is allowed to be generated without changing the production scheduling task, without causing the manufacturing execution unit to exceed the equipment capacity, and under the condition that reverse compensation can be completed; The signature action boundary is the set of action amplitude, action duration, action interval, cumulative delivery allowable deviation and corresponding tolerance that a conservative control signature is allowed to adopt within this time interval.

[0035] The central coordination controller calculates the drive load margin ratio, valve adjustment margin ratio, available buffer volume margin ratio, and action time margin ratio for each candidate conveying path p based on the equipment operating status sequence, and determines the path signature margin.

[0036]

[0037] In the formula, The path signature margin for candidate delivery path p; The ratio of the difference between the allowed continuous drive load and the current drive load to the allowed continuous drive load; It is the smaller of the ratios of the motion margin obtained by the actuator adjusting forward and backward relative to the current operating state; The ratio of the available volume of the buffer execution unit to the estimated maximum instantaneous delivery deviation of the conservative control signature; It is the ratio of the current conflict-free period to the time required to complete the identification action, compensation action, and the recovery of the executing agency.

[0038] when The signature enable window is opened when: the available volume of the relevant buffer execution unit is not less than the expected maximum instantaneous transport deviation; there is no higher priority abnormal control; there is no repeated control prohibition state of the same actuator from the previous control cycle; and the current transport topology remains unchanged. The signature enable window is closed when any of the conditions are no longer met, the topology version changes, critical data becomes invalid, or the remaining time is insufficient to complete the reverse compensation.

[0039] The central coordinating controller determines the upper limit of action amplitude, upper limit of action duration, action interval, allowable deviation of cumulative conveying volume, tolerance of action amplitude, tolerance of action time, and tolerance of recovery of baseline operating state based on the equipment basic configuration table and the resolution of the field controller, and sends the above results as the signature action boundary to S130.

[0040] S130. Generate a conservation-type control signature: A conservation-type control signature refers to applying at least one identification action and at least one compensation action in the opposite direction to the target pumping device or valve actuator within the signature permission window, ensuring that the deviation of the actual cumulative delivery volume from the planned cumulative delivery volume at the end of the signature period does not exceed the allowable deviation of the cumulative delivery volume, while simultaneously creating control action combinations with distinguishable response sequences for different delivery paths. Different action direction sequences, action durations, or adjacent action intervals are configured for different delivery paths, and two delivery paths with overlapping response ranges are prohibited from simultaneously executing the same signature.

[0041] The cumulative delivery deviation during the signature period is:

[0042]

[0043] In the formula, The cumulative delivery deviation during the signature period; and The cumulative quantity metering device for the export of the source manufacturing execution unit is located at the moment of signature start. and the end of the signature The reading; This refers to the planned cumulative delivery volume for the production scheduling task within the signature period.

[0044] During the compensation process, the central coordination controller continuously reads the cumulative delivery volume and... The signature should end when the cumulative delivery deviation is not greater than the allowable deviation and the actuator returns to the reference operating state. For example, if the pumping device generates a positive deviation of 12 metering pulses after increasing the frequency from 35Hz to 37Hz and holding it for 2 seconds, the compensation should not use a fixed open-loop control method of holding the frequency at 33Hz for 2 seconds. Instead, the cumulative delivery should be continuously read, and the compensation should be terminated when the remaining deviation is not greater than 2 metering pulses.

[0045] The central coordinating controller configures a signature identifier, source path, target actuator, action sequence, expected pressure response direction, expected drive load response direction, and expected response time range for each conservation control signature, forming a control signature instruction and signature response template.

[0046] The expected response time range is initially derived from equipment debugging data and updated by S530 based on valid feedback samples. S130 outputs control signature instructions, signature response templates, signature action boundaries, and cumulative transmission amounts corresponding to the start and end times of the signature, for S210 to confirm the execution status, and for use by S220, S230, and S310.

[0047] The dual-front identification module is used to identify the actual propagation path based on the actual action sequence and response of the conservation control signature, and to determine the control action front and the production segment front, specifically including the following:

[0048] S210, Confirm the actual execution status of the conservation control signature: The central coordinating controller reads the control signature command, signature action boundary, signature response template, cumulative delivery allowable deviation and topology version number output by S130, and reads the actual operating frequency and drive load from the frequency converter driver or motor controller of the pumping device, reads the valve feedback position from the position feedback interface of the valve actuator, reads the actual cumulative delivery from the cumulative quantity metering device at the outlet of the source manufacturing execution unit, and reads the acquisition timestamp and scan sequence number of each data from the field controller.

[0049] The actual action sequence is a set of the actual action direction, start and end times, amplitude, recovery results of the baseline operating state, and corresponding cumulative delivery deviation of the actuator, arranged in chronological order.

[0050] For pumping devices, when the actual operating frequency or driving load deviates from the baseline operating state and its fluctuation range for two consecutive sampling cycles, the first deviation time is determined as the signature propagation start point; for valve actuators, when the valve feedback position moves in the target direction for two consecutive sampling cycles, the first movement time is determined as the signature propagation start point.

[0051] The corresponding conservative control signature will be determined as a valid control signature only when the actual action direction is consistent with the control signature instruction, the action amplitude deviation does not exceed the action amplitude tolerance, the action start and end time deviation does not exceed the action time tolerance, the identification action and the compensation action have been completed, the actuator recovery error does not exceed the baseline operating state recovery tolerance, the cumulative delivery volume deviation of the signature cycle does not exceed the cumulative delivery volume allowable deviation, and the topology version number does not change during the execution period.

[0052] If an action has been performed but the compensation action has not been completed, it is identified as an incomplete control signature, and its cumulative transport deviation is sent to S410 as an in-transit occupancy correction. If the action direction is correct but the action sequence or duration exceeds the tolerance, it is identified as a distortion control signature. If the topology version number changes during execution, it is identified as a topology failure control signature and sent to S530. If the actuator fails to generate the prescribed feedback, it is identified as an execution failure control signature. None of the above control signatures are entered into S220.

[0053] For example, if a valve command requires the opening to increase from 40% to 44%, then return to 40%, and then decrease to 36%, but the valve actually only moves to 42% and then returns to 40%, then this control signature is an incomplete control signature and cannot be considered executed simply because the command has been issued. S210 outputs the actual action sequence of a valid control signature, the signature propagation start point, and the actual cumulative delivery volume corresponding to the start and end times of the signature, for use by S220, S230, and S310.

[0054] S220. Identifying the actual propagation path of valid control signatures: Before the signature propagation starting point, the central coordinating controller selects 10 consecutive pressure sampling cycles without equipment start / stop, valve switching, processing cycle switching, or other control signatures as a stable observation interval. The arithmetic mean is calculated for each pressure sensor and drive load sampling sequence, and the maximum absolute deviation of each sampled value within the stable observation interval from the arithmetic mean is determined as the background fluctuation amplitude. A response event refers to a record of changes in operating data relative to the baseline value of the stable observation interval that meet the requirements for response direction, response amplitude, duration, and expected response time range.

[0055] A response event is considered valid only if the direction of change matches the signature response template, the magnitude of change is greater than three times both the sensor's minimum resolution and the background fluctuation amplitude, the change is maintained continuously for at least two sampling periods, and the occurrence time falls within the expected response time range specified in the signature response template. The central coordinating controller, based on the current transport topology of S110, sequentially searches for valid response events along reachable connection segments and matches the response node, response direction, start time, duration, and order with the signature response template item by item.

[0056] The degree of signature path matching between signature s and candidate delivery path p is effectively controlled as follows:

[0057]

[0058] In the formula, To effectively control the degree of signature path matching between signature s and candidate delivery path p; This represents the total number of template response events that should appear on candidate delivery path p in the signature response template, and ; The number of actual response events that are matched in terms of response direction, response order, and response time range.

[0059] when If the value is not less than 0.8 and both the first and last response nodes of the candidate transmission path are matched, it is determined as a valid matching path; if only one candidate transmission path meets the condition, it is determined as the actual propagation path; if at least two candidate transmission paths meet the condition at the same time, the path ownership status is marked as not unique, and the unique actual propagation path is uncertain.

[0060] For shared pipe segments, the response within the shared pipe segment is only used to confirm that a valid control signature has entered the segment. Only when the pressure response, cumulative delivery increment, or manufacturing execution unit driven load response after the branch node completes the match is the specific branch path confirmed. The connection segment between the last valid matching node and the first node after it that does not produce a valid response within the longest expected response time is determined as the signature interruption location.

[0061] For example, if only the main pipe pressure responds while the sensor data after the branch node fails, only the confirmed shared pipe section is output, and the unique branch path is not output. S220 outputs the actual propagation path, the response time of each level, the path ownership status, and the signature interruption location for use by S230, S310, and S530.

[0062] S230. Identify the control action front and the production section front separately: The control action front is the furthest propagation position and corresponding arrival time of the pressure change or driving load change caused by the effective control signature in the actual propagation path; the production section front is the actual movement position and arrival time interval of the front boundary of the cumulative transport range bound to the effective control signature in the current transport topology. Both are bound to the same effective control signature but use different data sources and cannot be substituted for each other.

[0063] The central coordinating controller uses the farthest effective response node determined by S220 as the position of the control action front, the moment when the node first shows an effective pressure response as the earliest arrival time, and the moment when the pressure and drive load recover to the reference fluctuation range of the stable observation range and remain there for 3 consecutive sampling cycles as the latest duration. When only the inlet pressure responds and the target manufacturing execution unit drive load has not yet responded, the control action front is determined only to the inlet of the target manufacturing execution unit.

[0064] The central coordinating controller determines the cumulative transport amount corresponding to the signature propagation starting point output by S210 as the cumulative amount at the leading boundary of the production segment. For connection segments without branches or buffered execution units, when the following formula is true, it is determined that the leading edge of the production segment has reached the downstream node of the connection segment:

[0065]

[0066] In the formula, The reading of the cumulative quantity metering device installed at the entrance of the current connecting section of the transport path p at time t; The moment when the leading edge of the production section enters the current connecting section; This represents the effective occupied volume of the current connection segment; For a moment The corresponding cumulative metering device reading. The effective occupied volume is determined based on the pipeline inner diameter, effective length, valve cavity volume, and pump cavity volume. If complete filling cannot be guaranteed, the minimum and maximum values ​​calibrated on-site are used to form the volume range.

[0067] After the production section's leading edge enters the buffer execution unit, a buffer occupancy record is established based on the cumulative inlet and outlet conveying volumes and the volume of occupied space calculated from the equipment level. When passing through branch nodes, the readings of the independent cumulative metering devices for each branch are used preferentially. If reliable branch metering or estimation results are lacking, only the arrival of the production section's leading edge at the branch node is confirmed. For example, if the inlet pressure responds 4 seconds after the signature begins, but the production section's leading edge requires 19 seconds to arrive based on the effective occupied volume and cumulative conveying volume, then only the arrival of the control action leading edge is considered at 4 seconds.

[0068] Among them, the propagation time delay interval refers to the set of time spans from the effective control signature issuance time to the arrival of the control action front at each node; the transmission time delay interval refers to the set of time spans from the production section front to the actual movement of the source manufacturing execution unit output end to each downstream node.

[0069] The dual-front credibility state refers to the combined credibility of the control action front and the production segment front based on data quality tags, specifically including the aforementioned definite state, interval state, pending confirmation state, or abnormal state. S230 outputs the control action front state, production segment front state, propagation time delay interval, transmission time delay interval, and dual-front credibility state; the definite state and interval state are used by S310 to S330, the pending confirmation state continues to be tracked, and the abnormal state is sent to S530.

[0070] The segment interaction relationship construction module is used to generate target production segments based on the cumulative transport volume at the front and rear boundaries, determine stable occupancy intervals, and establish a set of prediction parameters for the control shadow area. Specifically, it includes the following:

[0071] S310. Generate production segment identity data bound to a valid control signature: A production segment refers to a continuous cumulative delivery range at the output end of the source manufacturing execution unit, defined by the cumulative delivery volume at the front boundary and the cumulative delivery volume at the back boundary, and bound to a valid control signature.

[0072] The central coordinating controller reads the valid control signature, signature propagation start point, actual action sequence, and actual cumulative transport volume corresponding to the start and end times of the signature output from S210. It also reads the actual propagation path output from S220, the production section leading edge state, transmission time delay interval, and double leading edge reliable state output from S230, and reads the cumulative transport volume from the cumulative quantity metering device at the source manufacturing execution unit outlet. When no metering device is configured, the calibrated instantaneous transport volume in S110 is accumulated according to the sampling period.

[0073] The central collaborative controller determines the cumulative transport volume corresponding to the signature propagation starting point as the front boundary cumulative transport volume, the cumulative transport volume corresponding to the completion of the compensation action and the return of the actuator to the baseline operating state as the back boundary cumulative transport volume, and the continuous cumulative transport volume range between the front boundary cumulative transport volume and the back boundary cumulative transport volume as the target production section.

[0074] The cumulative delivery range represents the actual delivery range through the output of the source manufacturing execution unit within the signature cycle. The cumulative delivery deviation represents the difference between the actual cumulative delivery and the planned cumulative delivery. These two represent completely different physical parameters. For example, if the cumulative delivery at the beginning and end of the signature cycle are 1000 and 1120 metering pulses respectively, and the planned cumulative delivery also increases by 120 metering pulses, then the cumulative delivery deviation is 0, but the target production section still corresponds to 1000 to 1120 metering pulses.

[0075] The central collaborative controller generates a segment identifier based on the signature identifier, the source manufacturing execution unit identifier, the cumulative transport volume at the front boundary, the topology version number, and the segment generation sequence number. This forms production segment identity data that includes the segment identifier, signature identifier, cumulative transport volume at the front boundary, cumulative transport volume at the rear boundary, current transport path, current connection segment, topology version number, leading edge status of the production segment, trailing edge status of the production segment, and the segment's trusted status. The trailing edge of the production segment represents the movement position and arrival time interval of the cumulative transport volume at the rear boundary within the current transport topology, and it is tracked using the same cumulative transport volume conservation method as the leading edge of the production segment in S230.

[0076] When the target production section passes through a branch node, it is preferentially divided into sub-production sections based on the cumulative delivery increment of the cumulative metering devices of each branch. Each sub-production section inherits the signature identifier and topology version number of the parent production section and is configured with an independent sub-section identifier. When there is no independent metering data, the central collaborative controller reads the actual operating frequency of the pumping device and the valve feedback position of each branch, and calculates the current instantaneous flow estimate of each branch in conjunction with the equipment basic configuration table. The ratio of the instantaneous flow estimate of a single branch to the sum of the instantaneous flow estimates of all connected branches is used as the allocation ratio of that branch. Based on this, the cumulative delivery increment of the sub-production section is divided, and the reliable status of the section is marked as the interval status. When the allocation ratio cannot be determined, it is only recorded that the target production section has reached the branch node.

[0077] When multiple sub-production segments enter the same buffer execution unit, a composite production segment record is established, retaining the segment identifier, remaining occupancy, entry time interval, and binding signature identifier of each sub-production segment, without merging them into a single production segment without a source. S310 outputs production segment identity data for use by S320, S410, S510, and S520.

[0078] S320. Determine the expected equipment occupancy range and stable occupancy range of the target production section: The expected equipment occupancy range is the entire time range during which the target production section may occupy the target manufacturing execution unit; the stable occupancy range is the time range during which it can be confirmed that the target manufacturing execution unit is completely occupied by the target production section. The central coordination controller reads the production section identity data output from S310 and the transmission time delay range output from S230, and reads the cycle start signal, cycle end signal, inlet cumulative conveying volume, outlet cumulative conveying volume, and equipment operating status from the target manufacturing execution unit field controller.

[0079] The expected equipment occupancy range of the target production section z in manufacturing execution unit e is:

[0080]

[0081] In the formula, For the expected equipment occupancy range; This is the moment when the production section's leading edge first enters manufacturing execution unit e; This is the latest moment when the production section leaves manufacturing execution unit e.

[0082] The stable occupied range of the target production section z in the manufacturing execution unit e is:

[0083]

[0084] In the formula, To ensure stable occupancy of the interval; The latest moment when the production section's leading edge enters manufacturing execution unit e; This is the earliest moment when the trailing edge of the production section leaves the manufacturing execution unit e. When the stable occupied interval is empty, no control commands that can fully apply to the production section are generated.

[0085] For continuously processing manufacturing execution units (MEPs), the departure time interval is determined based on the cumulative inlet and outlet conveying volume at the time of entry into the production section and the effective occupied volume inside the equipment. For periodically processing MEPs, the processing cycle that has not yet been completed after entry is included in the departure time. When the target production section enters the buffer execution unit, an occupancy segment is established. The occupancy segment is a record of the occupancy volume that has not yet been reduced (or emptied) by the cumulative outlet conveying volume, including the section identifier, entry time interval, existing occupancy volume ahead, current remaining occupancy volume, and estimated departure time interval.

[0086] For example, when the target production section enters the buffer execution unit, there is already 7.6 to 8.4 meters ahead. 3 Occupancy: Its own occupancy is 1.2m. 3 The cumulative export volume increased by 7.6m. 3 At that time, it could only be determined that the leading edge might have begun to move away, reaching 8.4m. 3 It was then determined that the front had moved away and continued to reach 8.8–9.6 m. 3 At each time, determine whether the trailing edge may have left or has already left.

[0087] When critical data fails, the arrival range is expanded according to the maximum and minimum conveying capacity in the equipment basic configuration table. If the expanded range still cannot cover the reliable location range, the corresponding production segment is marked as an occupied range failure and sent to S530. S320 outputs the expected equipment occupied range, stable occupied range, and occupied segment for use by S330, S410, S420, and S510.

[0088] S330. Establish a set of prediction parameters for the control action shadow area: The set of prediction parameters for the control action shadow area is a combination of the shortest propagation time, the longest propagation time, and the residual action duration that are pre-established for the target actuator, action direction, and discrete action amplitude. It is used by S420 to calculate the specific control action shadow area after generating candidate control actions.

[0089] The central coordinating controller reads the control action leading-edge state and propagation time delay interval output by S230, and calls up historical instruction execution records output by S520 that have been confirmed as valid feedback samples by S530. Only historical records with the same target actuator, the same direction of action, the same action amplitude within the same discrete amplitude interval, the same current conveying topology, the same pumping status and valve connection status, and the same processing cycle type of the target manufacturing execution unit are selected.

[0090] The shortest propagation time is the minimum difference between the start time of the actual control action and the start time of the actual action of the actuator in the historical records that meet the conditions, and the longest propagation time is the maximum value of the corresponding difference; the residual action duration is the maximum value of the time required for the pipeline pressure, the target manufacturing execution unit drive load, and the cumulative delivery increment within the unit sampling window to recover to the reference fluctuation range before the action and be maintained continuously for 3 sampling cycles after the actuator action ends.

[0091] When the first run or the number of qualified historical records is less than 3, the shortest and longest propagation times are taken from the values ​​recorded during the equipment commissioning phase, and the residual action duration is taken as the maximum value among the maximum stabilization time of the actuator, the recovery time of the on-site step test, and one equipment processing cycle. S330 outputs a set of control action shadow area prediction parameters indexed by the target actuator, direction of action, and discrete action amplitude, which is used by S420 to calculate the specific control action shadow area and continuously updated by S530 based on valid feedback samples.

[0092] The collaborative instruction generation module is used to generate candidate control instructions based on equipment capability boundaries and withdrawal paths, calculate the control action area and the allowed effective window, and use the ant colony algorithm to generate a set of conditional control instructions and a control execution sequence in the control instruction feasible connection graph. Specifically, it includes the following:

[0093] S410. Constructing Equipment Capability Boundaries and Withdrawal Paths: Equipment capability boundaries refer to the maximum range within which a manufacturing execution unit is allowed to add occupancy for candidate control actions within a predetermined control time domain, after deducting the current actual occupancy, the occupancy of production segments in transit, the committed occupancy, and the occupancy of incomplete control actions. The predetermined control time domain is determined by the latest departure time of the production scheduling task and the maximum coordination time in the equipment basic configuration table.

[0094] The central coordinating controller reads the allowable continuous load, maximum occupancy, maximum inlet conveying capacity, and maximum outlet conveying capacity from the equipment basic configuration table; it reads the drive load from the frequency converter or motor controller, reads the equipment liquid level from the liquid level sensor and calculates the current occupied volume according to the liquid level-volume correspondence table, and reads the equipment processing cycle from the field controller; at the same time, it calls the production section identity data of S310, the expected equipment occupancy range and occupancy segment of S320, the control action shadow area prediction parameter set of S330, the incomplete control signature corresponding to the in-transit occupancy correction amount of S210, and the conditional control commands that have been issued but not yet completed.

[0095] The central coordinating controller calls the frequency-conveying increment calibration table, opening-conveying increment calibration table, and load-allowable inlet conveying volume calibration table established during the production line commissioning phase, and converts the pumping frequency, valve opening, drive load, and incomplete control actions into the equipment capacity occupancy within the predetermined control time domain.

[0096] The calibration table establishment process is as follows: During the offline commissioning phase, the control valve actuator or pumping device is in multiple discrete basic operating states. Each state sends a step control command with a preset amplitude to the field controller, recording the average instantaneous delivery volume and cumulative metering pulse change after the system reaches steady state. The amplitudes of the multiple step commands are then fitted with the corresponding measured delivery increments to form a discrete mapping table. During actual system operation, for intermediate continuous values ​​not directly recorded in the calibration table, the system uses linear interpolation to calculate the corresponding equipment capacity utilization.

[0097] The remaining equipment capacity of manufacturing execution unit e at time t is:

[0098]

[0099] In the formula, For remaining equipment capacity; This represents the maximum occupancy. This represents the current actual occupancy level. The occupancy of the in-transit production segment is calculated based on the overlap between the expected equipment occupancy interval and the calculation step size at time t. The space reserved for conditional control instructions that have been issued but not yet executed; This refers to the remaining occupancy expected to result from actual control actions that are currently being executed or have not yet ended. All the above parameters are expressed as equipment capacity occupancy within the same predetermined control time domain, without directly adding data of different dimensions.

[0100] The central coordination controller calculates the remaining equipment capacity stepwise from the current moment to the predetermined control time domain end, using a 1-second calculation step, and merges the time periods that continuously meet the new occupancy requirements into the equipment capacity availability interval. The withdrawal path is the reverse control action and execution sequence that restores the pumping device, valve actuator, and buffer actuator to a safe operating state after the candidate control action is executed, in case of reduced equipment capacity, changes in the current conveying topology, or shifts in the production section position.

[0101] The withdrawal time is obtained by adding the maximum response time of the actuator, the longest propagation time, and the buffer switching time. S410 outputs the withdrawal action and withdrawal time as withdrawal path attributes; S420 verifies whether the complete withdrawal time is retained at the end of the window when determining the allowed effective window. S410 outputs the device capability boundary, the available range of device capabilities, and the withdrawal path for use by S420, S430, and S510.

[0102] S420: Generate candidate control instructions bound to the target production section: The central coordinating controller reads the target manufacturing execution unit, target production section, target entry time, target exit time, target processing cycle, allowable control range, and task priority from the production planning interface, and compares them with the production section identity data in S310 and the stable occupancy interval in S320 to determine the control direction of increasing conveying capacity, reducing conveying capacity, connecting buffer execution units, or adjusting the load of manufacturing execution units.

[0103] Based on the current state of the target actuator, candidate control actions are generated discretely according to the minimum action amount that the field controller can distinguish within the allowable range of equipment operation. The pumping device generates a combination of frequency adjustment and duration, the valve actuator generates a combination of opening adjustment and duration, the buffer actuator generates a combination of engagement, hold, and disengagement, and the manufacturing actuator generates a combination of target load and hold time.

[0104] For candidate control action a, the prediction parameters corresponding to its target actuator, action direction, and discrete action amplitude in S330 are called to calculate the control action shadow area:

[0105]

[0106] In the formula, The control effect shadow area of ​​candidate control action a; and These are the expected start and end times of the action, respectively. , and These are the shortest propagation time, the longest propagation time, and the duration of residual effect, respectively.

[0107] The allowable effective window is determined in reverse based on the stable occupancy interval, ensuring that the beginning of the control effect area is no earlier than the start of the stable occupancy interval and the end is no later than the end of the stable occupancy interval, and that the entire equipment capacity occupancy period is within the equipment capacity availability interval. If the earliest start time of the control effect area is earlier than the start time of the stable occupancy interval, an early action risk is identified; if the latest end time is later than the end time of the stable occupancy interval, a cross-segment leakage risk is identified.

[0108] When there is a risk of premature action, the expected execution time of the action is shifted backward by one field controller scan cycle. When there is a risk of cross-section leakage, the action amplitude is gradually reduced according to the minimum action resolution of the actuator. If this is still not sufficient, continuous actions are broken down into multiple short-duration actions, with the interval between adjacent actions not less than the minimum settling time of the actuator. After each adjustment, the control action shadow area, equipment capacity occupancy, and withdrawal path are recalculated. If a valid allowable effective window cannot be formed after 10 consecutive adjustments, the candidate control action is deleted.

[0109] For example, if the stable occupancy period is 10:00:20 to 10:00:42, the candidate valve action lasts for 4 seconds, the shortest propagation time is 6 seconds, the longest propagation time is 9 seconds, and the residual effect duration is 5 seconds, then the start time of the action must not be earlier than 10:00:14 and not later than 10:00:24. S420 encapsulates the candidate control actions that meet the conditions into candidate control instructions, including the target production section, target actuator, control action, control amplitude, allowed effective window, control effect shadow area, equipment capacity occupancy, withdrawal path, topology version number, and failure condition, for use by S430.

[0110] S430. Control execution sequence generation based on ant colony algorithm: The feasible connection graph of control instructions is a directed graph with combinations of "candidate control instructions - discrete start time" as control nodes and feasible connection edges for two control nodes that can execute continuously and satisfy all industrial control constraints. The central coordinating controller discretely sets the allowable effective window according to the scanning cycle of the field controllers; a feasible connection edge is established only when the execution time and recovery time of the preceding and following control nodes do not overlap, the superimposed equipment capacity occupancy does not exceed the equipment capacity boundary, their respective control shadow areas are located in the corresponding stable occupancy intervals, the withdrawal path is valid, and the topology version number is consistent.

[0111] The ant colony algorithm is only used to select the control execution sequence and does not directly calculate the control amplitude. The probability that the k-th ant will choose control node j from control node i is:

[0112]

[0113] In the formula, To select feasible connection edges The probability of; The pheromone corresponding to the feasible connecting edge; To obtain heuristic information, it is obtained through multi-objective weighted calculation, and the specific calculation formula is as follows:

[0114]

[0115] in, The Min-Max Normalization function linearly maps parameters of different dimensions to the (0,1] interval by appending a system-preset positive minimum value to the regular mapping result, thus preventing zero values ​​from appearing in the ant colony algorithm calculation; Pri represents the task priority. To control the time margin of the affected area, To allow for a time buffer before withdrawal, To account for the remaining equipment capacity after execution, Waiting time for the production section; to These are the system's preset positive weighting coefficients, and their sum is 1.

[0116] and These are pheromone weights and heuristic information weights, respectively. Let h be the set of feasible successor nodes for the k-th ant at control node i. Any candidate node in, and These are the pheromone and heuristic information for feasible connections between the corresponding control node i and candidate node h. Nodes with risks of premature action, cross-segment leakage, equipment exceeding capacity limits, or failed withdrawal paths are not allowed to enter the set.

[0117] During the initial run, all feasible connection edges use the same initial pheromone. A maximum of 30 ants and 50 iterations are preferred, and the process terminates if no better sequence is found after 10 consecutive iterations. If a feasible sequence exists, the control execution sequence that completes the most high-priority production scheduling tasks, has zero risks in both categories, and maximizes the remaining capacity of the equipment after execution is selected. If no feasible sequence exists, low-priority candidate control instructions are deleted, and the search is restarted. If no solution is found, the current safe operating state is maintained, and unfinished tasks are carried over to the next control cycle.

[0118] The S430 outputs a set of conditional control instructions and a control execution sequence, including the planned start and end times, precondition control instructions, allowed effective window, control shadow area, withdrawal path, failure condition, and execution priority, for the S510 to verify and execute; the pheromone is updated by the S530 based on the actual control hit results.

[0119] The distributed execution and feedback correction module is used to execute conditional control commands. It determines the control hit status based on the actual control action range and the actual occupied range of the target production section, and updates the propagation delay range, transmission delay range, control action shadow area prediction parameter set, equipment capacity boundary, and control command feasible connection diagram pheromones based on valid feedback samples. Specifically, it includes the following:

[0120] S510, Execute conditional control instructions and generate instruction execution records: The central coordination controller sends conditional control instructions to the corresponding field controllers according to the control execution sequence output by S430, and attaches the instruction sequence number, instruction check code, latest confirmation time, and repeated execution prohibition flag before sending. The field controller verifies the instruction check code, instruction sequence number, topology version number, and the status of the prerequisite conditional control instructions; instructions that have been executed, have exceeded the latest confirmation time, have failed verification, or whose prerequisite conditional control instructions have not been completed are not included in the execution queue.

[0121] The field controller sequentially places conditional control commands into the states of pending verification, execution lock, execution, residual action waiting, and completion. Commands exceeding the allowed effective window, experiencing topology version changes, or encountering anomalies in the target production section are transferred to the failure state. If the equipment capability boundary fails, the command is transferred to the withdrawal state. An execution lock is an exclusive occupancy identifier set by the field controller for a target actuator. When the same actuator is in the execution lock, execution, or withdrawal state, it will not receive other conditional control commands.

[0122] Before execution, the field controller retrieves the production section identity data from S310, the updated stable occupancy range from S320, the equipment capacity boundary and withdrawal path from S410, and reads the current time, actuator feedback, drive load, equipment level, processing cycle time, and the current conveyor topology and topology version number from S110. Based on the latest double-leading reliable state from S230 and the latest prediction parameters from S330, the control effect shadow area of ​​this conditional control command is recalculated.

[0123] An execution lock is set and an action is issued only when the current time is within the allowed effective window, the recalculated control action shadow area is completely within the stable occupancy interval, the remaining equipment capacity during the expected occupancy period is not less than the equipment capacity occupancy, the withdrawal path is valid, and the topology version number is consistent; it is not required that the target production section has entered the target manufacturing execution unit at the start of the action.

[0124] For conditional control commands of the concentration manufacturing execution unit or spray drying manufacturing execution unit, the field controller adopts a PID closed-loop control method. The target operating state determined by the conditional control command is used as the setpoint, and the temperature, vacuum, pressure, or flow rate collected by online monitoring instruments is used as the feedback value. The control output of the corresponding actuator is generated based on the deviation between the setpoint and the feedback value. The PID control parameters are pre-stored in the field controller and determined based on the production line debugging results. The field controller synchronously records the setpoint, feedback value, and actual feedback from the corresponding actuator to form a command execution record.

[0125] The action confirmation time is obtained by adding the field controller's scan cycle, communication cycle, and the actuator's allowable response time. When the actual feedback value of the actuator enters the allowable error range of the command target value within the action confirmation time, the confirmation condition control command begins execution; otherwise, it is marked as execution failure and the execution lock is released. For example, if the target valve opening is 44% and the allowable error is 0.5%, execution can be confirmed if the feedback reaches 43.8%, but not if the feedback is only 41%. S510 records the actual start and end times of the action, the actuator feedback value, equipment capacity utilization, withdrawal status, scan sequence number, and execution result, forming a command execution record and sending it to S520.

[0126] S520 Verification of Control Action and Actual Hit Status of Target Production Section: The central coordinating controller reads the instruction execution record from S510, as well as the pipeline pressure from the pressure sensor, the drive load of the frequency converter or motor controller, the cumulative delivery volume from the cumulative metering device, the equipment liquid level from the level sensor, and the processing cycle of the field controller. It also calls upon the dual leading-edge reliable status from S230 and the production section identity data from S310. All data are aligned according to the field acquisition timestamp, not the data reception time.

[0127] The actual control range refers to the time period from the moment the target manufacturing execution unit first experiences a pressure change or drive load change that satisfies condition S220, until the pipeline pressure, drive load, and cumulative delivery increment within the unit sampling window all return to the pre-execution baseline fluctuation range and remain there continuously for 3 sampling cycles. The actual occupancy range of the target production section consists of the moment when the leading edge of the production section actually enters the target manufacturing execution unit to the moment when the trailing edge of the production section actually leaves.

[0128] The overlap time between the two is:

[0129]

[0130] In the formula, The overlap time between the actual control range of conditional control instruction c and the actual occupied range of the target production section z; and These are the start and end times of the actual control action, respectively. and These are the actual entry and exit times of the target production section, respectively.

[0131] A complete control hit is determined when the entire actual control action range is located within the actual occupied range of the target production section and the propagation path is consistent with the path specified in the instruction; a partial control hit is determined when there is overlap but the control is not fully contained; an early control action is determined when the start time of the actual control action is earlier than the actual entry time of the production section; a cross-section leakage is determined when the end time is later than the actual departure time of the production section; a control miss is determined when the overlap time is 0; and an abnormal propagation path is determined when the response appears in a manufacturing execution unit that is unreachable in the current transport topology or in a non-predetermined branch.

[0132] When the control action front has been reached but the production section front has not yet been reached, an in-transit control lock is established for the combination of "target production section - target manufacturing execution unit - control direction". During the existence of the in-transit control lock, S420 and S430 must not generate duplicate control in the same direction. The in-transit control lock is released when any one of the following conditions is met: the actual control action disappears, control hit verification is completed, the withdrawal path execution is completed, or the propagation path abnormality is completed and isolated.

[0133] For example, if the actual occupied interval of the target production section is 10:00:20 to 10:00:42, and the actual control action interval is 10:00:18 to 10:00:35, then partial control hit and 2-second advance action should be recorded, and a complete control hit should not be determined. S520 outputs the actual control action interval, actual occupied interval, control hit result, actual propagation path, and abnormal status for use by S530.

[0134] S530, Execution Feedback Correction and Abnormal State Switching: The central coordinating controller only identifies records where the conditional control command is executed completely, the topology version number remains unchanged, key sensor data is valid, the actual propagation path is unique, and there are no other overlapping control responses as valid feedback samples. Records of propagation path abnormalities, execution failures, topology switching, and data failures are only used for anomaly statistics and do not participate in parameter updates.

[0135] Propagation delay, transmission delay, residual effect duration, and equipment capacity utilization are each updated independently using a bounded recursive method.

[0136]

[0137] In the formula, , and These are the values ​​before the update, the measured values ​​of the valid feedback samples, and the updated values ​​for the same type of parameter, respectively. The updated coefficients are set to values ​​greater than 0 and no greater than 0.2.

[0138] Each parameter retains its own physical unit, and no mixed calculations are performed between parameters of different dimensions. The update results are limited to the minimum and maximum safe values ​​determined during the equipment commissioning phase; the corresponding parameter interval is only narrowed when three consecutive valid feedback samples of the same type change in the same direction, and only widened when the change directions are inconsistent.

[0139] The actual propagation delay is used to update the propagation delay interval of S230. The actual entry and exit times of the production section into each manufacturing execution unit are used to update the transmission delay interval of S230 and the expected equipment occupancy interval and stable occupancy interval of S320. The actual start, end and disappearance times of the control action are used to update the control action shadow area prediction parameter set of S330. The equipment capacity occupancy is used to update the equipment capacity boundary of S410.

[0140] The actual response device sequence, response direction, and response time in the valid feedback samples are fed back to S130 to update the expected response time range in the signature response template. The pheromone in the control command feasible connection diagram is updated using an interval definition rule with a volatilization mechanism. The update formula is:

[0141]

[0142] In the formula, and These are the pheromones of feasible connecting edges before and after the update; The preset pheromone evaporation coefficient is used to prevent premature convergence and stagnation of the algorithm, and its value range is (0,1). and These are the system's preset lower and upper limits for pheromones, respectively. The pheromone increment is calculated as follows: a positive gain is applied when a complete control hit occurs, while a negative penalty is applied when a partial control hit occurs, the action is premature, or there is a cross-segment leak. When multiple instructions are associated with the same feasible connection edge, their pheromone increments are combined using an arithmetic accumulation mechanism. Simultaneously, for partial hits or leaks, the system automatically increases the time safety margin for similar candidate control actions.

[0143] Each update generates a feedback version number, recording parameters before and after the update, sample source, and associated instructions. If two consecutive instances of premature action, cross-section leakage, or refusal to execute occur, the previous feedback version is restored. In the event of a propagation path anomaly, unexecuted conditional control instructions on the abnormal path are frozen, and branch valve feedback is verified. If equipment capacity fails, a withdrawal path is executed. If an actuator malfunctions, a backup actuator is switched; if no backup actuator is available, the isolation valve is closed, and the upstream pumping device is stopped. When field controller communication is interrupted, the field remains in a safe state, and generating new instructions based on expired data is prohibited.

[0144] After the anomaly is handled, S530 feeds back the corrected production segment leading edge, production segment trailing edge, occupied segment and segment reliability status to S110, updates the current transport topology and topology version number, clears the old version of unexecuted instructions, and re-executes S120 to S530.

[0145] When the temperature, vacuum, pressure, or flow rate of the concentration manufacturing execution unit or the spray drying manufacturing execution unit exceeds the allowable operating range specified in the equipment basic configuration table, the field controller generates a parameter over-tolerance alarm and uploads the corresponding status to the central coordination controller. When the automatic control loop is abnormal, key online monitoring data fails, or communication is interrupted, it stops receiving new conditional control commands and allows the field controller to switch to manual standby mode. It will re-enter the automatic control state after the automatic control conditions are restored and the status verification is completed.

[0146] Except for parameters that are explicitly required to be updated recursively while maintaining their original physical units, all physical parameters involved in multi-objective weighting and probability transfer in the above algorithm model must be normalized or dimensionless before being substituted into the calculation. The relevant formulas are empirical models that approximate the real situation based on a large amount of data collection and software simulation fitting. The preset parameters and thresholds involved in the formulas can be conventionally set and adjusted by those skilled in the art according to the physical constraints of the actual application scenario.

[0147] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0148] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0149] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An automatic control system for manufacturing skim milk powder, characterized in that, include: Central coordinated controller, multiple field controllers, and pumping devices, valve actuators, and buffer actuators controlled by the field controllers; The central coordination controller includes: The control signature generation module is used to determine the signature allowable window and signature action boundary based on the equipment operating status sequence and the current conveying topology, control the target actuator to perform identification and compensation actions with opposite execution directions and limited cumulative conveying volume deviation, and generate a conservation control signature. The dual-front identification module is used to identify the actual propagation path based on the actual action sequence and response of the conservation control signature, and to determine the control action front and the production segment front. The segment action relationship construction module is used to generate target production segments based on the cumulative transport volume of the front boundary and the cumulative transport volume of the rear boundary, determine the stable occupied interval, and establish a set of prediction parameters for the control action shadow area. The collaborative instruction generation module is used to generate candidate control instructions based on the equipment capability boundary and withdrawal path, calculate the control action shadow area and the allowed effective window, and use the ant colony algorithm to generate a set of conditional control instructions and a control execution sequence in the control instruction feasible connection graph.

2. The automatic control system for manufacturing skim milk powder according to claim 1, characterized in that, It also includes: a distributed execution and feedback correction module, which is used to execute conditional control commands, determine the control hit status based on the actual control action range and the actual occupied range of the target production section, and update the propagation time delay range, transmission time delay range, control action shadow area prediction parameter set, equipment capacity boundary and control command feasible connection diagram pheromones based on effective feedback samples.

3. The automatic control system for manufacturing skim milk powder according to claim 1, characterized in that, The signature generation control module includes: State construction unit, window determination unit, and signature generation unit; The status construction unit is used to collect the operating frequency and drive load of the pumping device, valve command opening degree and feedback position, pipeline pressure, equipment liquid level, cumulative delivery volume and equipment processing cycle time, generate the equipment operating status sequence, and generate the current conveying topology and topology version number according to the operating status and connection relationship of each device; The window determination unit is used to determine the signature permission window and signature action boundary based on the drive load margin, valve adjustment margin, buffer available volume margin, and action time margin.

4. The automatic control system for manufacturing skim milk powder according to claim 3, characterized in that, Also includes: The signature generation unit is used to control the target actuator to sequentially execute the identification action and the compensation action in the opposite direction within the signature permission window, so that the cumulative delivery deviation of the signature cycle does not exceed the allowable deviation of the cumulative delivery, and generates control signature instructions and signature response templates.

5. An automatic control system for manufacturing skim milk powder according to claim 1, characterized in that, The dual-front identification module includes: The system includes an execution confirmation unit, a path identification unit, and a leading edge identification unit. The execution confirmation unit is used to confirm the valid control signature and determine the signature propagation starting point based on the actual direction of action, amplitude of action, action time, baseline operating state recovery result, cumulative delivery deviation and topology version number of the target actuator; The path identification unit is used to extract valid response events of pipeline pressure and driving load, match the valid response events with the signature response template, and determine the actual propagation path; The leading edge identification unit is used to determine the leading edge of control action based on the farthest effective response node in the actual propagation path, and to determine the leading edge of the production section based on the cumulative amount of the leading edge of the production section, the effective occupied volume of the connecting section, and the cumulative increase in the transport volume. It outputs the reliable state of the dual leading edges, the propagation time delay interval, and the transmission time delay interval.

6. The automatic control system for manufacturing skim milk powder according to claim 1, characterized in that, The module for constructing segment interaction relationships includes: Segment generation unit, occupied interval determination unit, and parameter establishment unit; The segment generation unit is used to determine the cumulative delivery volume corresponding to the signature propagation start point and the completion of the compensation action as the front boundary cumulative delivery volume and the back boundary cumulative delivery volume, respectively, to generate the target production segment and production segment identity data bound to the effective control signature, and to split the sub-production segment according to the branch cumulative delivery increment. The occupancy interval determination unit is used to determine the expected equipment occupancy interval, stable occupancy interval, and occupancy segment based on the leading edge of the production section, the trailing edge of the production section, the cumulative conveying volume at the inlet, the cumulative conveying volume at the outlet, and the effective occupancy volume inside the equipment.

7. An automatic control system for manufacturing skim milk powder according to claim 6, characterized in that, Also includes: The parameter establishment unit is used to establish a set of predicted parameters for the control action shadow area based on the control action front, propagation time delay interval, and effective feedback samples, indexed by the target actuator, action direction, and discrete action amplitude.

8. An automatic control system for manufacturing skim milk powder according to claim 1, characterized in that, The collaborative instruction generation module includes: Capability boundary construction unit, candidate instruction generation unit, and sequence generation unit; The capacity boundary construction unit is used to convert the current actual occupancy, the occupancy of the production section in transit, the committed occupancy, and the occupancy of the uncompleted control action into the equipment capacity occupancy, and to determine the equipment capacity boundary, the equipment capacity available range, and the withdrawal path. The candidate instruction generation unit is used to generate candidate control actions according to the minimum action resolution of the target actuator, calculate the control action shadow area based on the control action shadow area prediction parameter set, and determine the allowed effective window by combining the stable occupancy interval, the equipment capacity available interval and the withdrawal path, thus forming candidate control instructions. The sequence generation unit is used to establish a feasible connection graph of control commands and to generate a set of conditional control commands and a control execution sequence from control nodes that meet the constraints of equipment capability, scope of action, withdrawal path and topology version using the ant colony algorithm.

9. An automatic control system for manufacturing skim milk powder according to claim 2, characterized in that, The distributed execution and feedback correction module includes: Instruction execution unit, hit verification unit, and feedback correction unit; The instruction execution unit is used to verify the allowed effective window, control action area, equipment capability boundary, withdrawal path, topology version number and the status of the preceding conditional control instruction, lock the target execution mechanism and form an instruction execution record based on the actual feedback; The hit verification unit is used to determine the actual control action range and the actual occupied range of the target production section based on pipeline pressure, drive load, cumulative delivery volume and double leading edge confidence status, and accordingly determine complete control hit, partial control hit, premature action, cross-section leakage, control miss or abnormal propagation path.

10. An automatic control system for manufacturing skim milk powder according to claim 9, characterized in that, Also includes: The feedback correction unit is used to update the propagation time delay interval, transmission time delay interval, control action shadow area prediction parameter set, equipment capability boundary and pheromone based on valid feedback samples, and to execute the withdrawal path and update the topology version number in case of anomalies.

Images