Electroplating technological process self-adaptive regulation and control method based on user product requirements

Abstract

The invention discloses an electroplating process flow self-adaptive regulation and control method based on user product requirements, particularly relates to the field of industrial control, and is used for solving the problems that an existing electroplating process flow is difficult to automatically generate according to the user product requirements and process stability and requirement matching are difficult to consider. The method comprises the following steps: acquiring target product demand data, constructing a user demand model on a process time axis, calculating a basic process unit response track in a process knowledge base, forming a candidate process unit sequence, and calculating a process stability index and a demand matching index for a candidate electroplating process flow based on process deduction; and a process demand cooperation coefficient is constructed in the two-dimensional plane, the process demand cooperation coefficient is compared with a preset threshold value obtained through historical statistics, the candidate process unit sequence and the key control parameters are adjusted in a self-adaptive mode, and finally the target electroplating process flow is determined and a control instruction is generated.

Description

Technical Field

[0001] This invention relates to the field of industrial control, and more specifically, to an adaptive control method for electroplating processes based on user product requirements. Background Technology

[0002] On electroplating and surface treatment production lines, industrial control computers, control programs, and process databases are commonly installed. During production, operators identify product information via barcodes or QR codes, retrieve the corresponding process flow and parameters such as time, current, and voltage for each step from the database, and then issue these parameters to the field equipment for execution. Some existing solutions also automatically record processing parameters for subsequent quality traceability and querying, thereby reducing the workload of manual copying and data entry. In this application model, process engineers often pre-establish a complete process formula for each product model. During operation, control and recording can be completed simply by matching the product code. Daily operations on the production site are relatively stable, and usage habits have already been formed.

[0003] Current methods generally suffer from the following technical problem: the relationship between product codes and process formulas is static. They rely solely on product codes to invoke fixed process flows and parameters, making it difficult to directly respond to detailed user requirements regarding different quality levels, operating environments, and subsequent assembly methods for the same product. For parts with the same structure, once a user raises new performance requirements, the production site can only create multiple approximate formulas by manually copying and modifying existing process records, resulting in a large number of process versions. Operators are prone to confusion when selecting formulas, and changes to new requirements during process execution are difficult to reflect in a timely and accurate manner at the control level. Existing industrial control software lacks a mechanism to automatically generate or adjust electroplating process flows and key parameters based on user product requirements, failing to achieve adaptive process control based on user product needs. This represents a significant gap from the trend of electroplating production moving towards on-demand customization and precise control.

[0004] To address the aforementioned problems, a technical solution is provided. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide an adaptive control method for electroplating process flow based on user product requirements. This method involves collecting target product requirement data and constructing a user requirement model on the process timeline. It calculates the response trajectories of basic process units in a process knowledge base and forms a sequence of candidate process units. Based on process deduction, it calculates the process stability index and demand matching index for the candidate electroplating process flow. A process demand coordination coefficient is constructed in a two-dimensional plane, and this coefficient is compared with a preset threshold obtained from historical statistics. The method adaptively adjusts the candidate process unit sequence and key control parameters, ultimately determining the target electroplating process flow and generating control commands to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: S1: Collect target product demand data, discretize it into a demand trajectory sequence on the process time axis, and establish a corresponding user demand model; S2: Based on the user demand model, calculate the response trajectory of each basic process unit in the process knowledge base, sort them by the degree of overlap between the demand trajectory and each response trajectory, and select the candidate process unit sequence. S3: Generate an initial electroplating process flow based on the candidate process unit sequence, perform process deduction to obtain candidate flow, calculate the process stability index and demand matching index respectively, and obtain the process demand synergy coefficient by the geometric projection result of the two relatively balanced straight lines on the plane. S4: Compare the process requirement coordination coefficients of each candidate process with the preset thresholds. If none of them meet the standards, adjust the candidate process unit sequence or key control parameters and repeat steps S3 and S4. If the standards are met, determine the corresponding process as the target process and generate control instructions.

[0007] Furthermore, in step S1, target product demand data is collected, the average number of processes for similar products is obtained through the historical process formula set, the total planned process time is normalized to obtain the process time axis, and the time discrete step size is set using the reciprocal of the average number of processes. A discrete time node set is constructed from the process start point to the process end point with a fixed step size.

[0008] Furthermore, in step S1, a set of demand categories is established. After determining the normalized time interval for each demand record, the demand coverage count is statistically analyzed on the combination of the discrete time node set and the demand category set. The demand intensity coefficient is calculated based on the total count of each demand category across all time nodes. The demand intensity threshold is determined through the demand intensity distribution of historical qualified batches. Based on the demand intensity threshold, a demand trajectory sequence is formed and together with the discrete time node set and the demand category set, constitutes the user demand model.

[0009] Furthermore, in step S2, the basic process units in the process knowledge base are subjected to time normalization processing to obtain the normalized time interval and time center position. Combined with the demand category set, the basic response intensity coefficient, and the time position correction factor, the response intensity coefficient is calculated on the discrete time node and demand category combination to form the response trajectory of the basic process unit on the normalized time axis and demand category set.

[0010] Furthermore, in step S2, the demand intensity coefficient and the response intensity coefficient of the basic process unit are used in the user demand model to calculate the total overlap contribution and the total coverage of the baseline at all time nodes and all demand category combinations, so as to obtain the overlap index of the basic process unit. The basic process units are screened by the overlap threshold and sorted according to the time center position to generate a candidate process unit sequence.

[0011] Furthermore, in step S3, the process sequence and nominal control parameters of the basic process unit are read according to the candidate process unit sequence. Within the parameter fluctuation range of historical qualified production batches, a finite perturbation combination is constructed for each control parameter to generate multiple candidate electroplating process flows. The prediction results of each candidate electroplating process flow on the set of quality indicators are obtained through the process deduction module.

[0012] Furthermore, in step S3, the total mass change is calculated for each candidate electroplating process under positive and negative disturbances of a single control parameter. The total mass change is summed with the corresponding disturbance amplitude to form a disturbance sensitivity metric. The total disturbance sensitivity metric is obtained by summing the disturbance sensitivity metric of all control parameters. Then, the process stability index is obtained by normalizing within the total disturbance sensitivity range of all candidate electroplating processes.

[0013] Furthermore, in step S3, the deviation between the predicted quality index and the target quality value is calculated for each candidate electroplating process. The upper limit of the allowable deviation obtained from the statistics of historical qualified batches is used to form the relative deviation, and the maximum relative deviation is taken among all quality indicators. Normalization is performed within the maximum relative deviation range of all candidate electroplating processes to obtain the demand matching index. Then, the projection length and vertical distance relative to the equilibrium line are calculated using the process stability index and the demand matching index as plane coordinates. The process demand coordination coefficient is obtained based on the ratio of the projection length to the sum of the two.

[0014] Furthermore, in step S4, the process requirement coordination coefficients of the candidate electroplating processes are sorted by value, and a set of compliant processes indexes is formed using a preset coordination threshold set based on the quantile level of the coordination coefficients of historical qualified production batches. When the set of compliant processes indexes is not empty, the candidate electroplating process with the largest process requirement coordination coefficient is selected as the target electroplating process and a corresponding control instruction sequence is generated.

[0015] Furthermore, in step S4, the arithmetic mean of the process stability index and the demand matching index of all candidate electroplating processes is calculated and the difference between the two is constructed. When the average process stability index is greater than the average demand matching index, the candidate process unit sequence is adjusted by the demand matching gap. When the average process stability index is less than the average demand matching index, the key control parameters are adjusted by the disturbance sensitivity and the statistical results of historical high-cooperation batch parameters, and then steps S3 and S4 are executed again.

[0016] The technical effects and advantages of this invention's adaptive control method for electroplating processes based on user product requirements are as follows: This invention constructs user demand models, basic process unit response trajectories, candidate process unit sequences, and candidate electroplating process flows sequentially within a unified process timeline and demand category space. This transforms user product requirements from mere textual descriptions or product models into measurable demand trajectories, which are then compared and combined one by one with the basic process units in the process knowledge base. This solidifies the traditional formula design process, which relies on manual experience to select which process segments and in what order, into a calculable process. It reduces reliance on fixed process formulas and manual trial-and-error parameter adjustments, enabling the electroplating process flow to be automatically generated and modified around specific product requirements.

[0017] This invention calculates both the process stability index and the demand matching index for candidate electroplating processes simultaneously, constructs a process demand synergy coefficient in a planar space, and then uses a synergy threshold obtained from historical qualified batch statistics for screening. This transforms process stability and demand satisfaction from two mutually restrictive objectives into a unified and measurable evaluation scale. The judgment of the quality of a process no longer relies solely on local quality indicators or a single safety margin, but rather selects a process with a more balanced synergy level under the premise of comprehensively considering parameter disturbance sensitivity and quality deviation. This reduces process schemes on the production floor that are stable in formula but do not meet the requirements or meet the requirements but lack stability.

[0018] This invention automatically identifies the main sources of bottlenecks when the synergy coefficient fails to meet the standard, based on the average difference between the process stability index and the demand matching index. It then replaces basic process units with large demand matching gaps at the candidate process unit sequence level, or adjusts the disturbance range and nominal value at the key control parameter level. After each adjustment, it re-performs process simulation and synergy evaluation, thus forming a closed loop for electroplating process optimization from demand modeling, process generation, performance evaluation to adaptive adjustment of structure and parameters. Finally, it outputs the corresponding control command sequence, reducing the workload of repeated trial plating and manual comparison of solutions, and improving the speed of electroplating process deployment and overall production consistency under different product requirements. Attached Figure Description

[0019] Figure 1This is a schematic diagram of the adaptive control method for electroplating process based on user product requirements 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 1: Figure 1 This invention presents an adaptive control method for electroplating process flow based on user product requirements, comprising: S1: Collect target product demand data, discretize it into a demand trajectory sequence on the process time axis, and establish a corresponding user demand model.

[0022] S2: Based on the user demand model, calculate the response trajectory of each basic process unit in the process knowledge base, sort them by the degree of overlap between the demand trajectory and each response trajectory, and select the candidate process unit sequence.

[0023] S3: Generate an initial electroplating process flow based on the candidate process unit sequence, perform process deduction to obtain candidate flow, calculate the process stability index and demand matching index respectively, and obtain the process demand synergy coefficient by the geometric projection result of the two relatively balanced straight lines on the plane.

[0024] S4: Compare the process requirement coordination coefficients of each candidate process with the preset thresholds. If none of them meet the standards, adjust the candidate process unit sequence or key control parameters and repeat steps S3 and S4. If the standards are met, determine the corresponding process as the target process and generate control instructions.

[0025] In the adaptive control method of electroplating process based on user product requirements, it is necessary to characterize the distribution of user requirements in each process stage along the process time dimension. Otherwise, in subsequent steps, it is impossible to construct response trajectories based on the time correspondence between user requirements and basic process units, nor is it possible to select a suitable process unit sequence based on trajectory overlap. The purpose of step S1 is to organize the original product requirement records into a user requirement model with a time axis structure, so that the occurrence position and intensity of user requirements on the process time axis are clearly expressed, providing a unified reference benchmark for step S2 to calculate the response trajectory of basic process units and perform overlap sorting.

[0026] 1-1 Recording original requirements and obtaining the total planned process time.

[0027] Collect the set of original requirement records corresponding to the target product. Each requirement record includes three items: requirement category identifier, requirement level identifier, and process stage identifier. The requirement category identifier is used to distinguish different requirement dimensions, such as surface quality requirements, environmental tolerance requirements, and appearance requirements; the requirement level identifier is used to indicate the user's target requirements in that requirement dimension; and the process stage identifier is used to indicate which time interval in the electroplating process the requirement applies to.

[0028] Based on the product type of the target product, process formulas for similar products are selected from the historical process formula set. For each historical process formula, the time lengths of all processes in that formula are summed to obtain the total time for a complete process flow corresponding to each formula. The average total process flow time is then calculated to obtain an average total process time. This average total process time is used as the planned total process time for the target product in subsequent calculations.

[0029] 1-2 Construction of normalized process time axis and discrete time nodes.

[0030] Based on the planned total process time, the actual physical time is mapped to a normalized process time. The normalized process time ranges from zero to one, where zero corresponds to the start of the process flow and one corresponds to the end of the process flow. The ratio of the physical time at a given moment to the planned total process time is taken as the normalized time value at that moment.

[0031] To facilitate subsequent calculations on the timeline, a set of discrete time nodes on the normalized process timeline is constructed. First, the number of processes for each formula in the historical process recipes of similar products is averaged to obtain the average number of processes. The reciprocal of the average number of processes is taken as the time discretization threshold. The value of the time discretization threshold is greater than zero and does not exceed one. Using the time discretization threshold as a fixed time step, starting from the zero point of the normalized time, the steps are incremented sequentially to generate multiple normalized time nodes. When the last generated time node exceeds one, the last time node is adjusted to one, thus obtaining a sequence of discrete time nodes from zero to one.

[0032] 1-3 Determine the set of demand categories and the demand time interval.

[0033] Based on the company's process specifications, the required dimensions that need to be considered during the electroplating process are categorized, forming a set of requirement categories. Each requirement category in the set of requirement categories maintains a fixed definition throughout the subsequent trajectory calculations.

[0034] For each record in the original demand log, the process stage identifier is mapped to an actual physical time interval using the process structure of similar products in historical process formulas. This time interval is determined by the start and end physical times of the process stage. The ratio of the start physical time to the planned total process time is then used as the normalized start time position for that demand record, and the ratio of the end physical time to the planned total process time is used as the normalized end time position, thus obtaining the effective time interval of each demand record on the normalized time axis.

[0035] 1-4 Construction of the requirement coverage count table.

[0036] To describe the frequency of requirement occurrences at different time points and for different requirement categories, a requirement coverage count table is constructed. Each row of the requirement coverage count table corresponds to a discrete time point, and each column corresponds to a requirement category.

[0037] For any combination of discrete time point and demand category, count the number of demand records in the original demand record set that satisfy the following two conditions: First, the demand category identifier of the demand record is the same as the demand category. Second, the normalized time interval corresponding to the demand record covers the discrete time point, i.e., the discrete time point lies between the normalization start time and normalization end time of the record. The number of demand records satisfying these two conditions is recorded as the demand coverage count for that combination of time point and demand category. By performing the above statistics for all time points and all demand categories, a complete demand coverage count table is obtained.

[0038] Total demand count for categories 1-5 and handling of zero-demand categories.

[0039] To reflect the total demand for each demand category across all time points, the sum of each column in the demand coverage count table is calculated. The sum of each column reflects the total demand count for the corresponding demand category across all time points.

[0040] When the total demand count for a certain demand category is zero across all time points, it indicates that there are no demand records for that category in the original demand log. In subsequent demand intensity calculations, the demand intensity for this type of demand category is set to zero at all time points to ensure consistent processing rules for different demand categories in subsequent calculations.

[0041] 1-6 Calculation of demand intensity coefficient.

[0042] A demand intensity coefficient is introduced to represent the relative demand intensity at a given discrete time point and a combination of demand categories. For a demand category with a non-zero total demand count, at each time point within that category, the demand coverage count at that time point is divided by the total demand count for that demand category to obtain the demand intensity coefficient for that time point and the combination of that demand category. The demand intensity coefficient is a dimensionless value, ranging from zero to one.

[0043] For demand categories with a total demand count of zero, the demand intensity coefficient at all time points is directly set to zero. By this definition, for any demand category, the sum of its demand intensity coefficients at all time points equals one, thus forming an intensity distribution along the time axis within each category.

[0044] 1-7 Obtaining the Demand Intensity Threshold and Filtering the Demand Trajectory Sequence.

[0045] To eliminate demand points with minimal impact on process unit selection, a demand intensity threshold is introduced. The demand intensity threshold is greater than zero and less than one. It is determined using demand intensity samples from historical qualified batches.

[0046] The specific acquisition process involves selecting qualified batches from the historical process formula set whose output quality meets user requirements. All demand intensity coefficients from these qualified batches are then aggregated to form a demand intensity sample set. The values ​​in the demand intensity sample set are sorted in ascending order. A scaling factor between zero and one is set, for example, half or three-quarters. Based on this scaling factor, the sorted demand intensity samples are divided into low-intensity and high-intensity portions, and the smallest demand intensity value in the high-intensity portion is selected as the demand intensity threshold. In this way, the demand intensity threshold corresponds to the demand intensity level in the qualified batches that can actually influence process selection.

[0047] After obtaining the demand intensity threshold, all combinations of time points and demand categories are filtered. When the demand intensity coefficient of a combination is not less than the demand intensity threshold, the time point location, demand category identifier, and corresponding demand intensity coefficient are combined to form a demand trajectory point. All combinations that meet the conditions are collected to form a demand trajectory sequence. Each trajectory point in the demand trajectory sequence clearly records the normalized time location, demand category, and demand intensity, providing accurate time and demand information for subsequent steps to calculate the overlap with the process unit response trajectory.

[0048] 1-8 User Requirements Model Construction.

[0049] The user demand model is constructed by combining a set of discrete time nodes, a set of demand categories, a demand intensity coefficient table for all combinations of time nodes and demand categories, and a demand trajectory sequence. The set of discrete time nodes provides the structure of the normalized time axis, the set of demand categories provides the structure of the demand dimensions, the demand intensity coefficient table describes the relative demand intensity of each combination of time node and demand category, and the demand trajectory sequence provides the key demand points after being filtered by the demand intensity threshold. The user demand model is used as input in the subsequent step S2 to calculate the response trajectory of each basic process unit on the time axis and to generate a candidate process unit sequence based on the trajectory overlap.

[0050] Step S1 constructs a user requirement model that includes a set of discrete time nodes, a set of requirement categories, and a sequence of requirement trajectories. This transforms the original product requirements from scattered fields into a structured expression on the process timeline, clearly defining the role and intensity of each type of user requirement at different process stages. This provides a unified time reference and requirement intensity benchmark for subsequent basic process unit response trajectory calculations, avoiding reliance solely on static binding of process formulas to product codes. It enables the direct transmission of user product requirements to the electroplating process flow generation stage, reducing version accumulation and selection errors caused by manual experience-based process selection.

[0051] Step S1 has yielded the user demand model, which includes a set of discrete time nodes on the normalized process time axis, a set of demand categories, and a demand intensity matrix for combinations of time nodes and demand categories. The user demand distribution has already been given in both the time and demand dimensions. However, the user demand model itself does not contain any specific process information. To enable subsequent steps to automatically combine electroplating processes based on user demands, step S2 requires mapping the basic process units in the process knowledge base to a time axis and demand category dimension that are completely consistent with the user demand model. By calculating the response trajectories of the basic process units and performing overlap analysis with the user demand trajectories, candidate process unit sequences that both cover the user demand distribution and satisfy the process time order are selected for direct use in step S3 when generating the initial electroplating process flow.

[0052] 2-1 Time normalization of basic process units.

[0053] The process knowledge base contains multiple basic process units, each with a clearly defined start and end time on the physical timeline. First, the planned total process time for the target product is retrieved. This planned total process time is derived from the average sum of the times for each step in historical process formulas for similar products. For each basic process unit, its physical start time is divided by the planned total process time to obtain the normalized start time, and its physical end time is divided by the planned total process time to obtain the normalized end time. Both the normalized start and end times fall within the range of zero to one. The normalized time interval length is obtained by subtracting the normalized start time from the normalized end time. The arithmetic mean of the normalized start and end times is used as the time center position of this basic process unit, providing a basis for subsequent sorting.

[0054] 2-2 Determination of basic response intensity and construction of time location correction factor.

[0055] The set of demand categories given in the user demand model is used to constrain the demand dimensions that the basic process unit needs to respond to. For each combination of basic process unit and each demand category, a basic response intensity coefficient is defined. The basic response intensity coefficient takes a value between zero and one, is a dimensionless value, and is determined by quality inspection data and engineering experience evaluation from the historical process formulas in which the basic process unit participated in the execution. It represents the processing capacity of the basic process unit in the corresponding demand category.

[0056] To characterize the impact distribution of basic process units at different time positions within their coverage time interval, a time position correction factor needs to be constructed. For each discrete time node in the user requirement model, the distance between that time node and the time center position of each basic process unit is calculated, with the distance taken as the absolute difference on the normalized time axis. The time position correction factor is determined based on the relationship between distance and the normalized time interval length. When the distance does not exceed half the normalized time interval length, the time position correction factor is equal to one minus twice the ratio of distance to time interval length; when the distance exceeds half the normalized time interval length, the time position correction factor is zero. Thus, the time position correction factor approaches one near the time center position of the basic process unit, gradually decreases towards the ends of the time interval, and is zero outside the interval, reflecting the scope and centrality of the basic process unit's influence on the time axis.

[0057] 2-3 Formation of response intensity distribution and definition of response trajectory.

[0058] After obtaining the basic response intensity coefficient and the time position correction factor, the response intensity coefficient is calculated for each combination of basic process unit, each discrete time node, and each demand category. The response intensity coefficient is defined as the smaller of the basic response intensity coefficient and the time position correction factor. When the time node is near the time center of the basic process unit, the time position correction factor is close to one, and the response intensity coefficient is close to the basic response intensity coefficient. When the time node is near the edge of the time interval of the basic process unit, the time position correction factor decreases, and the response intensity coefficient decreases accordingly. When the time node is outside the normalized time interval, the time position correction factor is zero, and the response intensity coefficient is also zero. In this way, a complete set of response intensity distributions is obtained on a two-dimensional plane of discrete time nodes and demand categories. This set of distributions is regarded as the response trajectory of the basic process unit. Each point in the response trajectory contains three pieces of information: time node position, demand category identifier, and response intensity coefficient.

[0059] 2-4 Calculation logic of overlapping contribution and coverage benchmark.

[0060] The user demand model provides a demand intensity coefficient for each time point and each demand category combination, while the basic process unit provides a response intensity coefficient for the same combination based on the response trajectory. To measure the degree of overlap between the two in both time and demand category dimensions, an overlap contribution value and a coverage baseline value are constructed for each time point and demand category combination. The overlap contribution value is defined as the smaller of the demand intensity coefficient and the response intensity coefficient for that combination, representing the intersection of user demand and basic process unit response. The coverage baseline value is defined as the larger of the demand intensity coefficient and the response intensity coefficient for that combination, representing the greater coverage of demand and response.

[0061] For a basic process unit, iterate through all time points and all demand categories, summing the overlapping contribution values ​​across all combinations to obtain the total overlapping contribution of the basic process unit; simultaneously, sum the coverage baseline values ​​across all combinations to obtain the total coverage baseline of the basic process unit. When the total coverage baseline is greater than zero, divide the total overlapping contribution by the total coverage baseline to obtain the overlap index of the basic process unit; when the total coverage baseline is equal to zero, the overlap index is defined as zero. The overlap index ranges between zero and one. An overlap index close to one indicates that the response trajectory of the basic process unit is highly consistent with the distribution of the user demand trajectory in both time and demand category dimensions. An overlap index close to zero indicates that the user demand and the basic process unit response lack overlap in most time points and demand category combinations.

[0062] 2-5 Determination of overlap threshold and screening of candidate basic process units.

[0063] To select candidate process units from all basic process units that have proven to meet user needs in historical practice, an overlap threshold needs to be introduced. The overlap threshold ranges from zero to one and is determined using historical qualified batch data. Specifically, qualified production batches that meet electroplating quality inspection standards and whose user evaluations show no serious defects are selected from the historical process formula set. For each qualified batch, the basic process unit actually involved in execution is used to calculate the overlap index based on the corresponding batch's user demand model using the aforementioned method, resulting in a set of overlap index samples.

[0064] All overlap index samples are sorted in ascending order of value. Assuming the total number of samples is a certain natural number, a quantile position is determined using a pre-defined scaling factor (between zero and one). A position number is obtained by multiplying the sample size by the scaling factor. The overlap index samples near this position number are used as the overlap threshold. This selected overlap threshold corresponds to the basic process units with medium-to-high responsiveness in historical qualified batches. Then, for each basic process unit in the process knowledge base, its overlap index is checked to see if it is greater than or equal to the overlap threshold. If the overlap index is not less than the overlap threshold, the basic process unit is added to the candidate basic process unit set. Each basic process unit in the candidate basic process unit set has been proven through its overlap index to have sufficient overlap with the user's demand trajectory.

[0065] 2-6 Construction and output of candidate process unit sequences.

[0066] After obtaining the set of candidate basic process units, they need to be sorted according to their time center positions to form the subsequent candidate process unit sequence. For each basic process unit in the candidate basic process unit set, its time center position on the normalized time axis is read, and all candidate basic process units are arranged in ascending order of their time center positions to obtain an ordered sequence. This ordered sequence is the candidate process unit sequence. Each element in the candidate process unit sequence contains a basic process unit identifier, a normalized time interval, and response trajectory information, providing the process units arranged in chronological order as input for step S3 when generating the initial electroplating process flow.

[0067] In step S2, response trajectories are constructed for the basic process units in the process knowledge base on the normalized process time axis and the set of demand categories. The overlap between these trajectories and the demand trajectories in the user demand model is calculated. This eliminates the reliance on the static binding relationship between product codes and fixed process formulas. Instead, basic process units are selected based on the actual distribution of user demands in the time and demand dimensions. Candidate process unit sequences are then generated by sorting them according to their time center positions. This allows subsequent electroplating process flow construction steps to automatically select process units around the user demand trajectory, reducing the problems of formula version accumulation and selection errors caused by manually combining process flows based on experience.

[0068] In step S2, the normalized start time and normalized end time are calculated for each basic process unit in the process knowledge base based on the total planned process time. The normalized time interval length and time center position are obtained and aligned with the set of discrete time nodes on the normalized process time axis, providing a basis for sorting the time center position for the time order arrangement of the candidate process unit sequence.

[0069] In step S2, the basic response intensity coefficient is determined based on the combination of basic process units and demand category sets. A time position correction factor is introduced to generate a response intensity coefficient distribution. The overlap contribution value and coverage benchmark value are calculated point by point with the demand intensity coefficient distribution in the user demand model to obtain the overlap index of each basic process unit. Based on the overlap threshold, the basic process units are screened and sorted according to the time center position to generate a candidate process unit sequence.

[0070] Step S2 calculates the response trajectory of the basic process unit on the same time axis and demand category set, and selects candidate process unit sequences through overlap index. At this point, a set of process units that match the target product in terms of time sequence and demand matching has been obtained, but it is still at the level of "which process units can be used and how to arrange them", without forming a specific executable electroplating process flow, and without evaluating the process quality from the perspective of parameter perturbation and quality index deviation. Step S3 needs to generate multiple candidate electroplating process flows based on the candidate process unit sequences, obtain the quality index prediction results through process deduction, and construct process stability index and demand matching index respectively on this basis, and then synthesize them into process demand synergy coefficient through geometric method, so that each candidate electroplating process flow corresponds to an evaluation value that simultaneously reflects stability and demand satisfaction, for use in step S4 for threshold judgment and target process selection.

[0071] 3-1 Construct a set of candidate electroplating process flows from the sequence of candidate process units.

[0072] The candidate process unit sequence is obtained from step S2. The order of each basic process unit in the sequence on the normalized process time axis is determined, and it is associated with a set of nominal control parameter values ​​in the process knowledge base. The control parameters include at least time parameters, current parameters, voltage parameters, temperature parameters, and stirring intensity parameters. First, the nominal control parameters corresponding to each basic process unit are read sequentially from front to back according to the candidate process unit sequence. These nominal control parameters are then combined in sequence to obtain a complete initial electroplating process flow, which includes process sequence information and the set of nominal control parameters for each process.

[0073] To reflect the adjustment range of control parameters in actual production, multiple candidate electroplating processes are generated based on the initial electroplating process flow. Specifically, for each control parameter in the initial electroplating process flow, based on historical qualified production batch records, the maximum upward and downward adjustment range of the control parameter near its nominal value are statistically analyzed while ensuring product quality meets requirements. These two ranges are then limited to the allowable range of the process. Subsequently, a finite number of disturbance ratio values ​​are selected between the upward and downward adjustment ranges, for example, several ratios are selected in the upward direction and several ratios in the downward direction, forming a discrete disturbance set corresponding to each control parameter. By combining the disturbance sets of each control parameter, multiple sets of control parameter combinations are obtained. Each set of control parameter combinations is paired with the initial process sequence to form a candidate electroplating process flow. This forms a set of candidate electroplating process flows, each flow containing a uniform process sequence and a set of defined control parameter values.

[0074] 3-2 Process simulation and quality index prediction results acquisition.

[0075] In step S3, it is necessary to evaluate the degree to which each candidate electroplating process meets the user's requirements in the quality index space. First, the set of quality indicators is determined. Each quality indicator in the set corresponds to a characteristic of the electroplated product, such as coating thickness uniformity, in-hole coverage, coating adhesion, corrosion resistance, etc.

[0076] In the user requirement model, requirement categories are mapped to one or more quality indicators in a set of quality indicators through a pre-defined mapping relationship. This transforms user requirements for surface quality, durability, appearance, etc., into target values ​​for these quality indicators. Based on this mapping relationship, a target value is determined for each quality indicator, forming a target quality vector. This target quality vector serves as a reference in the subsequent calculation of the requirement matching index.

[0077] For each process in the candidate electroplating process flow set, the process deduction module is invoked for simulation calculations. The process deduction module utilizes an established electroplating mechanism model or an empirical model trained based on historical data to predict each quality indicator under given process sequence and control parameters, obtaining a set of quality prediction results for that process flow. The quality prediction result set is a set of numerical values, with each value corresponding to a quality indicator. Through this process, a quality performance vector for each candidate electroplating process flow is formed in the quality indicator space, providing basic data for subsequent calculations of the process stability index and demand matching index.

[0078] 3-3 Calculation logic and processing of process stability index.

[0079] The process stability index is used to reflect the sensitivity of quality indicators to parameter changes when control parameters of a candidate electroplating process change within a limited range, thus reflecting the process's ability to withstand process disturbances.

[0080] For any candidate electroplating process, firstly, all control parameters included in the process are extracted to form a control parameter set for that process. For each control parameter in the control parameter set, based on historical qualified production batch data, the allowable positive and negative disturbance amplitudes for that control parameter are determined, provided that all quality indicators meet the requirements. The positive disturbance amplitude represents the maximum allowable change in the direction of increase from the nominal value, and the negative disturbance amplitude represents the maximum allowable change in the direction of decrease from the nominal value. Both amplitudes are within the limits of equipment capacity and process specifications.

[0081] After determining the disturbance magnitude of a single control parameter, keeping other control parameters in the process unchanged, only adjust this control parameter to two values: nominal value plus the positive disturbance magnitude and nominal value minus the negative disturbance magnitude. For each adjusted control parameter combination, call the process deduction module to recalculate the quality index prediction results. For each quality index, calculate the absolute value of the difference between the predicted quality value under positive disturbance and the predicted quality value without disturbance, and sum the absolute values ​​of the differences for all quality indices to obtain the total quality change under positive disturbance. Use the same method to calculate the total quality change for the negative disturbance case.

[0082] After obtaining the total positive disturbance mass change, the total negative disturbance mass change, and the positive and negative disturbance amplitudes, the two total mass changes are added together, and the absolute values ​​of the two disturbance amplitudes are added together. The former is then divided by the latter to obtain the disturbance sensitivity metric value corresponding to the control parameter. The disturbance sensitivity metric value describes the total change in all quality indicators caused by a unit change in the parameter. For all control parameters in a candidate electroplating process, the disturbance sensitivity metric values ​​are calculated sequentially, and the disturbance sensitivity metric values ​​of all control parameters are added together to obtain the total disturbance sensitivity of the process.

[0083] In all candidate electroplating processes, the total disturbance sensitivity of each process is collected, and the maximum and minimum values ​​are found. For any process, the minimum total disturbance sensitivity is subtracted from the total disturbance sensitivity. The difference between the maximum and minimum total disturbance sensitivity is then used as the denominator to calculate the ratio between the two. Finally, this ratio is subtracted from one to obtain the process stability index of the process. When the total disturbance sensitivity of all processes is the same, the process stability index of each process is directly set to one. In this way, the process stability index takes a value between zero and one. The closer the total disturbance sensitivity is to the minimum value in the set, the closer the corresponding process stability index value is to one.

[0084] 3-4 The calculation logic and processing of the demand matching index.

[0085] The demand matching index is used to reflect the degree to which candidate electroplating processes approach the target quality requirements within the quality index space.

[0086] For any candidate electroplating process and any quality indicator, first calculate the absolute value of the difference between the predicted value and the target quality value for that process, and use this difference as the deviation for that quality indicator. To characterize the position of the deviation within the acceptable range, an upper limit for tolerance deviation needs to be set for each quality indicator. The upper limit for tolerance deviation is determined as follows: collect the absolute values ​​of the deviation between the predicted value and the corresponding target value for that quality indicator from historical qualified production batches to form a deviation sample set. Sort the deviation samples from smallest to largest, and select the deviation value closest to the upper quantile position in the sorting results as the upper limit for tolerance deviation, while ensuring that the upper limit for tolerance deviation does not exceed the deviation limit specified in the product technical specifications. The upper limit for tolerance deviation obtained in this way covers most qualified samples and meets the product specification requirements.

[0087] For each quality indicator, the deviation of the process on that indicator is divided by the upper limit of the allowable deviation for that indicator to obtain the relative deviation for that indicator. Then, the largest relative deviation value is selected from all quality indicators as the maximum relative deviation for that process. A maximum relative deviation of less than or equal to one indicates that the deviations of all quality indicators have not exceeded the corresponding upper limit of the allowable deviation, while a maximum relative deviation of more than one indicates that at least one quality indicator exceeds the allowable deviation range.

[0088] In all candidate electroplating processes, the maximum relative deviation for each process is collected, and the maximum and minimum values ​​are identified. For any given process, the minimum relative deviation is subtracted from the maximum relative deviation. The ratio of the minimum relative deviation to the maximum relative deviation across all processes is then calculated using the difference between the maximum and minimum relative deviations as the denominator. Finally, this ratio is subtracted from one to obtain the demand matching index for that process. When the maximum relative deviations of all processes are the same, the demand matching index for each process is uniformly set to one. Through this normalization method, the demand matching index takes a value between zero and one; the closer the maximum relative deviation is to the minimum value in the set, the closer the corresponding demand matching index is to one.

[0089] 3-5 Geometric construction and processing of process requirement coordination coefficient.

[0090] The process stability index reflects the stability of the process from the perspective of parameter disturbance, while the demand matching index reflects the degree to which the process meets the target quality from the perspective of quality deviation. In order to simultaneously represent these two dimensions in a single scalar, a geometric relationship needs to be established on a two-dimensional coordinate plane to conduct a comprehensive analysis of the two indices.

[0091] Using the process stability index of each candidate electroplating process as the x-axis and the demand matching index as the y-axis, a point is formed on the plane. This point represents the overall position of the process in terms of stability and demand fulfillment. An analytical equilibrium line is then established on the plane. Any point on this line where the x-coordinate and y-coordinate are equal indicates that the process stability index and demand matching index have reached the same level.

[0092] Calculate the vertical distance from the above coordinate point to the equilibrium line. The vertical distance reflects the degree of difference between the process stability index and the demand matching index. At the same time, calculate the projection length of the coordinate point in the direction of the equilibrium line. The projection length reflects the overall horizontal magnitude of the two indices along the direction of balanced growth.

[0093] After obtaining the vertical distance and projected length, the projected length is used as the numerator, and the sum of the projected length and the vertical distance is used as the denominator. The ratio of the two is then calculated to obtain the process requirement coordination coefficient for this process. When the sum of the projected length and the vertical distance is zero, the process requirement coordination coefficient is set to zero. The process requirement coordination coefficient ranges from zero to one. The closer the projected length is to the maximum horizontal position in the equilibrium direction, the closer the process requirement coordination coefficient is to one. With the projected length remaining constant, the closer the vertical distance is to zero, the closer the process requirement coordination coefficient is to one. Therefore, the process requirement coordination coefficient requires both the process stability index and the requirement matching index to reach certain numerical levels, while also ensuring that the difference between them is within an acceptable range.

[0094] Through the above calculation process, the corresponding process stability index, demand matching index and process demand synergy coefficient are obtained for each process in the candidate electroplating process flow set, forming a process demand synergy coefficient set, which provides a unified evaluation basis for the preset threshold comparison and target electroplating process flow selection in step S4.

[0095] Step S3 constructs multiple candidate electroplating processes based on the candidate process unit sequence, and uses process deduction to obtain quality index prediction results. Then, the process stability index and demand matching index are calculated separately, and the process demand synergy coefficient is formed geometrically. This allows each candidate electroplating process to reflect the stability under parameter disturbances and the degree of compliance with user quality objectives with a single value. It no longer relies on experience-based trial matching and static process card trial and error. This transforms the electroplating process optimization process from subjective human judgment to quantitative screening based on the synergistic evaluation of stability and demand satisfaction, which is more in line with complex demand scenarios.

[0096] Step S3 has constructed multiple candidate electroplating processes based on the candidate process unit sequence, and calculated the process stability index, demand matching index, and the process demand synergy coefficient composed of the two for each candidate electroplating process, thus obtaining a set of candidate electroplating processes with evaluation results. However, in engineering applications, it is necessary to make screening decisions among these candidate electroplating processes to ensure that the final target electroplating process not only meets the synergy level reflected in historical qualified batches, but also maintains a balance between process stability and demand satisfaction, while converting the target electroplating process into a sequence of control instructions that can be directly issued. Step S4, focusing on this objective, constitutes a convergence process from evaluation results to control instructions, from determining the synergy coefficient threshold, single-round judgment, adjustment for non-compliance, to the implementation of the target process.

[0097] 4-1 Organizing and sorting the set of process requirement coordination coefficients.

[0098] In step S3, a process requirement coordination coefficient has been calculated for each candidate electroplating process. First, all process requirement coordination coefficients are extracted from the candidate electroplating process set to form a set of process requirement coordination coefficients. The value of each process requirement coordination coefficient is neither less than zero nor greater than one. Then, all process requirement coordination coefficients are sorted in descending order of value to obtain an ordered sequence, where the first element of the sequence corresponds to the process with the highest process requirement coordination coefficient in the current candidate electroplating process set. The sorting result is used for subsequent comparison with a preset coordination threshold, and also provides a priority basis for selecting the target electroplating process when multiple compliant processes exist.

[0099] 4-2 Determination logic and value range of preset collaborative threshold.

[0100] The preset threshold for the process requirement synergy coefficient is used to determine whether the candidate electroplating process meets the overall synergy level. Since the process requirement synergy coefficient ranges from zero to one, the preset synergy threshold is strictly greater than zero and strictly less than one. The preset synergy threshold is derived from the synergy performance of historical qualified production batches.

[0101] Specifically, firstly, multiple qualified production batches confirmed through quality inspection and user feedback are selected from the production history. Each qualified batch corresponds to a previously actually running electroplating process. For each qualified process, the process stability index, demand matching index, and process demand synergy coefficient are calculated according to the method in step S3, forming a historical process demand synergy coefficient sample set. Then, this historical sample set is sorted in ascending order of process demand synergy coefficient values.

[0102] To select a representative reference level from the sorted historical samples, a scaling factor needs to be set. The scaling factor, with a value greater than zero and not exceeding one, is predetermined by process engineers based on long-term production experience. For example, when aiming to position the preset collaboration threshold near the median level of historical qualified batches, the scaling factor can be selected in the middle; when a more stringent screening result is desired, the scaling factor can be selected close to one. A sorting position number is calculated using the scaling factor and the number of historical samples. If this number is less than one, it is set to one; if the number is greater than the number of historical samples, the number of historical samples is used. Finally, in the sorted historical samples, the historical process requirement collaboration coefficient value corresponding to this position number is used as the preset collaboration threshold.

[0103] 4-3 Determination of single-round synergy coefficient and construction of target process candidate set.

[0104] After obtaining the preset collaboration threshold, the collaboration coefficient of each process in the current candidate electroplating process set is determined one by one. For any candidate electroplating process, if its collaboration coefficient is not less than the preset collaboration threshold, the process number is added to the compliant process index set; if its collaboration coefficient is less than the preset collaboration threshold, the process is not included in the compliant set. After the determination, a compliant process index set is obtained, corresponding to a group of candidate electroplating processes that have reached the historical experience threshold in terms of comprehensive collaboration level.

[0105] If the set of compliant processes contains at least one process, the process requirement coordination coefficient values ​​are compared among these processes, and the process with the highest coordination coefficient value is selected as the target electroplating process. The process sequence and key control parameters of this process will be directly used in subsequent steps to generate control instructions.

[0106] If the set of indexes for compliant processes is empty, it means that none of the current candidate electroplating processes have reached the preset compliant threshold in terms of the compliant coefficient of the process requirements. It is necessary to adjust the process structure or key control parameters, generate a new set of candidate electroplating processes, and re-execute steps S3 and S4 until a compliant process appears or the number of iterations reaches the preset upper limit.

[0107] 4-4 Adjustment decisions and candidate process unit sequence processing when the synergy coefficient does not meet the standard.

[0108] When no candidate electroplating process's process demand synergy coefficient reaches the preset synergy threshold in a single round of evaluation, it is necessary to analyze the overall level of the process stability index and demand matching index in the current candidate electroplating process set to determine the adjustment direction. First, the arithmetic mean of the process stability indices of all candidate electroplating processes is calculated to obtain an average stability index. Then, the arithmetic mean of the demand matching indices of all candidate electroplating processes is calculated to obtain an average demand matching index. Finally, the average stability index is subtracted from the average demand matching index to obtain the stability matching difference.

[0109] When the stability matching difference is greater than zero, it indicates that in the current set of candidate electroplating processes, the average level of the process stability index is biased towards a higher range, while the average level of the demand matching index is biased towards a lower range. Demand matching capability becomes the main factor limiting the improvement of the process demand synergy coefficient. At this time, it is necessary to prioritize the adjustment of the candidate process unit sequence from the process structure level.

[0110] In this scenario, for each basic process unit in the candidate process unit sequence, all process numbers containing that basic process unit in the candidate electroplating process flow set are counted, and the demand matching index of these processes is read. For each process containing that basic process unit, the demand matching index of that process is subtracted by 1 to obtain the corresponding demand matching gap. Then, the arithmetic mean of these gaps is calculated to obtain the demand matching gap amount of that basic process unit. The larger the value, the more significantly the process containing that unit deviates from the target in terms of demand satisfaction.

[0111] After obtaining the demand matching gap amounts for all basic process units, the basic process units in the candidate process unit sequence are sorted from largest to smallest gap amount, and several basic process units with the largest gap amounts are selected as priority replacement targets. In the process knowledge base, from other basic process units close to the normalized time center position of these basic process units, candidate basic process units with smaller demand matching gap amounts are selected, and corresponding positions in the candidate process unit sequence are replaced one-to-one, thus obtaining a new candidate process unit sequence. The new candidate process unit sequence will serve as the new input for step S3, used to regenerate the candidate electroplating process flow set and recalculate the process demand synergy coefficient.

[0112] 4-5 Key control parameter handling and iteration termination conditions when the synergy coefficient fails to meet the standard.

[0113] When the stable matching difference is less than zero, it indicates that the average level of the demand matching index in the current candidate electroplating process set is close to the upper limit, while the average level of the process stability index is low. Insufficient process stability becomes the main factor limiting the improvement of the process demand synergy coefficient. At this time, it is necessary to prioritize the adjustment of key control parameters.

[0114] Step S3 has calculated the disturbance sensitivity metric value for each candidate electroplating process and each control parameter. Now, all disturbance sensitivity metric values ​​are compiled to form a sensitivity sample set, and the sensitivity samples are sorted from largest to smallest. Based on a pre-set sensitivity ratio threshold, several control parameters are selected from the top-ranked samples to form a high-sensitivity parameter set. The control parameters in the high-sensitivity parameter set appear repeatedly in multiple candidate electroplating processes and have a significant impact on the quality indicators.

[0115] For each control parameter in the set of highly sensitive parameters, the operating range of that parameter under quality-compliant conditions is recalculated using historical qualified production batch data. This operating range is then used to redefine the boundaries between positive and negative disturbance amplitudes, and the disturbance ratio is redefined within this new range. This compresses the parameter disturbance range into the historically stable operating area, thereby reducing disturbance sensitivity. Regarding nominal parameter values, corresponding control parameter values ​​are extracted from historical batches where the historical process demand coordination coefficient is among the top few percentiles. These values ​​are then averaged, and the current nominal parameter values ​​are adjusted towards this average value, making the settings of key control parameters closer to historical high-coordination levels.

[0116] After each process structure or parameter adjustment, the new candidate process unit sequence and its corresponding control parameter set are sent to step S3. The candidate electroplating process flow set is reconstructed, and the process stability index, demand matching index, and process demand synergy coefficient are calculated. Then, the process returns to the decision-making stage in step S4. To avoid infinite loops, an upper limit for the number of iterations is pre-defined during implementation, and the iteration count is incremented after each completion of steps S3 and S4. When multiple rounds of adjustments fail to produce any candidate electroplating process flow with a process demand synergy coefficient reaching the preset synergy threshold, and the upper limit for iterations has been reached, the process flow with the highest process demand synergy coefficient from all candidate electroplating process flows is selected as the alternative target solution, and the decision is manually reviewed by process engineers.

[0117] 4-6 Generation of control instructions for the target electroplating process.

[0118] When at least one candidate electroplating process has a process requirement coordination coefficient that is not less than a preset coordination threshold in a certain round of evaluation, the process with the highest process requirement coordination coefficient value is selected from the set of qualified processes as the target electroplating process. The target electroplating process includes basic process units arranged in normalized time order and the set of final control parameters corresponding to each basic process unit.

[0119] Based on the target electroplating process flow, a control instruction sequence for the electroplating production line is constructed. For each basic process unit in the process, a control instruction record is generated. The control instruction record includes at least the process number, start execution time, execution duration, current setpoint, voltage setpoint, temperature setpoint, stirring intensity setpoint, and condition parameters for interlocking protection. All control instruction records are arranged in chronological order of execution time to form the control instruction sequence for the target process flow. The control instruction sequence serves as the input to the industrial control device, enabling the target electroplating process flow selected based on user product requirements to be executed on the production site.

[0120] Step S4 compares the process requirement coordination coefficient of the candidate electroplating process with the preset coordination threshold obtained from the statistics of historical qualified batches. Within the process set, the target electroplating process that meets the joint requirements of stability and demand matching is selected. When the process requirement coordination coefficient does not meet the standard, the candidate process unit sequence or key control parameters are selectively adjusted according to the overall difference between the process stability index and the demand matching index. This achieves closed-loop linkage between the evaluation results and the process structure and parameters, reduces human experience trial and error, and improves the robustness and implementation efficiency of the industrial control software when automatically generating electroplating process flow and control instruction sequence.

[0121] Specifically, the above are merely preferred embodiments of this application and are not intended to limit this application.

[0122] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0123] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. An adaptive control method for electroplating process flow based on user product requirements, characterized in that, Including the following steps: S1: Collect target product demand data, discretize it into a demand trajectory sequence on the process time axis, and establish a corresponding user demand model; S2: Based on the user demand model, calculate the response trajectory of each basic process unit in the process knowledge base, sort them by the degree of overlap between the demand trajectory and each response trajectory, and select the candidate process unit sequence. S3: Generate an initial electroplating process flow based on the candidate process unit sequence, perform process deduction to obtain candidate flow, calculate the process stability index and demand matching index respectively, and obtain the process demand synergy coefficient by the geometric projection result of the two relatively balanced straight lines on the plane. S4: Compare the process requirement coordination coefficients of each candidate process with the preset thresholds. If none of them meet the standards, adjust the candidate process unit sequence or key control parameters and repeat steps S3 and S4. If the standards are met, determine the corresponding process as the target process and generate control instructions.

2. The adaptive control method for electroplating process flow based on user product requirements according to claim 1, characterized in that: In step S1, target product demand data is collected, the average number of processes for similar products is obtained through the historical process formula set, the total planned process time is normalized to obtain the process time axis, and the time discrete step size is set by using the reciprocal of the average number of processes. A discrete time node set is constructed from the process start point to the process end point with a fixed step size.

3. The adaptive control method for electroplating process flow based on user product requirements according to claim 2, characterized in that: In step S1, a set of demand categories is established. After determining the normalized time interval for each demand record, the demand coverage count is calculated on the combination of the discrete time node set and the demand category set. The demand intensity coefficient is calculated based on the total count of each demand category at all time nodes. The demand intensity threshold is determined by the demand intensity distribution of historical qualified batches. Based on the demand intensity threshold, a demand trajectory sequence is formed and together with the discrete time node set and the demand category set, it constitutes the user demand model.

4. The adaptive control method for electroplating process flow based on user product requirements according to claim 3, characterized in that: In step S2, the basic process units in the process knowledge base are subjected to time normalization to obtain the normalized time interval and time center position. Combined with the demand category set, the basic response intensity coefficient and the time position correction factor, the response intensity coefficient is calculated on the discrete time node and demand category combination to form the response trajectory of the basic process unit on the normalized time axis and demand category set.

5. The adaptive control method for electroplating process flow based on user product requirements according to claim 4, characterized in that: In step S2, the demand intensity coefficient and the response intensity coefficient of the basic process unit in the user demand model are used to calculate the total overlapping contribution and the total coverage of the baseline at all time nodes and all demand category combinations to obtain the overlap index of the basic process unit. The basic process units are screened by the overlap threshold and sorted according to the time center position to generate a candidate process unit sequence.

6. The adaptive control method for electroplating process flow based on user product requirements according to claim 5, characterized in that: In step S3, the process sequence and nominal control parameters of the basic process unit are read according to the candidate process unit sequence. Within the parameter fluctuation range of historical qualified production batches, a finite perturbation combination is constructed for each control parameter to generate multiple candidate electroplating process flows. The prediction results of each candidate electroplating process flow on the set of quality indicators are obtained through the process deduction module.

7. The adaptive control method for electroplating process flow based on user product requirements according to claim 6, characterized in that: In step S3, the total mass change is calculated for each candidate electroplating process under positive and negative disturbances of a single control parameter. The total mass change is summed with the corresponding disturbance amplitude to form a disturbance sensitivity metric. The total disturbance sensitivity is obtained by summing the disturbance sensitivity metric values ​​of all control parameters. Then, the process stability index is obtained by normalizing within the total disturbance sensitivity range of all candidate electroplating processes.

8. The adaptive control method for electroplating process flow based on user product requirements according to claim 7, characterized in that: In step S3, the deviation between the predicted quality index and the target quality value is calculated for each candidate electroplating process. The upper limit of the allowable deviation obtained from the statistics of historical qualified batches is used to form the relative deviation. The maximum relative deviation is taken among all quality indicators. The demand matching index is obtained by normalization within the maximum relative deviation range of all candidate electroplating processes. Then, the projection length and vertical distance relative to the equilibrium line are calculated using the process stability index and the demand matching index as plane coordinates. The process demand coordination coefficient is obtained based on the ratio of the projection length to the sum of the two.

9. The adaptive control method for electroplating process flow based on user product requirements according to claim 8, characterized in that: In step S4, the process requirement coordination coefficients of the candidate electroplating processes are sorted by value, and a set of compliant processes indexes is formed by using a preset coordination threshold set based on the quantile level of the coordination coefficients of historical qualified production batches. When the set of compliant processes indexes is not empty, the candidate electroplating process with the largest process requirement coordination coefficient is selected as the target electroplating process and a corresponding control instruction sequence is generated.

10. The adaptive control method for electroplating process flow based on user product requirements according to claim 9, characterized in that: In step S4, the arithmetic mean of the process stability index and the demand matching index of all candidate electroplating processes is calculated and the difference between them is constructed. When the average process stability index is greater than the average demand matching index, the candidate process unit sequence is adjusted by the demand matching gap. When the average process stability index is less than the average demand matching index, the key control parameters are adjusted by the disturbance sensitivity and the statistical results of historical high-cooperation batch parameters, and then steps S3 and S4 are executed again.

Images