Intelligent prediction method and system for electroplating processing parameters

By obtaining the response sensitivity and impact scale of the electroplating processing parameters, determining the comprehensive quality score, and selecting the target parameter group, the problem of inaccurate prediction of electroplating processing parameters is solved, and the stability and accuracy of product quality are improved.

CN118685848BActive Publication Date: 2024-12-06DALIAN YONGGUANG IND & TRADE CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411187204.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-28
Publication Date
2024-12-06
Estimated Expiration
2044-08-28

AI Technical Summary

Technical Problem

During the electroplating process, it is difficult for the prior art to accurately predict processing parameters, resulting in uneven product quality and large errors in single parameter prediction, which reduces the accuracy of parameter prediction.

Method used

By obtaining the response sensitivity and impact scale of multiple parameters to the target product quality score, the comprehensive quality score of each parameter group is determined, and the target parameter group is then selected to improve prediction accuracy and ensure product quality.

Benefits of technology

It improves the prediction accuracy of processing parameters, reduces parameter errors, and ensures high-quality product production in subsequent parameter regulation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118685848B_ABST
    Figure CN118685848B_ABST
Patent Text Reader

Abstract

The present invention relates to the technical field of processing parameter prediction, and specifically to an intelligent prediction method and system for electroplating processing parameters, including: obtaining multiple parameters corresponding to a target product in electroplating processing, and obtaining a quality score set for each parameter to the target product; obtaining the first response sensitivity of each parameter type to each quality score in the quality score set, and obtaining the first influence scale of each parameter group to each quality score, wherein each parameter group includes at least two parameter types; determining the comprehensive quality score of each parameter group using the first response sensitivity and the first influence scale; determining the target parameter group of the target product using the comprehensive quality score, and then determining the prediction result of the electroplating processing parameters through the target parameter group. The present invention can effectively improve the prediction accuracy of the processing parameters by analyzing and processing the processing parameter group, and ensure the high quality of the product in the subsequent parameter regulation processing process based on the prediction.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of processing parameter prediction, and in particular to an electroplating processing parameter intelligent prediction method and system. Background Art

[0002] Electroplating is a common surface treatment technology used to improve the appearance, corrosion resistance and other properties of metal products. During the electroplating process, metal ions are deposited onto the surface of the conductor by electrochemical methods to form a uniform metal coating. Various parameters need to be regulated during the processing. For example, various process parameters need to be verified and selected during the electroplating process. Selecting the most suitable parameters can obtain better and higher quality finished products. Usually, the selection of parameters is based on a variety of practical results, that is, by constantly changing the parameters, the electroplated finished products corresponding to different parameters are obtained. This process wastes a lot of materials and time. Therefore, it is necessary to predict and adjust the processing parameters according to the demand and the feedback results of the finished products to improve production quality and efficiency.

[0003] However, in the process of predicting actual parameters, any assessment indicator is affected by multiple different parameters, and the impact of different parameter changes on the actual detection results of the indicators is different. Therefore, when making intelligent predictions on multiple parameters, the actual adjustment requirements of different parameters are different. At the same time, the finished product quality scores obtained from the historical processing process are highly overlapped, and the same quality score includes a variety of different processing parameter combinations. Predicting the parameters based solely on the difference of a single parameter is likely to result in large errors in the parameters that do not meet the corresponding processing conditions, reducing the accuracy of parameter prediction and resulting in uneven final product quality. Summary of the invention

[0004] In order to solve the technical problem of inaccurate prediction of processing parameters, the purpose of the present invention is to provide an intelligent prediction method and system for electroplating processing parameters, which can effectively improve the prediction accuracy of processing parameters, reduce the errors of processing parameters, and improve the production quality of products.

[0005] The technical solution adopted is as follows: Provide an intelligent prediction method for electroplating processing parameters, the method comprising:

[0006] Acquire multiple parameters corresponding to the target product in the electroplating process, and obtain a quality score set for each parameter to the target product; determine multiple parameter types using multiple parameters, and obtain a first response sensitivity of each parameter type to each quality score in the quality score set, and obtain a first influence scale of each parameter group to each quality score, wherein each parameter group includes at least two parameter types, and each parameter type includes at least one parameter; determine a comprehensive quality score for each parameter group using the first response sensitivity and the first influence scale; determine a target parameter group for the target product using the comprehensive quality score, and then determine a prediction result of the electroplating process parameters through the target parameter group.

[0007] In one embodiment of the present invention, the step of obtaining a plurality of parameters corresponding to a target product in an electroplating process, and obtaining a quality score set for each parameter for the target product, includes: obtaining a plurality of parameters that affect the quality score of the target product during the electroplating process; obtaining a plurality of quality scores corresponding to each parameter during the change process, and then determining a quality score set for each parameter for the target product using the plurality of quality scores.

[0008] In one embodiment of the present invention, obtaining the first response sensitivity of each parameter type to each quality score in the quality score set includes: obtaining the probability of external environment impact on each parameter; dividing multiple parameters into several parameter types by type, obtaining the maximum value and minimum value of the adjustment range of each parameter type, and obtaining the difference between two adjacent parameters within the adjustment range; and determining the first response sensitivity corresponding to each parameter type by using the external environment impact probability, the maximum value, the minimum value and the difference.

[0009] In one embodiment of the present invention, obtaining the probability of the external environment affecting each of the parameters includes: obtaining the number of quality scores corresponding to each of the parameters; obtaining the frequency of each quality score appearing in each of the parameters, and determining the first range of the quality score corresponding to each of the parameters; and determining the probability of the external environment affecting each of the parameters using the number of quality scores, the frequency, and the first range.

[0010] In one embodiment of the present invention, obtaining the first impact scale of each parameter group on each of the quality scores includes: obtaining a weighted mean of the quality scores corresponding to each of the parameters, and using the weighted mean of the quality scores to determine the quality evaluation mean difference between two adjacent parameters; using the first range to determine the second range of the quality scores corresponding to each parameter group; and using the quality evaluation mean difference and the second range to determine the first impact scale of each parameter group on each of the quality scores.

[0011] In an embodiment of the present invention, the method further includes: traversing multiple parameter groups of each of the parameters, and determining a second impact scale of each of the parameters on the quality score using the first impact scale.

[0012] In one embodiment of the present invention, the use of the first response sensitivity and the first influence scale to determine the comprehensive quality score of each parameter group includes: using the first response sensitivity and the second influence scale corresponding to each parameter to determine the amplification weight of each parameter; obtaining the first control value and the second control value corresponding to each parameter in different historical electroplating processes; using the amplification weight, the first control value and the second control value to determine the consistency of the quality score impact of different historical electroplating processes, and then determining the comprehensive quality score of each parameter group.

[0013] In one embodiment of the present invention, it also includes: normalizing the consistency of the quality score impact, determining the comprehensive quality score corresponding to each parameter group, and for several different historical electroplating processes with the same comprehensive quality score, treating the historical electroplating processes greater than a threshold as the same cluster; traversing all historical electroplating processes, and obtaining corresponding multiple different clusters for different comprehensive quality scores.

[0014] In one embodiment of the present invention, the method of determining the target parameter group of the target product using the comprehensive quality score includes: determining the actual values ​​of several parameters in the parameter group using the comprehensive quality score of each parameter group; matching the parameters with the intra-cluster data of different clusters using the parameters whose actual values ​​are obtained, and determining several matching parameter groups when the difference in values ​​between the same parameter types is the smallest; sorting the matching parameter groups using the comprehensive quality scores corresponding to them as a reference standard to determine the predicted values ​​of the remaining parameters in the parameter group; and regulating using the predicted values ​​to determine the target parameter group of the target product.

[0015] In order to solve the above technical problems, another technical solution adopted by the present invention is to provide an intelligent prediction system for electroplating processing parameters, including: a first acquisition module, used to obtain multiple parameters corresponding to the target product in the electroplating process, and obtain a quality score set of each parameter for the target product; a second acquisition module, using multiple parameters to determine multiple parameter types, and obtain the first response sensitivity of each parameter to each quality score in the quality score set, and obtain the first influence scale of each parameter group on each quality score, wherein each parameter group includes at least two parameter types, and each parameter type includes at least one parameter; a first determination module, using the first response sensitivity and the first influence scale to determine the comprehensive quality score of each parameter group; a second determination module, using the comprehensive quality score to determine the target parameter group of the target product, and then determining the prediction result of the electroplating processing parameters through the target parameter group.

[0016] The present invention has the following beneficial effects: the present invention controls the same type of parameters as variables, analyzes the impact of the same parameter changes on product quality one by one, and then obtains several processing processes with the same impact, and obtains corresponding parameter combinations, thereby avoiding the problem of predicting parameters based solely on the difference of a single parameter, which easily leads to large errors in the parameters that do not meet the corresponding processing conditions, improves the accuracy of parameter prediction, and ensures the high quality of the product in the subsequent parameter control processing process based on the prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0018] Figure 1 A schematic diagram of a flow chart of an intelligent prediction method for electroplating processing parameters provided by an embodiment of the present invention;

[0019] Figure 2 A schematic diagram of the structure of an intelligent prediction system for electroplating processing parameters provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0020] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the electroplating processing parameter intelligent prediction method and system proposed by the present invention, its specific implementation, structure, features and effects, in conjunction with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0021] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0022] The specific scheme of the intelligent prediction method of electroplating processing parameters provided by the present invention is described in detail below with reference to the accompanying drawings.

[0023] See also Figure 1 , which shows a flow chart of an intelligent prediction method for electroplating processing parameters provided by an embodiment of the present invention.

[0024] like Figure 1 As shown, a method for intelligent prediction of electroplating processing parameters includes the following steps:

[0025] S10, obtaining a plurality of parameters corresponding to the target product in the electroplating process, and obtaining a quality score set of each parameter for the target product.

[0026] Among them, the target product refers to the product produced by the electroplating process. Because the impact on product quality during the electroplating process involves multiple parameters, it is necessary to obtain multiple parameters corresponding to the electroplating process; the quality score set represents the score set for the quality of the target product when the parameters are at different values.

[0027] Specifically, for the target products that need to be produced, multiple parameters involved in the electroplating process of producing the target products are obtained, and the preset historical processing data are used to obtain different values ​​of each parameter to score the quality of the target products, and then the multiple quality scores of each parameter are determined as a quality score set for the corresponding parameter.

[0028] S20. Determine multiple parameter types using multiple parameters, and obtain a first response sensitivity of each parameter type to each quality score in the quality score set, and obtain a first impact scale of each parameter group on each quality score, wherein each parameter group includes at least two parameter types, and each parameter type includes at least one parameter.

[0029] Among them, parameter type refers to parameters divided by type, that is, the same parameter type refers to parameters that have the same impact on the product. There can be multiple parameters. For example, the size parameter type can include multiple parameters such as length, width, length, etc., which affect the size of the product; the first response sensitivity refers to the extent to which the parameter is affected by external factors during the adjustment process, that is, the greater the first response sensitivity, the greater the impact of the parameter on the quality score; the first impact scale refers to the impact of the parameter group on the product quality, that is, the larger the impact scale, the greater the impact of the parameter group on the product quality.

[0030] Specifically, after obtaining multiple parameters corresponding to the target product in the electroplating process and obtaining a quality score set for each parameter to the target product, for the multiple parameters, the parameters are divided into multiple parameter types using the type of influence on product quality, and the first response sensitivity of each parameter type to each quality score in the quality score set is obtained based on the quality score set corresponding to each parameter; because different products correspond to different processing conditions, different products have their own corresponding parameter groups during the processing, so it is necessary to determine the parameter group corresponding to the target product, each parameter group can include multiple parameter types, and then use the first response sensitivity to determine the first impact scale of each parameter group on each quality score.

[0031] S30. Determine a comprehensive quality score for each parameter group using the first response sensitivity and the first impact scale.

[0032] Among them, because different products correspond to different processing conditions, different products have their own corresponding parameter groups during the processing. Therefore, it is necessary to determine the parameter group corresponding to the target product, and the values ​​of different parameters are different, so different parameter groups can also be divided; the comprehensive quality score refers to the quality score of the target product for each parameter group determined by the first response sensitivity and the first impact scale.

[0033] Specifically, after obtaining the first response sensitivity and the first impact scale, corresponding to multiple parameter groups of the target product, the first response sensitivity and the first impact scale are used to determine the comprehensive quality score of each parameter group corresponding to the target product to obtain multiple comprehensive quality scores corresponding to the multiple parameter groups.

[0034] S40, determining a target parameter group of the target product using the comprehensive quality score, and then determining a prediction result of the electroplating processing parameters through the target parameter group.

[0035] The target parameter group refers to the optimal parameter group selected using the comprehensive quality score, that is, the value of each parameter in the target parameter group so that the quality of the target product is the best.

[0036] Specifically, after obtaining multiple comprehensive quality scores corresponding to the target product, the multiple comprehensive quality scores are sorted, the optimal target parameter group is selected, and then the value of each parameter in the target parameter group is used as the prediction result of the electroplating processing parameter.

[0037] In this embodiment, by obtaining the quality score of each parameter for the target product, the quality score set corresponding to each parameter is determined, and after dividing the parameters into multiple parameter types, the first response sensitivity of the parameter type to each quality score is determined, and after forming a parameter group including multiple parameters, the first influence scale of each parameter group on the quality score is determined, and then the comprehensive quality score corresponding to each parameter group is determined, and then the target parameter group is determined based on the comprehensive quality score, thereby improving the prediction accuracy of the processing parameters and ensuring the high quality of the product in the subsequent parameter control processing process based on the prediction.

[0038] In some embodiments, step S10 obtains a plurality of parameters corresponding to the target product in the electroplating process, and obtains a quality score set of each parameter for the target product, which may include the following operations:

[0039] First, multiple parameters that affect the quality score of the target product during the electroplating process are obtained.

[0040] Among them, each target product may have multiple parameters in multiple electroplating processes, and each parameter may affect the quality score of the product, so it is necessary to obtain multiple parameters.

[0041] Specifically, corresponding to the target product to be produced, based on preset prior knowledge, that is, based on the historical production process of electroplating processing, multiple parameters that affect the quality score of the target product during the electroplating processing are determined.

[0042] Next, a plurality of quality scores corresponding to each parameter in the changing process are obtained, and then a quality score set of each parameter for the target product is determined based on the plurality of quality scores.

[0043] During the numerical change of each parameter, each value can correspond to a quality score. Therefore, each parameter can correspond to multiple quality scores at different values. Therefore, multiple quality scores corresponding to each parameter in the numerical change process can be obtained. Because each parameter can have multiple quality scores, a quality score set for each parameter to the target product is formed.

[0044] Furthermore, historical data is screened to obtain multiple parameter types, specifically, one parameter is allowed to change, while other parameters are guaranteed to remain unchanged; it should be noted that each parameter type may include multiple parameters. At this time, when any other parameter changes, there may still be a data set that makes the current parameter change and correspond to multiple scores, and other parameters do not change; therefore, for any parameter, there may be multiple parameter groups, so quality score sets corresponding to several different parameter groups can be obtained.

[0045] In some embodiments, there may be multiple parameters that need to be adjusted during the electroplating process, or a combination of several of them, including, for example, current density controls the ratio of current to surface area, affecting the uniformity and thickness of the coating; electrolyte composition and concentration, including metal salts, pH regulators, etc., directly affecting the composition and quality of the coating; temperature, affecting the electrochemical reaction rate and the crystallinity and uniformity of the coating; the pH value of the plating solution, affecting the progress of the electrochemical reaction and the quality of the coating; the content of additives in the plating solution, such as plasticizers, wetting agents, etc., can adjust the surface tension and liquid fluidity during the electroplating process, affecting the smoothness and uniformity of the coating; electroplating time, controlling the thickness and uniformity of the coating, usually interacts with the current density and electrolyte concentration.

[0046] The value of each of the above parameters determines whether the various evaluation indicators of the final electroplating product meet the requirements, that is, whether the quality score of the target product of the electroplating process meets the requirements.

[0047] Furthermore, the actual values ​​of various parameters of the electroplating process and the corresponding target products can be obtained based on the prior electroplating related production process; at this time, the inspection indicators for the target products include appearance gloss and whether there are defects, coating thickness, adhesion between coating and substrate, corrosion resistance, dimensional accuracy, etc.; in the process of judging the final quality of the target product, it should be ensured that the above-mentioned various inspection indicators are within the acceptance range, so the quality of the finished product can be inspected as a whole based on the existing neural network algorithm. The specific operation process is as follows: 1. Select inspection indicators for various indicators of different finished products, such as the corresponding gloss, thickness, adhesion between coating and substrate, and corrosion resistance; 2. The above-mentioned feature information is scored manually, and the higher the score, the higher the quality of the product in the current indicator; 3. The classification network structure is a five-layer fully connected neural network model, and the loss function is the cross entropy function; 4. The training ratio set in the historical data is 7:3, and the classification network is trained by selecting link data. The input of the classification network is: the multidimensional characteristics of the assessment indicators of various electroplating finished products; the output is: the overall quality score of any target product; during the classification network training process, the gradient descent method is used for training until the loss function converges and the classification network training is completed. The quality score of the target product is between 0 and 100. The higher the score, the better the quality of the target product of electroplating processing.

[0048] Therefore, for all historical processing processes, in each production process, the actual parameter values ​​of new products produced by different types of parameter changes can be obtained, as well as the quality score of the final product. The relevant data of several historical production processes can be saved to the terminal to facilitate subsequent processing and operation.

[0049] In some embodiments, step S20 of obtaining the first response sensitivity of each parameter to each quality score in the quality score set, and obtaining the first impact scale of each parameter group on each quality score, may include the following operations:

[0050] Obtain the probability of the external environment affecting each parameter.

[0051] According to prior knowledge, different parameters control different evaluation indicators of the product, and their suitability determines the quality of the final product; however, any evaluation indicator is affected by a variety of different parameters, and the numerical changes of different parameters have different effects on the actual test results of the evaluation indicator. Therefore, when making intelligent predictions on multiple parameters, the actual adjustment requirements of different parameters are different.

[0052] Furthermore, the following operations may also be included:

[0053] Get the number of quality scores corresponding to each parameter.

[0054] Because each parameter may have different quality scores at different values, the quality scores at different values ​​of each parameter may be obtained, and the number of quality scores of each parameter may be determined.

[0055] The frequency of each quality score of each parameter is obtained, and the first range of the quality score corresponding to each parameter is determined.

[0056] According to the analysis, there may be a situation in which one parameter in the data set corresponds to multiple scores. The reason for this situation is generally due to external environmental problems in the actual operation process of the production process, such as temperature, humidity and other factors, or the operator's opinion of a series of uncontrollable factors caused by the operation process; and the greater the fluctuation of the score in a data set corresponding to any parameter, the greater the probability of error caused by external factors in the current parameter; therefore, the number of quality scores corresponding to any parameter in any set of data sets in the current parameter is obtained. ; and obtain the frequency of each quality score for each parameter, that is, the frequency of multiple different scores under the current parameter , and determine the first range of the quality score corresponding to each parameter, that is, obtain the first range of the quality score .

[0057] Furthermore, the probability of external environment influence on each parameter is determined by using the number, frequency and first range of quality scores.

[0058] The probability of external environmental impact is calculated as follows:

[0059] ;

[0060] in, represents the probability of external environment influence of the i-th parameter type, i represents the i-th parameter type, j represents the j-th quality score set, r represents the actual quality score corresponding to the r-th parameter, and x represents the x-th quality score size type. Indicates the number of quality scores; Indicates the first range between the values ​​of the quality score under the current parameters. The larger the value, the greater the fluctuation of the quality score of the target product under the current parameters, and the greater the possibility of error in the current parameters; Indicates the total number of different quality score sets obtained by the current i-th parameter type after controlling the variables. Indicates the total number of actual values ​​of the current parameter of the current i-th parameter type in the current quality score set. Represents a logarithmic function.

[0061] Indicates the frequency of occurrence of the xth quality score under the current parameters, is the existing information entropy calculation formula. The larger the value, the more complex the data is, and the greater the external influence on the current parameter is. Weighted, and traverse all different parameters and all parameter groups in the current array, thereby obtaining the probability of external environment influence The larger the value, the greater the probability of error in the current parameter.

[0062] Next, for the current parameters, at the same time, the required rated parameters of different assessment indicators adjust the indicator range and precise units. For any type of parameter, the smaller the error range and the smaller the precision unit, the smaller the selection range of the current parameter, the greater the demand for its precision, which also indicates that the extent of possible external influence is greater.

[0063] The multiple parameters are divided into several parameter types by using the type, and the maximum value and the minimum value of the adjustment range of each parameter type are obtained, and the difference between two adjacent parameters within the adjustment range is obtained.

[0064] The parameters may be divided into different parameter types, each of which may include one or more parameters. For example, electrolyte composition and concentration is a parameter type, which may include multiple different parameters such as metal salts, pH regulators, etc.

[0065] Specifically, the maximum and minimum values ​​of the adjustment range of each parameter type are obtained, and the difference between two adjacent parameters within the adjustment range is obtained. For example, the maximum value of the adjustment range of the current i-th parameter type is obtained. and minimum value ; and get the difference between two adjacent optional parameters within the adjustment range , the difference represents the exact unit of the current parameter type.

[0066] The first response sensitivity corresponding to each parameter type is determined by using the external environment influence probability, maximum value, minimum value and difference.

[0067] That is, by combining the external environment influence probability of the error obtained based on the data performance, the first response sensitivity of any type of parameter is obtained.

[0068] The first response sensitivity is calculated as follows:

[0069] ;

[0070] in, represents the first response sensitivity, Indicates the difference between two adjacent optional parameters within the adjustment range of the i-th parameter type; Indicates the maximum value of the adjustment range of the i-th parameter type; Indicates the minimum value of the adjustment range of the current i-th parameter type; represents the probability of external environmental influence on the i-th parameter type, and e represents the base of the exponential function.

[0071] In some embodiments, the second response sensitivity corresponding to each parameter group may also be determined, that is, the average of the first response sensitivities corresponding to different parameter types in the parameter group is used as the second response sensitivity.

[0072] In the formula, the difference The smaller the value, the higher the parameter adjustment accuracy requirement, and the greater the possibility that the parameter will be affected by external factors during the adjustment process. The smaller it is, the greater the possibility of being affected by external factors. Combine the above two equations and normalize them using the exponential function. The larger the value, the greater the external influence on the current parameter type.

[0073] Combined with the external environment impact probability obtained in the previous step, the first response sensitivity is obtained The larger the value, the higher the first response sensitivity of any type of parameter. It also means that the current parameter type has a higher adjustment accuracy requirement among all parameter types, the greater the impact on quality, and the greater its weight.

[0074] Next, a first impact scale of each parameter group on each quality score is obtained.

[0075] Furthermore, the following operations may also be included:

[0076] The weighted mean of the quality score corresponding to each parameter is obtained, and the weighted mean of the quality score is used to determine the difference in the quality evaluation means between two adjacent parameters.

[0077] Among them, the actual impact of different parameters on quality is different; therefore, it is necessary to obtain the ability of different parameters to affect the quality of the final product; and according to the analysis, the change of one parameter may cause the product quality to increase or decrease nonlinearly. When other parameters are kept unchanged, the impact trend and degree of any parameter change on product quality can be obtained.

[0078] Therefore, if the current parameters of any group are arranged from small to large, the order of the corresponding actual score data will also change; since any parameter corresponds to multiple quality score values, the weighted average of the quality scores within any parameter is obtained according to the frequency of occurrence of the corresponding quality score. ; Get the score value of any parameter at a time; and then use the weighted mean of the quality score to determine the mean difference in quality evaluation between two adjacent parameters .

[0079] Next, the first range is used to determine the second range of the quality score corresponding to each parameter group.

[0080] Specifically, the first range between the values ​​of the quality score under the current parameters is used , obtain the mean of the first range corresponding to each parameter in the parameter group, and use the mean of the first range to determine the second range of the quality score corresponding to each parameter group .

[0081] Furthermore, the first impact scale of each parameter group on each quality score is determined using the mean difference and the second range of the quality evaluation.

[0082] That is, for the current parameter type i (the i-th parameter type), the first impact scale of the quality score of the target product corresponding to any set of parameter change numbers is obtained.

[0083] The first effect size was calculated as follows:

[0084] ;

[0085] in, represents the first impact size; It represents the mean difference of quality evaluation corresponding to the rth parameter in the jth quality score set of the i-th parameter type; Indicates the second range of the quality score corresponding to each parameter group. Indicates the absolute value function; at this time, by accumulating all the score differences and the absolute values ​​of the differences in the current parameter group, it is expressed as a fraction, thus obtaining , the value range is [-1,1]. By obtaining its absolute value, when the value approaches 1, it means that the change of the current parameter in the current j-th quality score set has an obvious trend in the impact on the quality score, and the larger the first impact scale.

[0086] In some embodiments, when the calculated value of the first impact scale is 0, the calculation is performed again, and when the calculated values ​​are 0 multiple times, it indicates that the current parameter group can be used as the target parameter group.

[0087] The second worst combined quality score The larger the second range is, the greater the impact of the current parameter change on the quality score is. By weighting the above trend, we can get the first impact scale. , which represents the impact scale of the quality score corresponding to any set of parameter changes. The larger the value, the larger the scale.

[0088] In some embodiments, it may also include:

[0089] The plurality of parameter groups for each parameter are traversed, and a second impact scale of each parameter on the quality score is determined using the first impact scale.

[0090] Specifically, by traversing multiple combination data sets of the current parameter, the second impact scale of the parameter in the current parameter type i on the quality is obtained. .

[0091] The second effect size was calculated as follows:

[0092] ;

[0093] in, represents the second effect size; represents the first impact size; Indicates the total number of different quality score sets obtained for the current i-th parameter type after controlling the variables.

[0094] Second impact scale The larger it is, the greater the impact of the parameters in the current parameter type i on the quality score, which also indicates that the current parameter type has a greater impact on the quality score of the target product during the adjustment process.

[0095] At this time, according to prior knowledge, there are situations where the quality scores are the same but the actual process parameters are different. In order to make the parameter control in the future electroplating process meet the actual needs and produce smaller process errors to obtain high-quality target products, in the process of predicting and controlling the subsequent processing parameters, we can try to avoid the situation where the adjustment range of different parameters is too large or too small, which has different impacts on the quality of the final target product, resulting in a decrease in the quality of the finished target product.

[0096] According to the analysis, in the process of predicting multiple actual parameters, the differences between the same parameter types should be made as small as possible, and the actual adjustment requirements of different parameter types have different impacts on product quality; therefore, the differences are weighted based on the characteristic values ​​obtained by the above operations, and then the combinations of historical multiple parameter type control parameter groups are clustered.

[0097] In some embodiments, step S30 determines the comprehensive quality score of each parameter group using the first response sensitivity and the first impact scale, which may include the following operations:

[0098] The second impact scale of each parameter in each parameter group on the quality score is obtained by using the first response scale, as described above and will not be repeated here.

[0099] The amplification weight of each parameter is determined by using the first response sensitivity and the second impact scale corresponding to each parameter.

[0100] Among them, the amplification weight It refers to the actual control weight of any type of parameter. The larger the value, the greater the impact on the quality score during the current parameter adjustment process, and the higher the amplification weight of the difference generated in the subsequent similarity judgment.

[0101] Specifically, using response sensitivity The second effect size corresponding to each parameter , determine the amplification weight of each parameter , which is calculated as follows:

[0102] ;

[0103] in, Indicates the amplification weight; represents the second effect size; It represents the first response sensitivity; Indicates normalization by norm function.

[0104] Obtain the first control value and the second control value of each parameter corresponding to different historical electroplating processes, that is, the first control value is the actual value of the current parameter in the first historical electroplating process, and the second control value is the actual value of the current parameter in the second historical electroplating process.

[0105] The amplification weight, the first control value and the second control value are used to determine the consistency of the quality score impact of different historical electroplating processes, and then determine the comprehensive quality score of each parameter group.

[0106] The quality score impact consistency calculation is as follows:

[0107] ;

[0108] in, It shows that the quality score affects the consistency; a and b represent two different historical electroplating processes, , Respectively represent the actual first control value and the second control value of the same parameter type corresponding to the two processing processes a and b; by obtaining the value difference between the same parameters in the two historical electroplating processing processes , and use the amplification weight of the corresponding parameter Logarithmic Difference Weighting is performed to amplify the possible impact of the difference in parameter values, and all parameter types are traversed to obtain The smaller the value, the more it means that the two historical electroplating processes have the same impact on the quality score, and the two are clustered into one cluster. H represents the total number of parameter types of historical electroplating processes, and exp represents the calculation of the exponential function.

[0109] Furthermore, it also includes:

[0110] The consistency of the quality score influence is normalized to determine the comprehensive quality score corresponding to each parameter group. For several different historical electroplating processes with the same comprehensive quality score, the historical electroplating processes with a score greater than a threshold are regarded as the same cluster.

[0111] That is, the exponential function is used to normalize it, and the threshold range is set to 0.5 to 1, and the threshold can be 0.8. For several different historical electroplating processes with the same comprehensive quality score, two historical electroplating processes with scores greater than the threshold are considered to be the same cluster.

[0112] All historical electroplating processes are traversed, and multiple different clusters corresponding to different comprehensive quality scores are obtained.

[0113] Specifically, the optimal k is obtained by the elbow method, and all historical data are traversed in combination with the above-mentioned discrimination method, so as to obtain several different clusters under several different comprehensive quality scores. Among them, the implementation process of the elbow method is a prior art and is not within the scope of protection of the present invention, and will not be elaborated here.

[0114] In some embodiments, step S40 determines the target parameter group of the target product using the comprehensive quality score, which may include the following operations:

[0115] The comprehensive quality score of each parameter group is used to determine the actual values ​​of several parameters in the parameter group. That is, the comprehensive quality score of each parameter group is obtained, and the quality scores of the corresponding parameters in the comprehensive quality score are sorted. From the historical data of the historical electroplating process, the actual value corresponding to each parameter in the parameter group can be determined.

[0116] The parameters with actual values ​​are matched with the intra-cluster data of different clusters, and when the difference in values ​​between the same parameter types is minimized, several matching parameter groups are determined. Specifically, the parameters with actual values ​​determined are matched with the intra-cluster data of different clusters obtained in the above steps one by one, so that the difference in values ​​between the same parameter types is minimized, thereby obtaining several matching parameter combinations.

[0117] The quality score ranking corresponding to the matched parameter group is used as a reference standard to determine the predicted values ​​of the remaining parameters in the parameter group; the predicted values ​​are then used for regulation to determine the target parameter group for the target product.

[0118] Specifically, according to the above method, several different parameter combination methods are obtained, and each combination method corresponds to a quality score; in the actual electroplating process, the actual values ​​of several parameters have been determined. At this time, one or more parameters may not be determined. In order to avoid quality problems caused by inaccurate parameter settings, the actual parameter values ​​that have been determined are matched one by one with the cluster data obtained in the above steps, so that the difference in values ​​between the same parameter types is minimized, thereby obtaining several matching parameter combinations; at this time, among these combinations, several combinations with the highest quality scores in the cluster where the combinations are located are selected, and based on the parameter combinations with the highest comprehensive quality scores in the historical data, for example, The three with the highest quality scores in the cluster and the three with the highest quality scores in the historical data, and the one with the highest frequency are taken as the reference standard for the remaining parameters in the final current processing process, and then the prediction of the remaining parameters is completed. For example, when the values ​​of the remaining parameters are uncertain, the values ​​corresponding to the quality scores with the highest frequency in the reference standard are taken as the values ​​of the remaining parameters; based on the final predicted values, the remaining parameter values ​​are adjusted, that is, after obtaining the specific values ​​of the remaining parameters, the most appropriate parameters with determined values ​​can be selected according to the comprehensive quality scores, and then a target parameter group is formed; based on the target parameter group, the target product is electroplated in combination with the electroplating operation process to obtain the finished product of the target product.

[0119] In this embodiment, by obtaining the quality score of each parameter for the target product, the quality score set corresponding to each parameter is determined, and after dividing the parameters into multiple parameter types, the response sensitivity of the parameter type to each quality score is determined, and after the multiple parameters are formed into parameter groups, the first influence scale of each parameter group on the quality score is determined, and then the comprehensive quality score corresponding to each parameter group is determined, and then the target parameter group is determined based on the comprehensive quality score, thereby improving the prediction accuracy of the processing parameters and ensuring the high quality of the product in the subsequent parameter control processing process based on the prediction.

[0120] The present invention also provides an electroplating processing parameter intelligent prediction system 500, such as Figure 2As shown, it includes: a first acquisition module 510, a second acquisition module 520, a first determination module 530 and a second determination module 540; wherein, the first acquisition module is used to obtain multiple parameters corresponding to the target product in the electroplating process, and obtain a quality score set of each parameter for the target product; the second acquisition module uses multiple parameters to determine multiple parameter types, and obtains the first response sensitivity of each parameter to each quality score in the quality score set, and obtains the first influence scale of each parameter group to each quality score, wherein each parameter group includes at least two parameter types, and each parameter type includes at least one parameter; the first determination module uses the first response sensitivity and the first influence scale to determine the comprehensive quality score of each parameter group; the second determination module uses the comprehensive quality score to determine the target parameter group of the target product, and then determines the prediction result of the electroplating processing parameters through the target parameter group.

[0121] In the present invention, the method of controlling variables with parameters of the same type is adopted to analyze the influence of changes in the same parameters on product quality one by one, and then obtain several processing processes with the same influence based on the clustering method, and obtain the corresponding parameter combination, so as to avoid that the parameter prediction is easily caused by relying on the difference of a single parameter, which does not meet the corresponding processing conditions, thereby improving the accuracy of parameter prediction; at the same time, by analyzing the regularity of a single parameter corresponding to a variety of scoring data in the process of controlling variables, the error probability of any parameter is obtained, and the response sensitivity of the current parameter is obtained based on the probability, which avoids the high overlap of the finished product quality scores obtained in the historical processing process, resulting in the actual prediction results being unable to correspond to the influence of parameter changes, thereby ensuring the high quality of the product in the subsequent parameter regulation processing process based on the prediction.

[0122] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0123] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. An intelligent prediction method for electroplating processing parameters, characterized in that: The electroplating processing parameter intelligent prediction method comprises: Obtaining a plurality of parameters corresponding to a target product in an electroplating process, and obtaining a quality score set of each parameter for the target product; Determine a plurality of parameter types using a plurality of parameters, and obtain a first response sensitivity of each parameter type to each quality score in the quality score set, and obtain a first impact scale of each parameter group to each quality score, wherein each parameter group includes at least two parameter types, and each parameter type includes at least one parameter; determining a comprehensive quality score for each parameter group using the first response sensitivity and the first impact scale; Determine a target parameter group of the target product by using the comprehensive quality score, and then determine a prediction result of an electroplating processing parameter by using the target parameter group; The obtaining of the first response sensitivity of each parameter type to each quality score in the quality score set includes: Obtaining the probability of external environment influence on each of the parameters; Divide the plurality of parameters into a plurality of parameter types by type, obtain a maximum value and a minimum value of an adjustment range of each parameter type, and obtain a difference between two adjacent parameters within the adjustment range; Determine a first response sensitivity corresponding to each parameter type by using the external environment influence probability, the maximum value, the minimum value and the difference; The first response sensitivity is calculated as follows: ; in, represents the first response sensitivity, Indicates the difference between two adjacent optional parameters within the adjustment range of the i-th parameter type; Indicates the maximum value of the adjustment range of the i-th parameter type; Indicates the minimum value of the adjustment range of the current i-th parameter type; represents the probability of external environmental influence of the i-th parameter type, and e represents the base of the exponential function; The obtaining of the probability of the external environment affecting each of the parameters includes: Obtaining the number of quality scores corresponding to each of the parameters; Obtaining the frequency of each quality score of each parameter, and determining the first range of the quality score corresponding to each parameter; Determine the probability of external environment influence on each of the parameters by using the number of quality scores, the frequency and the first range; The probability of external environmental impact is calculated as follows: ; in, represents the probability of external environment influence of the i-th parameter type, i represents the i-th parameter type, j represents the j-th quality score set, r represents the actual quality score corresponding to the r-th parameter, and x represents the x-th quality score size type. Indicates the number of quality scores; Indicates the first range between the values ​​of the quality score under the current parameters; Indicates the total number of different quality score sets obtained by the current i-th parameter type after controlling the variables. Indicates the total number of actual values ​​of the current parameter of the current i-th parameter type in the current quality score set. represents the logarithmic function; The obtaining of a first impact scale of each parameter group on each quality score includes: Obtaining a weighted mean of the quality scores corresponding to each of the parameters, and using the weighted mean of the quality scores to determine the mean difference in quality evaluation between two adjacent parameters; Determine a second range of the quality score corresponding to each parameter group using the first range; Determine a first impact scale of each parameter group on each of the quality scores using the quality evaluation mean difference and the second range; ; in, represents the first impact size; It represents the mean difference of quality evaluation corresponding to the rth parameter in the jth quality score set of the i-th parameter type; Indicates the second range of the quality score corresponding to each parameter group. Represents the absolute value function.

2. The electroplating processing parameter intelligent prediction method according to claim 1, characterized in that: The step of obtaining a plurality of parameters corresponding to the target product in the electroplating process, and obtaining a quality score set of each parameter for the target product, includes: Acquire multiple parameters that affect the quality score of the target product during the electroplating process; A plurality of the quality scores corresponding to each of the parameters during the change process are obtained, and then a quality score set of each parameter for the target product is determined based on the plurality of the quality scores.

3. The intelligent prediction method for electroplating processing parameters according to claim 1, characterized in that: Also includes: The plurality of parameter groups of each of the parameters are traversed, and a second impact scale of each of the parameters on the quality score is determined using the first impact scale.

4. The intelligent prediction method for electroplating processing parameters according to claim 3 is characterized in that: The step of determining a comprehensive quality score of each parameter group by using the first response sensitivity and the first impact scale includes: Determine an amplification weight of each of the parameters by using the first response sensitivity and a second impact scale corresponding to each of the parameters; Obtaining a first control value and a second control value of each parameter corresponding to different historical electroplating processing processes; The amplification weight, the first control value and the second control value are used to determine the consistency of the quality score impact of different historical electroplating processes, and then determine the comprehensive quality score of each parameter group.

5. The electroplating processing parameter intelligent prediction method according to claim 4, characterized in that: Also includes: Normalizing the consistency of the quality score influence, determining the comprehensive quality score corresponding to each parameter group, and for a number of different historical electroplating processes with the same comprehensive quality score, treating the historical electroplating processes with a score greater than a threshold as the same cluster; All historical electroplating processes are traversed, and multiple different clusters corresponding to different comprehensive quality scores are obtained.

6. The intelligent prediction method for electroplating processing parameters according to claim 5, characterized in that: The step of determining a target parameter group of the target product by using the comprehensive quality score comprises: Using the comprehensive quality score of each parameter group, determining actual values ​​of several parameters in the parameter group; Using the parameters whose actual values ​​are obtained to match the intra-cluster data of different clusters, and determining a plurality of matching parameter groups when the value difference between the same parameter types is the smallest; Using the comprehensive quality score ranking corresponding to the matched parameter group as a reference standard, determining the predicted values ​​of the remaining parameters in the parameter group; The predicted value is used for regulation to determine a target parameter group for a target product.

7. An intelligent prediction system for electroplating processing parameters, characterized in that: include: A first acquisition module is used to acquire a plurality of parameters corresponding to a target product in an electroplating process, and to acquire a quality score set of each parameter for the target product; a second acquisition module, which uses the plurality of parameters to determine a plurality of parameter types, and acquires a first response sensitivity of each parameter to each quality score in the quality score set, and acquires a first impact scale of each parameter group to each quality score, wherein each parameter group includes at least two parameter types, and each parameter type includes at least one parameter; a first determination module, determining a comprehensive quality score of each parameter group using the first response sensitivity and the first impact scale; A second determination module, using the comprehensive quality score to determine a target parameter group for the target product, and then determining a prediction result of an electroplating processing parameter through the target parameter group; The obtaining of the first response sensitivity of each parameter type to each quality score in the quality score set includes: Obtaining the probability of external environment influence on each of the parameters; Divide the plurality of parameters into a plurality of parameter types by type, obtain a maximum value and a minimum value of an adjustment range of each parameter type, and obtain a difference between two adjacent parameters within the adjustment range; Determine a first response sensitivity corresponding to each parameter type by using the external environment influence probability, the maximum value, the minimum value and the difference; The first response sensitivity is calculated as follows: ; in, represents the first response sensitivity, Indicates the difference between two adjacent optional parameters within the adjustment range of the i-th parameter type; Indicates the maximum value of the adjustment range of the i-th parameter type; Indicates the minimum value of the adjustment range of the current i-th parameter type; represents the probability of external environmental influence of the i-th parameter type, and e represents the base of the exponential function; The obtaining of the probability of the external environment affecting each of the parameters includes: Obtaining the number of quality scores corresponding to each of the parameters; Obtaining the frequency of each quality score of each parameter, and determining the first range of the quality score corresponding to each parameter; Determine the probability of external environment influence on each of the parameters by using the number of quality scores, the frequency and the first range; The probability of external environmental impact is calculated as follows: ; in, represents the probability of external environment influence of the i-th parameter type, i represents the i-th parameter type, j represents the j-th quality score set, r represents the actual quality score corresponding to the r-th parameter, and x represents the x-th quality score size type. Indicates the number of quality scores; Indicates the first range between the values ​​of the quality score under the current parameters; Indicates the total number of different quality score sets obtained by the current i-th parameter type after controlling the variables. Indicates the total number of actual values ​​of the current parameter of the current i-th parameter type in the current quality score set. represents the logarithmic function; The obtaining of a first impact scale of each parameter group on each quality score includes: Obtaining a weighted mean of the quality scores corresponding to each of the parameters, and using the weighted mean of the quality scores to determine the mean difference in quality evaluation between two adjacent parameters; Determine a second range of the quality score corresponding to each parameter group using the first range; Determine a first impact scale of each parameter group on each of the quality scores using the quality evaluation mean difference and the second range; ; in, represents the first impact size; It represents the mean difference of quality evaluation corresponding to the rth parameter in the jth quality score set of the i-th parameter type; Indicates the second range of the quality score corresponding to each parameter group. Represents the absolute value function.