Production control method and related device, storage medium and program product

Abstract

Production control method and related devices, storage media and program products. The embodiment of the present application discloses a production control method and related devices and a medium program product, including the following steps: obtaining at least one benchmark operating condition version, at least one benchmark operating condition version is an operating condition version inferred by a machine learning model; generating a working condition benchmark value scorecard for each benchmark operating condition version in at least one benchmark operating condition version; performing mass production verification according to each benchmark operating condition version in at least one benchmark operating condition version, and updating the working condition benchmark value scorecard of each benchmark operating condition version according to the design parameters of the mass production verification; evaluating the availability of each benchmark operating condition version according to the working condition benchmark value scorecard of each benchmark operating condition version, and determining the mass production version in at least one benchmark operating condition version. The embodiment of the present application can scientifically judge the mass production quality of each operating condition version, determine the optimal production process parameters, perform quality control on mass-produced products, and thus ensure the optimal yield of the product.

Description

Technical Field

[0001] The present invention relates to the general field of data processing, and in particular to a production control method and related devices, storage media and program products. Background Art

[0002] For traditional process manufacturing companies to achieve production process optimization, actual operators need to infer and summarize experience in actual production and manufacturing. For example, in traditional chemical plant companies, the methods derived from this cannot be widely verified in mass production, excellent experience cannot be shared, and cannot be solidified and promoted. There is a lack of a systematic management of these experiences for the optimal production process parameters. At the same time, traditional experience lacks the support of big data and precise statistical analysis, and cannot adapt to the needs of current intelligent production.

[0003] During the production process of manufacturing enterprises, the operating boundary conditions of production equipment are constantly changing: the quality of incoming materials is variable, the climate is variable, the equipment is variable, and the load is variable, which makes it difficult to apply excellent experience in equipment operation. At the same time, there is often only one set of large and comprehensive production process standards in the production process. The quality indicators of products produced according to this standard fluctuate greatly, and the quality is very unstable. Engineers are required to adjust process parameters according to their own experience. There is a lack of a mechanism for multi-condition classification management of equipment operation under optimal parameters to ensure that product quality, cost, and efficiency indicators are achieved. Summary of the invention

[0004] In response to the above problems, the present application example provides a production control method based on multiple production working conditions and related devices, storage media and program products, and creates a new working condition benchmark value scorecard for multiple working condition versions obtained through machine learning model reasoning, thereby improving the refinement of production and improving product quality. Multiple working condition versions are verified for mass production, and multiple design parameters obtained from mass production are recorded in the working condition benchmark value scorecard. By determining the version of the corresponding training results under each working condition, and recommending each version of the working condition version according to the version type, there is a scientific basis for judging the production quality. Through the method of mass production verification of the working condition version, the optimal mass production working condition version is selected, which ensures that the setting of the production process parameters is always in the optimal state. The mass production version of the benchmark working condition is dynamically updated through the sampling and filtering algorithm, which realizes product quality control in the mass production stage and ensures the optimal product yield.

[0005] To achieve the above objectives, in a first aspect, an embodiment of the present application provides a production control method based on multiple production conditions, the method comprising:

[0006] Acquire at least one benchmark operating condition version, wherein the at least one benchmark operating condition version is an operating condition version inferred by a machine learning model; generate an operating condition benchmark value scorecard for each benchmark operating condition version in the at least one benchmark operating condition version, wherein the operating condition benchmark value scorecard is used to record the design parameters of each benchmark operating condition version; perform mass production verification according to each benchmark operating condition version in the at least one benchmark operating condition version, and update the operating condition benchmark value scorecard of each benchmark operating condition version according to the design parameters of the mass production verification; evaluate the availability of each benchmark operating condition version according to the operating condition benchmark value scorecard of each benchmark operating condition version, and determine a mass production version in the at least one benchmark operating condition version.

[0007] In combination with the first aspect, in a possible embodiment, obtaining at least one benchmark operating condition version includes:

[0008] Obtain a first working condition version, where the first working condition version is the working condition version with the highest priority among the working condition versions that have achieved the learning goal through machine learning model reasoning; obtain a second working condition version, where the second working condition version is M working condition versions randomly selected from the top N working condition versions with the highest priority, excluding the working condition version with the highest priority among the working condition versions that have achieved the learning goal through machine learning model reasoning, where N≥M; use the first working condition version and the second working condition version as at least one benchmark working condition version.

[0009] In combination with the first aspect, in a possible embodiment, obtaining at least one benchmark operating condition version includes:

[0010] If the benchmark working condition version fails to reach the learning goal after being inferred by the machine learning model, the top T working condition versions with the highest priority will be used as at least one benchmark working condition version.

[0011] In combination with the first aspect, in a possible embodiment, the priority ranking of the operating condition versions is determined according to whether the design parameters in the operating condition versions reach the target values.

[0012] In combination with the first aspect, in a possible embodiment, the availability of each benchmark operating condition version is evaluated according to the operating condition benchmark value scorecard of each benchmark operating condition version, including: the operating condition benchmark value scorecard of each benchmark operating condition version updates the batch data of the product parameters of the corresponding benchmark operating condition version according to the mass production verification results; obtains the upper and lower limits of the product parameter specifications of the operating condition benchmark value scorecard; calculates the product parameter mean and the product parameter standard deviation according to the batch data of the product parameters; and calculates the process capability index CPK of the product parameters of each benchmark operating condition version according to the product parameter mean, the product parameter standard deviation and the upper and lower limits of the product parameter specifications.

[0013] In combination with the first aspect, in a possible embodiment, the availability of each benchmark operating condition version is evaluated according to the CPK value of the product parameters, including: if the CPK of the product parameters of the first target operating condition version in at least one benchmark operating condition version is not greater than the first preset value, then the first target operating condition version is determined to be a version unavailable for mass production; if the CPK of the product parameters of the second target operating condition version in at least one benchmark operating condition version is greater than the first preset value, then the second target operating condition version is determined to be a version available for mass production; if the CPK of the product parameters of the third target operating condition version in at least one benchmark operating condition version is greater than the second preset value, then the current product is determined to be the optimal version for mass production.

[0014] In combination with the first aspect, in a possible embodiment, determining a mass production version in at least one benchmark operating condition version includes: if it is determined that at least one benchmark operating condition version includes a unique mass production optimal version, determining the unique mass production optimal version as the mass production version in at least one benchmark operating condition version; and / or if it is determined that at least one benchmark operating condition version includes multiple mass production optimal versions, determining a sorting according to the parameter priority and parameter value corresponding to each version in the multiple mass production optimal versions in the production data table, and determining the mass production version according to the sorting result, wherein the production data table consists of design parameters in the operating condition benchmark value scorecard; and / or if it is determined that at least one benchmark operating condition version does not include a unique mass production optimal version and includes at least one mass production available version, determining a sorting according to the parameter priority and parameter value corresponding to each version in the multiple mass production available versions in the production data table, and determining the mass production version according to the sorting result.

[0015] In a second aspect, an embodiment of the present application provides a production control device based on multiple production conditions, the production control device based on multiple production conditions comprising:

[0016] An acquisition unit: used to acquire at least one benchmark operating condition version, where at least one benchmark operating condition version is an operating condition version inferred by a machine learning model;

[0017] A recording unit: used to generate a working condition benchmark value scorecard for each benchmark working condition version in at least one benchmark working condition version, the working condition benchmark value scorecard being used to record the design parameters of each benchmark working condition version; performing mass production verification according to each benchmark working condition version in at least one benchmark working condition version, and updating the working condition benchmark value scorecard of each benchmark working condition version according to the design parameters of the mass production verification;

[0018] Evaluation unit: used to evaluate the availability of each benchmark operating condition version according to the operating condition benchmark value scorecard of each benchmark operating condition version, and determine a mass production version of at least one benchmark operating condition version.

[0019] In a third aspect, an embodiment of the present application provides an electronic device, including a processor, a memory, a communication interface, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the processor, and one or more instructions are suitable for being loaded by the processor and executing the following steps:

[0020] Obtain at least one benchmark operating condition version, where at least one benchmark operating condition version is an operating condition version inferred by a machine learning model;

[0021] Generate a working condition benchmark value scorecard for each benchmark working condition version in at least one benchmark working condition version, where the working condition benchmark value scorecard is used to record the design parameters of each benchmark working condition version;

[0022] Performing mass production verification on each benchmark operating condition version of at least one benchmark operating condition version, and updating the operating condition benchmark value scorecard of each benchmark operating condition version according to the design parameters of the mass production verification;

[0023] The availability of each benchmark operating condition version is evaluated according to the operating condition benchmark value scorecard of each benchmark operating condition version, and a mass production version of at least one benchmark operating condition version is determined.

[0024] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium storing a computer program for electronic data exchange, wherein the computer program enables a computer to execute the method of the first aspect.

[0025] Implementing the embodiments of the present application has the following beneficial effects:

[0026] New working condition benchmark value scorecards are created for multiple working condition versions obtained through machine learning model inference. The increased recorded process parameters improve the refinement of production and enhance product quality. Multiple working condition versions are verified for mass production, and multiple process parameters obtained from mass production are recorded in the working condition benchmark value scorecard. By determining the version corresponding to the training results under each working condition, and recommending each version of the working condition version based on the version type, there is a scientific basis for judging the production quality. By verifying the working condition version for mass production, the optimal mass production working condition version is selected to ensure that the setting of the production process parameters is always in the optimal state, thereby ensuring the optimal product yield. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0028] Figure 1 A schematic diagram of an application scenario of a production control method based on multiple production conditions provided in an embodiment of the present application;

[0029] Figure 2 A flowchart of a production control method based on multiple production conditions provided in an embodiment of the present application;

[0030] Figure 3A A schematic diagram of a product scorecard provided in an embodiment of the present application;

[0031] Figure 3B A schematic diagram of a component scorecard provided in an embodiment of the present application;

[0032] Figure 3C A schematic diagram of a process equipment scorecard provided in an embodiment of the present application;

[0033] Figure 4 A schematic diagram of the structure of a production control device based on multiple production conditions provided in an embodiment of the present application;

[0034] Figure 5 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0035] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0036] The terms "first", "second", etc. in the specification and claims of this application and the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes steps or units that are not listed, or optionally includes other steps or units inherent to these processes, methods, products or devices.

[0037] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0038] First, the terms involved in the embodiments of the present application are explained.

[0039] Multiple working conditions: The definition of working conditions is the combination of the change range of characteristic values ​​of the input factors of the production process, such as production workers, production machines, production materials, production methods, production environment, etc. Different working condition combinations will have a significant impact on the control parameters of the production process. Multiple working conditions means that the production process is classified and managed according to different working conditions.

[0040] Working condition version: In the embodiment of the present application, the working condition version includes various process parameters in the production method of a certain product, such as the personnel who manufacture the product, the machines used to manufacture the product, the raw materials used to manufacture the product, the method used to manufacture the product, the environment in which the product is manufactured, etc. The working condition version at different stages of the method will be continuously updated according to various process parameters.

[0041] Operating condition benchmark value version scorecard: In the embodiment of the present application, the operating condition benchmark value version scorecard includes a process equipment scorecard, a component scorecard and a product scorecard, which are used to record the design parameters of the above operating condition version and have the function of updating the numerical value.

[0042] Design parameters: including product parameters, component parameters and process equipment parameters, among which product parameters are various parameters used to evaluate the quality of the final product after actual production; component parameters are characteristic parameters of the supplier's incoming materials or intermediate products generated during the production process; process equipment parameters are process indicators (such as operation time, feed amount) and equipment parameter indicators (such as temperature, pressure, liquid level, etc.) that control each production process flow.

[0043] Upper Spec Limit (USL) and Lower Spec Limit (LSL) refer to the maximum and minimum values ​​within which the parameter value is allowed to float.

[0044] Capability Index of Process (Cpk): refers to the degree to which process capability meets product quality standard requirements (specification range, etc.).

[0045] The embodiments of the present application are described below in conjunction with the accompanying drawings.

[0046] See also Figure 1 , Figure 1 A schematic diagram of an application scenario of a production control method based on multiple production conditions provided in an embodiment of the present application; the application scenario 100 includes a machine learning module 101, a condition benchmark value version scorecard 102, an evaluation module 103 and a sampling and filtering module 104, the machine learning module 101 is used to perform machine learning on the benchmark condition version to adjust the process equipment parameters of the benchmark condition version, obtain a theoretically optimal combination of process equipment parameters, and obtain several trained benchmark condition versions, the condition benchmark value version scorecard 102 is used to record the design parameters of several completed benchmark condition versions at each stage, the evaluation module 103 is used to evaluate and classify the benchmark condition version according to the data recorded in the condition benchmark value version scorecard 102 to select the final mass production version, and the sampling and filtering module 104 is used to dynamically update the mass production version of the benchmark condition according to the sampling and filtering algorithm during the mass production stage.

[0047] See also Figure 2 , Figure 2 A flow chart of a production control method based on multiple production conditions provided in an embodiment of the present application can be based on Figure 1 The application environment shown is implemented as Figure 2 As shown, steps S201-S205 are included:

[0048] S201: Obtain at least one benchmark operating condition version, where at least one benchmark operating condition version is an operating condition version inferred by a machine learning model.

[0049] In an embodiment of the present application, after a new benchmark operating condition version is created, a preset target value for the learning task is set for the new benchmark operating condition version, and a theoretically optimal combination of process equipment parameters such as production time, raw materials used to manufacture products, and methods used to manufacture products is obtained through machine learning model reasoning to achieve the target value. The operating condition version reasoned by the machine learning model here includes the case where the target value is achieved through machine learning, and also includes the case where the target value is not achieved after machine learning; at this time, the operating condition version that reaches the learning goal or completes the number of training times can be called the benchmark operating condition version, or some operating condition versions can be selected from it and called the benchmark operating condition version. The benchmark operating condition version is selected to enter the mass production verification stage.

[0050] In a possible embodiment, obtaining at least one benchmark operating condition version includes: obtaining a first operating condition version, the first operating condition version being a working condition version with the highest priority ranking among the working condition versions that achieve the learning goal after machine learning model reasoning; obtaining a second operating condition version, the second operating condition version being M working condition versions randomly selected from the top N working condition versions with the highest priority ranking except the working condition version with the highest priority ranking among the working condition versions that achieve the learning goal after machine learning model reasoning, N ≥ M; and using the first operating condition version and the second operating condition version as at least one benchmark operating condition version.

[0051] Specifically, when some of the working condition versions that have achieved the learning goal or completed the number of training times are selected as benchmark working condition versions, the process of specifically determining the benchmark working condition versions may include: sorting according to the degree of achievement of the learning goal, first determining the highest-ranked benchmark working condition version as one of the above-mentioned at least one benchmark working condition version; secondly, randomly selecting M benchmark working condition versions from the N benchmark working condition versions that are not the highest-ranked but ranked high to determine them together with the benchmark working condition version with the highest priority as the above-mentioned at least one benchmark working condition version. Where M≤N, for example, N=5, M=3.

[0052] In a possible embodiment, obtaining at least one benchmark operating condition version includes:

[0053] If the benchmark working condition version fails to reach the learning goal after being inferred by the machine learning model, the top T working condition versions with the highest priority will be used as at least one benchmark working condition version.

[0054] Specifically, if the benchmark operating condition version fails to reach the preset target value after being inferred by the machine learning model, the benchmark operating condition version that fails to reach the target value can no longer be optimized through model training. Then, the multiple benchmark operating condition versions that have completed machine learning are directly ranked according to the degree of achievement of the learning target. After obtaining the ranked list, the top T benchmark operating condition versions with a higher degree of achievement of the learning target are determined as at least one benchmark operating condition version, and T can be 3, 5 or 2.

[0055] In a possible embodiment, the priority ranking of the operating condition version is determined according to the degree of matching between the operating condition version and a preset operating condition version.

[0056] Specifically, the aforementioned method for sorting multiple benchmark working condition versions that have completed machine learning according to the degree of achievement of learning objectives is that the machine learning model adjusts the values ​​of other process equipment parameters according to preset target parameters such as production time, and the degree of match between the actual value of the benchmark working condition version that has completed machine training and the target value is finally obtained, and the multiple benchmark working condition versions that have completed machine learning are sorted according to the degree of match.

[0057] Possibly, the priority ranking of the benchmark operating condition version can also be completed according to a single or multiple process parameters in the benchmark operating condition version. The content of substance A can be determined as the product parameter with the highest priority for measuring product quality. The content of substance A can be used to prioritize the benchmark operating condition version, and the benchmark operating condition version with the best amount of substance A among all operating condition versions can be selected. The better the design parameters, the higher the priority ranking of the benchmark operating condition version.

[0058] It can be seen that in the example of the present application, by sorting the benchmark operating condition versions that have completed machine learning according to different conditions, several benchmark operating condition versions that can be used for subsequent steps are screened out, and benchmark operating condition versions for subsequent mass production verification steps are added. Before outputting the final operating condition version, the data of each version has an object for parallel comparison, which increases the number of operating condition versions involved in the verification, records a large number of process parameters, improves the degree of refinement of production, and thereby improves product quality.

[0059] S202: Generate a working condition benchmark value scorecard for each benchmark working condition version in at least one benchmark working condition version, where the working condition benchmark value scorecard is used to record design parameters of each benchmark working condition version.

[0060] In this embodiment, a corresponding working condition benchmark value scorecard is newly created for each benchmark working condition version, and the working condition benchmark value scorecard here includes a product scorecard, a component scorecard and a process equipment scorecard. Figure 3A As shown, Figure 3A A schematic diagram of a product scorecard provided in an embodiment of the present application, the product scorecard is used to record multiple design parameter indicators, these product indicators will reflect the product quality, thereby reflecting the performance and quality of the production process equipment and intermediate steps in actual production, wherein the upper specification limit USL and the lower specification limit LSL of the product parameter are set and recorded in the working condition benchmark scorecard when the parameter learning target value is set when the benchmark is newly created. The product parameter process capability index CPK is calculated by the parameter mean, parameter standard deviation and parameter specification upper and lower limits; Figure 3B As shown, Figure 3B A schematic diagram of a component scorecard provided in an embodiment of the present application, wherein the component scorecard is used to record multiple component parameter indicators, where the component refers to the characteristic parameters of the supplier's incoming materials or the intermediate products produced in the production process that may appear in a benchmark operating condition version, and is used to reflect the quality of the supplier's incoming materials, intermediate products, and intermediate steps; Figure 3C As shown, Figure 3CA schematic diagram of a process equipment scorecard provided in an embodiment of the present application; the process equipment scorecard is used to record parameters such as the equipment used for the product, the amount of raw materials used, and the raw material use environment. The process equipment scorecard has been determined in the learning stage, and the calculation method of the parameter upper specification limit USL in the component scorecard and the process equipment scorecard is the sum of the target value of the parameter and A times the standard deviation; the calculation method of the parameter lower specification limit LSL is the difference between the target value of the parameter and A times the standard deviation, where A is a positive number greater than zero, which can be 1, 2.5, 4, etc. The parameter process capability index CPK is calculated by the parameter mean, parameter standard deviation, and the upper and lower limits of the parameter specifications.

[0061] S203: Perform mass production verification on each benchmark operating condition version of at least one benchmark operating condition version, and update the operating condition benchmark value scorecard of each benchmark operating condition version according to the design parameters of the mass production verification.

[0062] In this embodiment, the mass production verification according to each benchmark working condition version in at least one benchmark working condition version refers to: according to the actual input generation of the production method and process parameters recorded in the benchmark working condition version, a batch limit for mass production verification can be set at this time, and the preset value of the batch limit can be adjusted according to the actual mass production verification requirements. For example, after obtaining three benchmark working condition versions that can be used for mass production verification, 30 batches of actual production are arranged for each benchmark working condition version, and the product parameters of the final product are collected after each production is completed, and the product score card in the working condition benchmark value score card of each benchmark working condition version is updated with the collected product parameters.

[0063] S204: Evaluate the availability of each benchmark operating condition version according to the operating condition benchmark value scorecard of each benchmark operating condition version, and determine to put a mass production version of at least one benchmark operating condition version into mass production.

[0064] In the embodiments of the present application, the obtained benchmark operating condition versions can theoretically be used in production, but it is still necessary to select the operating condition version that is most suitable for mass production. Therefore, when these benchmark operating condition versions are used for mass production verification, the scores of the corresponding operating condition benchmark value scorecard can be used to determine the availability of each benchmark operating condition version, and the mass production version finally selected for the mass production stage can be determined.

[0065] In a possible embodiment, the availability of each benchmark operating condition version is evaluated according to the operating condition benchmark value scorecard of each benchmark operating condition version, including: the operating condition benchmark value scorecard of each benchmark operating condition version updates the batch data of the product parameters of the corresponding benchmark operating condition version according to the mass production verification results; the upper and lower limits of the product parameter specifications of the operating condition benchmark value scorecard are obtained; the product parameter mean and the product parameter standard deviation are calculated according to the batch data of the product parameters; the process capability index CPK of the product parameters of each benchmark operating condition version is calculated according to the product parameter mean, the product parameter standard deviation and the upper and lower limits of the product parameter specifications; and the availability of each benchmark operating condition version is evaluated according to the CPK value.

[0066] Specifically, after each benchmark operating condition version completes the actual production of a batch, the product scorecard updates the batch data of the benchmark operating condition version, calculates the parameter mean and product parameter standard deviation of the product parameters of the operating condition version; calculates the process capability index CPK of the batch product parameters through the product parameter mean, product parameter standard deviation, and upper and lower limits of product parameter specifications and records it in the product scorecard, where the product parameter mean μ is the mean of the data generated by a certain parameter in all batches currently produced; the product parameter standard deviation σ is the product parameter standard deviation of all batches; the upper limit of the product parameter specification here is the benchmark When setting the target value of learning when creating a new learning target, the working condition version is set and recorded in the working condition benchmark scorecard at the same time, which can be directly called; process capability index CPK = min{Cpl, Cpu}, where Cpl = (μ-LSL) / 3σ, Cpu = (USL-μ) / 3σ. After all batches of this benchmark working condition version are produced, the CPK average of all batches is recorded as the CPK data of this benchmark working condition version. The availability of each benchmark working condition version is evaluated according to the CPK value, and the benchmark working condition version of the mass production stage is selected according to the evaluation results for actual production. The upper and lower limits of the parameter specifications here can also be used to exclude erroneous data according to the upper and lower limits of the parameter specifications. In the production process, parameters, and other input elements that have been determined for mass production verification, the obtained design parameters will only fluctuate between the upper and lower limits of the specifications. In the case of possible human errors or machine failures, erroneous design parameters that are not within the range of the upper and lower limits of the specifications will be generated. In this case, the production batch is marked as an erroneous batch to exclude the erroneous parameters of the production batch.

[0067] In a possible embodiment, the method of updating the working condition benchmark value scorecard according to the design parameters of the benchmark working condition also includes: performing cluster analysis on the collected design parameters through DBSCAN (Density-Based Spatial Clustering of Applications with Noise); dividing the same production condition parameters into multiple clusters according to preset parameters; including the cluster with the largest number of parameters as the design parameters in the working condition benchmark value scorecard, and defining other clusters as noise and excluding them.

[0068] Specifically, in the actual mass production process, the production conditions of each production batch cannot be completely consistent, and the obtained design parameters will also have slight differences. If there are unexpected situations such as operator errors, sudden changes in the production environment, and production equipment failures, abnormal production conditions will occur, and the obtained design parameters will be greatly different from the normal values. The number of production batches used for mass production verification is not large, and the density of the obtained parameter sample set is relatively uniform, and the cluster spacing difference is not much different. At this time, the DBSCAN algorithm can be used to detect unexpected situations such as equipment failure and abnormal intermediate products. Therefore, the design parameters of the benchmark working condition version of this production batch are excluded, and only normal design parameters are recorded.

[0069] In the embodiment of the present application, the DBSCAN algorithm is more suitable for processing the benchmark operating condition version. It can accurately distinguish abnormal data belonging to outliers when there is not much data to be processed, exclude the abnormal data, improve the credibility of the data recorded in the operating condition benchmark value scorecard, and reduce the problem of inaccurate evaluation results caused by abnormal data.

[0070] In a possible embodiment, the availability of each benchmark operating condition version is evaluated according to the product parameter CPK value, including: if the product parameter CPK of the first target operating condition version in at least one benchmark operating condition version is not greater than the first preset value, the first target operating condition version is determined to be a version unavailable for mass production; if the product parameter CPK of the second target operating condition version in at least one benchmark operating condition version is greater than the first preset value, the second target operating condition version is determined to be a version available for mass production; if the product parameter CPK of the third target operating condition version in at least one benchmark operating condition version is greater than the second preset value, the current product is determined to be the optimal version for mass production.

[0071] Specifically, as shown in Table 1, the CPK value here specifically refers to the CPK value of the product parameter. The CPK preset value of the product parameter can be determined according to actual needs. The first preset value can be set to 1.0, and the second preset value can be set to 1.33. The process capability index CPK refers to the degree to which the process capability meets the product quality standard requirements (specification range, etc.), also known as the process capability index, which refers to the actual processing capability of the process under a controlled state (stable state) for a certain period of time. When the CPK of at least one product parameter of the benchmark process version is not greater than 1.0, it means that the process capability of the benchmark process version needs to be improved, and in serious cases, production needs to be stopped for rectification, then the benchmark process version is determined as a version unavailable for mass production; when the CPK of all product parameters of the benchmark process version are greater than 1.0, and the CPK of at least one product parameter is less than 1.33, it means that the process capability of the benchmark process version is within the normal range, then the benchmark process version is a version available for mass production; if the CPK values ​​of all product parameters of the third target operating condition version in at least one benchmark operating condition version are greater than 1.33, it means that the process capability of the benchmark process version is good, then the benchmark process version is confirmed to be the optimal version for mass production.

[0072] Table 1

[0073]

[0074] In a possible embodiment, determining a mass production version in at least one benchmark operating condition version includes: if it is determined that at least one benchmark operating condition version includes a unique mass production optimal version, determining the unique mass production optimal version as the mass production version in the at least one benchmark operating condition version; and / or if it is determined that at least one benchmark operating condition version includes multiple mass production optimal versions, determining a sorting according to parameter priorities and parameter values ​​corresponding to each of the multiple mass production optimal versions in a production data table, and determining the mass production version according to the sorting result, wherein the production data table consists of design parameters in a working condition benchmark value scorecard; and / or if it is determined that at least one benchmark operating condition version does not include a unique mass production optimal version and includes at least one mass production available version, determining a sorting according to parameter priorities and parameter values ​​corresponding to each of the multiple mass production available versions in the production data table, and determining the mass production version according to the sorting result.

[0075] Specifically, the CPK calculation results of the product parameters of multiple benchmark operating condition versions put into mass production verification may be different. If the CPK of the product parameters of only one benchmark operating condition version among all the multiple benchmark operating condition versions put into mass production verification exceeds 1.33, then the benchmark operating condition version can be determined as the final mass production version; if the CPK of the product parameters of more than one benchmark operating condition version among all the multiple benchmark operating condition versions put into mass production verification exceeds 1.3, the CPK of the single product parameters of the multiple benchmark operating condition versions are compared, and the product parameters are sorted into high, medium and low priorities. The priority of the product parameters here can be set according to actual needs, and the multiple versions are sorted according to the CPK value of the product parameter with the highest priority, and the benchmark operating condition version ranked first is determined as the final mass production version. If among all benchmark operating condition versions there is only a mass production version in which the CPK of all product parameters is greater than 1.0 and the CPK of at least one product parameter is less than 1.33, similarly, the sorting can be determined based on the priority of the product parameters corresponding to each version in the multiple mass production available versions in the production data table and the CPK value of each product parameter, and the final mass production version can be determined based on the sorting result.

[0076] It can be seen that in the embodiment of the present application, multiple operating condition versions are subjected to mass production verification, and multiple design parameters obtained from multiple mass production batches are recorded in an operating condition benchmark value scorecard. The version is determined by the corresponding training results under each operating condition, and the most suitable operating condition version for mass production is recommended according to the version type, so that there is a scientific basis for judging the production quality.

[0077] S205: According to the design parameters generated by the benchmark operating condition version in the mass production stage, the mass production version of the benchmark operating condition is dynamically updated through a sampling and filtering algorithm and put into mass production.

[0078] In the embodiment of the present application, after the mass production version of the benchmark operating condition is actually put into production, the mass production version of the benchmark operating condition is dynamically updated through a sampling filtering algorithm. The sampling and filtering algorithm here includes: obtaining the mass production version of the product scorecard of the benchmark working condition for X batches in the actual mass production stage, and calculating the CPK values ​​of the product parameters of the mass production version of the benchmark working condition for X batches; when the CPK values ​​of all product parameters are greater than the preset threshold, determining the upper and lower limits of the parameter specifications according to the mass production version of the product scorecard of the benchmark working condition for X batches; the upper and lower limits of the specifications are used, when the product parameters obtained from a certain mass production batch in the mass production stage exceed the upper and lower limits of the specifications, it can be determined that unexpected conditions such as equipment abnormalities or intermediate product abnormalities that may exist in the production process have caused the product parameters obtained from the production to exceed the upper and lower limits of the specifications, so the products of the mass production batch are excluded, and the mass production version of the benchmark working condition is updated, and the upper and lower limits of the specifications of the corresponding process parameters and component parameters are calculated according to the process equipment scorecard and component scorecard corresponding to the filtered samples, and the upper and lower limits of the specifications of the parameters are used to control the process quality. The calculation method of the CPK value here can refer to the relevant content of step S204 and will not be repeated here. Here, X is a positive integer, such as 20, 30, etc., and the preset threshold here can be 1.33, 1.0, etc.

[0079] In a possible embodiment, if the CPK values ​​of X batches of the mass production version of the benchmark working condition do not exceed the preset threshold, the CPK values ​​of the product parameters of another X batches are calculated. If the CPK values ​​of the other X batches of products still do not exceed the preset threshold, the upper and lower limits of the specifications are calculated according to the product scorecard of the mass production version of the benchmark working condition of all current batches, and the mass production version of the benchmark working condition is updated according to the obtained upper and lower limits of the specifications.

[0080] In the example of the present application, a mass production version of a product scorecard of a benchmark operating condition of X mass production batches whose CPK of a product parameter is greater than a preset threshold is collected, and the obtained upper and lower limits of specifications are used as filtering conditions. The upper and lower limits of specifications of the process equipment parameters and component parameters corresponding to the process equipment scorecard and component scorecard corresponding to the filtered samples are used, and the upper and lower limits of specifications of the parameters are used to perform process quality control, so that equipment anomalies and intermediate product anomalies are discovered in time and corresponding improvement measures are taken to ensure the yield of product production.

[0081] By implementing the method of the embodiment of the present invention, multiple working condition versions are verified for mass production, and the working condition benchmark value scorecard is updated according to the design parameters obtained from the mass production verification, and the product parameter CPK value of each working condition version is calculated. The version of the corresponding training results under each working condition is determined, and the working condition version that is most suitable for mass production is recommended, so that there is a scientific basis for judging the production quality. Through the method of mass production verification of the working condition version, the optimal mass production working condition version is selected, which ensures that the setting of the production process parameters is always in the optimal state, and the mass production version of the benchmark working condition is dynamically updated through the sampling and filtering algorithm, and the quality of the mass-produced products is controlled, thereby ensuring the optimal product yield.

[0082] Based on the description of the configuration method embodiment, the present application further provides a production control device 400 based on multiple production conditions, which can be a computer program (including program code) running in a terminal. The production control device 400 based on multiple production conditions can execute Figure 1 , Figure 2 , Figure 3A , Figure 3B and Figure 3C See the method shown in Figure 4 , the device comprises:

[0083] Acquisition unit 401: used to acquire at least one benchmark operating condition version, where at least one benchmark operating condition version is an operating condition version inferred by a machine learning model;

[0084] The recording unit 402 is used to generate a working condition benchmark value scorecard for each benchmark working condition version in at least one benchmark working condition version, where the working condition benchmark value scorecard is used to record the design parameters of each benchmark working condition version;

[0085] Performing mass production verification on each benchmark operating condition version of at least one benchmark operating condition version, and updating the operating condition benchmark value scorecard of each benchmark operating condition version according to the design parameters of the mass production verification;

[0086] Evaluation unit 403 is used to evaluate the availability of each benchmark operating condition version according to the operating condition benchmark value scorecard of each benchmark operating condition version, and determine a mass production version in at least one benchmark operating condition version.

[0087] In a possible embodiment, in terms of obtaining the benchmark operating condition version, the obtaining unit is further specifically used to execute instructions of the following steps:

[0088] Obtain a first working condition version, where the first working condition version is the working condition version with the highest priority among the working condition versions that achieve the learning goal after machine learning model reasoning; obtain a second working condition version, where the second working condition version is the working condition version that achieves the learning goal after machine learning model reasoning, and M working condition versions are randomly selected from the top N working condition versions with the highest priority, excluding the working condition version with the highest priority, where N≥M; use the first working condition version and the second working condition version as at least one benchmark working condition version.

[0089] In a possible embodiment, in terms of obtaining the benchmark operating condition version, the obtaining unit is further specifically used to execute instructions of the following steps:

[0090] If the benchmark working condition version fails to reach the learning goal after being inferred by the machine learning model, the top T working condition versions with the highest priority will be used as at least one benchmark working condition version.

[0091] In a possible embodiment, the priority ranking of the operating condition versions is determined according to whether the design parameters in the operating condition versions reach the target values.

[0092] In a possible embodiment, in terms of evaluating the availability of each benchmark operating condition version according to the operating condition benchmark value scorecard of each benchmark operating condition version, the recording unit is further used to execute instructions of the following steps:

[0093] The working condition benchmark value scorecard of each benchmark working condition version updates the batch data of the product parameters of the corresponding benchmark working condition version according to the mass production verification results; obtains the upper and lower limits of the product parameter specifications of the working condition benchmark value scorecard; calculates the product parameter mean and product parameter standard deviation based on the batch data of the product parameters; calculates the process capability index CPK of the product parameters of each benchmark working condition version based on the product parameter mean, product parameter standard deviation and the upper and lower limits of the product parameter specifications; and evaluates the availability of each benchmark working condition version based on the CPK value.

[0094] In a possible embodiment, in terms of evaluating the availability of each benchmark operating condition version according to the CPK value of the product parameter, the evaluation unit is further used to execute instructions for the following steps:

[0095] If the CPK of the product parameters of the first target operating condition version in at least one benchmark operating condition version is not greater than the first preset value, the first target operating condition version is determined to be a version unavailable for mass production; if the CPK of the product parameters of the second target operating condition version in at least one benchmark operating condition version is greater than the first preset value, the second target operating condition version is determined to be a version available for mass production; if the CPK of the product parameters of the third target operating condition version in at least one benchmark operating condition version is greater than the second preset value, the current product is determined to be the optimal version for mass production.

[0096] In a possible embodiment, in determining the mass production version in at least one benchmark operating condition version, the evaluation unit is further configured to execute instructions of the following steps:

[0097] If it is determined that at least one benchmark operating condition version includes a unique mass production optimal version, the unique mass production optimal version is determined to be the mass production version in the at least one benchmark operating condition version; and / or if it is determined that at least one benchmark operating condition version includes multiple mass production optimal versions, the sorting is determined according to the parameter priority and parameter value corresponding to each version in the multiple mass production optimal versions in the production data table, and the mass production version is determined according to the sorting result, wherein the production data table consists of the design parameters in the operating condition benchmark value scorecard; and / or if it is determined that at least one benchmark operating condition version does not include the unique mass production optimal version and includes at least one mass production available version, the sorting is determined according to the parameter priority and parameter value corresponding to each version in the multiple mass production available versions in the production data table, and the mass production version is determined according to the sorting result.

[0098] Based on the description of the above method embodiment and device embodiment, please refer to Figure 5 , Figure 5 Schematic diagram of the structure of an electronic device 500 provided in an embodiment of the present application. The electronic device 500 described in this embodiment is as follows: Figure 5 As shown, the electronic device 500 includes a processor 501, a memory 502, a communication interface 503, and one or more programs. The one or more programs are stored in the memory in the form of application code and are configured to be executed by the processor. In the embodiment of the present application, the program includes instructions for executing the following steps:

[0099] Obtain at least one benchmark operating condition version, where at least one benchmark operating condition version is an operating condition version inferred by a machine learning model;

[0100] Generate a working condition benchmark value scorecard for each benchmark working condition version in at least one benchmark working condition version, where the working condition benchmark value scorecard is used to record the design parameters of each benchmark working condition version;

[0101] Performing mass production verification on each benchmark operating condition version of at least one benchmark operating condition version, and updating the operating condition benchmark value scorecard of each benchmark operating condition version according to the design parameters of the mass production verification;

[0102] The availability of each benchmark operating condition version is evaluated according to the operating condition benchmark value scorecard of each benchmark operating condition version, and a mass production version of at least one benchmark operating condition version is determined.

[0103] In a possible embodiment, obtaining at least one benchmark operating condition version includes:

[0104] Obtain a first working condition version, where the first working condition version is the working condition version with the highest priority among the working condition versions that achieve the learning goal after machine learning model reasoning; obtain a second working condition version, where the second working condition version is the working condition version that achieves the learning goal after machine learning model reasoning, and M working condition versions are randomly selected from the top N working condition versions with the highest priority, excluding the working condition version with the highest priority, where N≥M; use the first working condition version and the second working condition version as at least one benchmark working condition version.

[0105] In a possible embodiment, obtaining at least one benchmark operating condition version includes:

[0106] If the benchmark working condition version fails to reach the learning goal after being inferred by the machine learning model, the top T working condition versions with the highest priority will be used as at least one benchmark working condition version.

[0107] In a possible embodiment, the priority ranking of the operating condition versions is determined according to whether the design parameters in the operating condition versions reach the target values.

[0108] In a possible embodiment, the availability of each benchmark operating condition version is evaluated according to the operating condition benchmark value scorecard of each benchmark operating condition version, including: the operating condition benchmark value scorecard of each benchmark operating condition version updates the batch data of the product parameters of the corresponding benchmark operating condition version according to the mass production verification results; the upper and lower limits of the product parameter specifications of the operating condition benchmark value scorecard are obtained; the product parameter mean and the product parameter standard deviation are calculated according to the batch data of the product parameters; the process capability index CPK of the product parameters of each benchmark operating condition version is calculated according to the product parameter mean, the product parameter standard deviation and the upper and lower limits of the product parameter specifications; and the availability of each benchmark operating condition version is evaluated according to the CPK value.

[0109] In a possible embodiment, the availability of each benchmark operating condition version is evaluated according to the CPK value of the product parameter, including: if the CPK of the product parameter of the first target operating condition version in at least one benchmark operating condition version is not greater than the first preset value, then the first target operating condition version is determined to be a version unavailable for mass production; if the CPK of the product parameter of the second target operating condition version in at least one benchmark operating condition version is greater than the first preset value, then the second target operating condition version is determined to be a version available for mass production; if the CPK of the product parameter of the third target operating condition version in at least one benchmark operating condition version is greater than the second preset value, then the current product is determined to be the optimal version for mass production.

[0110] In a possible embodiment, determining a mass production version in at least one benchmark operating condition version includes: if it is determined that at least one benchmark operating condition version includes a unique mass production optimal version, determining the unique mass production optimal version as the mass production version in the at least one benchmark operating condition version; and / or if it is determined that at least one benchmark operating condition version includes multiple mass production optimal versions, determining a sorting according to parameter priorities and parameter values ​​corresponding to each of the multiple mass production optimal versions in a production data table, and determining the mass production version according to the sorting result, wherein the production data table consists of design parameters in a working condition benchmark value scorecard; and / or if it is determined that at least one benchmark operating condition version does not include a unique mass production optimal version and includes at least one mass production available version, determining a sorting according to parameter priorities and parameter values ​​corresponding to each of the multiple mass production available versions in the production data table, and determining the mass production version according to the sorting result.

[0111] Exemplarily, the electronic device may include but is not limited to a processor, a memory, a communication interface, and one or more programs, and may also include a memory, a power supply, an application client module, etc. Those skilled in the art will appreciate that the schematic diagram is merely an example of an electronic device and does not constitute a limitation on the electronic device, and may include more or fewer components than shown in the figure, or combine certain components, or different components.

[0112] The embodiment of the present application also provides a computer storage medium (Memory), which is a memory device in an information processing device, an information sending device, or an information receiving device, for storing programs and data. It is understandable that the computer storage medium here can include both the built-in storage medium in the terminal and the extended storage medium supported by the terminal. The computer storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by the processor are also stored in the storage space, and these instructions can be one or more computer programs (including program codes). It should be noted that the computer storage medium here can be a high-speed RAM memory, or a non-volatile memory (Non-volatile Memory), such as at least one disk memory; optionally, it can also be at least one computer storage medium located away from the aforementioned processor. In one embodiment, the processor can load and execute one or more instructions stored in the computer storage medium to implement the corresponding steps of the above-mentioned production control method based on multiple production conditions.

[0113] The embodiments of the present application are introduced in detail above. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method and core idea of ​​the present application. At the same time, for general technical personnel in this field, according to the idea of ​​the present application, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A production control method based on multiple production conditions, characterized in that: include: Obtain at least one benchmark operating condition version, wherein the at least one benchmark operating condition version is an operating condition version inferred by a machine learning model, specifically comprising: obtaining a first operating condition version, wherein the first operating condition version is an operating condition version with the highest priority among the operating condition versions that have achieved the learning goal through the machine learning model inference; obtaining a second operating condition version, wherein the second operating condition version is M operating condition versions randomly selected from the top N operating condition versions with the highest priority, except for the operating condition version with the highest priority among the operating condition versions that have achieved the learning goal through the machine learning model inference, where N≥M; using the first operating condition version and the second operating condition version as the at least one benchmark operating condition version; Generating a working condition benchmark value scorecard for each benchmark working condition version in the at least one benchmark working condition version, wherein the working condition benchmark value scorecard is used to record the design parameters of each benchmark working condition version; Performing mass production verification according to each benchmark operating condition version of the at least one benchmark operating condition version, and updating the operating condition benchmark value scorecard of each benchmark operating condition version according to the design parameters of the mass production verification; The availability of each benchmark operating condition version is evaluated according to the operating condition benchmark value scorecard of each benchmark operating condition version after parameter update, and the mass production version of at least one benchmark operating condition version is determined, specifically including: the operating condition benchmark value scorecard of each benchmark operating condition version updates the batch data of the product parameters of the corresponding benchmark operating condition version according to the mass production verification result; the upper and lower limits of the product parameter specifications of the operating condition benchmark value scorecard are obtained; the mean value and the standard deviation of the product parameters are calculated according to the batch data of the product parameters; the process capability index of the product parameters of each benchmark operating condition version is calculated according to the mean value, the standard deviation of the product parameters and the upper and lower limits of the product parameter specifications; CPK value; the availability of each benchmark operating condition version is evaluated according to the CPK value of the product parameter, specifically including: if the CPK of the product parameter of the first target operating condition version in the at least one benchmark operating condition version is not greater than the first preset value, then the first target operating condition version is determined to be a version unavailable for mass production; if the CPK of the product parameter of the second target operating condition version in the at least one benchmark operating condition version is greater than the first preset value, then the second target operating condition version is determined to be a version available for mass production; if the CPK of the product parameter of the third target operating condition version in the at least one benchmark operating condition version is greater than the second preset value, then the current product is determined to be the optimal version for mass production.

2. The method according to claim 1, characterized in that The obtaining of at least one benchmark operating condition version includes: If the benchmark operating condition version fails to reach the learning objective after being inferred by the machine learning model, the T operating condition versions with the highest priority rankings are used as the at least one benchmark operating condition version.

3. The method according to claim 2, characterized in that The priority ranking of the operating condition versions is determined according to whether the design parameters in the operating condition versions reach the target values.

4. The method according to claim 1, characterized in that The determining of the mass production version of the at least one benchmark operating condition version includes: If it is determined that the at least one benchmark operating condition version includes a unique mass-produced optimal version, then determining the unique mass-produced optimal version to be the mass-produced version in the at least one benchmark operating condition version; and / or If it is determined that the at least one benchmark operating condition version includes multiple mass production optimal versions, then determining the sorting according to the parameter priority and parameter value corresponding to each version in the multiple mass production optimal versions in the production data table, and determining the mass production version according to the sorting result, wherein the production data table is composed of the design parameters in the operating condition benchmark value scorecard; and / or If it is determined that the at least one benchmark operating condition version does not include the only optimal version for mass production, and includes at least one available version for mass production, the sorting is determined according to the parameter priority and parameter value corresponding to each version in the multiple available versions for mass production in the production data table, and the mass production version is determined according to the sorting result.

5. A production control device based on multiple production conditions, the device is used to execute a production control method based on multiple production conditions, characterized in that include: An acquisition unit: used to acquire at least one benchmark operating condition version, wherein the at least one benchmark operating condition version is an operating condition version that has been adjusted through machine learning model reasoning and has ended the learning phase, specifically comprising: acquiring a first operating condition version, wherein the first operating condition version is an operating condition version with the highest priority ranking among the operating condition versions that have reached the learning target through the machine learning model reasoning; acquiring a second operating condition version, wherein the second operating condition version is M operating condition versions randomly selected from the top N operating condition versions with the highest priority ranking, except for the operating condition version with the highest priority ranking among the operating condition versions that have reached the learning target through the machine learning model reasoning, where N≥M; and using the first operating condition version and the second operating condition version as the at least one benchmark operating condition version; A recording unit: used for generating a working condition benchmark value scorecard for each benchmark working condition version in the at least one benchmark working condition version, wherein the working condition benchmark value scorecard is used for recording design parameters of each benchmark working condition version; Performing mass production verification according to each benchmark operating condition version of the at least one benchmark operating condition version, and updating the operating condition benchmark value scorecard of each benchmark operating condition version according to the design parameters of the mass production verification; Evaluation unit: used to evaluate the availability of each benchmark operating condition version according to the operating condition benchmark value scorecard of each benchmark operating condition version after parameter update, and determine the mass production version of at least one benchmark operating condition version, specifically including: updating the operating condition benchmark value scorecard of each benchmark operating condition version with the batch data of the product parameters of the corresponding benchmark operating condition version according to the mass production verification result; obtaining the upper and lower limits of the product parameter specifications of the operating condition benchmark value scorecard; calculating the product parameter mean and the product parameter standard deviation according to the batch data of the product parameters; and calculating the process of obtaining the product parameters of each benchmark operating condition version according to the product parameter mean, the product parameter standard deviation and the upper and lower limits of the product parameter specifications. Capability index CPK; evaluating the availability of each benchmark operating condition version according to the CPK value of the product parameter, specifically including: if the CPK of the product parameter of the first target operating condition version in the at least one benchmark operating condition version is not greater than the first preset value, determining that the first target operating condition version is a version unavailable for mass production; if the CPK of the product parameter of the second target operating condition version in the at least one benchmark operating condition version is greater than the first preset value, determining that the second target operating condition version is a version available for mass production; if the CPK of the product parameter of the third target operating condition version in the at least one benchmark operating condition version is greater than the second preset value, determining that the current product is the optimal version for mass production.

6. An electronic device, characterized in that: The method comprises a processor, a memory, a communication interface, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the processor, and the programs include instructions for executing the steps in the method according to any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program for electronic data exchange, wherein the computer program enables a computer to execute the method according to any one of claims 1 to 4.

Images