Cross-process defect root cause tracing method and system

Abstract

The invention relates to the technical field of defect detection, in particular to a cross-process defect root cause tracing method and system. According to the method, the data feature matrix covering multiple dimensions is formed by integrating the process parameters, the equipment state and the quality detection information, so that the performance evaluation of each process is more comprehensive, the interaction and influence paths among the processes can be clearly described by constructing the process relation graph, and the performance evaluation efficiency is improved. Meanwhile, basic data support is provided for quantifying the relation between the procedures through introduction of procedure attenuation factors, the shortest propagation path and the propagation probability of the defects can be accurately recognized by analyzing a procedure relation graph, root cause tracing of the cross-procedure defects becomes systematized in combination with construction of a knowledge graph, and the defect tracing efficiency is improved. The knowledge graph not only can effectively integrate and display data, but also is convenient for quickly positioning problems, and by utilizing an adaptive correlation analysis technology, the system can intelligently adjust an analysis model and a path and continuously optimize a defect detection and tracing process when facing new data.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of defect detection, in particular to a cross-process defect root cause tracing method and system. BACKGROUND

[0002] Cross-process defects refer to defects generated in a certain process that affect the quality of subsequent processes, even the final product. Such defects are not generated in a single process, but are transmitted between multiple processes, thereby affecting the quality of the entire production process. This is because the output of each process is usually the input of the next process. If there is a defect in the previous process, the subsequent process will be difficult to avoid being affected.

[0003] Currently, the cross-process defect root cause tracing method in the prior art is to monitor the quality data in the production process in real time by statistical process control method, and to use statistical methods to find fluctuations and abnormalities in the production process, and then to analyze the root cause of cross-process quality problems. However, statistical process control method is usually used to statistically analyze the quality data of a single process. For the complex interaction between multiple processes, it may not have enough ability to reveal the source of defects. In the root cause tracing of cross-process defects, the correlation between different processes is very complex, and it is difficult to effectively trace the root cause of cross-process defects by simply relying on SPC to analyze data fluctuations in a single process. SUMMARY

[0004] The main purpose of the present application is to provide a cross-process defect root cause tracing method to solve the technical problems in the prior art.

[0005] The present application provides a cross-process defect root cause tracing method, comprising:

[0006] Obtaining full-process chain data of a target product, and obtaining production quality characteristic parameters of each process according to the full-process chain data, wherein the production quality characteristic parameters include process parameter information, equipment state information and quality detection information;

[0007] Obtaining process data characteristic matrix of the corresponding process according to each of the process parameter information, equipment state information and quality detection information;

[0008] Obtaining process data characteristic matrix of the corresponding process according to each of the process parameter information, equipment state information and quality detection information;

[0009] Constructing a process relationship diagram according to each process and cross-process correlation strength, and obtaining defect propagation data according to the process relationship diagram, wherein the defect propagation data includes the shortest propagation path, defect propagation probability and defect type;

[0010] According to the full-process-chain data, the plurality of cross-process correlation strengths, the shortest propagation path, the defect propagation probability, and the defect type, a knowledge graph is constructed, and cross-process root cause tracing is performed according to the knowledge graph.

[0011] Preferably, the step of obtaining a process data feature matrix of a corresponding process according to each of the process parameter information, the equipment state information, and the quality detection information comprises:

[0012] According to the process parameter information, a plurality of process temperature data, process pressure data, and process speed data are obtained, and according to each of the process temperature data, the process pressure data, and the process speed data, a corresponding temperature skewness coefficient, a pressure skewness coefficient, and a speed skewness coefficient are obtained;

[0013] According to each of the temperature skewness coefficient, the pressure skewness coefficient, and the speed skewness coefficient, a corresponding process stability feature vector is obtained;

[0014] According to the equipment state information, a plurality of equipment efficiency data and equipment load data are obtained, and according to each of the equipment efficiency data and the equipment load data, a corresponding efficiency skewness coefficient and a load skewness coefficient are obtained;

[0015] According to each of the efficiency skewness coefficient and the load skewness coefficient, a corresponding equipment stability feature vector is obtained;

[0016] According to the quality detection information, a plurality of size deviation data, surface quality data, and defect area data are obtained, and according to each of the size deviation data, the surface quality data, and the defect area data, a corresponding size skewness coefficient, a quality skewness coefficient, and a defect skewness coefficient are obtained;

[0017] According to each of the size skewness coefficient, the quality skewness coefficient, and the defect skewness coefficient, a corresponding quality stability feature vector is obtained;

[0018] Each of the quality stability feature vector, the equipment stability feature vector, and the process stability feature vector is normalized to obtain a corresponding standard quality stability feature vector, a standard equipment stability feature vector, and a standard process stability feature vector;

[0019] According to a plurality of the standard quality stability feature vectors, the standard equipment stability feature vectors, and the standard process stability feature vectors, a process data feature matrix of a corresponding process is constructed.

[0020] Preferably, the step of obtaining a process data feature matrix of a corresponding process according to each of the process parameter information, the equipment state information, and the quality detection information comprises:

[0021] According to the difference values of the corresponding features in the process data feature matrices of any two processes, a difference matrix is obtained, and the probability of each difference value is obtained according to the difference matrix;

[0022] According to the probability of the plurality of difference values, the frequency of each difference value is obtained, and according to the frequency of the plurality of difference values, the quality index entropy value is obtained;

[0023] According to the difference matrix, the mean value and the maximum difference value are obtained, and according to the mean value and the maximum difference value, the equipment health degree index is obtained;

[0024] According to the partial derivative of the corresponding feature in the process data feature matrix of any two processes, the process influence feature matrix is obtained, and according to the process influence feature matrix, the process complexity is obtained;

[0025] According to the process complexity, the equipment health degree index and the quality index entropy value, the process attenuation factor is calculated, and the calculation formula is:

[0026]

[0027] Wherein, G(SJ) represents the process attenuation factor, Z(BS) represents the quality index entropy value, G(FZ) represents the process complexity, and S(JZ) represents the equipment health degree index.

[0028] Preferably, the step of obtaining the cross-process correlation strength according to the process attenuation factor comprises:

[0029] According to the process data feature matrix of the first process, the first installation layout data and the parameter fluctuation amount are obtained, and according to the process data feature matrix of the second process, the quality fluctuation amount corresponding to the parameter fluctuation amount is obtained;

[0030] According to the quality fluctuation amount and the parameter fluctuation amount, the parameter influence coefficient corresponding to any two processes is obtained;

[0031] According to the first installation layout data, the first product installation position data and the first product layout data are obtained;

[0032] According to the process data feature matrix of the second process, the second installation layout data is obtained, and according to the second installation layout data, the second product installation position data and the second product layout data are obtained;

[0033] According to the first product installation position data and the second product installation position data, the installation position correlation coefficient is obtained, and according to the first product layout data and the second product layout data, the product layout correlation coefficient is obtained;

[0034] The process attenuation factor is obtained, and according to the process attenuation factor, the parameter influence coefficient, the product layout correlation coefficient and the installation position correlation coefficient, the cross-process correlation strength is calculated, and the calculation formula is:

[0035] K(GQ)=D(SY)*e -G(SJ)*[1-B(JX)*A(WX)] ;

[0036] wherein K(GQ) represents a cross-process correlation strength, G(SJ) represents a process attenuation factor, D(SY) represents a parameter influence coefficient, B(JX) represents a product layout correlation coefficient, and A(WX) represents an installation position correlation coefficient.

[0037] Preferably, the step of constructing a process relationship graph according to each process and cross-process correlation strength, and obtaining defect propagation data according to the process relationship graph comprises:

[0038] all cross-process correlation strengths of each process are obtained, and the maximum cross-process correlation strength is selected as a dependent correlation strength;

[0039] a most dependent process is determined according to each dependent correlation strength and a corresponding process, and a process dependency relationship is obtained;

[0040] each process is taken as a node, the process dependency relationship is taken as an edge, and a corresponding cross-process correlation strength is taken as an edge weight, and a process relationship graph is constructed;

[0041] all propagation paths of each process are obtained according to edges of the process relationship graph, and a shortest propagation path is determined according to the all propagation paths;

[0042] a historical defect probability of each process is obtained, and a propagation probability is obtained according to the historical defect probability and the shortest propagation path;

[0043] a defect propagation probability is obtained according to the propagation probability and the historical defect probability.

[0044] Preferably, the construction module comprises:

[0045] a selection unit configured to obtain all cross-process correlation strengths of each process, and select the maximum cross-process correlation strength as a dependent correlation strength;

[0046] a first determination unit configured to determine a most dependent process according to each dependent correlation strength and a corresponding process, and obtain a process dependency relationship;

[0047] a construction unit configured to take each process as a node, take the process dependency relationship as an edge, and take a corresponding cross-process correlation strength as an edge weight, and construct a process relationship graph;

[0048] a second determination unit configured to obtain all propagation paths of each process according to edges of the process relationship graph, and determine a shortest propagation path according to the all propagation paths;

[0049] a first obtaining unit configured to obtain a historical defect probability of each process, and obtain a propagation probability according to the historical defect probability and the shortest propagation path;

[0050] A second acquisition unit is configured to acquire a defect propagation probability according to the propagation probability and the historical defect probability.

[0051] The application also provides a cross-process defect root cause tracing system, comprising:

[0052] A first acquisition module is configured to acquire full-process chain data of a target product and acquire production quality characteristic parameters of each process according to the full-process chain data, wherein the production quality characteristic parameters comprise process parameter information, equipment state information and quality detection information.

[0053] A second acquisition module is configured to acquire process data characteristic matrices of corresponding processes according to each of the process parameter information, the equipment state information and the quality detection information.

[0054] A third acquisition module is configured to acquire process attenuation factors of corresponding processes according to process data characteristic matrices of any two processes and acquire cross-process correlation strengths according to the process attenuation factors.

[0055] A construction module is configured to construct a process relationship graph according to each process and the cross-process correlation strengths and acquire defect propagation data according to the process relationship graph, wherein the defect propagation data comprises shortest propagation paths, defect propagation probabilities and defect types.

[0056] A tracing module is configured to construct a knowledge graph according to the full-process chain data, the cross-process correlation strengths, the shortest propagation paths, the defect propagation probabilities and the defect types and perform cross-process root cause tracing according to the knowledge graph.

[0057] Preferably, the step of performing cross-process root cause tracing according to the knowledge graph comprises:

[0058] acquiring a defect type and selecting a plurality of associated processes from the knowledge graph according to the defect type to obtain a process set;

[0059] acquiring cross-process correlation strengths of each process according to the process set and inversely deducing corresponding shortest propagation paths from the knowledge graph according to each of the cross-process correlation strengths;

[0060] deducing corresponding defect propagation probabilities according to each of the shortest propagation paths;

[0061] acquiring quality index entropy values of each process in the process set and acquiring corresponding defect occurrence probabilities according to each of the quality index entropy values and the corresponding defect propagation probabilities;

[0062] sorting a plurality of the defect occurrence probabilities in order of size and selecting a process corresponding to a largest defect occurrence probability as a root cause process.

[0063] The application further provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the cross-process defect root cause tracing method when executing the computer program.

[0064] The application further provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the steps of the cross-process defect root cause tracing method.

[0065] The application has the following beneficial effects: the application greatly improves the understanding of the interaction between multiple processes by obtaining the full-process chain data of the target product, forms a multi-dimensional data feature matrix by integrating process parameters, equipment states and quality detection information, so that the performance evaluation of each process is more comprehensive, the interaction and influence path between processes can be clearly depicted by constructing a process relationship diagram, this graphical method makes the complex interaction more intuitive, which helps the technical personnel better understand the potential causes of defects, the introduction of the process attenuation factor provides basic data support for quantifying the relationship between processes, avoids the fuzziness and limitations of the correlation strength analysis in the previous method, the shortest propagation path and propagation probability of the defect can be accurately identified by analyzing the process relationship diagram, and the risk in the production process is effectively managed instead of post-tracing, so as to reduce the economic loss and time cost caused by the defect, the construction of the knowledge graph makes the root cause tracing of the cross-process defect systematic, the knowledge graph can effectively integrate and display data, and is convenient for quickly locating problems, especially when facing complex multi-process interaction, the technical personnel can quickly find the problem root cause and take corresponding measures, the adaptive correlation analysis technology is used, so that the system can intelligently adjust the analysis model and path when facing new data, continuously optimizes the defect detection and tracing process, this flexibility is particularly important in the rapidly changing production environment, and the quality management efficiency can be effectively improved, the relationship between complex processes is revealed by the fusion analysis of multi-dimensional data, and an efficient, accurate and systematic defect troubleshooting and management method is provided for enterprises. BRIEF DESCRIPTION OF DRAWINGS

[0066] Figure 1 The method flowchart of an embodiment of the application is shown.

[0067] Figure 2 The system structure schematic diagram of an embodiment of the application is shown.

[0068] Figure 3 The internal structure schematic diagram of the computer device of an embodiment of the application is shown.

[0069] The implementation, functional features and advantages of the application will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0070] It should be understood that the specific embodiments described herein are merely illustrative of the present application and do not limit the present application.

[0071] As shown in Figure 1 The present application provides a cross-process defect root cause tracing method, comprising:

[0072] S1, obtaining full-process chain data of a target product, and obtaining production quality characteristic parameters of each process according to the full-process chain data, wherein the production quality characteristic parameters include process parameter information, equipment state information and quality detection information;

[0073] S2, obtaining process data characteristic matrix of the corresponding process according to each of the process parameter information, equipment state information and quality detection information;

[0074] S3, obtaining process attenuation factors of the corresponding processes according to the process data characteristic matrices of any two processes, and obtaining cross-process correlation strengths according to the process attenuation factors;

[0075] S4, constructing a process relationship graph according to each process and the cross-process correlation strengths, and obtaining defect propagation data according to the process relationship graph, wherein the defect propagation data includes the shortest propagation path, defect propagation probability and defect type;

[0076] S5, constructing a knowledge graph according to the full-process chain data, the plurality of cross-process correlation strengths, the shortest propagation path, the defect propagation probability and the defect type, and performing cross-process root cause tracing according to the knowledge graph.

[0077] As described in steps S1-S5 above, the present application obtains the whole process chain data of the target product, and obtains the process parameter information, equipment state information and quality detection information of the production quality characteristic parameters of each process according to the whole process chain data. By collecting all process data in the whole production process, the overall situation of each process in the production process can be obtained, ensuring that the production process information of all processes is covered, eliminating information blind spots, providing the basis for cross-process analysis. According to the whole process chain data, the process parameters, equipment states and quality detection information of each process are extracted, which can refine the data of each process to process parameters, equipment states and quality detection, accurately capture the specific source of quality fluctuation, and more finely identify potential problems in the process, providing the basis for subsequent process optimization and quality improvement. Through each process parameter information, equipment state information and quality detection information, the process data characteristic matrix of the corresponding process is obtained, the process parameters, equipment states and quality detection information are integrated into the process data characteristic matrix, and the data of each process is represented in the form of matrix, providing a structured data format for subsequent analysis. Matrix processing makes the data of different processes structured and standardized, facilitating subsequent processing and calculation. Converting process data into matrix form can effectively perform subsequent operations through numerical analysis methods, improve efficiency, and obtain process attenuation factors through process data characteristic matrices of any two processes, and obtain cross-process correlation strength according to the process attenuation factors. Wherein, obtaining the corresponding process attenuation factor through the process data characteristic matrix of any two processes in the whole process chain of the target product refers to calculating and obtaining the corresponding process attenuation factor through the process data characteristic matrix of any two processes in the whole process chain of the target product. Each process has a corresponding process attenuation factor relative to other multiple processes, and a corresponding cross-process correlation strength can be obtained through the process attenuation factor. Therefore, each process has multiple cross-process correlation strengths. By calculating the attenuation factor between any two processes, the influence degree between processes is reflected. This factor considers the changes of process parameters, the fluctuations of equipment states and the quality detection results. The attenuation factor can quantify the influence degree between different processes, reveal the weight and role of each process in the whole production process, and through the introduction of process attenuation factor, the influence of a certain process on defects can be more accurately understood, providing key clues for root cause tracing. The attenuation factor is used to calculate the correlation strength between processes, which helps to reveal the mutual influence between different processes. Cross-process correlation strength provides a quantitative analysis tool for revealing the interaction between multiple processes, which helps to identify key correlations between processes. Through the correlation strength between processes, it can effectively identify which processes have a greater impact on defects, and then perform effective tracing. A process relationship diagram is constructed through each process and cross-process correlation strength, and the shortest propagation path, defect propagation probability and defect type of defect propagation data are obtained according to the process relationship diagram. The process relationship diagram is constructed by using the process and its correlation strength, showing the relationship and influence path between processes,The process relationship diagram can clearly show the relationship between each process, so that the defect propagation path and the influence chain are obvious at a glance, which helps to intuitively understand the complex interaction between processes, facilitates analysis and decision-making, and according to the process relationship diagram, the shortest path of defect propagation, the probability of defect propagation and the type of defect are calculated. These data help to analyze how defects propagate in the production process and their possible consequences. Accurate calculation of defect propagation path, probability and type can provide clear clues for root cause analysis. The shortest propagation path can quickly identify the root cause of the defect, and the defect propagation probability further improves the accuracy of the analysis. A knowledge graph is constructed by using full-process chain data, multiple cross-process correlation strengths, the shortest propagation path, defect propagation probability and defect type, and cross-process root cause tracing is performed according to the knowledge graph. The full-process chain data, cross-process correlation strength, defect propagation path and other information are summarized to construct a comprehensive knowledge graph. The graph not only describes the relationship between processes, but also includes the propagation process and prediction of defects. The knowledge graph integrates complex cross-process information and provides comprehensive knowledge support for root cause tracing, systematically manages complex production data, and provides data support and decision-making basis for production line optimization, defect prevention and continuous improvement through the graph presentation method. The constructed knowledge graph is used for cross-process defect root cause tracing, in-depth analysis of the relationship between each process and the source of defects, and through cross-process root cause tracing, the source of defects generated across multiple processes can be accurately identified, solving the problem that the prior art cannot effectively trace.

[0078] In one embodiment, the step S2 of obtaining the process data feature matrix of the corresponding process according to each of the process parameter information, equipment state information and quality detection information comprises:

[0079] S21, obtaining a plurality of process temperature data, process pressure data and process speed data according to the process parameter information, and obtaining a corresponding temperature skewness coefficient, pressure skewness coefficient and speed skewness coefficient according to each of the process temperature data, process pressure data and process speed data;

[0080] S22, obtaining a corresponding process stability feature vector according to each of the temperature skewness coefficient, pressure skewness coefficient and speed skewness coefficient;

[0081] S23, obtaining a plurality of equipment efficiency data and equipment load data according to the equipment state information, and obtaining a corresponding efficiency skewness coefficient and load skewness coefficient according to each of the equipment efficiency data and equipment load data;

[0082] S24, obtaining a corresponding equipment stability feature vector according to each of the efficiency skewness coefficient and load skewness coefficient;

[0083] S25, obtaining a plurality of size deviation data, surface quality data and defect area data according to the quality detection information, and obtaining corresponding size skewness coefficient, quality skewness coefficient and defect skewness coefficient according to each of the size deviation data, surface quality data and defect area data;

[0084] S26, obtaining corresponding quality stability characteristic vector according to each of the size skewness coefficient, quality skewness coefficient and defect skewness coefficient;

[0085] S27, performing normalization processing on each of the quality stability characteristic vector, equipment stability characteristic vector and process stability characteristic vector to obtain corresponding standard quality stability characteristic vector, standard equipment stability characteristic vector and standard process stability characteristic vector;

[0086] S28, constructing process data characteristic matrix of corresponding process according to a plurality of the standard quality stability characteristic vector, standard equipment stability characteristic vector and standard process stability characteristic vector.

[0087] As described in steps S21-S28 above, the present application obtains multiple process temperature data, process pressure data and process speed data through process parameter information, and obtains corresponding temperature skewness coefficients, pressure skewness coefficients and speed skewness coefficients according to each process temperature data, process pressure data and process speed data, and obtains corresponding process stability feature vectors through each temperature skewness coefficient, pressure skewness coefficient and speed skewness coefficient. By obtaining temperature, pressure and speed data of multiple processes, key process parameters in the production process can be comprehensively recorded and monitored, and by calculating the skewness coefficient of each process, the distribution characteristics of each process parameter in different processes can be revealed. Skewness coefficients can help identify deviations and abnormalities in process parameters, providing in-depth insights into process stability, which is crucial for quality control of multi-process complex systems. The process stability feature vector obtained based on the skewness coefficient is a mathematical abstraction of the state of each process, which provides a quantifiable standard for the correlation between multiple processes, making it easier to identify stability problems in process parameters by comparing feature vectors between processes. By collecting multiple device efficiency data and device load data through device state information, and obtaining corresponding efficiency skewness coefficients and load skewness coefficients according to each device efficiency data and device load data, and obtaining corresponding device stability feature vectors through each efficiency skewness coefficient and load skewness coefficient, the efficiency and load data of the device directly affect the quality and yield of the production process. By collecting these data, dynamic monitoring of device status can be achieved. Efficiency skewness coefficients and load skewness coefficients reflect the performance deviation of the device in actual operation, helping to identify performance abnormalities of the device under different working conditions. By quantifying the deviation of device performance, data support is provided for device maintenance, adjustment and optimization, which helps to reduce production instability caused by device problems. Through the calculation of device stability feature vectors, a complete performance index system can be formed at the device level, converting device performance data into feature vectors, which not only allows real-time monitoring of device stability, but also enables comparison with other process or quality data. By collecting multiple size deviation data, surface quality data and defect area data through quality detection information, and obtaining corresponding size skewness coefficients, quality skewness coefficients and defect skewness coefficients according to each size deviation data, surface quality data and defect area data, and obtaining corresponding quality stability feature vectors through each size skewness coefficient, quality skewness coefficient and defect skewness coefficient, the quality of the product can be thoroughly understood through the collection of quality detection information. Size deviation, surface quality and defect area data directly reflect the production precision and appearance quality of the product. Size skewness coefficients, quality skewness coefficients and defect skewness coefficients can reflect the abnormal distribution of different quality characteristics (such as size, surface quality, defects, etc.) in the production process. By calculating these skewness coefficients, the root cause of quality fluctuations can be revealed, especially the cross-process quality impact, helping to locate potential process or device problems.The quality stability eigenvector can convert different quality detection information into standardized characteristic parameters, provide a unified reference for subsequent analysis, and through the stable eigenvectors, more accurate monitoring and prediction of quality can be realized, and a more scientific basis for root cause tracing of cross-process defects can be provided. Through normalization processing of each quality stability eigenvector, equipment stability eigenvector and process stability eigenvector, corresponding standard quality stability eigenvector, standard equipment stability eigenvector and standard process stability eigenvector are obtained. The process data eigenvector matrix of the corresponding process is constructed through multiple standard quality stability eigenvectors, standard equipment stability eigenvectors and standard process stability eigenvectors. The normalized eigenvectors not only improve the comparability of the data, but also eliminate the dimensional differences that may exist in different data sources, providing a basis for multi-dimensional, multi-process cross-domain analysis. By integrating multiple standardized eigenvectors into a process data eigenvector matrix, the key features of each process can be fully displayed, and data support is provided for further analysis and prediction. The feature matrix not only reveals the correlation between processes, but also provides an intuitive data framework for root cause tracing of cross-process defects, greatly improving the accuracy and efficiency of interaction analysis between processes.

[0088] In one embodiment, the step S3 of obtaining the corresponding process attenuation factor according to the process data eigenvector matrix of any two processes comprises:

[0089] S31, obtaining a difference matrix according to the difference values of the corresponding characteristics in the process data eigenvector matrix of any two processes, and obtaining the probability of each difference value according to the difference matrix;

[0090] S32, obtaining the frequency of each difference value according to the probability of multiple difference values, and calculating the quality index entropy value according to the frequency of multiple difference values, wherein the calculation formula is:

[0091]

[0092] Wherein, Z(BS) represents the quality index entropy value, C(PL) i represents the frequency of the i-th difference value, i represents the serial number of the frequency of the difference value, and N represents the number of the frequency of the difference value;

[0093] S33, obtaining the mean value and the maximum difference value according to the difference matrix, and obtaining the equipment health degree index according to the mean value and the maximum difference value;

[0094] S34, obtaining a process influence characteristic matrix according to the partial derivative of the corresponding characteristics in the process data eigenvector matrix of any two processes, and obtaining the process complexity according to the process influence characteristic matrix;

[0095] S35、According to the process complexity, equipment health index and quality index entropy value, the process attenuation factor is calculated, and the calculation formula is:

[0096]

[0097] Wherein, G(SJ) represents the process attenuation factor, Z(BS) represents the quality index entropy value, G(FZ) represents the process complexity, and S(JZ) represents the equipment health index.

[0098] As described in the above steps S31-S35, wherein, in the calculation formula of the process attenuation factor and the quality index entropy value, the respective parameters need to be normalized first to eliminate the dimensional difference between different variables, the purpose is to ensure that all variables are on the same order of magnitude, so that the calculation is more stable and effective, the method for obtaining the difference matrix according to the difference value of the corresponding features in the process data feature matrix of any two processes is to first normalize the data features in the process data feature matrix of the two processes, and then calculate the difference value of each data feature by the corresponding data features in the process data feature matrix of the two processes, and then a plurality of difference values can be arranged according to the format of the process data feature matrix to obtain the difference matrix, the method for obtaining the frequency of each difference value according to the probability of a plurality of difference values is to first calculate the total probability by summing the probability of a plurality of difference values, and then obtain the frequency of each difference value by the ratio of the probability of each difference value to the total probability, the device health index is obtained according to the ratio of 1 minus the mean value to the maximum difference value, and the process complexity is obtained by weighted summation of all feature data (partial derivatives) in the process influence feature matrix, the present application obtains the difference matrix according to the difference value of the corresponding features in the process data feature matrix of any two processes, calculates the feature difference value of any two processes in the process data feature matrix, can accurately capture the mutual difference between different processes, helps to identify the potential defect source between different processes, the traditional method often only analyzes the data fluctuation of a single process, ignores the complex relationship between processes, the present application effectively improves the accuracy and comprehensiveness of cross-process defect tracing by constructing a multi-process feature difference matrix, and obtains the probability of each difference value according to the difference matrix, obtains the frequency of each difference value according to the probability of a plurality of difference values, and quantifies the possibility of occurrence of each process feature difference by calculating the probability of the difference value, so that the cross-process defect tracing not only depends on qualitative analysis, but also combines the probability and statistics method, provides a more objective and quantitative analysis method, by statistical frequency of a plurality of difference values, helps to evaluate which feature difference frequently appears between processes, further reveals the potential key factors of cross-process defects. This frequency analysis enhances the statistical significance of the difference value, making the defect source tracing more reliable, the traditional defect tracing method focuses on the fluctuation of a single data point, ignoring the difference features that appear multiple times, which may lead to some potential problems being ignored, the present application can more comprehensively identify the key factors that continuously affect the quality of the process by frequency analysis, and calculates the quality index entropy value according to the frequency of a plurality of difference values, because in the production process, each process has a certain feature value difference, these differences reflect the process fluctuation, equipment state, personnel operation and other factors, these difference values are usually random and present a certain probability distribution, the frequency of a plurality of difference values (i.e. the number of times or probability of different difference values) can help to understand the universality or abnormality of these differences in the production process,By calculating the entropy value of the quality index, the volatility of the overall quality in the production process can be reflected, that is, the uncertainty of product quality is quantified through the distribution of difference values. If the distribution of difference values is more uniform, the entropy value will be higher, indicating that the quality fluctuation is larger. If the probability of certain difference values is higher, the entropy value is lower, indicating that the quality is more stable. The calculation of frequency and entropy value can help to capture potential instability factors of the process, identify the source of quality fluctuation, and quantify the uncertainty and complexity of product quality through entropy value, providing a basis for subsequent quality optimization. Entropy value is an index for measuring system uncertainty. By calculating the entropy value of the quality index, the complexity and uncertainty between different processes can be evaluated. The calculation of the entropy value of the quality index provides a quantitative standard for defect tracing, which can effectively measure the abnormal fluctuations that may occur in the production process. Traditional methods often cannot provide a global quality control perspective. The introduction of entropy makes the quality monitoring of the entire production process more comprehensive and dynamic. Through entropy analysis, it can more accurately determine whether there is abnormal fluctuation or potential defect in the process. By obtaining the mean value and maximum difference value of the difference matrix, the standard deviation and extreme difference between processes can be clearly indicated, which provides a direct basis for evaluating the health status of the process and possible defects. Traditional methods usually focus on average data analysis of the process, while ignoring the influence of extreme data points. The present application more comprehensively reveals the key factors that may cause defects through analysis of the maximum difference value. The equipment health index is obtained according to the mean value and the maximum difference value. Equipment health is an important index for evaluating whether the production equipment is in good working condition. The combination of mean value and maximum difference value can provide decision basis for equipment maintenance and optimization. By accurately calculating the equipment health, equipment abnormalities can be detected in time, and cross-process defects caused by equipment problems can be reduced. The equipment health evaluation in the prior art usually relies on a single data source, lacking comprehensive and detailed monitoring of the equipment state. The present application considers multiple data characteristics in the equipment operation process, providing a more scientific and reliable health evaluation method. The process influence feature matrix is obtained by the partial derivative of the corresponding features in the process data feature matrix of any two processes. Partial derivative analysis can reveal the sensitivity relationship between different process features, identify the key factors affecting the process result, and provide more detailed process influence analysis, revealing the potential interaction between processes. Existing methods usually only analyze based on static data, lacking a deep understanding of the dynamic influence between processes. The present application can more accurately capture the interaction and influence between processes through partial derivative analysis, improving the depth and accuracy of analysis. The process complexity is obtained according to the process influence feature matrix. The evaluation of process complexity can help to identify complex processes that are difficult to control, and timely measures can be taken to improve them, making the entire process analysis more systematic and comprehensive, and effectively preventing cross-process defects caused by complex processes. Existing technologies usually cannot quantify process complexity.The application provides more targeted improvement suggestions by quantifying the process complexity evaluation, calculates a process attenuation factor through a process complexity, a device health index and a quality index entropy value, the process attenuation factor is a key parameter for evaluating process stability and sustainability, the process complexity is obtained through a process influence characteristic matrix and reflects the complexity of the process itself, a process with high complexity means that more parameters and variables need to be controlled, process fluctuations can be larger, thereby affecting the stability of subsequent processes and product quality, therefore, the higher the process complexity, the greater the impact on subsequent processes and the higher the attenuation factor, the device health index directly affects the stability of the process, device failure, aging or instability can cause process fluctuations to intensify, thereby affecting the quality of the entire production process, the lower the device health index, the greater the impact of the device on the production process and the higher the attenuation factor, the quality index entropy value reflects the degree of quality fluctuation, a process with large quality fluctuations can mean that quality control is difficult, subsequent processes will also be affected by these fluctuations, the higher the quality index entropy value, the greater the quality fluctuation of the process and the higher the attenuation factor, meaning that the impact of the previous process on the subsequent process gradually increases, the process complexity, device health and quality fluctuation (reflected by the entropy value) can comprehensively reflect the influence relationship between processes from different angles, the attenuation factor is calculated through these factors, which can comprehensively consider the complexity of the process, the state of the device and the quality fluctuation, thereby obtaining a more accurate attenuation factor, through the calculation of the attenuation factor, it can be identified which processes have a greater impact on the entire production process, helping managers to maintain, adjust or optimize key processes and key devices.

[0099] By combining the process complexity, the device health and the quality index entropy value, the degradation trend of the process can be comprehensively evaluated, providing a feasible basis for prevention and repair, the prior art lacks a systematic method for comprehensive evaluation of multiple factors, the application can effectively predict the possible degradation of the process by multi-factor analysis, preventing problems from occurring in advance, by introducing probability statistics, entropy analysis and partial derivative analysis, the application can more comprehensively and objectively evaluate abnormal factors in the production process, improving the accuracy and reliability of defect tracing.

[0100] In one embodiment, the step S3 of obtaining the cross-process correlation strength according to the process attenuation factor comprises:

[0101] S36, obtaining first installation layout data and a parameter fluctuation amount according to the process data characteristic matrix of the first process, and obtaining a quality fluctuation amount corresponding to the parameter fluctuation amount according to the process data characteristic matrix of the second process;

[0102] S37, obtaining a parameter influence coefficient corresponding to any two processes according to the quality fluctuation amount and the parameter fluctuation amount;

[0103] S38, obtaining first product installation position data and first product layout data according to the first installation layout data;

[0104] S39, obtaining second installation layout data according to the process data feature matrix of the second process, and obtaining second product installation position data and second product layout data according to the second installation layout data;

[0105] S310, obtaining an installation position correlation coefficient according to the first product installation position data and the second product installation position data, and obtaining a product layout correlation coefficient according to the first product layout data and the second product layout data;

[0106] S311, obtaining a process attenuation factor, and calculating a cross-process correlation strength according to the process attenuation factor, a parameter influence coefficient, the product layout correlation coefficient and the installation position correlation coefficient, wherein the calculation formula is:

[0107] K(GQ) = D(SY) * e -G(SJ)*[1-B(JX)*A(WX)] ;

[0108] Wherein, K(GQ) represents the cross-process correlation strength, G(SJ) represents the process attenuation factor, D(SY) represents the parameter influence coefficient, B(JX) represents the product layout correlation coefficient, and A(WX) represents the installation position correlation coefficient.

[0109] As described in steps S36-S311 above, the present application obtains the first installation layout data and the parameter fluctuation amount through the process data feature matrix of the first process, obtains the first installation layout data and the parameter fluctuation amount through the process data feature matrix, can accurately capture the data features related to the installation layout and the parameter fluctuation in the first process, provides basic data support for subsequent quality analysis and cross-process relationship establishment, can effectively reveal the key factors affecting product quality by quantifying and extracting the features in the process, and lays a solid data foundation for defect tracing, obtains the quality fluctuation amount corresponding to the parameter fluctuation amount according to the process data feature matrix of the second process, and the quality fluctuation amount refers to the fluctuation difference of product quality in the second process caused by the parameter fluctuation amount in the first process, correlates the parameter fluctuation amount and the quality fluctuation amount by analyzing the process data feature matrix of the second process, so that the parameter fluctuation between different processes can be directly reflected on the quality fluctuation, thereby the influence of parameter change on quality fluctuation can be effectively identified, and a clear causal relationship chain is provided for cross-process defect analysis, the parameter influence coefficient corresponding to any two processes is obtained through the quality fluctuation amount and the parameter fluctuation amount, in multiple processes, the fluctuation of parameters is often the key factor leading to defects, by calculating the parameter influence coefficient, the mutual influence relationship between parameters in different processes can be quantitatively described, the mutual dependency between different processes can be revealed, and a quantitative analysis tool is provided for cross-process defect root cause tracing, the first product installation position data and the first product layout data are obtained through the first installation layout data, the second installation layout data is obtained through the process data feature matrix of the second process, and the second product installation position data and the second product layout data are obtained according to the second installation layout data, the installation position data and the layout data of the product are obtained through the first installation layout data, the installation method of the product can be accurately understood in the space and structure level, data support is provided for subsequent cross-process defect analysis in the space and structure level, which is helpful for more comprehensive analysis of potential problems in the installation process, the installation layout data of the second process provides an installation data source different from the first process, by extracting the second product installation position data and the layout data, the differences in installation in different processes can be compared, not only more dimensional data support is provided for cross-process defect analysis, but also the potential influence of installation method and layout difference on the final product quality can be revealed, the installation position correlation coefficient is obtained through the first product installation position data and the second product installation position data, the installation position correlation coefficient can quantitatively describe the similarity or difference of the product installation position in different processes, so as to evaluate the influence of installation position change on product quality, the introduction of the installation position correlation coefficient provides spatial correlation analysis for defect root cause analysis between different processes, and the quality problems caused by installation position difference can be revealed.The product layout correlation coefficient is obtained through the first product layout data and the second product layout data, the product layout correlation coefficient quantifies the similarity or difference of the product layout in two processes, reveals the influence of the layout design on the product quality, especially in the case of multiple processes being carried out at the same time, the layout difference in different processes can be compared to more clearly find out the contribution of the layout problem to the quality fluctuation, the process attenuation factor is obtained, and the cross-process correlation strength is calculated according to the process attenuation factor, the parameter influence coefficient, the product layout correlation coefficient and the installation position correlation coefficient, by introducing the process attenuation factor, the attenuation effect that each process may produce in the product quality process can be effectively considered, combined with the parameter influence coefficient, the product layout correlation coefficient and the installation position correlation coefficient, the correlation strength between multiple processes can be comprehensively analyzed, not only the accuracy of cross-process quality analysis is improved, but also the cross-process defect tracing is no longer limited to a single process, but can more comprehensively reveal the root cause of quality fluctuation by comprehensively considering the complex relationship between different processes, therefore, the interaction and quality fluctuation factors between processes are accurately extracted through multi-dimensional data matrix analysis, so that the root cause of cross-process defects can be effectively traced, unlike the analysis method of traditional method which depends on a single process data, the method can reveal the complex dependence relationship between processes and accurately locate the defect source through the correlation coefficient and influence coefficient between processes, the process layout and installation position can be optimized, so as to improve the quality control ability of the production process, which is suitable for complex production processes, can be flexibly adjusted and optimized according to different processes and different parameters, has wide application potential, the cross-process correlation strength is calculated through the process attenuation factor, the parameter influence coefficient, the product layout correlation coefficient and the installation position correlation coefficient, the purpose is to comprehensively quantify and optimize the interaction and influence between different processes, so as to improve the product quality and production efficiency, the parameter influence coefficient reflects the influence degree of the input parameter of a process on the output result (such as product quality, installation effect, etc.) of other processes, by quantifying the influence of different parameters, the influence of the parameters of a process on the quality fluctuation of subsequent processes can be deeply understood, the product layout correlation coefficient describes the influence of the layout design and installation position of products between different processes on the interaction between processes, if the product layout of the previous process and the subsequent process is similar or related, the process and quality fluctuation between them may be more closely linked, the layout correlation coefficient makes the influence between processes not only based on the characteristics and parameters of process data, but also considers the factors in the physical space, the installation position of the workpiece will affect the stress, temperature distribution and other factors in the machining process, and further affect the degree of quality fluctuation, and the installation position correlation coefficient reflects the similarity or difference of the installation position of products between different processes, the installation position of products may affect the interaction between processes, by considering the correlation of the installation position between processes, the spatial correlation between processes is predicted,Especially in multi-process operation, the change of a position can affect the execution effect of multiple processes, and therefore the installation position correlation coefficient is used to reflect the influence of spatial relationship on cross-process correlation strength.

[0110] In one embodiment, the step S4 of constructing a process relationship graph according to each process and cross-process correlation strength, and obtaining defect propagation data according to the process relationship graph, comprises:

[0111] S41, obtaining all cross-process correlation strengths of each process, and selecting the maximum cross-process correlation strength as a dependent correlation strength;

[0112] S42, determining a most dependent process according to each dependent correlation strength and a corresponding process, to obtain a process dependency relationship;

[0113] S43, constructing a process relationship graph by taking each process as a node, the process dependency relationship as an edge, and the corresponding cross-process correlation strength as an edge weight;

[0114] S44, obtaining all propagation paths of each process according to the edges of the process relationship graph, and determining a shortest propagation path according to the all propagation paths;

[0115] S45, obtaining a historical defect probability of each process, and obtaining a propagation probability according to the historical defect probability and the shortest propagation path;

[0116] S46, obtaining a defect propagation probability according to the propagation probability and the historical defect probability.

[0117] As described in steps S41-S46, the propagation path is obtained by using a graph algorithm through the edges in the process relationship graph. Specifically, all possible paths starting from a process starting point can be found by depth-first search. The present application obtains all cross-process correlation strengths of each process and selects the maximum cross-process correlation strength as the dependent correlation strength. By obtaining all cross-process correlation strengths of each process, the interaction relationship between different processes can be accurately identified, avoiding the limitation of traditional methods that can only analyze a single process. Selecting the maximum cross-process correlation strength as the dependent correlation strength can ensure that the key inter-process influence is focused on during the traceability process, thereby improving the traceability accuracy. The most dependent process is determined through each dependent correlation strength and the corresponding process, and the process dependency relationship is obtained. By clarifying the relationship between each process and the most dependent process, the complex dependency network between cross-processes can be simplified, facilitating further analysis and operation, which helps to determine which processes have a significant impact on defect propagation, facilitates the discovery of potential defect sources, and optimizes monitoring and preventive measures in subsequent production processes. By taking each process as a node, the process dependency relationship as an edge, and the corresponding cross-process correlation strength as an edge weight, a process relationship graph is constructed. By converting the process relationship into a graph model, the mutual relationship between processes can be visually displayed and effectively analyzed through graph theory methods. This modeling approach has high visualization and operability, which is conducive to intuitively identifying the defect propagation path between processes. By taking the cross-process correlation strength as the weight of the edge, the size of the influence between different processes is further emphasized, providing a more accurate basis for subsequent defect traceability. All propagation paths of each process are obtained through the edges of the process relationship graph, and the shortest propagation path is determined according to all propagation paths. By analyzing all propagation paths and selecting the shortest path, the interference of redundant information can be reduced, the core path of defect propagation can be accurately locked, and the speed and accuracy of defect positioning can be improved. The determination of the shortest propagation path helps to simplify the defect traceability process, enabling the traceability method to still perform efficiently when facing complex inter-process interactions. By obtaining the historical defect probability of each process and obtaining the propagation probability according to the historical defect probability and the shortest propagation path, the historical defect probability of each process is introduced, which can be based on past data for prediction, thereby improving the prediction accuracy of the defect propagation probability and providing a reference basis for future production processes. Based on the combination of historical defect probability and propagation path, dynamic analysis of defect propagation for each process can be realized, which is more timely and targeted, facilitating the development of preventive measures. By obtaining the defect propagation probability through the propagation probability and the historical defect probability, the combination of the propagation probability and the historical defect probability can comprehensively consider the generation and propagation mechanism of defects, providing a comprehensive quantitative analysis of the mutual influence between processes. Through accurate defect propagation probability prediction, enterprises can identify high-risk processes in advance and take appropriate preventive measures to reduce the occurrence and propagation of defects, improve production quality and efficiency, and not only improve the accuracy of defect traceability,It can also help enterprises better control quality in the production process and reduce losses caused by defects.

[0118] In one embodiment, the step S5 of performing cross-process root cause tracing according to the knowledge graph comprises:

[0119] S51, acquire a defect type, and select a plurality of associated processes from the knowledge graph according to the defect type to obtain a process set;

[0120] S52, acquire the cross-process correlation strength of each process according to the process set, and acquire the corresponding shortest propagation path from the knowledge graph according to each cross-process correlation strength;

[0121] S53, according to each of the shortest propagation path, the corresponding defect propagation probability is deduced;

[0122] S54, acquire the quality index entropy value of each process in the process set, and acquire the corresponding defect occurrence probability according to each quality index entropy value and the corresponding defect propagation probability;

[0123] S55, sort a plurality of defect occurrence probabilities in order of size, and select the process corresponding to the largest defect occurrence probability as the root cause process.

[0124] As described in steps S51-S55, the present application obtains the defect type, selects multiple associated processes from the knowledge graph according to the defect type, and obtains the process set. By accurately identifying the defect type, multiple processes associated with the defect can be selected from the knowledge graph. This not only improves the depth of understanding of the defect, but also accurately locks the process group that may be affected, avoiding the problem of incomplete or missed tracing caused by the traditional method of starting from a single process. Therefore, comprehensive analysis across processes can be realized, providing more comprehensive and accurate data support for subsequent steps. The cross-process correlation strength of each process is obtained through the process set. The cross-process correlation strength reveals the mutual relationship and dependency between different processes. The analysis of cross-process correlation strength data can provide a quantitative basis for subsequent root cause tracing. Through the quantitative cross-process correlation strength, it can effectively identify which processes have a relatively close relationship, thereby helping to locate the possible propagation path and source of the defect. According to each cross-process correlation strength, the corresponding shortest propagation path is obtained from the knowledge graph. Traditional methods usually rely on intuitive assumptions or single data points for analysis. However, this step obtains the shortest propagation path through reverse deduction, and derives the shortest propagation path through the knowledge graph. This can effectively reduce unnecessary path analysis, ensure the efficiency and accuracy of the tracing process, reduce the complexity of calculation, avoid interference from irrelevant factors, and provide a clear defect propagation trajectory. Through the defect propagation probability corresponding to each shortest propagation path, the defect propagation probability of each path is quantified. This not only realizes the correlation analysis across processes, but also more accurately predicts the propagation possibility of the defect between processes. This method effectively makes up for the deficiency of the existing technology that cannot accurately quantify the propagation process, thereby improving the accuracy of prediction and the ability to identify abnormal processes. By obtaining the quality indicator entropy value of each process in the process set, and obtaining the corresponding defect occurrence probability according to each quality indicator entropy value and the corresponding defect propagation probability, the introduction of quality indicator entropy value provides quantitative analysis of the quality fluctuation of each process, enabling the quality difference of different processes to be evaluated in a more scientific way. Combining the defect propagation probability to calculate the defect occurrence probability can consider the intrinsic quality characteristics of the process and its role in defect propagation, thereby forming a comprehensive quality control framework and effectively improving the accuracy of root cause analysis. By sorting multiple defect occurrence probabilities in order of size and selecting the process corresponding to the largest defect occurrence probability as the root cause process, the step of sorting and selecting the largest defect occurrence probability is an important part of prioritizing all possible process roots. It can effectively identify the most likely root cause process, helping relevant personnel quickly focus on the process that may cause the problem, avoiding redundant analysis and excessive speculation in traditional methods, and improving the efficiency of problem solving. Through the quantitative cross-process correlation strength and the shortest propagation path,The defect propagation analysis between multiple processes can be completed in a short time, high-precision defect prediction can be performed in a multivariate environment by combining the quality index entropy value and the defect propagation probability, blind analysis in the traditional method is avoided, process sorting and root cause selection make the tracing process more concise, intuitive, easy to operate and apply, and the method is suitable for various complex manufacturing processes.

[0125] As shown in Figure 2 The application also provides a cross-process defect root cause tracing system, comprising:

[0126] A first acquisition module is configured to acquire full-process chain data of a target product and acquire production quality characteristic parameters of each process according to the full-process chain data, wherein the production quality characteristic parameters include process parameter information, equipment state information and quality detection information;

[0127] A second acquisition module is configured to acquire process data characteristic matrices of corresponding processes according to each of the process parameter information, the equipment state information and the quality detection information;

[0128] A third acquisition module is configured to acquire process attenuation factors of corresponding processes according to the process data characteristic matrices of any two processes and acquire cross-process correlation strengths according to the process attenuation factors;

[0129] A construction module is configured to construct a process relationship graph according to each process and the cross-process correlation strengths and acquire defect propagation data according to the process relationship graph, wherein the defect propagation data includes a shortest propagation path, a defect propagation probability and a defect type;

[0130] A tracing module is configured to construct a knowledge graph according to the full-process chain data, the multiple cross-process correlation strengths, the shortest propagation path, the defect propagation probability and the defect type and perform cross-process root cause tracing according to the knowledge graph.

[0131] In one embodiment, the construction module comprises:

[0132] A selection unit is configured to acquire all cross-process correlation strengths of each process and select a maximum cross-process correlation strength as a dependent correlation strength;

[0133] A first determination unit is configured to determine a most dependent process according to each of the dependent correlation strengths and a corresponding process and obtain a process dependency relationship;

[0134] A construction unit is configured to construct a process relationship graph by taking each process as a node, the process dependency relationship as an edge and the corresponding cross-process correlation strength as an edge weight;

[0135] The second determining unit is configured to acquire all propagation paths of each process according to edges of the process relationship diagram, and determine a shortest propagation path according to the all propagation paths;

[0136] The first acquiring unit is configured to acquire a historical defect probability of each process, and acquire a propagation probability according to the historical defect probability and the shortest propagation path;

[0137] The second acquiring unit is configured to acquire a defect propagation probability according to the propagation probability and the historical defect probability.

[0138] It should be noted that each module and unit in the cross-process defect root cause tracing system corresponds to each step in the cross-process defect root cause tracing method.

[0139] As shown in Figure 3 The computer device can be a server, and the internal structure thereof can be as shown in Figure 3 The computer device includes a processor, a memory, a network interface and a database connected through a system bus. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for running the operating system and the computer program in the non-volatile storage medium. The database of the computer device is configured to store all data required in the process of the cross-process defect root cause tracing method. The network interface of the computer device is configured to communicate with an external terminal through a network connection. The computer program is executed by the processor to implement the cross-process defect root cause tracing method.

[0140] Those skilled in the art can understand that Figure 3 The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied.

[0141] An embodiment of the present application further provides a computer readable storage medium, which stores a computer program. The computer program is executed by a processor to implement any one of the above cross-process defect root cause tracing methods.

[0142] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when the computer program is executed, the processes of the above-mentioned embodiment methods can be included. Any reference to memory, storage, databases, or other media in this application and in examples provided herein, unless specifically stated otherwise, can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM is available in many forms such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0143] It should be noted that in this document, the terms "comprising", "including", or any other variant thereof are intended to cover a non-exclusive inclusion, such that a process, device, article, or method that comprises a list of elements does not only include those elements, but can also include other elements not expressly listed or inherent to such process, device, article, or method. Without more limitations, the element defined by the statement "comprising a" does not exclude the presence of additional identical elements in the process, device, article, or method that includes the element.

[0144] The above description is only the preferred embodiment of the present application, and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation, or direct or indirect application in other related technical fields, based on the content of the present application specification and drawings, is also included in the patent protection scope of the present application.

Claims

1. A cross-process defect root cause tracing method, characterized by: include: Acquire full process chain data of the target product, and acquire production quality characteristic parameters of each process based on the full process chain data, wherein the production quality characteristic parameters include process parameter information, equipment status information, and quality inspection information; Obtaining a process data feature matrix of a corresponding process according to each of the process parameter information, equipment status information, and quality inspection information; Obtaining corresponding process attenuation factors according to process data feature matrices of any two processes, and obtaining cross-process correlation strengths of any two processes according to the process attenuation factors; Constructing a process relationship diagram based on the full process chain data and cross-process correlation strength of each process, and obtaining defect propagation data based on the process relationship diagram, wherein the defect propagation data includes the shortest propagation path, defect propagation probability, and defect type; A knowledge graph is constructed based on the full process chain data, multiple cross-process association strengths, the shortest propagation path, the defect propagation probability and the defect type, and cross-process root cause tracing is performed based on the knowledge graph.

2. The cross-process defect root cause tracing method according to claim 1 is characterized in that: The step of obtaining a process data feature matrix of a corresponding process according to each of the process parameter information, equipment status information, and quality inspection information includes: Acquire a plurality of process temperature data, process pressure data, and process speed data according to the process parameter information, and acquire a corresponding temperature skewness coefficient, pressure skewness coefficient, and speed skewness coefficient according to each of the process temperature data, process pressure data, and process speed data; Obtaining a corresponding process stability characteristic vector according to each of the temperature skewness coefficient, pressure skewness coefficient, and velocity skewness coefficient; Acquire a plurality of equipment efficiency data and equipment load data according to the equipment status information, and acquire a corresponding efficiency skewness coefficient and load skewness coefficient according to each of the equipment efficiency data and equipment load data; Obtaining a corresponding equipment stable characteristic vector according to each of the efficiency skewness coefficients and the load skewness coefficients; Acquire a plurality of dimensional deviation data, surface quality data, and defect area data according to the quality inspection information, and acquire a corresponding dimensional skewness coefficient, a mass skewness coefficient, and a defect skewness coefficient according to each of the dimensional deviation data, surface quality data, and defect area data; Obtaining a corresponding quality stability characteristic vector according to each of the size skewness coefficient, mass skewness coefficient, and defect skewness coefficient; Normalizing each of the quality stability feature vectors, equipment stability feature vectors, and process stability feature vectors to obtain corresponding standard quality stability feature vectors, standard equipment stability feature vectors, and standard process stability feature vectors; A process data feature matrix of a corresponding process is constructed according to a plurality of the standard quality stability feature vectors, the standard equipment stability feature vectors and the standard process stability feature vectors.

3. The cross-process defect root cause tracing method according to claim 1 is characterized in that: The step of obtaining the corresponding process attenuation factor according to the process data feature matrix of any two processes includes: Obtaining a difference matrix according to the difference values ​​of corresponding features in the process data feature matrix of any two processes, and obtaining the probability of each difference value according to the difference matrix; Obtaining the frequency of each difference value according to the probability of the multiple difference values, and obtaining the quality indicator entropy value according to the frequency of the multiple difference values; Obtaining a mean value and a maximum difference value according to the difference matrix, and obtaining a device health index according to the mean value and the maximum difference value; Obtaining a process influence feature matrix according to partial derivatives of corresponding features in a process data feature matrix of any two processes, and obtaining process complexity according to the process influence feature matrix; The process attenuation factor is calculated based on the process complexity, equipment health index and quality index entropy value, wherein the calculation formula is: Among them, G(SJ) represents the process attenuation factor, Z(BS) represents the quality index entropy value, G(FZ) represents the process complexity, and S(JZ) represents the equipment health index.

4. The cross-process defect root cause tracing method according to claim 1 is characterized in that: The step of obtaining the cross-process association strength according to the process attenuation factor includes: Obtaining first installation layout data and parameter fluctuation amount according to the process data characteristic matrix of the first process, and obtaining quality fluctuation amount corresponding to the parameter fluctuation amount according to the process data characteristic matrix of the second process; Obtaining parameter influence coefficients corresponding to any two processes according to the quality fluctuation amount and the parameter fluctuation amount; acquiring first product installation position data and first product layout data according to the first installation layout data; Acquire second installation layout data according to the process data feature matrix of the second process, and acquire second product installation position data and second product layout data according to the second installation layout data; Acquire an installation position correlation coefficient based on the first product installation position data and the second product installation position data, and acquire a product layout correlation coefficient based on the first product layout data and the second product layout data; Obtain the process attenuation factor, and calculate the cross-process correlation strength based on the process attenuation factor, parameter influence coefficient, product layout correlation coefficient, and installation position correlation coefficient. The calculation formula is: K(GQ)=D(SY)*e -G(SJ)*[1-B(JX)*A(WX)] ; Among them, K(GQ) represents the cross-process correlation strength, G(SJ) represents the process attenuation factor, D(SY) represents the parameter influence coefficient, B(JX) represents the product layout correlation coefficient, and A(WX) represents the installation position correlation coefficient.

5. The cross-process defect root cause tracing method according to claim 1 is characterized in that: The step of constructing a process relationship diagram based on the full process chain data and cross-process association strength of each process, and obtaining defect propagation data based on the process relationship diagram includes: Obtain all cross-process correlation strengths of each process and select the largest cross-process correlation strength as the dependent correlation strength; Determine the most dependent process according to each dependency association strength and the corresponding process, and obtain the process dependency relationship; Each process is regarded as a node, the process dependency as an edge, and the corresponding cross-process association strength as the edge weight to construct a process relationship graph; Obtaining all propagation paths of each process according to the edges of the process relationship graph, and determining the shortest propagation path according to all the propagation paths; Obtaining historical defect probabilities and historical defect types for each process, and obtaining propagation probabilities based on the historical defect probabilities and the shortest propagation path; The defect propagation probability is obtained according to the propagation probability and the historical defect probability.

6. The cross-process defect root cause tracing method according to claim 1 is characterized in that: The step of tracing the root cause across processes according to the knowledge graph includes: Obtain the defect type, and select multiple related processes from the knowledge graph according to the defect type to obtain a process set; Obtaining the cross-process association strength of each process according to the process set, and obtaining the corresponding shortest propagation path from the knowledge graph according to each cross-process association strength; Obtaining a corresponding defect propagation probability according to each of the shortest propagation paths; Obtaining a quality index entropy value for each process in the process set, and obtaining a corresponding defect occurrence probability based on each quality index entropy value and a corresponding defect propagation probability; The plurality of defect occurrence probabilities are sorted in order of magnitude, and the process corresponding to the largest defect occurrence probability is selected as the root cause process.

7. A cross-process defect root cause tracing system, characterized by: include: A first acquisition module is configured to acquire full process chain data of a target product and obtain production quality characteristic parameters of each process based on the full process chain data, wherein the production quality characteristic parameters include process parameter information, equipment status information, and quality inspection information; A second acquisition module is used to acquire a process data feature matrix of a corresponding process according to each of the process parameter information, equipment status information and quality inspection information; A third acquisition module is used to obtain corresponding process attenuation factors according to the process data feature matrix of any two processes, and obtain cross-process correlation strength according to the process attenuation factors; A construction module is used to construct a process relationship diagram based on each process and the correlation strength across processes, and obtain defect propagation data based on the process relationship diagram, wherein the defect propagation data includes the shortest propagation path, the defect propagation probability and the defect type; The tracing module is used to construct a knowledge graph based on the full process chain data, multiple cross-process association strengths, the shortest propagation path, the defect propagation probability and the defect type, and to perform cross-process root cause tracing based on the knowledge graph.

8. The cross-process defect root cause tracing system according to claim 7 is characterized in that: The building blocks include: The selection unit is used to obtain all cross-process correlation strengths of each process and select the largest cross-process correlation strength as the dependent correlation strength; A first determining unit is configured to determine the most dependent process according to each dependency association strength and the corresponding process, and obtain a process dependency relationship; The construction unit is used to construct a process relationship graph by taking each process as a node, process dependency as an edge, and the corresponding cross-process association strength as the edge weight; a second determining unit, configured to obtain all propagation paths of each process according to the edges of the process relationship graph, and determine the shortest propagation path according to all the propagation paths; A first acquisition unit is configured to acquire a historical defect probability of each process and acquire a propagation probability based on the historical defect probability and the shortest propagation path; The second acquiring unit is configured to acquire the defect propagation probability according to the propagation probability and the historical defect probability.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images