Root cause analysis method, device and electronic equipment for product quality

Abstract

This disclosure provides a method, apparatus, and electronic device for root cause analysis of product quality, relating to the field of artificial intelligence technology, particularly machine learning and data processing technology. The specific implementation involves: acquiring target data of the product; acquiring multiple candidate features of the product, whereby these candidate features influence product quality; obtaining a total importance parameter of the candidate features based on the target data, whereby the total importance parameter characterizes the importance of the candidate features to product quality; and selecting the target feature from the multiple candidate features based on the total importance parameter. Therefore, based on the target data of the product, the total importance parameter of the candidate features can be obtained to select the target feature from multiple candidate features. Compared to related technologies that largely rely on user subjective experience for root cause analysis, this method quantifies the importance of candidate features to product quality and obtains the total importance parameter for root cause analysis, thus improving the accuracy of root cause analysis of product quality.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of artificial intelligence, in particular to the technical field of machine learning and data processing, and in particular to a product quality root cause analysis method and device, an electronic device, a storage medium, and a computer program product. BACKGROUND

[0002] At present, product quality root cause analysis is crucial for improving product quality. However, with the increasing complexity of product manufacturing processes, such as the increase in process steps and the increase in the number of raw materials, product quality root cause analysis is difficult. In related technologies, root cause analysis is mostly dependent on user subjective experience, which requires high user experience accumulation and has low accuracy. SUMMARY

[0003] The present disclosure provides a product quality root cause analysis method and device, an electronic device, a storage medium, and a computer program product.

[0004] According to an aspect of the present disclosure, a product quality root cause analysis method is provided, including: obtaining target data of a product; obtaining a plurality of candidate features of the product, wherein the candidate features affect the quality of the product; based on the target data, obtaining a total importance parameter of the candidate features, wherein the total importance parameter is used to represent the importance of the candidate features to the quality of the product; and based on the total importance parameter, screening a target feature from the plurality of candidate features.

[0005] According to another aspect of the present disclosure, a product quality root cause analysis device is provided, including: a first obtaining module configured to obtain target data of a product; a second obtaining module configured to obtain a plurality of candidate features of the product, wherein the candidate features affect the quality of the product; an analysis module configured to obtain a total importance parameter of the candidate features based on the target data, wherein the total importance parameter is used to represent the importance of the candidate features to the quality of the product; and a screening module configured to screen a target feature from the plurality of candidate features based on the total importance parameter.

[0006] According to another aspect of the present disclosure, an electronic device is provided, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform a product quality root cause analysis method.

[0007] According to another aspect of the present disclosure, there is provided a non-transitory computer readable storage medium storing computer instructions for causing a computer to execute a root cause analysis method of product quality.

[0008] According to another aspect of the present disclosure, there is provided a computer program product comprising a computer program, wherein the computer program, when executed by a processor, implements the steps of a root cause analysis method of product quality.

[0009] It should be understood that the contents described in this section are not intended to identify key or important features of the embodiments of the present disclosure, nor are they used to limit the scope of the present disclosure. Other features of the present disclosure will become apparent from the following description. BRIEF DESCRIPTION OF DRAWINGS

[0010] The accompanying drawings are used to better understand the present scheme, and do not limit the present disclosure. Among them:

[0011] Figure 1 is a flowchart of a root cause analysis method of product quality according to a first embodiment of the present disclosure;

[0012] Figure 2 is a flowchart of a root cause analysis method of product quality according to a second embodiment of the present disclosure;

[0013] Figure 3 is a flowchart of a root cause analysis method of product quality according to a third embodiment of the present disclosure;

[0014] Figure 4 is a flowchart of a root cause analysis method of product quality according to a fourth embodiment of the present disclosure;

[0015] Figure 5 is a flowchart of a root cause analysis method of product quality according to a fifth embodiment of the present disclosure;

[0016] Figure 6 is a schematic diagram of a cause-effect diagram in a root cause analysis method of product quality according to a fifth embodiment of the present disclosure;

[0017] Figure 7 is a block diagram of a root cause analysis device of product quality according to a first embodiment of the present disclosure;

[0018] Figure 8 is a block diagram of an electronic device for implementing a root cause analysis method of product quality according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0019] Exemplary embodiments of the present disclosure are described herein with reference to the accompanying drawings, which are cited as illustrative examples. Various details of the embodiments of the present disclosure are described herein in order to provide a thorough understanding thereof. It will be understood by those of ordinary skill in the art that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Also, in the following description, descriptions of well-known functions and constructions are omitted for clarity and conciseness.

[0020] AI (Artificial Intelligence) is a technical science that studies and develops theories, methods, technologies and application systems for simulating, extending and expanding human intelligence. Currently, AI technology has the advantages of high automation, high accuracy and low cost, and has been widely applied.

[0021] ML (Machine Learning) is a science that studies how to use computers to simulate or implement human learning activities, and is one of the most intelligent and most advanced research fields in artificial intelligence. It involves probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory and other disciplines.

[0022] The basic purpose of data processing is to extract and derive data that is valuable and meaningful to certain people from a large amount of data that may be disorganized and difficult to understand, including data collection, storage, retrieval, processing, transformation and transmission.

[0023] Figure 1 is a flowchart of a product quality root cause analysis method according to the first embodiment of the present disclosure.

[0024] As shown in Figure 1 , the product quality root cause analysis method of the first embodiment of the present disclosure comprises:

[0025] S101, obtaining target data of a product.

[0026] It should be noted that the execution subject of the product quality root cause analysis method of the embodiments of the present disclosure can be a hardware device with data information processing capability and / or necessary software for driving the hardware device to work. Alternatively, the execution subject can include workstations, servers, computers, user terminals and other intelligent devices. Among them, the user terminal includes but is not limited to mobile phones, computers, intelligent voice interaction devices, smart home appliances, vehicle-mounted terminals, etc.

[0027] It should be noted that there are no strict limitations on the products or target data. For example, products can include chemical products, machinery and equipment, electronic products, etc. Target data can include raw material data, process data, environmental data, test data, etc. Raw material data can include material batch number, material number, material parameters, etc. Material parameters can include inner diameter, diameter, height, elasticity, strength, etc. Test data can include test data at at least one stage of the product's life cycle, where stages can include production stages, finished product testing stages, etc.

[0028] In one implementation, a mapping relationship or mapping table between the target data of a product and the product can be established in advance. When performing root cause analysis on the product quality, the above mapping relationship or mapping table can be queried based on the product, and the queried target data can be determined as the target data of the product.

[0029] In one implementation, obtaining target data for the product may include obtaining raw data of the product, performing data preprocessing on the raw data, and generating target data for the product. It should be noted that the specific method of data preprocessing is not limited in many ways; for example, at least one data preprocessing method from related technologies may be used, including data cleaning, missing value imputation, normalization, discretization, etc.

[0030] S102, obtain multiple candidate features of the product, where the candidate features affect the quality of the product.

[0031] It should be noted that candidate features affect product quality, and there are no strict limitations on candidate features. For example, they can include material parameters, process parameters, environmental parameters, etc.

[0032] It should be noted that there is no strict limit on the number of candidate features; for example, there can be 10 or 20 candidate features.

[0033] S103. Based on the target data, obtain the total importance parameter of the candidate features, whereby the total importance parameter is used to characterize the importance of the candidate features to the quality of the product.

[0034] It should be noted that there is only one total importance parameter for each candidate feature, and different candidate features can correspond to different total importance parameters. There are no strict restrictions on the range of values ​​for the total importance parameter; for example, the range can be -1 to 1.

[0035] It should be noted that the total importance parameter is used to characterize the importance of candidate features to product quality. For example, the total importance parameter may be positively or negatively correlated with the importance of candidate features to product quality. In one implementation, the larger the total importance parameter, the stronger the importance of the corresponding candidate feature to product quality; conversely, the smaller the total importance parameter, the weaker the importance of the corresponding candidate feature to product quality.

[0036] In one implementation, the total importance parameter of candidate features is obtained based on the target data, including constructing an importance evaluation index system for candidate features and obtaining the total importance parameter based on the importance evaluation index system. It should be noted that the importance evaluation index system is not subject to excessive limitations.

[0037] In one implementation, based on the target data, the total importance parameter of the candidate features is obtained. This includes inputting the target data into a root cause analysis model, which then outputs the total importance parameter of the candidate features. It should be noted that the root cause analysis model is not overly limited; for example, it can include machine learning models, causal models, statistical models, etc. Machine learning models may include GBDT (Gradient Boosting Decision Tree), causal models may include Bayesian tree models, and statistical models may include chi-square distribution models, correlation coefficient models, etc. Correlation coefficient models may include Spearman's rank correlation coefficient models. The root cause analysis model can be pre-trained or trained in real-time.

[0038] S104, based on the total importance parameter, selects the target feature from multiple candidate features.

[0039] It should be noted that the target feature is of great importance to the quality of the product, that is, the target feature is the key feature that affects the quality of the product. There is no limit to the number of target features, for example, there can be at least one target feature.

[0040] In one implementation, a target feature is selected from multiple candidate features based on a total importance parameter. This includes sorting the multiple candidate features in descending order according to the total importance parameter, and determining the top N candidate features as the target features. It should be noted that N is not subject to any restrictions; N is a positive integer, for example, N = 10. Therefore, this method can determine at least one candidate feature with the largest total importance parameter as the target feature.

[0041] In summary, the root cause analysis method for product quality according to the embodiments of this disclosure obtains target data of the product and multiple candidate features of the product, wherein the candidate features affect the product quality. Based on the target data, a total importance parameter of the candidate features is obtained, which characterizes the importance of the candidate features to the product quality. Based on the total importance parameter, the target feature is selected from the multiple candidate features. Therefore, based on the target data of the product, the total importance parameter of the candidate features can be obtained to select the target feature from multiple candidate features. Compared with related technologies that mostly rely on the subjective experience of users for root cause analysis, this solution can quantify the importance of candidate features to the product quality and obtain a total importance parameter for root cause analysis, thus improving the accuracy of root cause analysis of product quality.

[0042] Figure 2 This is a schematic flowchart of a root cause analysis method for product quality according to a second embodiment of the present disclosure.

[0043] like Figure 2 As shown, the root cause analysis method for product quality according to the second embodiment of this disclosure includes:

[0044] S201, Obtain target data for the product.

[0045] S202, obtain multiple candidate features of the product, where the candidate features affect the quality of the product.

[0046] The details of steps S201-S202 can be found in the above embodiments and will not be repeated here.

[0047] S203, based on the target data, obtain the correlation between candidate features and product quality labels.

[0048] It should be noted that there are no excessive restrictions on the quality label of the product. For example, the quality label may include "good" or "poor", or it may include "excellent", "good", "medium", or "poor", or it may include "qualified" or "unqualified".

[0049] It should be noted that there are no excessive restrictions on the relationship. For example, the relationship can be numerical, meaning that the relationship can be quantified in this solution; the larger the value, the stronger the relationship.

[0050] In one implementation, the target data includes multiple sets of data, each set including candidate features and product quality labels. Based on the target data, the correlation between candidate features and product quality labels is obtained, including obtaining the first set of numbers where candidate features and quality labels belong to the same set of data, and obtaining the correlation based on the first set of numbers.

[0051] For example, if the target data includes (candidate feature 1, quality label A), (candidate feature 2, quality label A), (candidate feature 1, quality label B), and (candidate feature 1, quality label B), that is, the target data includes 4 sets of data, then the number of the first set of data in which candidate feature 1 and quality label A belong is 1, the number of the first set of data in which candidate feature 1 and quality label B belong is 2, the number of the first set of data in which candidate feature 2 and quality label A belong is 1, and the number of the first set of data in which candidate feature 2 and quality label B belong is 0.

[0052] In some examples, association relationships are derived based on the first set of numbers. If the first set of numbers is greater than or equal to a set threshold, the association is classified as a first association; conversely, if the first set of numbers is less than the set threshold, the association is classified as a second association. The first association is considered stronger than the second association.

[0053] In some examples, relationships are derived based on the first set of numbers, including relationships derived based on the correspondence between the defined intervals in which the first set of numbers are located and the relationships themselves. It is understandable that the first set of numbers can be pre-divided into multiple defined intervals, with any two defined intervals corresponding to different relationships.

[0054] S204, based on the association relationship, obtain the total importance parameter.

[0055] It should be noted that the total importance parameter of the candidate features may differ depending on the association relationship.

[0056] In one implementation, the total importance parameter is obtained based on the association relationship, including obtaining the total importance parameter based on the correspondence between the association relationship and the total importance parameter.

[0057] In one implementation, the correlation is a numerical value. Based on the correlation, the total importance parameter is obtained, including obtaining the total importance parameter based on the correspondence between the set interval in which the correlation is located and the total importance parameter.

[0058] In one implementation, there are multiple quality labels. Based on the correlation relationships, a total importance parameter is obtained, including obtaining the total importance parameter based on the correlation relationships between candidate features and multiple quality labels. Therefore, this method comprehensively considers the correlation relationships between candidate features and multiple quality labels to obtain the total importance parameter, thus improving the accuracy of the total importance parameter.

[0059] In some examples, the total importance parameter is obtained based on the correlation between candidate features and multiple quality labels. This includes obtaining sub-importance parameters of candidate features based on the correlation between candidate features and quality labels, and then taking a weighted average of the multiple sub-importance parameters to obtain the total importance parameter.

[0060] For example, if the association between candidate feature 1 and quality label A is the first association, then the sub-importance parameter of candidate feature 1 can be determined to be 0.5. If the association between candidate feature 1 and quality label B is the second association, then the sub-importance parameter of candidate feature 1 can be determined to be 1. The sub-importance parameters of candidate feature 1 include 0.5 and 1. The weighted average of 0.5 and 1 yields the total importance parameter of candidate feature 1 as 0.75.

[0061] S205, based on the total importance parameter, selects the target feature from multiple candidate features.

[0062] In summary, the root cause analysis method for product quality according to the embodiments of this disclosure can obtain the correlation between candidate features and product quality labels based on target data, so as to obtain the total importance parameter.

[0063] Figure 3 This is a schematic flowchart of a root cause analysis method for product quality according to a third embodiment of this disclosure.

[0064] like Figure 3 As shown, the root cause analysis method for product quality according to the third embodiment of this disclosure includes:

[0065] S301, Obtain target data for the product.

[0066] S302, Obtain multiple candidate features of the product, where the candidate features affect the quality of the product.

[0067] The details of steps S301-S302 can be found in the above embodiments and will not be repeated here.

[0068] S303, based on the target data, obtains multiple sub-importance parameters of the candidate features.

[0069] In one implementation, such as Figure 4 As shown, the root cause analysis model includes a machine learning model, a causal model, and a statistical model. The statistical model includes a chi-square distribution model and a correlation coefficient model. Based on the target data, multiple sub-importance parameters of the candidate features are obtained. This can include inputting the target data into the machine learning model, the causal model, the chi-square distribution model, and the correlation coefficient model respectively, and having the machine learning model, the causal model, the chi-square distribution model, and the correlation coefficient model output the first sub-importance parameter 1, the first sub-importance parameter 2, the second sub-importance parameter, and the third sub-importance parameter of the candidate features respectively.

[0070] In one implementation, multiple sub-importance parameters of candidate features are obtained based on the target data, which may include the following possible implementation methods:

[0071] Method 1: Based on the target data, obtain the first probability of the product's quality under the quality label, assuming the product features include candidate features. Based on the first probability, obtain the first sub-importance parameter of the candidate features.

[0072] Therefore, this method can obtain the first probability based on the target data, and thus obtain the first sub-importance parameter.

[0073] In some examples, the target data can be input into the machine learning model and the causal model respectively. The machine learning model obtains the first probability based on the target data, and the first sub-importance parameter 1 of the candidate feature is obtained based on the first probability. The causal model obtains the first probability based on the target data, and the first sub-importance parameter 2 of the candidate feature is obtained based on the first probability.

[0074] In some examples, the target data includes multiple sets of data, each set including candidate features and product quality labels. Based on the target data, the first probability of a product's quality under a quality label, assuming the product features include candidate features, is determined by obtaining the number of candidate features and quality labels belonging to the same set of data. The ratio of this first set number to the total number of sets of data included in the target data is then used to determine the first probability.

[0075] In some examples, based on the target data, the method obtains the following probabilities: a first probability that the product's quality includes candidate features, a second probability that the product's quality includes candidate features, a third probability that the product's quality includes candidate features, and a fourth probability that the product's quality includes candidate features. The first probability is then derived from these probabilities. It should be noted that the second and third probabilities are prior probabilities, while the first and fourth probabilities are posterior probabilities. Therefore, this method comprehensively considers the second, third, and fourth probabilities to obtain the first probability, improving its accuracy.

[0076] In some examples, the target data includes multiple sets of data, each set containing candidate features and product quality labels. The second probability is determined by obtaining the number of data points in the second set containing the candidate features, and the ratio of this number to the total number of data sets included in the target data. Similarly, the third probability is determined by obtaining the number of data points in the third set containing the quality labels, and the ratio of this number to the total number of data sets included in the target data. Finally, the fourth probability is determined by obtaining the number of data points in the first set where both candidate features and quality labels belong, and the ratio of this number to the total number of data sets included in the target data.

[0077] In some examples, the first sub-importance parameter of the candidate feature is obtained based on the first probability. This may include obtaining the first sub-importance parameter based on the correspondence between the set interval in which the first probability is located and the first sub-importance parameter.

[0078] In some cases, there are multiple quality labels. Based on the first probability, the first sub-importance parameter of the candidate feature is obtained, including obtaining the first sub-importance parameter of the candidate feature based on multiple first probabilities corresponding to the candidate feature. For example, the first sub-importance parameter can be obtained by weighted averaging of multiple first probabilities corresponding to the candidate feature.

[0079] Method 2: Based on the target data, obtain the product quality. In the case of quality labels, the product features include the fourth probability of candidate features. Based on the fourth probability, obtain the chi-square coefficient between candidate features and quality labels. Based on the chi-square coefficient, obtain the second sub-importance parameter of candidate features.

[0080] Therefore, this method can obtain a fourth probability based on the target data, and then obtain the chi-square coefficient based on the fourth probability to obtain the second sub-importance parameter.

[0081] It should be noted that the relevant content on obtaining the fourth probability based on the target data can be found in the above embodiments, and will not be repeated here.

[0082] In some examples, the target data can be input into the chi-square distribution model, which then obtains a fourth probability based on the target data. Based on the fourth probability, the chi-square coefficients of the candidate features are obtained, and based on the chi-square coefficients, the second sub-importance parameters of the candidate features are obtained.

[0083] In some examples, based on the target data, the fourth probability of the product's quality including candidate feature 1 under quality label A can be obtained. Based on the above fourth probability, the chi-square coefficient between candidate feature 1 and quality label A can be obtained. Alternatively, based on the target data, the fourth probability of the product's quality including candidate feature 1 under quality label B can be obtained. Based on the above fourth probability, the chi-square coefficient between candidate feature 1 and quality label B can be obtained.

[0084] It should be noted that the relevant content of obtaining the chi-square coefficient based on the fourth probability can be achieved by any of the relevant techniques for obtaining the chi-square coefficient, which will not be elaborated here.

[0085] In some examples, the second sub-importance parameter of the candidate feature is obtained based on the chi-square coefficient, which may include determining the chi-square coefficient as the second sub-importance parameter.

[0086] In some cases, there are multiple quality labels, and the second sub-importance parameter of the candidate feature is obtained based on the chi-square coefficient. This includes obtaining the second sub-importance parameter of the candidate feature based on multiple chi-square coefficients corresponding to the candidate feature. For example, the second sub-importance parameter can be obtained by weighted averaging of multiple chi-square coefficients corresponding to the candidate feature.

[0087] Method 3: Based on the target data, obtain the correlation coefficient between candidate features and product quality labels, and based on the correlation coefficient, obtain the third sub-importance parameter of the candidate features.

[0088] Therefore, this method can obtain the correlation coefficient based on the target data to obtain the third sub-importance parameter.

[0089] In some cases, the target data can be input into the correlation coefficient model, which then uses the target data to obtain the correlation coefficient, thus yielding the third sub-importance parameter.

[0090] It should be noted that the relevant content based on the target data to obtain the correlation coefficient can be achieved by any of the correlation coefficient acquisition methods in the relevant technology, which will not be elaborated here.

[0091] In some examples, a third sub-importance parameter for candidate features is obtained based on the correlation coefficient, which may include determining the correlation coefficient as the third sub-importance parameter.

[0092] In some cases, there are multiple quality labels. Based on the correlation coefficient, a third sub-importance parameter is obtained for the candidate feature. This includes obtaining the third sub-importance parameter based on multiple correlation coefficients corresponding to the candidate feature. For example, a weighted average of multiple correlation coefficients corresponding to the candidate feature can be used to obtain the third sub-importance parameter.

[0093] S304 calculates a weighted average of multiple sub-importance parameters to obtain the total importance parameter.

[0094] Continue with Figure 4 For example, the first sub-importance parameter 1, the first sub-importance parameter 2, the second sub-importance parameter, and the third sub-importance parameter can be weighted and averaged to obtain the total importance parameter.

[0095] S305, based on the total importance parameter, selects the target feature from multiple candidate features.

[0096] In summary, the root cause analysis method for product quality according to the embodiments of this disclosure can obtain the sub-importance parameters of candidate features multiple times based on target data to obtain multiple sub-importance parameters, and then perform a weighted average of the multiple sub-importance parameters to obtain the total importance parameter. This method can comprehensively consider the multiple sub-importance parameters of candidate features to obtain the total importance parameter, thereby improving the accuracy of the total importance parameter.

[0097] Figure 5 This is a schematic flowchart of a root cause analysis method for product quality according to the fifth embodiment of this disclosure.

[0098] like Figure 5 As shown, the root cause analysis method for product quality according to the fifth embodiment of this disclosure includes:

[0099] S501, Obtain target data for the product.

[0100] S502, obtain multiple candidate features of the product, where the candidate features affect the quality of the product.

[0101] The details of steps S501-S502 can be found in the above embodiments and will not be repeated here.

[0102] S503. Based on each candidate feature, construct a causal graph, where the first node of the causal graph is used to represent the candidate feature, the second node of the causal graph is used to represent the quality of the product, and the edge between the first node and the second node is used to represent the correlation between the candidate feature and the quality of the product.

[0103] It should be noted that there are no strict limitations on causal graphs. For example, causal graphs can include undirected graphs, directed graphs, isomorphic graphs, heteromorphic graphs, etc.

[0104] It is understandable that there is a one-to-one correspondence between the first node and the candidate feature, meaning the number of first nodes is equal to the number of candidate features. There are multiple first nodes, while there is only one second node. It should be noted that the relationship between candidate features and product quality is not overly restricted; for example, it can include the candidate feature influencing product quality (i.e., the candidate feature is the cause, and product quality is the result).

[0105] In one implementation, the edge between the first node and the second node is an edge from the first node to the second node.

[0106] In some examples, cause-effect graphs are as follows: Figure 6 As shown, the first node includes v1, v2, v3, v4, v5, and v6, and the second node is v7. The edges between the first node v1, v2, and v3 and the second node v7 are e1, e2, and e3, respectively. Edge e1 is the edge from the first node v1 to the second node v7, edge e2 is the edge from the first node v2 to the second node v7, and edge e3 is the edge from the first node v3 to the second node v7.

[0107] In one implementation, the node attributes of the first node include the total importance parameter and sub-importance parameters of the candidate features.

[0108] S504, based on the target data, the fifth probability is that the product features include the second candidate features, assuming the product features include the first candidate features.

[0109] In one implementation, the target data includes multiple sets of data, each set including candidate features and product quality labels. Based on the target data, if the product features include a first candidate feature, the fifth probability that the product features include a second candidate feature includes a second candidate feature. This includes obtaining a fourth group number in which the first and second candidate features belong to the same set of data, and determining the fifth probability as the ratio of the fourth group number to the total number of sets of data included in the target data.

[0110] In one implementation, when the product features include a first candidate feature based on the target data, the fifth probability that the product features include a second candidate feature includes a sixth probability, a seventh probability, and an eighth probability (when the product features include a second candidate feature). The fifth probability is obtained based on the sixth, seventh, and eighth probabilities. It should be noted that the sixth and seventh probabilities are prior probabilities, and the fifth and eighth probabilities are posterior probabilities. Therefore, this method comprehensively considers the sixth, seventh, and eighth probabilities to obtain the fifth probability, improving its accuracy.

[0111] It should be noted that the process of obtaining the sixth and seventh probabilities can be referred to the process of obtaining the second probability in the above embodiments, and the process of obtaining the eighth probability can be referred to the process of obtaining the fourth probability in the above embodiments, and will not be repeated here.

[0112] S505, based on the fifth probability, determines that there is a correlation between the first candidate feature and the second candidate feature.

[0113] It should be noted that the relationship between the first candidate feature and the second candidate feature is not limited in many ways. For example, it can include the first candidate feature influencing the second candidate feature (i.e., the first candidate feature is the cause and the second candidate feature is the result), and the second candidate feature influencing the first candidate feature (i.e., the second candidate feature is the cause and the first candidate feature is the result), etc.

[0114] In one implementation, a correlation is determined between a first candidate feature and a second candidate feature based on a fifth probability, including determining a correlation between the first candidate feature and the second candidate feature if the fifth probability is greater than or equal to a set threshold.

[0115] S506, add connecting edges between the nodes corresponding to the first candidate feature and the nodes corresponding to the second candidate feature to update the causal graph. The edge between any two second nodes is used to represent the association between any two candidate features.

[0116] In one implementation, adding a connection edge between the node corresponding to the first candidate feature and the node corresponding to the second candidate feature includes adding a connection edge in the causal graph pointing from the node corresponding to the first candidate feature to the node corresponding to the second candidate feature in response to the association between the first candidate feature and the second candidate feature affecting the second candidate feature; conversely, adding a connection edge in the causal graph pointing from the node corresponding to the second candidate feature to the node corresponding to the first candidate feature in response to the association between the first candidate feature and the second candidate feature affecting the first candidate feature.

[0117] Continue with Figure 6 For example, if the first nodes v1, v2, v3, v4, v5, and v6 are used to represent candidate features 1 to 6 (not shown in the figure), if there is a relationship between candidate feature 1 and candidate feature 4, then a connecting edge e4 is added between the first nodes v1 and v4; if there is a relationship between candidate feature 1 and candidate feature 5, then a connecting edge e5 is added between the first nodes v1 and v5; if there is a relationship between candidate feature 2 and candidate feature 6, then a connecting edge e6 is added between the first nodes v2 and v6.

[0118] If the relationship between candidate feature 1 and candidate feature 4 is that candidate feature 4 influences candidate feature 1, then edge e4 is the edge from the first node v4 to the first node v1. If the relationship between candidate feature 1 and candidate feature 5 is that candidate feature 5 influences candidate feature 1, then edge e5 is the edge from the first node v5 to the first node v1. If the relationship between candidate feature 2 and candidate feature 6 is that candidate feature 6 influences candidate feature 2, then edge e6 is the edge from the first node v6 to the first node v2.

[0119] In summary, according to the root cause analysis method for product quality according to the embodiments of this disclosure, a fifth probability is obtained based on target data to determine the correlation between the first candidate feature and the second candidate feature, and a connection edge is added between the node corresponding to the first candidate feature and the node corresponding to the second candidate feature to update the causal graph. Thus, the causal graph can reflect the correlation between the candidate feature and the product quality, and can also reflect the correlation between multiple candidate features.

[0120] The collection, storage, use, processing, transmission, provision, and disclosure of user personal information involved in the technical solution disclosed herein comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0121] According to embodiments of this disclosure, this disclosure also provides a root cause analysis apparatus for product quality, used to implement the above-described root cause analysis method for product quality.

[0122] Figure 7 This is a block diagram of a root cause analysis apparatus for product quality according to a first embodiment of the present disclosure.

[0123] like Figure 7 As shown, the product quality root cause analysis device 700 of this disclosure includes: a first acquisition module 701, a second acquisition module 702, an analysis module 703, and a screening module 704.

[0124] The first acquisition module 701 is used to acquire target data of the product;

[0125] The second acquisition module 702 is used to acquire multiple candidate features of the product, wherein the candidate features affect the quality of the product;

[0126] The analysis module 703 is used to obtain the total importance parameter of the candidate features based on the target data, wherein the total importance parameter is used to characterize the importance of the candidate features to the quality of the product;

[0127] The filtering module 704 is used to filter out target features from multiple candidate features based on the total importance parameter.

[0128] In one embodiment of this disclosure, the analysis module 703 is further configured to: obtain the correlation between the candidate features and the quality label of the product based on the target data; and obtain the total importance parameter based on the correlation.

[0129] In one embodiment of this disclosure, there are multiple quality labels, and the analysis module 703 is further configured to: obtain the total importance parameter based on the correlation between the candidate features and the multiple quality labels.

[0130] In one embodiment of this disclosure, the analysis module 703 is further configured to: obtain multiple sub-importance parameters of the candidate features based on the target data; and perform a weighted average of the multiple sub-importance parameters to obtain the total importance parameter.

[0131] In one embodiment of this disclosure, the analysis module 703 is further configured to: based on the target data, obtain a first probability of the product's quality under a quality label when the product's features include the candidate features; and based on the first probability, obtain a first sub-importance parameter of the candidate features.

[0132] In one embodiment of this disclosure, the analysis module 703 is further configured to: based on the target data, obtain a second probability that the product features include the candidate features, a third probability that the product quality is under the quality label, and a fourth probability that the product features include the candidate features when the product quality is under the quality label; and obtain the first probability based on the second probability, the third probability, and the fourth probability.

[0133] In one embodiment of this disclosure, the analysis module 703 is further configured to: construct a causal graph based on each candidate feature, wherein a first node of the causal graph is used to characterize the candidate feature, a second node of the causal graph is used to characterize the quality of the product, and an edge between the first node and the second node is used to characterize the correlation between the candidate feature and the quality of the product; based on the target data, obtain a fifth probability that the product's features include a second candidate feature when the product's features include the first candidate feature; based on the fifth probability, determine that there is a correlation between the first candidate feature and the second candidate feature; add a connection edge between the node corresponding to the first candidate feature and the node corresponding to the second candidate feature to update the causal graph, and an edge between any two second nodes is used to characterize the correlation between any two candidate features.

[0134] In one embodiment of this disclosure, the analysis module 703 is further configured to: based on the target data, obtain a fourth probability that the product's quality, under the condition of a quality label, includes the candidate feature; based on the fourth probability, obtain the chi-square coefficient between the candidate feature and the quality label; and based on the chi-square coefficient, obtain a second sub-importance parameter of the candidate feature.

[0135] In one embodiment of this disclosure, the analysis module 703 is further configured to: obtain a correlation coefficient between the candidate feature and the quality label of the product based on the target data; and obtain a third sub-importance parameter of the candidate feature based on the correlation coefficient.

[0136] In one embodiment of this disclosure, the target data includes multiple sets of data, each set of data including the candidate features and the quality label of the product.

[0137] In summary, the root cause analysis apparatus for product quality according to this disclosure acquires target data of the product and multiple candidate features of the product, wherein the candidate features affect the product quality. Based on the target data, a total importance parameter of the candidate features is obtained, which characterizes the importance of the candidate features to the product quality. Based on the total importance parameter, a target feature is selected from the multiple candidate features. Therefore, based on the target data of the product, the total importance parameter of the candidate features can be obtained to select the target feature from multiple candidate features. Compared to related technologies that mostly rely on the user's subjective experience for root cause analysis, this solution can quantify the importance of candidate features to the product quality and obtain a total importance parameter for root cause analysis, thus improving the accuracy of root cause analysis of product quality.

[0138] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0139] Figure 8 A schematic block diagram of an example electronic device 800 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0140] like Figure 8 As shown, the electronic device 800 includes a computing unit 801, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 802 or a computer program loaded from a storage unit 808 into a random access memory (RAM) 803. The RAM 803 may also store various programs and data required for the operation of the electronic device 800. The computing unit 801, ROM 802, and RAM 803 are interconnected via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0141] Multiple components in electronic device 800 are connected to I / O interface 805, including: input unit 806, such as keyboard, mouse, etc.; output unit 807, such as various types of displays, speakers, etc.; storage unit 808, such as disk, optical disk, etc.; and communication unit 809, such as network card, modem, wireless transceiver, etc. Communication unit 809 allows electronic device 800 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0142] The computing unit 801 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 801 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 801 performs the various methods and processes described above, such as... Figures 1 to 6 The product quality root cause analysis method is described above. For example, in some embodiments, the product quality root cause analysis method may be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 808. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 800 via ROM 802 and / or communication unit 809. When the computer program is loaded into RAM 803 and executed by computing unit 801, one or more steps of the product quality root cause analysis method described above may be performed. Alternatively, in other embodiments, computing unit 801 may be configured to perform the product quality root cause analysis method by any other suitable means (e.g., by means of firmware).

[0143] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0144] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0145] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0146] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0147] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with embodiments of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0148] Computer systems can include clients and servers. Clients and servers are generally geographically separated and typically interact via communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. A server can be a cloud server, also known as a cloud computing server or cloud host, a hosting product within the cloud computing service ecosystem, addressing the shortcomings of traditional physical hosts and VPS (Virtual Private Server, or simply "VPS") services, such as high management difficulty and weak business scalability. Servers can also be servers for distributed systems or servers incorporating blockchain technology.

[0149] According to embodiments of this disclosure, this disclosure also provides a computer program product, including a computer program, wherein when the computer program is executed by a processor, it implements the steps of the root cause analysis method for product quality described in the above embodiments of this disclosure.

[0150] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0151] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A root cause analysis method for product quality, comprising: Obtain target data for the product; Obtain multiple candidate features of the product, wherein the candidate features affect the quality of the product; Based on the target data, the total importance parameter of the candidate feature is obtained. The total importance parameter is used to characterize the importance of the candidate feature to the quality of the product. It is obtained by weighted averaging of multiple sub-importance parameters of the candidate feature. The multiple sub-importance parameters include a first sub-importance parameter, a second sub-importance parameter, and a third sub-importance parameter. Based on the total importance parameter, the target feature is selected from the multiple candidate features; The first sub-importance parameter is determined based on a first probability, which is the probability that the quality of the product is under the quality label when the product features include the candidate features, based on the target data. The second sub-importance parameter is determined based on the chi-square coefficient between the candidate feature and the quality label obtained from the target data; The third sub-importance parameter is determined based on the correlation coefficient between the candidate features and the quality labels obtained from the target data.

2. The method according to claim 1, wherein, The process of obtaining the total importance parameter of the candidate features based on the target data includes: Based on the target data, the correlation between the candidate features and the quality label of the product is obtained; Based on the aforementioned relationship, the total importance parameter is obtained.

3. The method according to claim 2, wherein, The quality labels are multiple, and the total importance parameter is obtained based on the association relationship, including: The total importance parameter is obtained based on the correlation between the candidate features and multiple quality labels.

4. The method according to claim 1, wherein, The process of obtaining the total importance parameter of the candidate features based on the target data includes: Based on the target data, multiple sub-importance parameters of the candidate features are obtained; The total importance parameter is obtained by taking a weighted average of the multiple sub-importance parameters.

5. The method according to claim 4, wherein, The process of obtaining multiple sub-importance parameters of the candidate features based on the target data includes: Based on the target data, the first probability of the product's quality under the quality label is obtained when the product's features include the candidate features. Based on the first probability, the first sub-importance parameter of the candidate feature is obtained.

6. The method according to claim 5, wherein, The first probability of the product's quality under the quality label, obtained based on the target data and considering that the product's features include the candidate features, includes: Based on the target data, a second probability is obtained that the product features include the candidate features, a third probability that the product quality is under the quality label, and a fourth probability that the product features include the candidate features when the product quality is under the quality label. The first probability is obtained based on the second probability, the third probability, and the fourth probability.

7. The method according to claim 5, wherein, Also includes: Based on each candidate feature, a causal graph is constructed, wherein a first node of the causal graph is used to characterize the candidate feature, a second node of the causal graph is used to characterize the quality of the product, and the edge between the first node and the second node is used to characterize the correlation between the candidate feature and the quality of the product; Based on the target data, if the product's features include the first candidate features, then the product's features include the second candidate features, which is a fifth probability. Based on the fifth probability, it is determined that there is a correlation between the first candidate feature and the second candidate feature; Add connecting edges between the node corresponding to the first candidate feature and the node corresponding to the second candidate feature to update the causal graph. The edge between any two second nodes is used to characterize the association between any two candidate features.

8. The method according to claim 4, wherein, The process of obtaining multiple sub-importance parameters of the candidate features based on the target data includes: Based on the target data, the quality of the product, under the condition of quality label, is obtained, and the characteristics of the product include the fourth probability of the candidate features; Based on the fourth probability, the chi-square coefficient between the candidate feature and the quality label is obtained; Based on the chi-square coefficient, the second sub-importance parameter of the candidate feature is obtained.

9. The method according to claim 4, wherein, The process of obtaining multiple sub-importance parameters of the candidate features based on the target data includes: Based on the target data, the correlation coefficient between the candidate features and the quality label of the product is obtained; Based on the correlation coefficient, the third sub-importance parameter of the candidate feature is obtained.

10. The method according to any one of claims 1-9, wherein, The target data includes multiple sets of data, each set of data including the candidate features and the quality label of the product.

11. A root cause analysis device for product quality, comprising: The first acquisition module is used to acquire target data for the product; The second acquisition module is used to acquire multiple candidate features of the product, the candidate features affecting the quality of the product; The analysis module is used to obtain the total importance parameter of the candidate feature based on the target data. The total importance parameter is used to characterize the importance of the candidate feature to the quality of the product. It is obtained by weighted averaging of multiple sub-importance parameters of the candidate feature. The multiple sub-importance parameters include a first sub-importance parameter, a second sub-importance parameter, and a third sub-importance parameter. A filtering module is used to filter target features from multiple candidate features based on the total importance parameter; The first sub-importance parameter is determined based on a first probability, which is the probability that the quality of the product is under the quality label when the product features include the candidate features, based on the target data. The second sub-importance parameter is determined based on the chi-square coefficient between the candidate feature and the quality label obtained from the target data; The third sub-importance parameter is determined based on the correlation coefficient between the candidate features and the quality labels obtained from the target data.

12. The apparatus according to claim 11, wherein, The analysis module is also used for: Based on the target data, the correlation between the candidate features and the quality label of the product is obtained; Based on the aforementioned relationship, the total importance parameter is obtained.

13. The apparatus according to claim 12, wherein, The quality labels are multiple, and the analysis module is also used for: The total importance parameter is obtained based on the correlation between the candidate features and multiple quality labels.

14. The apparatus according to claim 11, wherein, The analysis module is also used for: Based on the target data, multiple sub-importance parameters of the candidate features are obtained; The total importance parameter is obtained by taking a weighted average of the multiple sub-importance parameters.

15. The apparatus according to claim 14, wherein, The analysis module is also used for: Based on the target data, the first probability of the product's quality under the quality label is obtained when the product's features include the candidate features. Based on the first probability, the first sub-importance parameter of the candidate feature is obtained.

16. The apparatus according to claim 15, wherein, The analysis module is also used for: Based on the target data, a second probability is obtained that the product features include the candidate features, a third probability that the product quality is under the quality label, and a fourth probability that the product features include the candidate features when the product quality is under the quality label. The first probability is obtained based on the second probability, the third probability, and the fourth probability.

17. The apparatus according to claim 15, wherein, The analysis module is also used for: Based on each candidate feature, a causal graph is constructed, wherein a first node of the causal graph is used to characterize the candidate feature, a second node of the causal graph is used to characterize the quality of the product, and the edge between the first node and the second node is used to characterize the correlation between the candidate feature and the quality of the product; Based on the target data, if the product's features include the first candidate features, then the product's features include the second candidate features, which is a fifth probability. Based on the fifth probability, it is determined that there is a correlation between the first candidate feature and the second candidate feature; Add connecting edges between the node corresponding to the first candidate feature and the node corresponding to the second candidate feature to update the causal graph. The edge between any two second nodes is used to characterize the association between any two candidate features.

18. The apparatus according to claim 14, wherein, The analysis module is also used for: Based on the target data, the quality of the product, under the condition of quality label, is obtained, and the characteristics of the product include the fourth probability of the candidate features; Based on the fourth probability, the chi-square coefficient between the candidate feature and the quality label is obtained; Based on the chi-square coefficient, the second sub-importance parameter of the candidate feature is obtained.

19. The apparatus according to claim 14, wherein, The analysis module is also used for: Based on the target data, the correlation coefficient between the candidate features and the quality label of the product is obtained; Based on the correlation coefficient, the third sub-importance parameter of the candidate feature is obtained.

20. The apparatus according to any one of claims 11-19, wherein, The target data includes multiple sets of data, each set of data including the candidate features and the quality label of the product.

21. An electronic device, comprising: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the root cause analysis method for product quality as described in any one of claims 1-10.

22. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to perform the root cause analysis method for product quality as described in any one of claims 1-10.

23. A computer program product comprising a computer program that, when executed by a processor, implements the steps of the root cause analysis method for product quality as described in any one of claims 1-10.

Images