Product quality traceability analysis method, electronic equipment and storage medium

Abstract

The invention relates to the technical field of computers, in particular to a product quality traceability analysis method, electronic equipment and a storage medium, and the method comprises the steps: inputting a plurality of quality monitoring index data of a target product into a fault monitoring model, obtaining each fault problem corresponding to the target product and an initial score of each fault problem, and further obtaining a first fault cause analysis text corresponding to the target product according to the obtained result, respectively obtaining the association degree of each fault problem and the remaining fault problems, optimizing the initial score of each fault problem according to the association degree, and inputting the optimization result into a given large language model to obtain a fault cause analysis result. And finally, determining a fault tracing production link corresponding to the target product based on the first fault reason analysis text and the second fault reason analysis text. According to the invention, the accuracy and reliability of fault reason prediction can be improved, and the efficiency and accuracy of product quality traceability are further improved.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and particularly to a method for traceability analysis of product quality, an electronic device, and a storage medium. Background Art

[0002] Currently, in the production process of traditional manufacturing, generally multiple production links are involved, and different components may even involve different production enterprises. Due to the large number of production nodes and hierarchical distributions in the production cycle, product quality inspection is generally carried out at the end of the production cycle or before the product leaves the factory. When a failure occurs, it is impossible to trace back in time, that is, it is impossible to find the production link where the failure occurs in time. In actual operation, generally, based on the experience of the staff, it is guessed which link the problem appears in. However, limited by the technical level mastered by the inspectors, for example, the staff in different links cannot fully understand the production mechanisms of other links, which will lead to incorrect traceability and affect the traceability efficiency. And since the manufacturing link is still running, more defective products will be produced. Therefore, timely discovering the source of the failure is crucial for both the production of the enterprise and the improvement of product quality. Summary of the Invention

[0003] In view of the above technical problems, the present invention provides a method for traceability analysis of product quality, an electronic device, and a storage medium, which can improve the accuracy and reliability of predicting the cause of failure, and further improve the efficiency and accuracy of product quality traceability.

[0004] According to a first aspect of the present invention, there is provided a method for traceability analysis of product quality, including the following steps:

[0005] Obtain a plurality of quality monitoring index data of a target product, and input the plurality of quality monitoring index data into a trained fault monitoring model to obtain each fault problem corresponding to the target product and an initial score of each fault problem.

[0006] Input each fault problem corresponding to the target product and the initial score of each fault problem into a trained given large language model to obtain a first fault cause analysis text corresponding to the target product.

[0007] Respectively obtain the degree of association between each fault problem and the remaining fault problems; the remaining fault problems refer to other fault problems among all the fault problems corresponding to the target product.

[0008] Optimize the initial score of each fault problem according to the degree of association between each fault problem and the remaining fault problems to obtain the final score of each fault problem, and input the final score of each fault problem into the given large language model to obtain a second fault cause analysis text.

[0009] Based on the first failure cause analysis text and the second failure cause analysis text, the occurrence probability of each failure problem corresponding to the target product is obtained, and the failure traceability production link corresponding to the target product is determined according to the occurrence probability of each failure problem.

[0010] According to a second aspect of the present invention, there is provided a non-transitory computer-readable storage medium storing at least one instruction or at least one program segment, and the at least one instruction or the at least one program segment is loaded and executed by a processor to implement the above product quality traceability analysis method.

[0011] According to a third aspect of the present invention, there is provided an electronic device including a processor and the above non-transitory computer-readable storage medium.

[0012] The present invention has at least the following beneficial effects:

[0013] In the product quality traceability analysis method of the present invention, first, a plurality of quality monitoring index data of the target product obtained are input into a failure monitoring model to obtain each failure problem corresponding to the target product and an initial score of each failure problem, and the obtained results are input into a large language model to obtain a first failure cause analysis text corresponding to the target product. Then, the correlation degree between each failure problem and the remaining failure problems is respectively obtained, the initial score of each failure problem is optimized according to the correlation degree corresponding to each failure problem, and the optimized final score is input into a given large language model to obtain a second failure cause analysis text. By introducing the correlation degree between different failure problems and improving the scores of the failure problems with correlation degree, it is beneficial to improve the accuracy of failure cause prediction and make the obtained second failure cause analysis text more reliable. Finally, based on the first failure cause analysis text and the second failure cause analysis text, the failure traceability production link corresponding to the target product is determined. By comprehensively considering the two failure cause analysis texts, the accuracy and reliability of failure cause prediction can be improved, and thus the efficiency and accuracy of product quality traceability can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can also obtain other drawings based on these drawings without creative efforts.

[0015] Figure 1 It is a flowchart of the product quality traceability analysis method provided by the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0016] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of the present invention.

[0017] This embodiment provides a method for traceability analysis of product quality, as Figure 1 shown, the method includes the following steps:

[0018] S100, obtain a number of quality monitoring index data of the target product, and input the number of quality monitoring index data into the trained fault monitoring model to obtain each fault problem corresponding to the target product and the initial score of each fault problem; it can be understood that: the initial score of the fault problem can be the confidence of the fault problem output by the fault monitoring model.

[0019] Specifically, the quality monitoring index data refers to the data obtained by detecting the target product according to a number of preset quality monitoring indexes. For example, when the target product is rubber, the detection items include physical property detection, chemical property detection, electrical property detection, and environmental protection property detection, etc. Among them, the quality monitoring indexes involved in physical property detection are hardness, tensile strength, tear strength, etc., the quality monitoring indexes involved in chemical property detection are chemical corrosion resistance, heat resistance, oxidation resistance, etc., the quality monitoring indexes involved in electrical property detection are insulation resistance, dielectric constant, and dielectric loss, etc., and the quality monitoring indexes involved in environmental protection property detection are the content of toxic and harmful substances in rubber; when the target product is a rubber tire, in actual operation, it also includes appearance detection, uniformity detection, dynamic balance test, and X-ray detection of the rubber tire.

[0020] Furthermore, each fault problem corresponds to a quality monitoring index data. For example, when the hardness is unqualified, it is considered that there is a hardness fault, and when the tensile strength is unqualified, it is considered that there is a tensile fault.

[0021] Specifically, the fault monitoring model is a neural network prediction model, which is used to predict the fault problems of the target product according to a number of quality monitoring index data of the target product. Those skilled in the art know the specific training process of the neural network prediction model, which will not be elaborated here.

[0022] S200, input each fault problem corresponding to the target product and the initial score of each fault problem into the trained given large language model to obtain the first fault cause analysis text corresponding to the target product. In a specific implementation, the large language model is trained according to the scores and fault cause analysis results of historical fault products to obtain the trained given large language model.

[0023] As described above, several quality monitoring index data of the target product can be used to obtain the fault problems of the target product, and a first fault cause analysis text can be obtained based on several fault problems and the score of each fault problem. Furthermore, the predicted cause of the fault problem can be obtained from the first fault cause analysis text, which is beneficial to tracing the production link when the fault occurs in the target product.

[0024] S300, respectively obtain the correlation degree between each fault problem and the remaining fault problems; the remaining fault problems refer to other fault problems among all the fault problems corresponding to the target product; it can be understood that for any fault problem, obtain the correlation degree between each fault problem other than the any fault problem and the any fault problem.

[0025] Specifically, the correlation degree between each fault problem and the remaining fault problems is obtained through the following steps:

[0026] S301, obtain the historical fault product data set; the historical fault product data set includes several historical fault products and several fault problems corresponding to each historical fault product.

[0027] S302, respectively use any two fault problems as the first fault problem and the second fault problem, and determine the historical fault products that have both the first fault problem and the second fault problem based on the several fault problems corresponding to each historical fault product, and all of them are used as target fault products.

[0028] S303, when the proportion of the number of target fault products in the total number of all obtained historical fault products is less than the preset proportion threshold, determine that there is no correlation between the first fault problem and the second fault problem; among them, those skilled in the art set the preset proportion threshold according to actual needs. For example, 0.3, that is, when the proportion of the number of target fault products in the total number of all obtained historical fault products is less than the preset proportion threshold, it is considered that the first fault problem and the second fault problem are problems that occur simultaneously by chance and have no correlation.

[0029] S304, when the proportion of the number of target fault products in the total number of all obtained historical fault products is not less than the preset proportion threshold, determine the proportion of the number of target fault products in the total number of all obtained historical fault products as the correlation degree between the first fault problem and the second fault problem.

[0030] As mentioned above, for several fault problems, considering that in the actual production process, multiple fault problems may occur in the same production link. For example, for hardness and tear strength, when the hardness is low, the tear strength may also be relatively low, so the correlation degree between the two is high. It is also possible that the semi-finished products in the previous production link will affect the production effect of the next production link, and their correlation degree is also high. Therefore, the correlation degree of different fault problems is introduced, and on this basis, the scores of fault problems are optimized, which is beneficial to predicting more accurate and reliable fault causes.

[0031] S400. Optimize the initial score of each fault problem according to the correlation degree between each fault problem and the remaining fault problems to obtain the final score of each fault problem, and input the final score of each fault problem into the given large language model to obtain the second fault cause analysis text. It can be understood that both the first fault cause analysis text and the second fault cause analysis text include the determined final fault causes and the causes of each fault cause. For example, product faults caused by process parameters not meeting requirements in the manufacturing process or product faults caused by overheating or jitter after the equipment has run for too long, which helps the staff to analyze the faults and verify the traceability.

[0032] Further, step S400 includes the following steps:

[0033] S401. For any selected fault problem, obtain the correlation degree between the selected fault problem and each of the remaining fault problems. It can be understood that the remaining fault problems here are any fault problems corresponding to the target product except the selected fault problem. Since the correlation degree between each fault problem and other fault problems has been obtained in step S300, the correlation degree between the selected fault problem and each of the remaining fault problems can be known.

[0034] S402. Calculate the final score corresponding to the selected fault problem according to the correlation degree between the selected fault problem and each of the remaining fault problems. The final score corresponding to the selected fault problem meets the following conditions:

[0035] S = S 0 ×(1 + ∑ n j＝1 K j / η), where S is the final score corresponding to the selected fault problem, S 0 is the initial score corresponding to the selected fault problem, K j is the correlation degree between the selected fault problem and the j-th fault problem among the remaining fault problems, n is the number of fault problems among the remaining fault problems, and η is a preset score reduction reference threshold. It can be understood that ∑ nj＝1 K j Try to control the value of / η between 0 and 1 so that the final score obtained will not be overly amplified.

[0036] As described above, considering that when two fault problems with a degree of association occur simultaneously, it indicates that the prediction accuracy of these two fault problems is relatively high. Therefore, the scores of these two fault problems should be correspondingly increased. By improving the scores of the fault problems with a degree of association, it is beneficial to improve the accuracy of fault cause prediction and make the obtained second fault cause analysis text more reliable.

[0037] S500. Based on the first fault cause analysis text and the second fault cause analysis text, obtain the occurrence probability of each fault problem corresponding to the target product, and determine the fault tracing production link corresponding to the target product according to the occurrence probability of each fault problem. For example, for rubber tires, the production links include: the link of adding rubber preparations to natural rubber, the fiber cord production link, the bead and steel cord production link, the calendering link of the ply, the cutting link of the ply, the calendering link of the belt layer, the cutting link of the belt layer, the bead forming link, the inner liner pressing link, the tire vulcanization link, etc.

[0038] In a specific embodiment, the occurrence probability of any fault problem corresponding to the target product is obtained through the following steps:

[0039] S501. Based on the order of appearance of several fault problems in the first fault cause analysis text, assign the first accurate prediction probability to each fault problem in the first fault cause analysis text according to the preset probability distribution rule. In a specific implementation, since the fault problems with high prediction accuracy will be proposed preferentially at the front in the text, a larger accurate prediction probability should be assigned to the fault problems that appear earlier in the order.

[0040] Furthermore, the preset probability distribution rule means that as the order of appearance of the fault problems goes from front to back, the assigned accurate prediction probability becomes smaller and smaller, and the sum of the assigned accurate prediction probabilities is 1. In a specific implementation, the probability can also be assigned according to a pre-set probability distribution table, that is, the accurate prediction probability of each fault problem is pre-set under different numbers of fault problems. For example, when there are 3 fault problems, the probabilities are set to 0.5, 0.3, and 0.2 in sequence; when there are 2 fault problems, the probabilities are set to 0.7 and 0.3 in sequence.

[0041] S502. Based on the order of appearance of several fault problems in the second fault cause analysis text, assign the second accurate prediction probability to each fault problem in the second fault cause analysis text according to the preset probability distribution rule.

[0042] Specifically, the method for assigning the second accurate prediction probability to the fault problem is the same as the method for assigning the first accurate prediction probability to the fault problem, which will not be elaborated here.

[0043] S503. Calculate the occurrence probability P of any fault problem corresponding to the target product according to the first accurate prediction probability of each fault problem in the first fault cause analysis text, the second accurate prediction probability of each fault problem in the second fault cause analysis text, and the preset text weights corresponding to the first fault cause analysis text and the second fault cause analysis text respectively. The occurrence probability P of any fault problem corresponding to the target product meets the following conditions:

[0044] P = P 1 × w 1 + P 2 × w 2 where P 1 is the first accurate prediction probability corresponding to any fault problem, P 2 is the second accurate prediction probability corresponding to any fault problem, w 1 is the preset text weight corresponding to the first fault cause analysis text, and w 2 is the preset text weight corresponding to the second fault cause analysis text.

[0045] Specifically, the determining of the production link for fault tracing corresponding to the target product according to the occurrence probability of each fault problem includes the following steps:

[0046] S510. When the occurrence probability of any fault problem corresponding to the target product is greater than the preset probability threshold, determine the any fault problem as the target fault corresponding to the target product. In a specific implementation, the fault problems corresponding to the occurrence probability not greater than the preset probability threshold can be used as alternative faults. When there is no problem in the production link corresponding to the target fault after verification, the staff can then detect the production link corresponding to the alternative faults.

[0047] S520. Determine the production link corresponding to each target fault according to the preset relationship mapping table. The relationship mapping table includes several preset fault problems and the production links corresponding to each preset fault problem.

[0048] As described above, by comprehensively considering the first fault cause analysis text and the second fault cause analysis text, the prediction rationality of the occurrence probability of each fault problem can be improved. On this basis, the mapped production link is made more reliable, thereby improving the accuracy of product quality tracing. Moreover, only by inputting several quality inspection index data of the target product can the predicted production link be obtained, improving the efficiency of product quality tracing.

[0049] Furthermore, the method also determines the occurrence probability of each fault problem corresponding to the target product through the following steps:

[0050] S10. Input the initial scores of each fault problem corresponding to the target product into a given large language model respectively to obtain a number of third fault cause analysis texts.

[0051] S20. Based on the first fault cause analysis text, the second fault cause analysis text, and a number of third fault cause analysis texts, determine the occurrence probability of each fault problem corresponding to the target product. The obtaining method of the occurrence probability of the fault problem is the same as that in step S500 and will not be elaborated here.

[0052] As described above, by increasing the number of fault cause analysis texts, it is equivalent to improving the detection data samples. And the third fault cause analysis text corresponding to one fault problem may also involve other fault problems. Therefore, it can make the determined occurrence probability of each fault problem corresponding to the target product more reasonable and reliable, which is beneficial to tracing the production link of the fault problems of the target product.

[0053] In summary, for the product quality traceability analysis method of the present invention, first, a number of quality monitoring index data of the target product obtained are input into the fault monitoring model to obtain each fault problem corresponding to the target product and the initial score of each fault problem, and the obtained results are input into the large language model to obtain the first fault cause analysis text corresponding to the target product. Then, the correlation degree between each fault problem and the remaining fault problems is obtained respectively. The initial score of each fault problem is optimized according to the correlation degree corresponding to each fault problem, and the optimized final score is input into the given large language model to obtain the second fault cause analysis text. By introducing the correlation degree between different fault problems and improving the scores of the fault problems with correlation degree, it is beneficial to improve the accuracy of fault cause prediction and make the obtained second fault cause analysis text more reliable. Finally, based on the first fault cause analysis text and the second fault cause analysis text, the fault traceability production link corresponding to the target product is determined. By comprehensively considering the two fault cause analysis texts, the accuracy and reliability of fault cause prediction can be improved, and further the efficiency and accuracy of product quality traceability can be improved.

[0054] An embodiment of the present invention also provides a non-transitory computer-readable storage medium, which can be set in an electronic device to store at least one instruction or at least one program related to a method for implementing a method in the method embodiment. The at least one instruction or the at least one program is loaded and executed by the processor to implement the method provided in the above embodiment.

[0055] An embodiment of the present invention further provides an electronic device, including a processor and the aforementioned non-transitory computer-readable storage medium.

[0056] Although some specific embodiments of the present invention have been described in detail by way of examples, those skilled in the art should understand that the above examples are for illustrative purposes only and are not intended to limit the scope of the present invention. Those skilled in the art should also understand that various modifications can be made to the embodiments without departing from the scope and spirit of the present invention. The scope of the present invention is defined by the appended claims.

Claims

1. A product quality traceability analysis method, characterized in that: The method comprises the following steps: Obtaining a number of quality monitoring indicator data of the target product, and inputting the quality monitoring indicator data into the trained fault monitoring model to obtain each fault problem corresponding to the target product and an initial score of each fault problem; Input each fault problem corresponding to the target product and the initial score of each fault problem into the trained given large language model to obtain the first fault cause analysis text corresponding to the target product; Obtaining the degree of correlation between each fault problem and the remaining fault problems respectively; the remaining fault problems refer to other fault problems among all fault problems corresponding to the target product; The initial score of each fault problem is optimized according to the correlation degree between each fault problem and the remaining fault problems to obtain the final score of each fault problem, and the final score of each fault problem is input into a given large language model to obtain a second fault cause analysis text; Based on the first fault cause analysis text and the second fault cause analysis text, the occurrence probability of each fault problem corresponding to the target product is obtained, and the fault tracing production link corresponding to the target product is determined according to the occurrence probability of each fault problem.

2. The product quality traceability analysis method according to claim 1, characterized in that: Obtain the degree of correlation between each fault problem and the remaining fault problems by following the steps below: Acquire a historical faulty product data set; the historical faulty product data set includes a number of historical faulty products and a number of fault problems corresponding to each historical faulty product; Any two fault problems are respectively regarded as the first fault problem and the second fault problem, and according to a number of fault problems corresponding to each historical faulty product, historical faulty products having both the first fault problem and the second fault problem are determined and all are regarded as target faulty products; When the ratio of the number of target faulty products to the number of all acquired historical faulty products is less than a preset ratio threshold, it is determined that the first fault problem and the second fault problem have no correlation relationship; When the ratio of the number of target faulty products to the number of all acquired historical faulty products is not less than a preset ratio threshold, the ratio of the number of target faulty products to the number of all acquired historical faulty products is determined as the correlation degree between the first fault problem and the second fault problem.

3. The product quality traceability analysis method according to claim 1, characterized in that: The step of optimizing the initial score of each fault problem according to the degree of correlation between each fault problem and the remaining fault problems comprises the following steps: For any selected fault problem, obtaining a correlation degree between the selected fault problem and each of the remaining fault problems; According to the degree of association between the selected fault problem and each of the remaining fault problems, a final score corresponding to the selected fault problem is calculated, wherein the final score corresponding to the selected fault problem meets the following conditions: S=S0×(1+∑ n j＝1 K j / η), where S is the final score corresponding to the selected fault problem, S0 is the initial score corresponding to the selected fault problem, and K j is the correlation degree between the selected fault problem and the jth fault problem in the remaining fault problems, n is the number of fault problems in the remaining fault problems, and η is the preset score reduction reference threshold.

4. The product quality traceability analysis method according to claim 1, characterized in that: The probability of occurrence of any fault problem corresponding to the target product is obtained through the following steps: Based on the order of occurrence of the plurality of fault problems in the first fault cause analysis text, assigning a first accurate prediction probability to each fault problem in the first fault cause analysis text according to a preset probability assignment rule; Based on the order of occurrence of the plurality of fault problems in the second fault cause analysis text, assigning a second accurate prediction probability to each fault problem in the second fault cause analysis text according to a preset probability assignment rule; The probability of occurrence P of any fault problem corresponding to the target product is calculated based on the first accurate prediction probability of each fault problem in the first fault cause analysis text, the second accurate prediction probability of each fault problem in the second fault cause analysis text, and the preset text weights corresponding to the first fault cause analysis text and the second fault cause analysis text respectively; the probability of occurrence P of any fault problem corresponding to the target product meets the following conditions: P=P1×w1+P2×w2, where P1 is the first accurate prediction probability corresponding to any fault problem, P2 is the second accurate prediction probability corresponding to any fault problem, w1 is the preset text weight corresponding to the first fault cause analysis text, and w2 is the preset text weight corresponding to the second fault cause analysis text.

5. The product quality traceability analysis method according to claim 1, characterized in that: The method of determining the fault tracing production link corresponding to the target product according to the probability of occurrence of each fault problem includes the following steps: When the probability of occurrence of any fault problem corresponding to the target product is greater than a preset probability threshold, the any fault problem is determined as a target fault corresponding to the target product; Determine the production link corresponding to each target fault according to the preset relationship mapping table; The relationship mapping table includes a plurality of preset fault problems and a production link corresponding to each preset fault problem.

6. The product quality traceability analysis method according to claim 1, characterized in that: The method also determines the occurrence probability of each fault problem corresponding to the target product through the following steps: Inputting the initial score of each fault problem corresponding to the target product into a given large language model respectively, and obtaining a plurality of third fault cause analysis texts; Based on the first fault cause analysis text, the second fault cause analysis text and a plurality of third fault cause analysis texts, the occurrence probability of each fault problem corresponding to the target product is determined.

7. A non-transitory computer-readable storage medium, wherein at least one instruction or at least one program is stored in the storage medium, characterized in that: The at least one instruction or the at least one program is loaded and executed by the processor to implement the product quality traceability analysis method as described in any one of claims 1-6.

8. An electronic device, characterized in that: The invention comprises a processor and the non-transitory computer-readable storage medium as claimed in claim 7.

Images