Method and system for detecting production quality of lithium battery protection plate

Abstract

The invention provides a lithium battery protection board production quality detection method and system, and relates to the technical field of intelligent detection, and the method comprises the steps: obtaining a plurality of detection priorities according to a plurality of test times and a plurality of fault probability evaluations, carrying out the sequential optimization of a plurality of predetermined detection steps, and generating an optimized detection process; carrying out quality detection according to an optimized detection process, recording a detection data set meeting an expected index, and carrying out fault probability prediction on subsequent associated detection steps; and judging the fault prediction probability set according to an early warning probability threshold, and determining an early warning detection step set to perform queue jumping detection. According to the method, the technical problems of product detection resource waste and relatively low detection efficiency caused by insufficient adaptation degree of a detection step sequence and product characteristics due to product quality detection based on a fixed detection step sequence in a traditional lithium battery protection plate quality detection method can be solved, defective products can be rapidly detected, and the detection efficiency is improved. And unnecessary detection steps and detection time are reduced.

Description

Technical Field

[0001] The present application relates to the field of intelligent detection technology, and in particular to a production quality detection method and system for a lithium battery protection board. Background Art

[0002] With the continuous development of lithium battery technology and the expansion of its application fields, the importance of production quality inspection of lithium battery protection boards, as an important component of lithium batteries, has become increasingly prominent. In the actual production process, the quality of lithium battery protection boards is affected by many factors, such as raw material quality, production process, equipment accuracy, etc. In order to ensure that the quality of lithium battery protection boards meets the requirements, strict quality inspections are required.

[0003] At present, the traditional lithium battery protection board production quality inspection method is often based on a fixed inspection step sequence for product quality inspection, resulting in insufficient compatibility between the inspection step sequence and product characteristics, and failure to detect defective products in advance, resulting in technical problems such as waste of product inspection resources and low inspection efficiency. Summary of the invention

[0004] The purpose of this application is to provide a production quality inspection method and system for lithium battery protection boards, so as to solve the technical problems that traditional production quality inspection methods for lithium battery protection boards often perform product quality inspection based on a fixed sequence of inspection steps, resulting in insufficient adaptability between the inspection step sequence and product characteristics, and inability to detect defective products in advance, resulting in waste of product inspection resources and low inspection efficiency.

[0005] In view of the above problems, the present application provides a production quality detection method and system for a lithium battery protection board.

[0006] In a first aspect, the present application provides a production quality inspection method for a lithium battery protection board, which is implemented by a production quality inspection system for a lithium battery protection board, including: reading a predetermined inspection process for a target lithium battery protection board, wherein the predetermined inspection process includes multiple predetermined inspection steps, and the predetermined inspection steps are marked with a test time; obtaining a key processing parameter set and an attribute feature set for the target lithium battery protection board, and using the multiple predetermined inspection steps, key processing parameter sets, and attribute feature sets as constraints to perform positive sample retrieval based on the industrial Internet to obtain a sample fault data set; determining multiple fault probabilities based on analysis of the sample fault data set, and obtaining multiple fault probabilities based on multiple test times and the multiple fault probability assessments. According to the detection priority, the detection order of the multiple predetermined detection steps is optimized to generate an optimized detection process; through the predetermined detection equipment, the target lithium battery protection board is quality inspected according to the optimized detection process, the detection data set that meets the expected indicators is recorded, and the failure probability of subsequent related detection steps is predicted according to the detection data set, and a failure prediction probability set is output; the fault prediction probability set is judged according to the early warning probability threshold, and the subsequent related detection steps corresponding to the failure prediction probability that does not meet the early warning probability threshold are marked with early warning, and the early warning detection step set is determined; based on the early warning detection step set, the target lithium battery protection board is queued for detection, and the quality detection result is output.

[0007] In the second aspect, the present application also provides a production quality inspection system for a lithium battery protection board, which is used to execute a production quality inspection method for a lithium battery protection board as described in the first aspect, including: a predetermined inspection process reading module, which is used to read a predetermined inspection process for a target lithium battery protection board, wherein the predetermined inspection process includes multiple predetermined inspection steps, and the predetermined inspection steps are marked with a test time; a sample fault data set acquisition module, which is used to obtain a key processing parameter set and an attribute feature set of the target lithium battery protection board, and to perform positive sample retrieval based on the industrial Internet with the multiple predetermined inspection steps, key processing parameter sets and attribute feature sets as constraints to obtain a sample fault data set; a detection sequence optimization module, which is used to analyze and determine multiple fault probabilities based on the sample fault data set, and to optimize the detection sequence according to multiple test times and the multiple fault probabilities. The method comprises the following steps: a first step is to evaluate the detection rate and obtain multiple detection priorities, optimize the detection sequence of the multiple predetermined detection steps, and generate an optimized detection process; a subsequent fault probability prediction module is used to perform quality inspection on the target lithium battery protection board according to the optimized detection process through a predetermined detection device, record the detection data set that meets the expected indicators, predict the fault probability of subsequent related detection steps according to the detection data set, and output a fault prediction probability set; a warning detection step identification module is used to judge the fault prediction probability set according to a warning probability threshold, and to mark the subsequent related detection steps corresponding to the fault prediction probability that does not meet the warning probability threshold, and determine the warning detection step set; a queue-jumping detection module is used to perform queue-jumping detection on the target lithium battery protection board based on the warning detection step set, and output a quality detection result.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages: By taking multiple predetermined detection steps, key processing parameter sets and attribute feature sets of the target lithium battery protection board as constraints, a sample fault data set is retrieved and obtained; then, multiple fault probabilities are determined based on the analysis of the sample fault data set, and multiple detection priorities are obtained according to multiple test times and multiple fault probability assessments. The detection sequence of multiple predetermined detection steps is optimized to generate an optimized detection process, which can improve the adaptability of the product detection sequence and product characteristics; then, the target lithium battery protection board is quality inspected according to the optimized detection process through predetermined detection equipment, and the detection data set that meets the expected indicators is recorded. The failure probability of subsequent related detection steps is predicted based on the detection data set, and the output is A fault prediction probability set is generated; the fault prediction probability set is then judged according to the warning probability threshold, and subsequent related detection steps corresponding to the fault prediction probability that does not meet the warning probability threshold are warned, and the warning detection step set is determined; finally, based on the warning detection step set, a queue-jumping detection is performed on the target lithium battery protection board, and the quality detection result is output; by optimizing the detection step sequence according to the product failure probability prediction result, the detection steps with a higher failure probability can be executed in advance, the adaptability of the detection step sequence to the product characteristics can be improved, defective products can be detected quickly, and unnecessary detection steps and detection time can be reduced, thereby achieving the technical effect of improving the utilization rate of detection resources and detection efficiency.

[0009] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented according to the contents of the specification, and in order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are specifically cited below. It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easy to understand through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in the present application or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0011] Figure 1 A schematic diagram of a production quality inspection method for a lithium battery protection board for this application; Figure 2 A schematic diagram of a process for determining multiple fault probabilities in a production quality inspection method for a lithium battery protection board of the present application; Figure 3 This is a structural schematic diagram of a production quality inspection system for a lithium battery protection board according to the present application.

[0012] Description of reference numerals: A predetermined detection process reading module 11, a sample fault data set acquisition module 12, a detection sequence optimization module 13, a subsequent fault probability prediction module 14, a warning detection step identification module 15, and a queue-jumping detection module 16. DETAILED DESCRIPTION

[0013] This application provides a production quality inspection method and system for lithium battery protection boards, which solves the technical problems that the traditional production quality inspection method for lithium battery protection boards often performs product quality inspection based on a fixed inspection step sequence, resulting in insufficient adaptability between the inspection step sequence and product characteristics, and inability to detect defective products in advance, resulting in waste of product inspection resources and low inspection efficiency. By optimizing the inspection step sequence according to the product failure probability prediction results, the inspection steps with a higher failure probability can be executed in advance, the adaptability between the inspection step sequence and product characteristics can be improved, defective products can be detected quickly, and unnecessary inspection steps and inspection time can be reduced, thereby achieving the technical effect of improving the utilization rate of inspection resources and inspection efficiency.

[0014] Below, the technical solutions in the present application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments of the present application. It should be understood that the present application is not limited to the example embodiments described herein. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application. It should also be noted that, for the convenience of description, only the parts related to the present application are shown in the accompanying drawings, rather than all of them.

[0015] For example, please refer to the attached Figure 1 The present application provides a production quality inspection method for a lithium battery protection board, which is applied to a production quality inspection system for a lithium battery protection board, and specifically includes the following steps: Step 1: Read the predetermined detection process of the target lithium battery protection board, wherein the predetermined detection process includes a plurality of predetermined detection steps, and the predetermined detection steps are marked with a test time.

[0016] Specifically, the predetermined detection process of the target lithium battery protection board is read, and the predetermined detection process can be set according to the actual situation of the target lithium battery protection board, wherein the predetermined detection process includes multiple predetermined detection steps, for example, poor welding detection, appearance defect detection, open circuit voltage test, short circuit protection test, overcharge protection test, temperature detection and other steps, wherein each predetermined detection step is marked with a test time, and the test time refers to the test time required for each test step.

[0017] Step 2: Obtain the key processing parameter set and attribute feature set of the target lithium battery protection board, and perform positive sample retrieval based on the industrial Internet with the multiple predetermined detection steps, key processing parameter set and attribute feature set as constraints to obtain a sample fault data set.

[0018] Specifically, the key processing parameter set and attribute feature set of the target lithium battery protection board are obtained. The key processing parameter set includes the processing parameters of multiple key production processing nodes, such as welding parameters of welding nodes, etc.; the attribute feature set includes the size shape, material properties, electrical properties, mechanical properties and other features of the target lithium battery protection board. Then, the multiple predetermined detection steps, key processing parameter sets and attribute feature sets are used as retrieval constraint features, and positive sample retrieval is performed based on the industrial Internet. Positive sample retrieval is a commonly used technology in machine learning and data mining. It is used to quickly find sample data with specific characteristics from a large amount of data, which can improve the accuracy and efficiency of data retrieval, and obtain a sample fault data set, wherein the sample fault data is a fault record containing all predetermined detection steps. By obtaining the sample fault data set, data support is provided for the next step of product failure probability analysis corresponding to the predetermined detection steps.

[0019] Step three: Determine multiple fault probabilities based on the sample fault data set analysis, obtain multiple detection priorities according to multiple test times and the multiple fault probability evaluations, optimize the detection sequence of the multiple predetermined detection steps, and generate an optimized detection process.

[0020] Specifically, based on the sample fault data set, the fault frequencies corresponding to the multiple predetermined detection steps are counted to determine the multiple fault frequencies, and the ratio of the fault frequency to the total number of sample fault data is further set as the fault probability corresponding to the predetermined detection step to determine the multiple fault probabilities. Then, the detection priority of the multiple predetermined detection steps is evaluated according to the multiple test times and the multiple fault probabilities, wherein the detection priority is negatively correlated with the test time and positively correlated with the fault probability, that is, the shorter the test time, the higher the detection priority, and the longer the test time, the lower the detection priority; the greater the fault probability, the higher the detection priority, and the smaller the fault probability, the lower the detection priority.

[0021] Further, the detection order of the multiple predetermined detection steps is optimized according to the multiple detection priorities, that is, the multiple predetermined detection steps are arranged in order from high to low priority to generate an optimized detection process. By generating an optimized detection process, the adaptability of the detection step sequence to the product characteristics can be improved, unnecessary detection steps and detection time can be reduced, and a reference basis can be provided for subsequent product quality inspection.

[0022] Step 4: Use predetermined testing equipment to perform quality testing on the target lithium battery protection board according to the optimized testing process, record the test data set that meets the expected indicators, predict the failure probability of subsequent related testing steps based on the test data set, and output a failure prediction probability set.

[0023] Specifically, configure predetermined testing equipment, such as an X-ray machine, an electrical performance tester, and environmental testing equipment; then, use the predetermined testing equipment to perform quality testing on the target lithium battery protection board in accordance with the optimized testing process. First, perform quality testing on the target lithium battery protection board according to the first testing step in the optimized testing process to obtain first quality testing data. If the first quality testing data does not meet the first expected quality index corresponding to the first testing step, stop the subsequent testing process and mark the target lithium battery protection board as unqualified in the quality testing. If the first quality testing data meets the first expected quality index, record the first quality testing data that meets the expected index, and perform failure probability prediction on subsequent associated testing steps based on the first quality testing data. Subsequent associated steps refer to testing steps whose quality testing results are affected by the first quality testing data.

[0024] Based on the BP neural network, an associated fault prediction unit is constructed. The associated fault prediction unit is a neural network model that can be iteratively optimized in machine learning and is obtained through supervised training of a training data set. Then, the first quality detection data is input into the associated fault prediction unit, and a fault prediction probability set of subsequent associated detection steps is output. By obtaining the fault prediction probability set, a basis is provided for sequential optimization of subsequent detection steps.

[0025] Step 5: The fault prediction probability set is judged according to the warning probability threshold, and a warning mark is made for subsequent associated detection steps corresponding to the fault prediction probability that does not meet the warning probability threshold, so as to determine the warning detection step set.

[0026] Specifically, a warning probability threshold is configured, and the warning probability threshold can be set according to actual scenarios and needs; then the fault prediction probability set is judged according to the warning probability threshold, and if the fault prediction probability sets are all less than or equal to the warning probability threshold, subsequent detection is performed according to the detection step sequence in the optimized detection process; if there is a fault prediction probability greater than the warning probability threshold in the fault prediction probability set, a warning mark is issued for the subsequent associated detection steps corresponding to the fault prediction probability greater than the warning probability threshold, and a warning detection step set is determined, wherein the warning detection step represents the subsequent detection step with a larger fault probability predicted based on the current detection result.

[0027] Step 6: Based on the early warning detection step set, the target lithium battery protection board is subjected to queue-jump detection, and the quality detection result is output.

[0028] Specifically, if the number of early warning detection steps in the early warning detection step set is 1, then the target lithium battery protection board is subjected to queue-jump detection based on the early warning detection step, that is, the early warning detection step is detected first, and the same method is used for iterative judgment detection until the product quality is detected to be unqualified or the entire optimization detection process is completed; if the number of early warning detection steps in the early warning detection step set is not 1, then the multiple early warning detection steps in the early warning detection step set are arranged from large to small according to the fault prediction probability, and the target lithium battery protection board is subjected to queue-jump detection based on the early warning detection step sequence, and the quality detection result is output. Through the above steps, the order of subsequent detection steps can be optimized in time according to the product failure probability prediction results, and the detection steps with higher failure probability can be executed in advance, so as to quickly detect defective products, reduce unnecessary detection steps and detection time, and improve product detection resource utilization and detection efficiency.

[0029] The production quality inspection method of a lithium battery protection board is applied to a production quality inspection system of a lithium battery protection board, which can solve the technical problems that the traditional production quality inspection method of lithium battery protection boards often performs product quality inspection based on a fixed inspection step sequence, resulting in insufficient adaptability between the inspection step sequence and product characteristics, and inability to detect defective products in advance, resulting in waste of product inspection resources and low inspection efficiency. By optimizing the inspection step sequence according to the product failure probability prediction results, the inspection steps with a higher failure probability can be executed in advance, the adaptability between the inspection step sequence and product characteristics can be improved, defective products can be quickly detected, and unnecessary inspection steps and inspection time can be reduced, thereby achieving the technical effect of improving the utilization rate of inspection resources and inspection efficiency.

[0030] Further, to determine the key processing parameter set, step 2 of this application includes: The production and processing flow of the target lithium battery protection board is read to determine multiple processing nodes; based on the Pearson correlation coefficient, the impact of the target lithium battery protection board performance indicators and the multiple processing nodes are evaluated respectively, and multiple impact coefficients are obtained by weighted calculation; the processing nodes whose impact coefficients are greater than a predetermined impact threshold are set as key processing nodes, and processing parameters are collected based on the multiple key processing nodes to obtain the key processing parameter set.

[0031] Specifically, first, the production and processing flow of the target lithium battery protection board is read to determine multiple processing nodes; then the performance indicators of the target lithium battery protection board, such as charging and discharging efficiency, cycle life, safety performance, etc., are determined. Further, based on the Pearson correlation coefficient, the target lithium battery protection board performance indicators and the multiple processing nodes are evaluated for impact respectively. The Pearson correlation coefficient is a statistical algorithm used to measure the degree of linear correlation between two variables. The Pearson correlation coefficient is used to calculate the correlation between each processing node and each performance indicator. The higher the correlation coefficient, the greater the impact of the processing node on the performance indicator; then, according to the size of the correlation coefficient, an indicator influence coefficient is assigned to each processing node, and the influence coefficient can be used to indicate the importance of the processing node to the performance indicator. Further, a weight ratio is set for each performance indicator, and a weighted calculation is performed on the multiple indicator influence coefficients of each processing node according to the weight ratio to obtain multiple influence coefficients of multiple processing nodes. Then, the processing node with an influence coefficient greater than the predetermined influence threshold is set as a key processing node, and processing parameters are collected based on multiple key processing nodes to obtain the key processing parameter set.

[0032] Further, multiple fault probabilities are determined based on the analysis of the sample fault data set, as shown in the attached Figure 2 As shown, step three of this application includes: Based on the multiple predetermined detection steps, fault correlation analysis is performed in combination with historical detection logs to determine multiple rigid correlation fault mappings, wherein the rigid correlation fault mapping is that when a predecessor detection step fails, the probability of a failure in the rear-drive correlation detection step is 100%; the sample fault data set is screened according to the multiple rigid correlation fault mappings to obtain a standard fault data set; based on the standard fault data set, the failure frequency of the predetermined detection step is counted, the ratio of the failure frequency to the total amount of standard fault data is set as the failure probability, and the multiple fault probabilities are output.

[0033] Specifically, first, collect historical inspection log data of lithium battery protection boards produced in the past, where the inspection log includes the result of each inspection step (qualified or unqualified); then perform fault correlation analysis based on multiple predetermined inspection steps in combination with historical inspection logs. For example, use statistical methods (such as Pearson correlation coefficient, chi-square test, etc.) to analyze the correlation between different inspection steps and determine multiple rigid correlation fault mappings. Among them, the rigid correlation fault mapping is that when the predecessor inspection step fails, the probability of failure in the subsequent drive associated inspection step is 100%, that is, when the predecessor inspection step fails, the subsequent inspection steps directly related to it will inevitably fail.

[0034] Then, the sample fault data set is screened according to the multiple rigid-associated fault mappings, that is, according to the multiple rigid-associated fault mappings, the fault samples related thereto are screened out, and for each rigid-associated fault mapping, if a fault occurs in the predecessor detection step, the successor-drive associated detection step should also be included in the standard fault data set; then, the fault samples that do not conform to the rigid-associated fault mapping are removed, ensuring that the predecessor detection step and the successor-drive associated detection step of each rigid-associated fault mapping are included in the standard fault data set; and a standard fault data set is obtained. By screening the sample fault data set according to the rigid-associated fault mapping, a more accurate and useful standard fault data set can be obtained, thereby improving the accuracy and reliability of the standard fault data set, and thereby improving the accuracy of the fault probability setting.

[0035] Finally, based on the standard fault data set, the fault frequency of the predetermined detection step is counted, and the ratio of the fault frequency to the total amount of standard fault data is set as the fault probability, to obtain multiple fault probabilities of multiple predetermined detection steps.

[0036] Further, an optimized detection process is generated, and step three of this application includes: Based on the coefficient of variation method, the first weight and the second weight are configured, wherein the first weight is the test time weight and the second weight is the fault probability weight; according to the first weight and the second weight, the multiple test times and the multiple fault probabilities are evaluated to obtain multiple detection priorities, wherein the detection priority is negatively correlated with the test time and positively correlated with the fault probability; based on the multiple detection priorities, the multiple predetermined detection steps are arranged in descending order of priority to generate the optimized detection process.

[0037] Specifically, the coefficient of variation method is a method that uses statistical methods to calculate the degree of change of each indicator of the system. Each indicator is weighted according to the degree of variation between the current value of each evaluation indicator and the target value. If the numerical difference of a certain indicator is large, it can clearly distinguish the evaluated objects, indicating that the identification information of the indicator is rich, and thus the indicator should be given a larger weight; conversely, if the numerical difference of each evaluated object on a certain indicator is small, then the ability of this indicator to distinguish the evaluation objects is weak, and thus the indicator should be given a smaller weight. Based on the coefficient of variation method, the test time and the probability of failure are weighted to obtain the first weight and the second weight, where the first weight is the test time weight, and the second weight is the failure probability weight. The first weight determines the degree of influence of the test time on the detection priority based on the production efficiency and cost-effectiveness analysis; the second weight determines the degree of influence of the failure probability on the detection priority based on the degree of influence of the failure on the product quality and the production process.

[0038] Further, the detection priority of the multiple test times and multiple failure probabilities is evaluated, wherein the detection priority is negatively correlated with the test time and positively correlated with the failure probability, that is, the longer the test time, the lower the detection priority, and the greater the failure probability, the higher the detection priority; then, according to the first weight and the second weight, the evaluation results are weighted and calculated to obtain multiple detection priorities. Finally, based on the multiple detection priorities, the multiple predetermined detection steps are arranged from high to low according to the priority to generate the optimized detection process.

[0039] Further, the fault prediction probability set is outputted. Step 4 of this application includes: A first detection step is selected in the optimized detection process, wherein the first detection step is the first detection step in the optimized detection process; a quality detection is performed on the target lithium battery protection board according to the first detection step through a predetermined detection device, and a first quality detection value is output; if the first quality detection value meets the first expected quality index, the first quality detection value is input into an associated fault prediction unit, a fault probability prediction is performed on subsequent associated detection steps, and the fault prediction probability set is output.

[0040] Specifically, first, a first detection step is selected in the optimized detection process, wherein the first detection step is the first detection step in the optimized detection process; then, a matching detection device is called based on the first detection step, and quality detection is performed on the target lithium battery protection board according to the first detection step, and a first quality detection value is output; the first quality detection value is further judged according to a first expected quality indicator, and if the first quality detection value meets the first expected quality indicator, the first quality detection value is input into an associated fault prediction unit, wherein the associated prediction unit is constructed based on a BP neural network, and is used to predict the product failure probability of subsequent associated detection steps according to current measurement values, and output a fault prediction probability set of subsequent associated detection steps.

[0041] Further, to construct an associated fault prediction unit, the present application also includes the following steps: Perform association analysis on the multiple predetermined detection steps to determine multiple sets of associated detection steps that meet the predetermined degree of association; randomly select a first set of associated detection steps, and randomly select a first detection step in the first set of associated detection steps, using the quality detection value of the first detection step as the antecedent variable, and using the failure probability of other detection steps in the first set of associated detection steps as the after variables, to obtain multiple after variables; call historical detection logs, extract data based on the antecedent variable and the multiple after variables, and construct multiple sample prediction data sets, wherein the sample prediction data includes sample quality detection values ​​and sample failure probabilities; supervise the BP neural network through the multiple sample prediction data sets to obtain multiple first associated fault predictors that meet the expected convergence constraints, construct a first associated fault prediction branch based on the multiple first associated fault predictors, and sequentially obtain multiple associated fault prediction branches of multiple antecedent variables; based on the classification decision principle, build a first associated fault prediction channel according to the mapping relationship between the multiple antecedent variables and the multiple associated fault prediction branches, and construct the associated fault prediction unit based on the first associated fault prediction channel.

[0042] Specifically, first, an association analysis is performed on the multiple predetermined detection steps, such as analyzing historical data to find out which detection steps have dependencies or interactions, that is, the test results of the previous step will affect the test results of which detection steps, determine the degree of association between these steps, and obtain multiple correlation degrees between the multiple predetermined detection steps; further select detection steps with a correlation degree greater than the predetermined correlation degree and set them as associated detection steps, and determine multiple sets of associated detection steps.

[0043] Then, any one of the multiple associated detection step sets is randomly selected as the first associated detection step set, and the first detection step is randomly selected from the first associated detection step set; then, the quality detection value of the first detection step is used as the antecedent variable, and the failure probability of other detection steps in the first associated detection step set is used as the post-variable, to obtain multiple post-variables; further, the historical detection log is called, and data is extracted from the historical detection log based on the antecedent variable and the multiple post-variables, to construct multiple sample prediction data sets, wherein the sample prediction data includes the sample quality detection value of the antecedent variable and the sample failure probability of the post-variable.

[0044] Then, the sample quality detection value is used as input and the sample failure probability is used as output, and the BP neural network is supervised and trained through the multiple sample prediction data sets, such as using a back propagation algorithm, and adjusting the network weights and thresholds so that the output of the neural network is as close as possible to the actual failure probability, so as to obtain multiple first associated fault predictors that meet the expected convergence constraints; then, a first associated fault prediction branch is constructed according to the multiple first associated fault predictors, and multiple associated fault prediction branches of multiple previous variables are obtained in sequence using the same method.

[0045] The classification decision tree is a common machine learning algorithm, in which each internal node represents a feature, each branch represents the attribute value of a feature, and each leaf node represents a category. Based on the classification decision principle, the leaf nodes with the previous variables as child nodes and the associated fault prediction branches as child nodes are constructed according to the mapping relationship between the multiple previous variables and the multiple associated fault prediction branches. The multiple associated fault prediction channels are obtained in sequence using the same method, and the associated fault prediction unit is constructed according to the multiple associated fault prediction channels.

[0046] Further, based on the early warning detection step set, the target lithium battery protection board is subjected to queue-jump detection, and step six of the present application includes: If the early warning detection step set includes multiple early warning detection steps, identify the fault prediction probabilities of the multiple early warning detection steps, arrange the multiple early warning detection steps from large to small according to the fault prediction probabilities, and generate an early warning detection step sequence; based on the early warning detection step sequence, perform queue-jumping detection on the target lithium battery protection board.

[0047] Specifically, if the early warning detection step set includes multiple early warning detection steps, the fault prediction probabilities of the multiple early warning detection steps are identified, and the multiple early warning detection steps are arranged from large to small according to the fault prediction probabilities to generate an early warning detection step sequence; further, based on the early warning detection step sequence, the target lithium battery protection board is subjected to queue interruption detection.

[0048] Further, the quality inspection results are outputted. Step 6 of this application includes: If the first fault prediction probability sets of the first detection step all meet the warning probability threshold, a second detection step is selected in the optimized detection process, wherein the second detection step is an adjacent step of the first detection step; through a predetermined detection device, the target lithium battery protection board is quality inspected according to the second detection step, and a second quality detection value is output; if the second quality detection value meets the second expected quality index, a failure probability prediction is performed on subsequent associated detection steps based on the first quality detection value and the second quality detection value, and subsequent detection steps are executed based on the prediction results; if the second quality detection value does not meet the second expected quality index, the detection is stopped and the quality detection result is output.

[0049] Specifically, if the first fault prediction probability sets of the first detection step all meet the warning probability threshold, then the second detection step is selected in the optimized detection process, wherein the second detection step is an adjacent step of the first detection step; then the detection device is called based on the second detection step, and the target lithium battery protection board is quality inspected according to the second detection step, and a second quality inspection value is output; if the second quality inspection value does not meet the second expected quality index, the subsequent detection steps are stopped, and the quality inspection result is output as unqualified.

[0050] If the second quality detection value meets the second expected quality index, the failure probability prediction of the subsequent associated detection step is performed based on the first quality detection value and the second quality detection value, wherein a failure probability prediction model is constructed based on a BP neural network, the input of the failure probability prediction model is the first quality detection value and the second quality detection value, the output is the failure probability prediction result of the subsequent associated detection step, and the subsequent detection step is executed according to the prediction result. By constructing a BP neural network model, the failure probability prediction of the subsequent associated detection step is performed using the detected quality data set. As the number of detected quality data sets increases, the accuracy of the subsequent failure probability prediction is also increasing, thereby further improving the accuracy of the detection step sequence setting and quickly detecting defective products.

[0051] In summary, the production quality inspection method of a lithium battery protection board provided by the present application has the following technical effects: By taking multiple predetermined detection steps, key processing parameter sets and attribute feature sets of the target lithium battery protection board as constraints, a sample fault data set is retrieved and obtained; then, multiple fault probabilities are determined based on the analysis of the sample fault data set, and multiple detection priorities are obtained according to multiple test times and multiple fault probability assessments. The detection sequence of multiple predetermined detection steps is optimized to generate an optimized detection process, which can improve the adaptability of the product detection sequence and product characteristics; then, the target lithium battery protection board is quality inspected according to the optimized detection process through predetermined detection equipment, and the detection data set that meets the expected indicators is recorded. The failure probability of subsequent related detection steps is predicted based on the detection data set, and the output is A fault prediction probability set is generated; the fault prediction probability set is then judged according to the warning probability threshold, and subsequent related detection steps corresponding to the fault prediction probability that does not meet the warning probability threshold are warned, and the warning detection step set is determined; finally, based on the warning detection step set, a queue-jumping detection is performed on the target lithium battery protection board, and the quality detection result is output; by optimizing the detection step sequence according to the product failure probability prediction result, the detection steps with a higher failure probability can be executed in advance, the adaptability of the detection step sequence to the product characteristics can be improved, defective products can be detected quickly, and unnecessary detection steps and detection time can be reduced, thereby achieving the technical effect of improving the utilization rate of detection resources and detection efficiency.

[0052] Embodiment 2: Based on the production quality inspection method of a lithium battery protection board in the above embodiment, the present application also provides a production quality inspection system for a lithium battery protection board, please refer to the attached Figure 3 ,include: A predetermined detection process reading module 11 is used to read the predetermined detection process of the target lithium battery protection board, wherein the predetermined detection process includes multiple predetermined detection steps, and the predetermined detection steps are marked with test time; a sample fault data set acquisition module 12 is used to obtain the key processing parameter set and attribute feature set of the target lithium battery protection board, and perform positive sample retrieval based on the industrial Internet with the multiple predetermined detection steps, key processing parameter set and attribute feature set as constraints to obtain a sample fault data set; a detection sequence optimization module 13 is used to determine multiple fault probabilities based on the analysis of the sample fault data set, obtain multiple detection priorities according to multiple test times and the multiple fault probability evaluations, and perform detection order optimization on the multiple predetermined detection steps. Optimize and generate an optimized detection process; a subsequent fault probability prediction module 14 is used to perform quality inspection on the target lithium battery protection board according to the optimized detection process through a predetermined detection device, record the detection data set that meets the expected indicators, and predict the fault probability of subsequent related detection steps based on the detection data set, and output a fault prediction probability set; an early warning detection step identification module 15 is used to judge the fault prediction probability set according to the early warning probability threshold, and to mark the subsequent related detection steps corresponding to the fault prediction probability that does not meet the early warning probability threshold, and determine the early warning detection step set; a queue-jumping detection module 16 is used to perform queue-jumping detection on the target lithium battery protection board based on the early warning detection step set, and output the quality detection result.

[0053] Furthermore, the production quality inspection system of the lithium battery protection board is also used for: The production and processing flow of the target lithium battery protection board is read to determine multiple processing nodes; based on the Pearson correlation coefficient, the impact of the target lithium battery protection board performance indicators and the multiple processing nodes are evaluated respectively, and multiple impact coefficients are obtained by weighted calculation; the processing nodes whose impact coefficients are greater than a predetermined impact threshold are set as key processing nodes, and processing parameters are collected based on the multiple key processing nodes to obtain the key processing parameter set.

[0054] Furthermore, the production quality inspection system of a lithium battery protection board is also used for: Based on the multiple predetermined detection steps, fault correlation analysis is performed in combination with historical detection logs to determine multiple rigid correlation fault mappings, wherein the rigid correlation fault mapping is that when a predecessor detection step fails, the probability of a failure in the rear-drive correlation detection step is 100%; the sample fault data set is screened according to the multiple rigid correlation fault mappings to obtain a standard fault data set; based on the standard fault data set, the failure frequency of the predetermined detection step is counted, the ratio of the failure frequency to the total amount of standard fault data is set as the failure probability, and the multiple fault probabilities are output.

[0055] Furthermore, the production quality inspection system of a lithium battery protection board is also used for: Based on the coefficient of variation method, the first weight and the second weight are configured, wherein the first weight is the test time weight and the second weight is the fault probability weight; according to the first weight and the second weight, the multiple test times and the multiple fault probabilities are evaluated to obtain multiple detection priorities, wherein the detection priority is negatively correlated with the test time and positively correlated with the fault probability; based on the multiple detection priorities, the multiple predetermined detection steps are arranged in descending order of priority to generate the optimized detection process.

[0056] Furthermore, the production quality inspection system of a lithium battery protection board is also used for: A first detection step is selected in the optimized detection process, wherein the first detection step is the first detection step in the optimized detection process; a quality detection is performed on the target lithium battery protection board according to the first detection step through a predetermined detection device, and a first quality detection value is output; if the first quality detection value meets the first expected quality index, the first quality detection value is input into an associated fault prediction unit, a fault probability prediction is performed on subsequent associated detection steps, and the fault prediction probability set is output.

[0057] Furthermore, the production quality inspection system of a lithium battery protection board is also used for: Perform association analysis on the multiple predetermined detection steps to determine multiple sets of associated detection steps that meet the predetermined degree of association; randomly select a first set of associated detection steps, and randomly select a first detection step in the first set of associated detection steps, using the quality detection value of the first detection step as the antecedent variable, and using the failure probability of other detection steps in the first set of associated detection steps as the after variables, to obtain multiple after variables; call historical detection logs, extract data based on the antecedent variable and the multiple after variables, and construct multiple sample prediction data sets, wherein the sample prediction data includes sample quality detection values ​​and sample failure probabilities; supervise the BP neural network through the multiple sample prediction data sets to obtain multiple first associated fault predictors that meet the expected convergence constraints, construct a first associated fault prediction branch based on the multiple first associated fault predictors, and sequentially obtain multiple associated fault prediction branches of multiple antecedent variables; based on the classification decision principle, build a first associated fault prediction channel according to the mapping relationship between the multiple antecedent variables and the multiple associated fault prediction branches, and construct the associated fault prediction unit based on the first associated fault prediction channel.

[0058] Furthermore, the production quality inspection system of a lithium battery protection board is also used for: If the early warning detection step set includes multiple early warning detection steps, identify the fault prediction probabilities of the multiple early warning detection steps, arrange the multiple early warning detection steps from large to small according to the fault prediction probabilities, and generate an early warning detection step sequence; based on the early warning detection step sequence, perform queue-jumping detection on the target lithium battery protection board.

[0059] Furthermore, the production quality inspection system of a lithium battery protection board is also used for: If the first fault prediction probability sets of the first detection step all meet the warning probability threshold, a second detection step is selected in the optimized detection process, wherein the second detection step is an adjacent step of the first detection step; through a predetermined detection device, the target lithium battery protection board is quality inspected according to the second detection step, and a second quality detection value is output; if the second quality detection value meets the second expected quality index, a failure probability prediction is performed on subsequent associated detection steps based on the first quality detection value and the second quality detection value, and subsequent detection steps are executed based on the prediction results; if the second quality detection value does not meet the second expected quality index, the detection is stopped and the quality detection result is output.

[0060] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The production quality inspection method and specific examples of a lithium battery protection board in the aforementioned embodiment 1 are also applicable to a production quality inspection system for a lithium battery protection board in this embodiment. Through the aforementioned detailed description of the production quality inspection method for a lithium battery protection board, those skilled in the art can clearly understand the production quality inspection system for a lithium battery protection board in this embodiment, so for the sake of brevity of the specification, it will not be described in detail here. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method section.

[0061] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

[0062] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application belong to the scope of the present application and its equivalent technology, the present application is also intended to include these modifications and variations.

Claims

1. A production quality inspection method for a lithium battery protection board, characterized in that: include: Reading a predetermined detection process of a target lithium battery protection board, wherein the predetermined detection process includes a plurality of predetermined detection steps, and the predetermined detection steps are marked with a test time; Acquire the key processing parameter set and attribute feature set of the target lithium battery protection board, and perform positive sample retrieval based on the industrial Internet with the plurality of predetermined detection steps, key processing parameter set and attribute feature set as constraints to acquire a sample fault data set; Determine multiple fault probabilities based on the sample fault data set analysis, obtain multiple detection priorities according to multiple test times and the multiple fault probability evaluations, optimize the detection sequence of the multiple predetermined detection steps, and generate an optimized detection process; Through a predetermined testing device, the target lithium battery protection board is quality tested according to the optimized testing process, a test data set that meets the expected indicators is recorded, a failure probability prediction is performed for subsequent associated testing steps according to the test data set, and a failure prediction probability set is output; The fault prediction probability set is judged according to the warning probability threshold, a warning mark is made for subsequent associated detection steps corresponding to the fault prediction probability that does not meet the warning probability threshold, and a warning detection step set is determined; Based on the early warning detection step set, the target lithium battery protection board is subjected to queue-jump detection, and a quality detection result is output.

2. The production quality inspection method of a lithium battery protection board according to claim 1, characterized in that: Identify key processing parameter sets, including: Read the production process of the target lithium battery protection board and determine multiple processing nodes; Based on the Pearson correlation coefficient, the target lithium battery protection board performance index and the multiple processing nodes are evaluated for influence, and multiple influence coefficients are obtained by weighted calculation; The processing nodes whose influence coefficients are greater than a predetermined influence threshold are set as key processing nodes, and processing parameters are collected based on the multiple key processing nodes to obtain the key processing parameter set.

3. The production quality inspection method of a lithium battery protection board according to claim 1, characterized in that: Determining multiple failure probabilities based on the sample failure data set analysis includes: Based on the multiple predetermined detection steps, combined with historical detection logs, a fault correlation analysis is performed to determine multiple rigid correlation fault mappings, wherein the rigid correlation fault mapping is that when a front drive detection step fails, the probability of a rear drive correlation detection step failing is 100%; Screening the sample fault data set according to the multiple rigid-associated fault mappings to obtain a standard fault data set; Based on the standard fault data set, the fault frequency of the predetermined detection step is counted, the ratio of the fault frequency to the total amount of standard fault data is set as the fault probability, and the multiple fault probabilities are output.

4. The production quality inspection method of a lithium battery protection board according to claim 1, characterized in that: Generate optimized testing workflows, including: Based on the coefficient of variation method, a first weight and a second weight are configured, wherein the first weight is a test time weight and the second weight is a failure probability weight; According to the first weight and the second weight, the multiple test times and the multiple failure probabilities are evaluated to obtain multiple detection priorities, wherein the detection priority is negatively correlated with the test time and positively correlated with the failure probability; Based on the multiple detection priorities, the multiple predetermined detection steps are arranged in descending order of priority to generate the optimized detection process.

5. The production quality inspection method of a lithium battery protection board according to claim 1, characterized in that: Output fault prediction probability set, including: Selecting a first detection step in the optimization detection process, wherein the first detection step is the first detection step in the optimization detection process; By using a predetermined testing device, performing a quality test on the target lithium battery protection board according to the first testing step, and outputting a first quality test value; If the first quality detection value meets the first expected quality index, the first quality detection value is input into the associated fault prediction unit, the fault probability prediction is performed on the subsequent associated detection steps, and the fault prediction probability set is output.

6. A production quality inspection method for a lithium battery protection board according to claim 5, characterized in that: Construct a correlation fault prediction unit, including: Performing association analysis on the plurality of predetermined detection steps to determine a plurality of associated detection step sets that satisfy a predetermined degree of association; Randomly selecting a first associated detection step set, and randomly selecting a first detection step in the first associated detection step set, taking the quality detection value of the first detection step as a pre-variable, and taking the failure probability of other detection steps in the first associated detection step set as a post-variable, to obtain a plurality of post-variables; Calling historical detection logs, extracting data based on the antecedent variables and the multiple post-variables, and constructing multiple sample prediction data sets, wherein the sample prediction data includes sample quality detection values ​​and sample failure probabilities; Performing supervised training on the BP neural network through the multiple sample prediction data sets to obtain multiple first associated fault predictors that meet expected convergence constraints, constructing first associated fault prediction branches according to the multiple first associated fault predictors, and sequentially obtaining multiple associated fault prediction branches of multiple pre-variables; Based on the classification decision principle, a first associated fault prediction channel is established according to the mapping relationship between the multiple pre-variables and the multiple associated fault prediction branches, and the associated fault prediction unit is constructed based on the first associated fault prediction channel.

7. The production quality inspection method of a lithium battery protection board according to claim 1, characterized in that: Performing queue-jump detection on the target lithium battery protection board based on the early warning detection step set includes: If the early warning detection step set includes multiple early warning detection steps, identifying the failure prediction probabilities of the multiple early warning detection steps, arranging the multiple early warning detection steps from large to small according to the failure prediction probabilities, and generating an early warning detection step sequence; Based on the early warning detection step sequence, the target lithium battery protection board is subjected to queue-jump detection.

8. The production quality inspection method of a lithium battery protection board according to claim 5, characterized in that: Output quality inspection results, including: If the first fault prediction probability sets of the first detection step all meet the warning probability threshold, selecting a second detection step in the optimized detection process, wherein the second detection step is an adjacent step of the first detection step; By using a predetermined testing device, performing a quality test on the target lithium battery protection board according to the second testing step, and outputting a second quality test value; If the second quality detection value meets the second expected quality index, predicting the failure probability of subsequent associated detection steps based on the first quality detection value and the second quality detection value, and executing subsequent detection steps based on the prediction result; If the second quality detection value does not meet the second expected quality indicator, the detection is stopped and the quality detection result is output.

9. A production quality inspection system for lithium battery protection boards, characterized in that: The steps for implementing the production quality inspection method of a lithium battery protection board according to any one of claims 1 to 8 include: A predetermined detection process reading module is used to read the predetermined detection process of the target lithium battery protection board, wherein the predetermined detection process includes a plurality of predetermined detection steps, and the predetermined detection steps are marked with a test time; A sample fault data set acquisition module is used to acquire a key processing parameter set and an attribute feature set of the target lithium battery protection board, and to perform positive sample retrieval based on the industrial Internet with the plurality of predetermined detection steps, key processing parameter sets and attribute feature sets as constraints to acquire a sample fault data set; A detection sequence optimization module, configured to determine multiple fault probabilities based on the analysis of the sample fault data set, obtain multiple detection priorities according to multiple test times and the multiple fault probability evaluations, optimize the detection sequence of the multiple predetermined detection steps, and generate an optimized detection process; A subsequent fault probability prediction module is used to perform quality inspection on the target lithium battery protection board according to the optimized inspection process through a predetermined inspection device, record the inspection data set that meets the expected indicators, perform fault probability prediction on subsequent related inspection steps according to the inspection data set, and output a fault prediction probability set; An early warning detection step identification module is used to judge the fault prediction probability set according to the early warning probability threshold, to perform early warning identification on subsequent associated detection steps corresponding to the fault prediction probability that does not meet the early warning probability threshold, and to determine the early warning detection step set; The queue-jumping detection module is used to perform queue-jumping detection on the target lithium battery protection board based on the early warning detection step set and output a quality detection result.

Images