QMS defective product analysis and tracing method based on battery manufacturing industry
By establishing a multidimensional quality feature dataset and a root cause correlation matrix for defective products, we have achieved full-chain traceability and quality closed-loop optimization of defective products in the battery manufacturing industry. This solves the problems of data accuracy and efficiency in defective product analysis and traceability in existing technologies, and improves the systematicness and accuracy of quality management.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- ZHEJIANG TIANNENG NEW ENERGY CO LTD
- Filing Date
- 2025-11-26
- Publication Date
- 2026-04-24
AI Technical Summary
In the battery manufacturing industry, defective product analysis and traceability suffer from low data accuracy and incompleteness, lack of unified identification standards and real-time monitoring, leading to difficulties in tracing quality problems, high recall costs, and a lack of systematic quality management methods.
By quantifying the defect rate target, defining a multi-dimensional quality feature dataset, generating a defect root cause correlation matrix, establishing a full-chain traceability map of defective products, generating a quality closed-loop optimization instruction set, realizing full-process data collection and real-time monitoring, and accurately tracing the source and impact range of defective products.
It improved the efficiency and accuracy of defective product analysis and traceability, enhanced the comprehensiveness and pertinence of quality management, and enabled the prevention of defective products from the source and continuous improvement, thereby raising the quality management level of the battery manufacturing industry.
Smart Images
Figure CN121920648A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of battery manufacturing technology, and in particular to a QMS (Quality Management System) method for defective product analysis and traceability in the battery manufacturing industry. Background Technology
[0002] In the battery manufacturing industry, product quality directly affects safety, performance, and market competitiveness. With the rapid development of the new energy industry, battery products are becoming increasingly diverse, and production processes are becoming more complex, placing higher demands on quality management. However, the analysis and traceability of defective products in the current battery production process faces numerous challenges. Regarding data collection, traditional methods rely heavily on manual recording or fragmented system collection, resulting in low data accuracy, insufficient completeness, and difficulty in effectively linking data from different stages, creating "data silos." In defective product identification, the lack of unified identification standards and real-time monitoring mechanisms often leads to defects being discovered only during the finished product inspection stage, making it difficult to trace quality problems and determine the specific stage and cause of the problem.
[0003] In the analysis phase, the scattered and inconsistent formats of data make in-depth statistical analysis and trend prediction difficult, hindering the timely detection of potential quality issues. During traceability, traditional methods suffer from broken traceability chains, making it difficult to quickly and accurately track key information such as the source of raw materials, production batches, production equipment, and operators of defective products, resulting in high recall costs and delayed quality improvements. Furthermore, the poor integration of supplier quality control and production process quality control exacerbates the difficulty of defective product management. Current technologies lack a systematic approach that integrates quality planning, end-to-end data monitoring, intelligent analysis, and precise traceability. This is particularly true in high-precision, high-safety-requirement fields like battery manufacturing, where standardized defective product analysis and traceability systems are still underdeveloped. Summary of the Invention
[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a QMS (Quality Management System) method for defective product analysis and traceability in the battery manufacturing industry.
[0005] To achieve the above objectives, the present invention adopts the following technical solution: a QMS defect analysis and traceability method based on the battery manufacturing industry, comprising the following steps:
[0006] Based on the performance requirements and safety standards of battery products, the defect rate target is quantified, the standard for the purity of positive electrode material and the deviation of cell capacity is defined, the coating thickness deviation and welding strength parameters are obtained, and the equipment operation status information entered by the operator and the raw material batch number provided by the supplier are integrated. The collected parameters are compared with the preset standard in real time, and data exceeding the threshold are marked to establish a multi-dimensional quality feature dataset.
[0007] Based on the multidimensional quality feature dataset, by calculating the frequency of different defective product types and sorting them in descending order, the two defective product types with the highest frequency are identified. The production equipment number, operator number, raw material batch number and environmental temperature and humidity values associated with the identified defective product types are extracted. By calculating the correlation coefficient between each associated factor and the defective product incidence rate, factors with correlation coefficients higher than the preset correlation threshold are identified as key influencing factors, and a defective product root cause correlation matrix is generated.
[0008] Based on the key influencing factors in the defective product root cause correlation matrix, the source index retrieves the defective product batch number and associates it with raw material supplier information, warehousing inspection records, and transportation batch number. The process index matches the batch of products with all process equipment numbers, operator numbers, and process start and end timestamps in the production process. The source information and process information are linked together in chronological order to form an information link from supplier to finished product, and a full-link traceability map of defective products is established.
[0009] Preferably, the method further includes:
[0010] Using the aforementioned full-chain traceability map of defective products, the map interruption points and abnormal parameter nodes are located. Supplier audit instructions are generated for raw material nodes, parameter optimization instructions are generated for equipment parameter nodes, and skills training instructions are generated for operator nodes. The optimized equipment parameters are updated to the quality control point standard library, and the related raw material acceptance standards are adjusted. All instructions are summarized to obtain a quality closed-loop optimization instruction set.
[0011] Preferably, the steps for obtaining the multidimensional quality feature dataset are as follows:
[0012] Based on the performance requirements and safety standards of battery products, the capacity deviation of the cells is set with reference to industry documents, and the purity lower limit is set according to the technical specifications of the cathode material supplier. At the same time, the defect rate target is set. All the above numerical indicators and allowable fluctuation ranges are integrated to obtain a quantitative quality standard set.
[0013] Based on the quantitative quality standard set, coating thickness deviation data is collected, welding strength values are obtained, and the equipment operation status code entered by the operator is read from the MES system. The supplier batch number is obtained by scanning the raw material packaging barcode. All data is then timestamped and aggregated by batch to generate the original production parameter stream.
[0014] Based on the original production parameter stream, the values of each parameter in the stream are extracted and compared with the upper and lower limits of the corresponding indicators in the quantitative quality standard set. If the parameter value exceeds the range, an abnormal status code is assigned. The parameters with status codes, along with the associated equipment number and raw material batch number, are stored in a structured manner to establish a multidimensional quality feature dataset.
[0015] Preferably, the steps for obtaining the root cause correlation matrix of defective products are as follows:
[0016] Based on the multidimensional quality feature dataset, all data marked with abnormal status codes in the dataset are traversed, the defective product types associated with the abnormal status codes are counted and statistically analyzed, and sorted in descending order according to the count values. The first two defective product types in the sorting results are selected to obtain a list of main defective product types.
[0017] Based on the list of major defective product types, using the defective product types in the list as the filtering criteria, all matching records are filtered out in the multidimensional quality feature dataset, and the production equipment number, operator number, raw material batch number, and environmental temperature and humidity values are extracted from the matching records to establish a set of potential influencing factors.
[0018] Based on the set of potential influencing factors, the ratio of the number of defective products to the total number of productions for each influencing factor in different numerical ranges is calculated. The ratio is compared with a preset occurrence rate threshold. Influencing factor combinations that exceed the threshold are identified as key influencing factors, and a root cause correlation matrix for defective products is generated.
[0019] Preferably, the steps for obtaining the full-chain traceability map of defective products are as follows:
[0020] Based on the key influencing factors in the root cause correlation matrix of defective products, the defective product batch number is used as the query keyword to search the supplier management database for matching supplier name, inbound inspection item value and transportation batch number. The retrieved information is then arranged in chronological order to construct a raw material traceability path.
[0021] Based on the raw material traceability path, the defective batch number is used as the query index to retrieve all process equipment numbers, operator numbers, and start and end timestamp records of each process in the production execution database during the batch product flow process. The retrieved records are then sorted in ascending order by timestamp to construct the production process traceability path.
[0022] Preferably, the step of obtaining the defective product end-to-end traceability map further includes:
[0023] Based on the raw material traceability path and the production process traceability path, all information nodes in the two paths are merged and sorted according to timestamps to form a continuous time series from raw material warehousing inspection to finished product output. The series is then visualized and connected to establish a full-chain traceability map for defective products.
[0024] Preferably, the step of obtaining the quality closed-loop optimization instruction set is as follows:
[0025] Using the aforementioned full-chain traceability map of defective products, check whether the parameter value of each node in the map exceeds the corresponding range of the quantitative quality standard set. Nodes that exceed the range are marked as abnormal parameter nodes. The information of raw materials, equipment, and personnel to which the nodes belong is classified and organized to generate a list of nodes to be optimized.
[0026] Based on the list of nodes to be optimized, an instruction is created for each node in the list. The instruction content includes node identifier, abnormal parameters, target supplier number, target equipment number, and target personnel number. The instructions are then grouped into three categories: raw materials, equipment, and personnel to form a draft of categorized optimization instructions.
[0027] Preferably, the step of obtaining the quality closed-loop optimization instruction set further includes:
[0028] Based on the aforementioned draft classification optimization instructions, the optimized values of equipment parameters in the instructions are extracted and written into the corresponding fields of the quality control point standard library. The limits of related indicators in the raw material acceptance standards are adjusted synchronously. All updated instructions are integrated and encapsulated to obtain a quality closed-loop optimization instruction set.
[0029] Compared with the prior art, the advantages and positive effects of the present invention are as follows:
[0030] This invention integrates multiple dimensions, including quality planning, quality control, defective product analysis, traceability, and quality improvement, forming a systematic and standardized method for defective product analysis and traceability in the battery manufacturing industry's QMS (Quality Management System). By establishing clear goals and standards and a supplier management system during the quality planning phase, it prevents the generation of defective products at the source. During the quality control phase, it achieves full-process data collection and real-time monitoring to promptly identify defective products. Scientific analysis methods determine the causes and key influencing factors of defective products, and a complete traceability chain accurately tracks the source and scope of impact. Finally, continuous improvement is carried out based on the analysis and traceability results to constantly refine the quality management system. This not only improves the efficiency and accuracy of defective product analysis and traceability but also enhances the comprehensiveness and relevance of quality management, providing an effective solution for quality improvement in the battery manufacturing industry. Attached Figure Description
[0031] Figure 1 This is a schematic diagram of the steps of the present invention. Detailed Implementation
[0032] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.
[0033] Please see Figure 1This invention provides a technical solution: a method for QMS defect analysis and traceability in the battery manufacturing industry, comprising the following steps:
[0034] Based on the performance requirements and safety standards of battery products, the defect rate target is quantified, the standard for the purity of positive electrode material and the deviation of cell capacity is defined, the coating thickness deviation and welding strength parameters are obtained, and the equipment operation status information entered by the operator and the raw material batch number provided by the supplier are integrated. The collected parameters are compared with the preset standard in real time, and data exceeding the threshold are marked to establish a multi-dimensional quality feature dataset.
[0035] Based on the multidimensional quality feature dataset, the frequency of different defective product types is calculated and sorted in descending order. The two defective product types with the highest frequency are identified. The production equipment number, operator number, raw material batch number and environmental temperature and humidity values associated with the identified defective product types are extracted. By calculating the correlation coefficient between each associated factor and the defective product incidence rate, factors with correlation coefficients higher than the preset correlation threshold are identified as key influencing factors, and a defective product root cause correlation matrix is generated.
[0036] Based on the key influencing factors in the root cause correlation matrix of defective products, the source index is used to call the defective product batch number to associate with the raw material supplier information, warehousing inspection records and transportation batch number. The process index is used to match the equipment number, operator number and process start and end timestamp of all processes that the batch of products goes through in the production process. The source information and process information are linked in chronological order to form an information link from supplier to finished product and to establish a full-link traceability map of defective products.
[0037] By adopting a full-chain traceability map of defective products, the map interruption points and abnormal parameter nodes are located. Supplier audit instructions are generated for raw material nodes, parameter optimization instructions are generated for equipment parameter nodes, and skills training instructions are generated for operator nodes. The optimized equipment parameters are updated to the quality control point standard library, and the related raw material acceptance standards are adjusted. All instructions are summarized to obtain a quality closed-loop optimization instruction set.
[0038] The steps to obtain the multidimensional quality feature dataset are as follows:
[0039] Based on the performance requirements and safety standards of battery products, the capacity deviation of the cells is set with reference to industry documents, and the purity lower limit is set according to the technical specifications of the cathode material supplier. At the same time, the defect rate target is set. All the above numerical indicators and allowable fluctuation ranges are integrated to obtain a quantitative quality standard set.
[0040] Based on the quantitative quality standard set, coating thickness deviation data is collected, welding strength values are obtained, and the equipment operation status code entered by the operator is read from the MES system. The supplier batch number is obtained by scanning the raw material packaging barcode. All data is then timestamped and aggregated by batch to generate the original production parameter stream.
[0041] Based on the original production parameter stream, the values of each parameter in the stream are extracted and compared with the upper and lower limits of the corresponding indicators in the quantitative quality standard set. If the parameter value exceeds the range, an abnormal status code is assigned. The parameters with status codes, along with the associated equipment number and raw material batch number, are stored in a structured manner to establish a multidimensional quality feature dataset.
[0042] Specifically, based on the battery product performance requirements and safety standards, the clauses regarding cycle life and safety are broken down, and the regulations on cell capacity decay are converted into specific values. The upper limit of cell capacity deviation is set as 2% of the nominal capacity, and the lower limit is -2% of the nominal capacity. The purity declaration of LiNiMnCoO2 in the technical specifications provided by the cathode material supplier is read, and its minimum guaranteed value of 99.9% is extracted as the lower limit standard for purity. Then, the production history data of the past six months is analyzed to calculate the average defect rate of all batches. 80% of this average value is used as the new defect rate target. For example, if the historical average defect rate is 0.125%, the target is set to 0.1%. Finally, the upper and lower limits of cell capacity deviation, the lower limit of cathode material purity, and the defect rate target are compiled in a structured manner to form a lookup table containing parameter names, upper limits, lower limits, and units, resulting in a quantitative quality standard set.
[0043] Based on the quantitative quality standard set, a laser triangulation sensor is deployed at the winding position of the electrode coating process. Five measurement points are set along the width of the electrode sheet, located 5 mm from the edge, one-quarter of the width, at the center, three-quarters of the width, and 5 mm from the other edge. Thickness readings are collected simultaneously at these five points, and the difference between the maximum and minimum values of the five readings is calculated as the coating thickness deviation data. After the welding process, a tensile testing machine is used to perform a destructive tensile test on each weld point, and the maximum force it withstands at fracture is recorded as the weld strength value, in Newtons. Simultaneously, through OPC... The UA protocol interface reads equipment status codes in real time from the Manufacturing Execution System (MES) database. For example, 101 indicates normal operation, 301 indicates parameter drift, and 501 indicates mechanical failure. Then, an industrial barcode scanner is used to scan the QR code on the raw material packaging to parse out the supplier batch number, which consists of the supplier ID, material code, production date, and serial number. The collected coating thickness deviation data, welding strength values, equipment status codes, and supplier batch numbers are uniformly appended with a Coordinated Universal Time (UTC) timestamp and aggregated using the production work order number as a unique identifier to generate the original production parameter stream.
[0044] Based on the original production parameter stream, each data packet in the stream is parsed to extract the values of parameters such as coating thickness deviation, welding strength, equipment status, and raw material purity. Then, for each value, the upper and lower limits of the corresponding indicator are queried from the set of quantitative quality standards, and the values are directly compared. For example, if the coating thickness deviation is 10 micrometers, while the upper limit in the standard set is 8 micrometers, the parameter is determined to be out of range. For parameters that are out of range, a unique abnormal status code is assigned according to a predefined mapping table. For example, coating thickness exceeding the upper limit is E101, welding strength below the lower limit is E201, and raw material purity not meeting the standard is E301. If the value is within the standard range, the status code OK is assigned. Finally, the parameter with the status code, along with its original value, collection timestamp, associated equipment number, and raw material batch number, is inserted as a record into the database table. The collection of all records establishes a multidimensional quality feature dataset.
[0045] The steps to obtain the root cause correlation matrix of defective products are as follows:
[0046] Based on the multidimensional quality feature dataset, all data marked with abnormal status codes in the dataset are traversed, the defective product types associated with the abnormal status codes are counted and statistically analyzed, and sorted in descending order according to the count values. The top two defective product types in the sorted results are selected to obtain a list of major defective product types.
[0047] Based on the list of major defective product types, all matching records are filtered out from the multidimensional quality feature dataset using the defective product types in the list as the filtering criteria. The production equipment number, operator number, raw material batch number, and environmental temperature and humidity values are extracted from the matching records to establish a set of potential influencing factors.
[0048] Based on the set of potential influencing factors, the ratio of the number of defective products to the total number of productions for each influencing factor in different numerical ranges is calculated. The ratio is compared with a preset occurrence rate threshold. Influencing factor combinations that exceed the threshold are identified as key influencing factors, and a root cause correlation matrix for defective products is generated.
[0049] Specifically, based on the multidimensional quality feature dataset, a database query is first performed to filter out all records in the dataset whose status code field is not OK. Then, these filtered records are traversed to extract the abnormal status code field of each record. Based on a pre-established correspondence table between abnormal status codes and defective product types (e.g., E101 corresponds to uneven coating and E201 corresponds to cold solder joint), the status codes are converted into specific defective product type descriptions. Next, a frequency counting dictionary is created using the defective product type description as the key. The occurrence counts of all converted defective product types are accumulated and statistically analyzed. After the statistics are completed, the dictionary is sorted from high to low according to the count value. Finally, the names of the first and second ranked defective product types are selected from the sorted results and stored in a list to obtain the list of main defective product types.
[0050] Based on the list of major defective product types, two defective product types in the list, such as uneven coating and poor soldering, are used as filtering conditions for database queries. The system then searches the multidimensional quality feature dataset for all records corresponding to these two defective product types. Specifically, it finds the corresponding status codes E101 and E201 through the inverse mapping relationship between defective product types and abnormal status codes. Then, it filters out all data rows in the dataset with status code fields of E101 or E201, forming a subset containing all relevant defective product information. Finally, for each row in this subset, it extracts the production equipment number, operator number, raw material batch number, and the corresponding environmental temperature and humidity values retrieved from the associated environmental monitoring sensor data table. These extracted fields are combined into new records. All these new records form a set of potential influencing factors.
[0051] Based on the set of potential influencing factors, continuous influencing factors such as ambient temperature are first binned at two-degree Celsius intervals, for example, into intervals of less than 20 degrees Celsius, 20 to 22 degrees Celsius, and 22 to 24 degrees Celsius. Then, for each influencing factor and its respective binning interval, the frequency of its occurrence in the potential influencing factor set is calculated and recorded as the defect occurrence frequency. Simultaneously, the full production log is queried to count the total production frequency for that influencing factor and binning interval. Then, the ratio is calculated, i.e., the defect occurrence frequency divided by the total production frequency, to obtain the specific defect rate for each interval. Next, an occurrence rate threshold is set. The threshold is calculated by first calculating the overall baseline defect rate for the specific defect type. Then calculate the standard deviation of the specific defect rate for each bin interval of all influencing factors. threshold Set as For example, if the baseline defect rate is 0.2 percent and the standard deviation is 0.3 percent, then the threshold is 0.65 percent. The specific defect rate calculated for each interval is compared with this threshold. All influencing factors that are higher than the threshold and their corresponding binning intervals or specific values, such as equipment number A03, operator B07, and ambient temperature range of 24 to 26 degrees Celsius, are identified as key influencing factors. They are then stored together with the corresponding defect type in a two-dimensional table to generate a defect root cause correlation matrix.
[0052] The steps to obtain the full-chain traceability map of defective products are as follows:
[0053] Based on the key influencing factors in the root cause correlation matrix of defective products, the defective product batch number is used as the query keyword to search the supplier management database for matching supplier name, inbound inspection item value and transportation batch number. The retrieved information is then arranged in chronological order to construct a raw material traceability path.
[0054] Based on the raw material traceability path, using the defective batch number as the query index, the production execution database is used to retrieve all process equipment numbers, operator numbers, and start and end timestamp records of each process during the batch product flow. The retrieved records are then sorted in ascending order by timestamp to construct the production process traceability path.
[0055] Based on the raw material traceability path and the production process traceability path, all information nodes in the two paths are merged and sorted according to timestamps to form a continuous time series from raw material warehousing inspection to finished product output. The series is then visualized and connected to establish a full-chain traceability map for defective products.
[0056] Specifically, based on the key influencing factors identified in the defective product root cause correlation matrix, such as raw material batch number RN20230801A, the defective product batch number associated with this influencing factor is first extracted from the matrix, for example, BP20230915X. Then, this defective product batch number is used as the primary key to construct an SQL query statement and perform an exact match query in the incoming material inspection table of the supplier management database. From the query result set, the supplier name field, the measured values of all inspection items such as the purity and particle size distribution of the cathode material, and the transport tracking number associated with this batch of raw materials are extracted. Next, this retrieved information, including the supplier name, the values of each inspection item, and the transport batch number, is bound to a fixed timestamp, i.e., the time when the batch of raw materials was inspected upon entry into the warehouse, forming a structured data node containing the timestamp and multiple source information. Finally, all source information nodes related to this defective product are linked in chronological order of the time when the incoming inspection was completed to construct a raw material traceability path. Based on the defective batch number contained in the raw material traceability path, such as BP20230915X, this batch number is used as a query index to query the work order flow table in the production execution database. This query will return all process records that the batch of products has undergone in the entire production process. For each process record returned, the equipment number that performed the process is extracted, such as mixer JM05 or coating machine TB02, the personnel number that operated the equipment, such as OP1138, and the start and end timestamps of the process. These extracted equipment numbers, personnel numbers, start and end timestamps are used as an information node. Then, all process information nodes related to the defective batch number are collected and sorted in ascending order according to their respective process start timestamps to form an ordered sequence from material input to finished product output, thus constructing a production process traceability path.
[0057] Based on the raw material traceability path and the production process traceability path, all information nodes in the two paths are first extracted. Each node contains a timestamp and associated event information. For example, nodes in the raw material traceability path contain the warehousing inspection timestamp and supplier information, while nodes in the production process traceability path contain the process start timestamp and equipment number. Then, these two sets of nodes are placed into the same set, and all nodes in the set are uniformly sorted in ascending order according to their respective timestamps to form a single, chronologically ordered continuous time series. This series clearly shows every key step and status from the entry of raw materials into the factory to the completion of the final product. Finally, using graph database technology, each node in this time series is represented as a graph vertex, and the sequential relationship between nodes is represented as directed edges. Visual rendering is then performed to establish a full-chain traceability map for defective products.
[0058] The steps for obtaining the quality closed-loop optimization instruction set are as follows:
[0059] Using a full-chain traceability map of defective products, check whether the parameter value of each node in the map exceeds the corresponding range of the quantitative quality standard set. Nodes that exceed the range are marked as abnormal parameter nodes. The information of raw materials, equipment and personnel to which the nodes belong is classified and organized to generate a list of nodes to be optimized.
[0060] Based on the list of nodes to be optimized, create instructions for each node in the list. The instructions include node identifier, abnormal parameters, target supplier number, target equipment number and target personnel number. Group the instructions into three categories: raw materials, equipment and personnel to form a draft of classified optimization instructions.
[0061] Based on the draft classification optimization instructions, the optimized values of equipment parameters in the instructions are extracted and written into the corresponding fields of the quality control point standard library. The limits of related indicators in the raw material acceptance standards are adjusted simultaneously. All updated instructions are integrated and encapsulated to obtain a quality closed-loop optimization instruction set.
[0062] Specifically, a defective product end-to-end traceability map is used. Starting from the first node of the map, each information node in the map is traversed one by one. For each node, the recorded parameter value is read, such as the coating thickness deviation value or the welding strength value. Then, the quantitative quality standard set is called to find the standard range corresponding to the parameter, that is, the upper limit value and the lower limit value. The actual parameter value of the node is compared with the standard range. If the parameter value of the node is less than the lower limit value or greater than the upper limit value, the node is marked as an abnormal parameter node. After traversing all nodes, all nodes marked as abnormal parameters are summarized and classified according to the content of the node information. For example, if the node information involves the supplier name or raw material batch number, it is classified into the raw material dimension; if it involves the equipment number, it is classified into the equipment dimension; if it involves the operator number, it is classified into the personnel dimension. The classified and sorted results are used to generate a list of nodes to be optimized.
[0063] Based on the list of nodes to be optimized, each node in the list is traversed, and a structured instruction object is automatically generated for each node. This instruction object contains four core fields. The first field is the node identifier, used to uniquely identify the node, for example, using a combination of raw material batch number or equipment number and timestamp. The second field is the abnormal parameter, recording the specific parameter name and value that exceeds the standard for the node, such as a coating thickness deviation of 10.5 micrometers. The third and fourth fields are the target number, which are filled according to the dimension of the node. For example, for nodes in the raw material dimension, the target supplier number is filled; for nodes in the equipment dimension, the target equipment number is filled; and for nodes in the personnel dimension, the target personnel number is filled. After the instruction objects for all nodes are created, these instruction objects are grouped and stored according to three categories: raw materials, equipment, and personnel, forming a draft of categorized optimization instructions.
[0064] According to the draft classification optimization instructions, the instructions for all equipment categories are first screened out. From these instructions, the recommended values for equipment parameter optimization recorded in the abnormal parameter fields are extracted. For example, the blade gap of the coating machine is adjusted from 50 micrometers to 48 micrometers. Then, these new parameter values are written into the corresponding equipment parameter setting fields in the quality control point standard library through database update operations to achieve dynamic updates of the standards. Simultaneously, the instructions for all raw material categories are screened out, their abnormal parameters are analyzed, and the acceptance standards related to the raw materials are adjusted accordingly. For example, the upper limit of the particle size D50 of the cathode material is tightened from 15 micrometers to 13 micrometers. Finally, all instructions, including the updated equipment instructions and personnel training instructions, are integrated and assigned a unique batch number, archived and packaged to obtain the quality closed-loop optimization instruction set.
[0065] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.
Claims
1. A method for QMS defect analysis and traceability in the battery manufacturing industry, characterized in that, Includes the following steps: Based on the performance requirements and safety standards of battery products, the defect rate target is quantified, the standard for the purity of positive electrode material and the deviation of cell capacity is defined, the coating thickness deviation and welding strength parameters are obtained, and the equipment operation status information entered by the operator and the raw material batch number provided by the supplier are integrated. The collected parameters are compared with the preset standard in real time, and data exceeding the threshold are marked to establish a multi-dimensional quality feature dataset. Based on the multidimensional quality feature dataset, by calculating the frequency of different defective product types and sorting them in descending order, the two defective product types with the highest frequency are identified. The production equipment number, operator number, raw material batch number and environmental temperature and humidity values associated with the identified defective product types are extracted. By calculating the correlation coefficient between each associated factor and the defective product incidence rate, factors with correlation coefficients higher than the preset correlation threshold are identified as key influencing factors, and a defective product root cause correlation matrix is generated. Based on the key influencing factors in the defective product root cause correlation matrix, the source index retrieves the defective product batch number and associates it with raw material supplier information, warehousing inspection records, and transportation batch number. The process index matches the batch of products with all process equipment numbers, operator numbers, and process start and end timestamps in the production process. The source information and process information are linked together in chronological order to form an information link from supplier to finished product, and a full-link traceability map of defective products is established.
2. The method for QMS defective product analysis and traceability based on battery manufacturing as described in claim 1, characterized in that, The method further includes: Using the aforementioned full-chain traceability map of defective products, the map interruption points and abnormal parameter nodes are located. Supplier audit instructions are generated for raw material nodes, parameter optimization instructions are generated for equipment parameter nodes, and skills training instructions are generated for operator nodes. The optimized equipment parameters are updated to the quality control point standard library, and the related raw material acceptance standards are adjusted. All instructions are summarized to obtain a quality closed-loop optimization instruction set.
3. The method for QMS defect analysis and traceability in battery manufacturing according to claim 1, characterized in that, The steps for obtaining the multidimensional quality feature dataset are as follows: Based on the performance requirements and safety standards of battery products, the capacity deviation of the cells is set with reference to industry documents, and the purity lower limit is set according to the technical specifications of the cathode material supplier. At the same time, the defect rate target is set. All the above numerical indicators and allowable fluctuation ranges are integrated to obtain a quantitative quality standard set. Based on the quantitative quality standard set, coating thickness deviation data is collected, welding strength values are obtained, and the equipment operation status code entered by the operator is read from the MES system. The supplier batch number is obtained by scanning the raw material packaging barcode. All data is then timestamped and aggregated by batch to generate the original production parameter stream. Based on the original production parameter stream, the values of each parameter in the stream are extracted and compared with the upper and lower limits of the corresponding indicators in the quantitative quality standard set. If the parameter value exceeds the range, an abnormal status code is assigned. The parameters with status codes, along with the associated equipment number and raw material batch number, are stored in a structured manner to establish a multidimensional quality feature dataset.
4. The method for QMS defective product analysis and traceability based on battery manufacturing as described in claim 1, characterized in that, The steps for obtaining the root cause correlation matrix of defective products are as follows: Based on the multidimensional quality feature dataset, all data marked as abnormal status codes in the dataset are traversed, the defective product types associated with the abnormal status codes are counted and statistically analyzed, and sorted in descending order according to the count values. The first two defective product types in the sorting results are selected to obtain a list of main defective product types. Based on the list of major defective product types, using the defective product types in the list as the filtering criteria, all matching records are filtered out in the multidimensional quality feature dataset, and the production equipment number, operator number, raw material batch number, and environmental temperature and humidity values are extracted from the matching records to establish a set of potential influencing factors. Based on the set of potential influencing factors, the ratio of the number of defective products to the total number of productions for each influencing factor in different numerical ranges is calculated. The ratio is compared with a preset occurrence rate threshold. Influencing factor combinations that exceed the threshold are identified as key influencing factors, and a root cause correlation matrix for defective products is generated.
5. The method for QMS defective product analysis and traceability based on battery manufacturing according to claim 1, characterized in that, The steps for obtaining the full-chain traceability map of defective products are as follows: Based on the key influencing factors in the root cause correlation matrix of defective products, the defective product batch number is used as the query keyword to search the supplier management database for matching supplier name, inbound inspection item value and transportation batch number. The retrieved information is then arranged in chronological order to construct a raw material traceability path. Based on the raw material traceability path, the defective batch number is used as the query index to retrieve all process equipment numbers, operator numbers, and start and end timestamp records of each process in the production execution database during the batch product flow process. The retrieved records are then sorted in ascending order by timestamp to construct the production process traceability path.
6. The method for QMS defect analysis and traceability in battery manufacturing according to claim 5, characterized in that, The steps for obtaining the full-chain traceability map of defective products also include: Based on the raw material traceability path and the production process traceability path, all information nodes in the two paths are merged and sorted according to timestamps to form a continuous time series from raw material warehousing inspection to finished product output. The series is then visualized and connected to establish a full-chain traceability map for defective products.
7. The method for QMS defect analysis and traceability in battery manufacturing according to claim 2, characterized in that, The steps for obtaining the quality closed-loop optimization instruction set are as follows: Using the aforementioned full-chain traceability map of defective products, check whether the parameter value of each node in the map exceeds the corresponding range of the quantitative quality standard set. Nodes that exceed the range are marked as abnormal parameter nodes. The information of raw materials, equipment, and personnel to which the nodes belong is classified and organized to generate a list of nodes to be optimized. Based on the list of nodes to be optimized, an instruction is created for each node in the list. The instruction content includes node identifier, abnormal parameters, target supplier number, target equipment number, and target personnel number. The instructions are then grouped into three categories: raw materials, equipment, and personnel to form a draft of categorized optimization instructions.
8. The method for QMS defect analysis and traceability in battery manufacturing according to claim 7, characterized in that, The steps for obtaining the quality closed-loop optimization instruction set also include: Based on the aforementioned draft classification optimization instructions, the optimized values of equipment parameters in the instructions are extracted and written into the corresponding fields of the quality control point standard library. The limits of related indicators in the raw material acceptance standards are adjusted synchronously. All updated instructions are integrated and encapsulated to obtain a quality closed-loop optimization instruction set.