Mass root cause analysis and report generation system based on large model

Abstract

The invention is applicable to the technical field of quality traceability, and particularly relates to a mass root cause analysis and report generation system based on a large model, the system comprises a data module and a large model module, the data module is used for collecting data in various modes and storing different data by using different databases, and the large model module is used for generating a large model; and the large model module is used for identifying a problem type according to the problem description, generating a data query and analysis instruction according to the problem type, performing data query and data extraction according to the instruction, performing analysis according to the queried data, obtaining a root cause analysis result, performing visualization processing on the root cause analysis result, and generating a quality report. According to the method, multiple factors and data are comprehensively considered by utilizing the large model, and for complex quality problems, key points can be accurately found out and effective measures can be formulated when product quality problems involving multiple suppliers, multiple raw materials and complex processes are processed by comprehensively collecting quality data and processing the large model.

Description

Technical Field

[0001] The present invention belongs to the technical field of quality traceability, and in particular relates to a quality root cause analysis and report generation system based on a large model. Background Art

[0002] In the prior art, product quality root cause analysis is usually provided by the quality data traceability module in the QMS (quality management system). In the QMS, the process parameters in the product production process, the composition and characteristic data of the raw materials, the operating status parameters of the production equipment, the operation records of each production link, and the various test results of the quality inspection link (including dimensional measurement values, performance index test values, appearance defects, etc.) are recorded. After collecting quality problems in product quality inspection or customer complaints, the quality personnel will extract the corresponding data from the system for analysis based on certain rules and personal experience, and write a quality problem report.

[0003] Existing technologies focus on the collection and analysis of quality data and surface monitoring of quality issues, but lack the ability to deeply analyze the root causes of quality issues. For example, when a quality problem occurs, although data statistics can show that the product failure rate of a certain supplier has increased, it is difficult to determine whether the problem is caused by the supplier's production process, raw material quality, or transportation links, and it is impossible to deeply explore the root cause. Summary of the invention

[0004] The purpose of the present invention is to provide a quality root cause analysis and report generation system based on a large model, aiming to solve the problem that when quality problems occur, although it can be known through data statistics that the unqualified rate of a certain supplier's products has increased, it is difficult to determine whether the problem is caused by the supplier's production process, raw material quality or transportation links, and it is impossible to deeply explore the root cause.

[0005] The present invention is implemented as follows: a quality root cause analysis and report generation system based on a large model, the method includes a data module and a large model module, the data module is used to collect data in a variety of ways, and use different databases to store different data, the large model module is used to identify the problem type according to the problem description, generate data query and analysis instructions according to the problem type, perform data query and data extraction according to the instructions, analyze the queried data, obtain the root cause analysis results, visualize the root cause analysis results, and generate a quality report.

[0006] Preferably, the data module includes a data acquisition unit for collecting quality data in a variety of ways. In the production stage, sensors are used to collect physical parameter data in the production process in real time, and in the quality inspection stage, test result data is obtained from the inspection equipment and manually input data is obtained.

[0007] Preferably, the data module includes a data acquisition unit for collecting quality data in a variety of ways. In the production stage, sensors are used to collect physical parameter data in the production process in real time, and in the quality inspection stage, test result data is obtained from the inspection equipment and manually input data is obtained.

[0008] Preferably, the data module includes a data storage unit for storing the collected data in a corresponding database, and storing the data according to the data structure and read and write frequency. For sensor data with a clear structure, the data is stored in a specific table of the relational database. For temporary cache data or frequently read and written data, a non-relational database is used for storage.

[0009] Preferably, the large model module includes a problem identification unit, which is used to input quality problems into the large model in the form of natural language. The large model understands and identifies the input problems through a pre-trained language model, and matches and calculates the similarity between the input problems and the existing quality problem type knowledge base in the model to determine the category and scope of the problems.

[0010] Preferably, the large model module includes an instruction generation unit for generating corresponding data query and analysis instructions according to the category and scope of the problem, and the data query and analysis instructions are generated based on the rules and algorithms within the large model.

[0011] Preferably, the large model module includes a data query unit for receiving instructions generated from the large model, obtaining required data from the data module, and retrieving relevant data from the stored data according to the conditions in the instructions using a database query interface and corresponding query statements.

[0012] Preferably, the large model module includes a root cause analysis unit for analyzing the queried data, mining patterns and rules in the data through a neural network structure, and analyzing the root causes of quality problems.

[0013] Preferably, the large model module includes a result visualization unit for visualizing the root cause analysis results, using data visualization tools to display the distribution of quality problems, the relationship between influencing factors, and the analysis process and conclusions of the root causes through charts and graphs.

[0014] Preferably, the large model module includes a report generation unit for generating a quality report based on the visualization results and the root cause analysis results, and the content of the quality report includes a description of the quality problem, an analysis process, a root cause conclusion and improvement suggestions.

[0015] Preferably, the quality report is in word format or pdf format, the quality report is recorded in natural language, and the quality report is filled in using a preset template.

[0016] The quality root cause analysis and report generation system based on a large model provided by the present invention improves the accuracy of root cause analysis. By comprehensively considering multiple factors and data using a large model, the contribution of each factor to quality problems and their interrelationships can be analyzed, such as determining whether product quality problems are caused by differences in supplier raw material batches or improper production process parameters. For complex quality problems, by comprehensively collecting quality data and processing with a large model, complex problems can be decomposed into multiple sub-problems and analyzed one by one. When dealing with product quality problems involving multiple suppliers, multiple raw materials, and complex processes, the key can be accurately identified and effective measures can be formulated. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 An architectural diagram of a large model-based quality root cause analysis and report generation system provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0019] It is understood that the terms "first", "second", etc. used in this application may be used herein to describe various elements, but unless otherwise specified, these elements are not limited by these terms. These terms are only used to distinguish a first element from another element. For example, without departing from the scope of this application, a first xx script may be referred to as a second xx script, and similarly, a second xx script may be referred to as a first xx script.

[0020] like Figure 1 As shown, it is an architecture diagram of a quality root cause analysis and report generation system based on a big model provided in an embodiment of the present invention. The method includes a data module and a big model module. The data module is used to collect data in a variety of ways and use different databases to store different data. The big model module is used to identify the problem type according to the problem description, generate data query and analysis instructions according to the problem type, query and extract data according to the instructions, analyze the queried data, obtain the root cause analysis results, visualize the root cause analysis results, and generate a quality report.

[0021] When faced with complex quality issues, existing technologies are unable to accurately analyze problems due to the lack of the ability to comprehensively consider multiple factors. For example, when product quality is affected by multiple factors such as design, production process, raw materials, and personnel operations, it is difficult for existing systems to determine the contribution of each factor to the quality problem and the relationship between them, and it is impossible to effectively formulate solutions. Manually written quality problem reports are highly subjective, and differences in knowledge, experience, and thinking patterns among different quality personnel will lead to a lack of consistency in the content of the report, which can easily cause confusion. It is inefficient, and it is time-consuming and laborious to manually screen, organize information, and write reports from large amounts of data, which can easily lead to delays in report generation; it is prone to omissions or inaccuracies in information, and may omit key information or make wrong judgments due to misreading data.

[0022] In this embodiment, the data module includes:

[0023] Data collection unit: collects quality-related data in a variety of ways. In the production process, various sensors (such as temperature sensors, pressure sensors, displacement sensors, etc.) are used to collect physical parameter data in the production process in real time; in the quality inspection process, test result data is obtained from various inspection equipment (such as non-destructive testing instruments, dimensional measurement equipment, etc.); at the same time, some special data that cannot be automatically obtained (such as descriptions of appearance defects, etc.) can also be collected from manual entry channels. These collected data are all accompanied by identification information such as timestamps, equipment numbers, and production batches for subsequent processing.

[0024] Data storage unit: Store the collected data in a suitable database. You can choose a relational database (such as MySQL, Oracle, etc.) or a non-relational database (such as MongoDB, Redis, etc.) to store the data according to its data structure, read and write frequency, and other characteristics. For example, sensor data with a clear structure can be stored in a specific table of a relational database; for some temporary cache data or frequently read and written data, a non-relational database can be used.

[0025] The large model module includes:

[0026] Problem identification unit: used to input quality problems into the big model in the form of natural language. For example, when the defective rate of products increases, the quality management personnel will input the description of "increasing defective rate problem". The big model understands and identifies the input problems through the pre-trained language model. Its internal word vector representation and neural network structure can convert natural language text into a vector form that can be processed by computers, and match and calculate similarity with the existing quality problem type knowledge base in the model to determine the category and scope of the problem.

[0027] Instruction generation unit: used to generate corresponding data query and analysis instructions based on the problem identification results. These instructions are generated based on the rules and algorithms within the large model. For example, if it is identified that the size of a certain part of the product is abnormal, the large model will generate instructions to query all relevant dimensional measurement data, production equipment parameters, and corresponding process parameters in the production process of the part. The instructions are generated in a specific format so that the subsequent execution module can understand and execute them.

[0028] Data query unit: used to receive instructions (data query and analysis instructions) from the large model generation instruction module and obtain the required data from the data module. Retrieve relevant data from the stored data according to the conditions in the instruction using the database query interface and the corresponding query statement (such as SQL statement). For complex query requirements, it may be necessary to combine data warehouse technology and data mining algorithms to extract valuable information from massive data. For example, the potential connection between data related to quality issues can be found through association analysis algorithms.

[0029] Root cause analysis unit: conducts in-depth analysis of the queried data. The large model uses its internal neural network structure to mine patterns and regularities in the data. Combined with professional knowledge in the quality field and experience gained from model pre-training, the root causes of quality problems are analyzed. For example, by analyzing the relationship between the changing trend of production equipment parameters and the time when defective products appear, it is determined whether the quality problem is caused by equipment failure; by analyzing the distribution of product quality under different process parameters, it is determined whether there are defects in the process. During the analysis process, it will also be combined with existing quality report data for learning and comparison to further optimize the analysis results. Historical cases and analysis conclusions in the quality report data can be used as a reference to help the large model more accurately determine the root cause of the current quality problem.

[0030] Result visualization unit: used to display the root cause analysis results in an intuitive way. Data visualization tools can be used to display the distribution of quality problems, the relationship between influencing factors, and the root cause analysis process and conclusions through charts (such as bar charts, line charts, scatter plots, etc.) and graphs (such as flow charts, fishbone diagrams, etc.). For example, a bar chart can be used to display the proportion of defective products caused by different reasons, and a fishbone diagram can be used to display the causal relationship between quality problems and various factors in the production process, which is convenient for quality management personnel to understand and make decisions.

[0031] Report generation unit: used for the large model to automatically generate a quality report based on the visualized results and detailed conclusions of the root cause analysis. The content of the quality report includes the description of the quality problem, the analysis process, the root cause conclusions and improvement suggestions. The report can be generated in word or pdf format and stored in the quality report management module. The text content in the report is generated through natural language generation technology. The large model fills in the content according to the preset template and analysis results to ensure the standardization and professionalism of the report. After the report is generated, the report data will also be fed back to the large model for subsequent learning and model optimization, so that the large model can be more accurate and efficient when dealing with similar quality problems.

[0032] The present invention improves the accuracy of root cause analysis. By comprehensively considering multiple factors and data using a large model, the contribution of each factor to quality problems and their interrelationships can be analyzed, such as determining whether product quality problems are caused by differences in supplier raw material batches or improper production process parameters. For complex quality problems, by comprehensively collecting quality data and processing with a large model, complex problems can be decomposed into multiple sub-problems and analyzed one by one. For example, when dealing with product quality problems involving multiple suppliers, multiple raw materials, and complex processes, the key can be accurately identified and effective measures can be formulated.

[0033] The present invention improves the adaptability of root cause analysis. Based on the learning mechanism of neural networks, the large model can continuously learn new patterns from new data. When performing root cause analysis on quality problems, the large model can combine professional knowledge in the quality field and experience obtained from model pre-training to conduct analysis. For example, when faced with a new combination of production process parameters, the large model can learn the potential connection between these new parameters and quality problems by analyzing a large amount of data related to them.

[0034] The present invention improves the efficiency and accuracy of report generation. Using a large model to generate reports can objectively, accurately and efficiently analyze quality problems, generate standardized and professional reports, and cover quality problem descriptions, analysis processes, root cause conclusions and improvement suggestions, etc.; the report format can be unified according to preset templates; and the results can be intuitively displayed in a variety of visual ways, which is easy for quality management personnel to understand.

[0035] Overall, the automated process from data processing to report generation greatly shortens the quality analysis cycle, enabling companies to respond to quality issues more quickly and reduce losses caused by quality issues.

[0036] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

[0037] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A quality root cause analysis and report generation system based on a large model, characterized in that: The method includes a data module and a large model module. The data module is used to collect data in multiple ways and use different databases to store different data. The large model module is used to identify the problem type according to the problem description, generate data query and analysis instructions according to the problem type, query and extract data according to the instructions, analyze the queried data, obtain the root cause analysis results, visualize the root cause analysis results, and generate a quality report.

2. The large model-based quality root cause analysis and report generation system according to claim 1, characterized in that: The data module includes a data acquisition unit, which is used to collect quality data in a variety of ways in the production stage, use sensors to collect physical parameter data in the production process in real time, and in the quality inspection stage, obtain test result data from the inspection equipment and obtain manually entered data.

3. The quality root cause analysis and report generation system based on a large model according to claim 1, characterized in that: The data module includes a data storage unit, which is used to store the collected data in a corresponding database, and store the data according to the data structure and read and write frequency. For sensor data with a clear structure, it is stored in a specific table of the relational database. For temporary cache data or frequently read and written data, a non-relational database is used for storage.

4. The large model-based quality root cause analysis and report generation system according to claim 1, characterized in that: The large model module includes a problem identification unit, which is used to input quality problems into the large model in the form of natural language. The large model understands and identifies the input problems through a pre-trained language model, and matches and calculates the similarity between the input problems and the existing quality problem type knowledge base in the model to determine the category and scope of the problems.

5. The large model-based quality root cause analysis and report generation system according to claim 4, characterized in that: The large model module includes an instruction generation unit, which is used to generate corresponding data query and analysis instructions according to the category and scope of the problem. The data query and analysis instructions are generated based on the rules and algorithms inside the large model.

6. The large model-based quality root cause analysis and report generation system according to claim 4, characterized in that: The large model module includes a data query unit for receiving instructions generated from the large model, obtaining required data from the data module, and retrieving relevant data from the stored data according to the conditions in the instructions using a database query interface and corresponding query statements.

7. The large model-based quality root cause analysis and report generation system according to claim 4, characterized in that: The large model module includes a root cause analysis unit, which is used to analyze the queried data, mine patterns and rules in the data through a neural network structure, and analyze the root causes of quality problems.

8. The large model-based quality root cause analysis and report generation system according to claim 4, characterized in that: The large model module includes a result visualization unit for visualizing the root cause analysis results, using data visualization tools to display the distribution of quality problems, the relationship between influencing factors, and the analysis process and conclusions of the root causes through charts and graphs.

9. The large model-based quality root cause analysis and report generation system according to claim 4, characterized in that: The large model module includes a report generation unit, which is used to generate a quality report based on the visualization results and the root cause analysis results. The quality report content includes the description of the quality problem, the analysis process, the root cause conclusion and improvement suggestions.

10. The large model-based quality root cause analysis and report generation system according to claim 9, characterized in that: The quality report is in word format or pdf format, the quality report is recorded in natural language, and the quality report is filled in using a preset template.

Images