Intelligent quality inspection data authenticity judgment method, device and system and storage medium

Abstract

The invention relates to an intelligent quality inspection data authenticity judgment method, device and system and a storage medium, and the method comprises the steps: collecting the original quality inspection data of a product through detection equipment disposed on a production line, carrying out the multi-dimensional processing of the original quality inspection data, obtaining the first-dimensional data, and storing the first-dimensional data in a storage database; integrating the first dimension data according to a preset matrix decomposition rule to obtain second dimension data, and performing feature extraction calculation on the second dimension data to obtain corresponding dimension features; obtaining first historical dimension data from the storage database, and performing trend comparison on the dimension features according to the first historical dimension data based on the association relationship to obtain conformity; and comprehensively analyzing the abnormal information and the conformity to obtain comprehensive information, and generating a judgment result of the first dimension data according to the comprehensive information. According to the invention, multi-dimensional information can be effectively utilized, and the problem of information omission is avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of quality inspection data, and in particular to an intelligent method, device, system and storage medium for judging the authenticity of quality inspection data. Background Art

[0002] In the field of industrial manufacturing, product quality control is a key link to ensure stable product performance and improve user satisfaction. Traditional quality inspection methods usually rely on manual inspection or simple automated inspection equipment, which often have certain limitations. For example, on some complex product production lines, it is difficult to achieve comprehensive and in-depth quality inspection of products due to the limitations of inspection equipment and insufficient technical means. At the same time, due to the lack of effective data integration and analysis methods, it is difficult to accurately predict the trend of data changes, which affects the accurate judgment of product quality. Summary of the invention

[0003] The main purpose of the present invention is to provide an intelligent method, device, system and storage medium for judging the authenticity of quality inspection data, which can effectively utilize multi-dimensional information and avoid the problem of information omission.

[0004] To achieve the above object, the present invention provides an intelligent method for judging the authenticity of quality inspection data, comprising: Collecting original quality inspection data of the product through the detection equipment deployed on the production line, performing multi-dimensional processing on the original quality inspection data to obtain first dimensional data, and storing the first dimensional data in a storage database; Integrate the first dimensional data according to a preset matrix decomposition rule to obtain second dimensional data, perform feature extraction calculation on the second dimensional data to obtain corresponding dimensional features; Performing association analysis on the first dimensional data and the dimensional features according to a preset association rule to obtain an association relationship, and performing anomaly detection on the dimensional features according to the association relationship to obtain anomaly information; Acquire first historical dimension data from the storage database, and compare the trend of the dimension feature according to the first historical dimension data based on the association relationship to obtain a degree of conformity; The abnormal information and the conformity are comprehensively analyzed to obtain comprehensive information, and a judgment result of the first dimensional data is generated based on the comprehensive information.

[0005] Furthermore, the method of collecting original quality inspection data of the product through the detection equipment deployed on the production line, performing multi-dimensional processing on the original quality inspection data to obtain first dimensional data, and storing the first dimensional data in the storage database includes: Collect the original quality inspection data of the product through various inspection devices deployed on the production line, and classify and process the original quality inspection data to obtain original type data; Performing multidimensional processing on the original type data according to a preset multidimensional rule to obtain original multidimensional data, and performing standardization processing on the original multidimensional data to obtain the first dimensional data; The first dimensional data is stored in a storage database.

[0006] Furthermore, the first dimensional data is integrated according to a preset matrix decomposition rule to obtain second dimensional data, and feature extraction calculation is performed on the second dimensional data to obtain corresponding dimensional features, including: Performing matrix conversion on the first dimensional data to obtain a first dimensional data matrix, and performing non-negative matrix decomposition on the first dimensional data matrix to obtain a base matrix and a coefficient matrix; Reconstruct the first dimensional data according to the basis matrix and the coefficient matrix to obtain reconstructed data; The reconstructed data is subjected to difference calculation according to the preset standard first dimension data to obtain the reconstruction error; The first dimensional data is screened according to the reconstruction error to obtain screened first dimensional data; feature extraction is performed on the screened first dimensional data to obtain size dimensional features, material composition dimensional features, surface defect dimensional features, functional test dimensional features and time series dimensional features; Construct feature vectors of size dimension features, material composition dimension features, surface defect dimension features, functional test dimension features and time series dimension features to obtain feature vector matrix, calculate feature weights of the feature vector matrix to obtain weighted feature data; Based on the weighted feature data, comprehensive feature extraction is performed on size dimensional features, material composition dimensional features, surface defect dimensional features, functional test dimensional features and time series dimensional features to obtain the dimensional features.

[0007] Further, the first dimensional data is associated with the dimensional feature according to a preset association rule to obtain an association relationship, and anomaly detection is performed on the dimensional feature according to the association relationship to obtain anomaly information, including: Calculating the correlation coefficient matrix between the first dimensional data and the dimensional features to obtain an initial correlation strength table, applying an adaptive threshold screening method to the initial correlation strength table to obtain core correlation pairs; Based on the core association pairs, a multi-level association network is constructed to obtain a hierarchical association structure; a topological analysis is performed on the hierarchical association structure to obtain target nodes and association paths; Evaluate the importance of the dimension features according to the target node and the associated path to obtain a feature weight vector; Directingly setting a preset anomaly detection rule set according to the feature weight vector to obtain a dynamic detection criterion, and performing anomaly detection on the dimensional feature according to the dynamic detection criterion to obtain an original anomaly indicator; The original abnormal indicators are grouped and constructed by using a density clustering method to obtain an abnormal category matrix, and the abnormal category matrix is ​​used to screen and decompose the original abnormal indicators to extract abnormal trend characteristics; Based on the abnormal trend characteristics and hierarchical association structure, anomaly propagation simulation is performed to form abnormal impact characteristics. The abnormal impact characteristics are evaluated according to the preset evaluation rules to obtain the final abnormal level. An exception description is generated according to the final exception level and the hierarchical association structure to obtain the exception information.

[0008] Further, the acquiring of the first historical dimension data from the storage database, and performing trend comparison on the dimension feature according to the first historical dimension data based on the association relationship to obtain the degree of conformity include: Extracting first historical dimension data matching the time span of the dimension feature from the storage database; Performing time series decomposition on the first historical dimension data to obtain three components of periodicity, seasonality and randomness, and performing feature mapping on the dimension features and the three components of periodicity, seasonality and randomness according to a preset feature mapping rule to obtain a multi-dimensional mapping feature matrix; Based on the association relationship, feature importance evaluation is performed on the multidimensional mapping feature matrix to obtain a feature weight vector; Using a predefined adaptive time window, dynamically adjust the feature weight vector to obtain a time-varying weight sequence, and perform a correlation analysis on the time-varying weight sequence and the dimensional feature according to a preset nonlinear trend comparison algorithm to obtain a correlation index; The correlation index is dynamically adjusted based on a preset adaptive threshold mechanism to obtain an adjustment index, and the adjustment index is hierarchically evaluated according to a predefined multi-level evaluation standard to obtain the degree of compliance.

[0009] Furthermore, the performing a comprehensive analysis on the abnormal information and the compliance to obtain comprehensive information, and generating a judgment result of the first dimensional data according to the comprehensive information, includes: Performing weight calculation on the abnormal information to obtain an abnormal weight coefficient, performing numerical standardization on the compliance to obtain a standardized compliance, and performing a comprehensive evaluation based on the abnormal weight coefficient and the standardized compliance to obtain comprehensive information; Based on the preset evaluation threshold, the comprehensive information is graded to obtain preliminary classification results, and the preliminary classification results are intelligently analyzed to obtain optimized judgment parameters; Extracting historical judgment data from the storage database, and correcting the preliminary classification result using optimized judgment parameters to obtain a correction result; A judgment report is generated according to the correction result, wherein the judgment report includes authenticity evaluation and credibility analysis; based on the judgment report, feature identification is performed on the first dimensional data to obtain the final judgment result.

[0010] The present invention also provides an intelligent quality inspection data authenticity judgment device, which is applied to any of the above-mentioned intelligent quality inspection data authenticity judgment methods, comprising: A collection module, the collection module is used to collect original quality inspection data of the product through the detection equipment deployed on the production line, perform multi-dimensional processing on the original quality inspection data to obtain first dimensional data, and store the first dimensional data in a storage database; An analysis module, the analysis module is used to integrate the first dimensional data according to a preset matrix decomposition rule to obtain second dimensional data, perform feature extraction calculation on the second dimensional data to obtain corresponding dimensional features; An association module, the association module is used to perform association analysis on the first dimensional data and the dimensional features according to a preset association rule to obtain an association relationship, and perform anomaly detection on the dimensional features according to the association relationship to obtain anomaly information; A processing module, the processing module is used to obtain the first historical dimension data from the storage database, and perform trend comparison on the dimension feature according to the first historical dimension data based on the association relationship to obtain a degree of conformity; A control module is used to perform a comprehensive analysis on the abnormal information and the compliance to obtain comprehensive information, and generate a judgment result of the first dimensional data based on the comprehensive information.

[0011] The present invention also provides an intelligent quality inspection data authenticity judgment system, comprising: Memory, used to store programs; The processor is used to execute the program to implement each step of the intelligent method for determining the authenticity of quality inspection data as described in any one of the above.

[0012] The present invention also provides a storage medium storing computer instructions, wherein the computer instructions are used to enable a computer to execute any of the above methods.

[0013] The present invention provides the following beneficial effects: By deploying detection equipment on the production line to collect the original quality inspection data of the product in real time and perform multi-dimensional processing on it, multi-dimensional information can be effectively utilized, overcoming the problem of information omission caused by focusing only on single first-dimensional data in traditional methods. By integrating multiple first-dimensional data through preset matrix decomposition rules, and combining feature extraction technology to obtain dimensional features, and then using preset association rules for anomaly detection, potential quality problems can be quickly and accurately identified, improving the sensitivity and accuracy of anomaly detection. By comparing the trend changes of the first historical dimension data with the current dimension features, the conformity of the data can be obtained, which helps to predict quality problems that may occur in the future and provides a basis for taking preventive measures in advance. Comprehensive analysis of abnormal information and conformity to generate judgment results of the first-dimensional data can help quality inspectors make decisions faster and improve work efficiency. By performing multi-dimensional analysis and integration of data, not only the utilization rate of data is improved, but also more hidden valuable information can be excavated, which helps to optimize the production process and reduce resource waste. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 This is a flow chart of the intelligent method for judging the authenticity of quality inspection data provided by the present invention; Figure 2 This is a structural diagram of the intelligent quality inspection data authenticity judgment device provided by the present invention; Figure 3 It is a structural diagram of the intelligent quality inspection data authenticity judgment system provided by the present invention.

[0015] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0016] 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.

[0017] The present invention is further described below in conjunction with the accompanying drawings and specific implementation methods.

[0018] Reference Figure 1 As shown, the present invention provides an intelligent method for judging the authenticity of quality inspection data, comprising: Step S1: collecting original quality inspection data of the product through the detection equipment deployed on the production line, performing multi-dimensional processing on the original quality inspection data to obtain first dimensional data, and storing the first dimensional data in a storage database; Step S2: Integrate the first dimensional data according to a preset matrix decomposition rule to obtain second dimensional data, perform feature extraction calculation on the second dimensional data, and obtain corresponding dimensional features; Step S3: performing association analysis on the first dimension data and the dimension features according to the preset association rules to obtain the association relationship, and performing anomaly detection on the dimension features according to the association relationship to obtain anomaly information; Step S4: obtaining the first historical dimension data from the storage database, and performing trend comparison on the dimension features according to the first historical dimension data based on the association relationship to obtain the degree of conformity; Step S5: Comprehensively analyze the abnormal information and the degree of compliance to obtain comprehensive information, and generate a judgment result of the first dimension data based on the comprehensive information.

[0019] Based on the above steps, the detailed steps are as follows: Step S1: By deploying various inspection equipment (such as visual inspection system, pressure sensor, temperature sensor, etc.) on the production line, various quality inspection data of the product during the production process are collected in real time.

[0020] The collected raw quality inspection data is preprocessed by cleaning, denoising and standardization to ensure data quality. The preprocessed data is multi-dimensionally processed, including data aggregation, grouping, conversion and other operations, to extract higher-level features or statistics.

[0021] The data obtained after multidimensional processing is defined as "first dimension data", which usually contains key information about product quality.

[0022] The first dimension data will be stored in a specially designed storage database for subsequent analysis.

[0023] Step S2: According to a preset matrix decomposition rule (such as singular value decomposition SVD, principal component analysis PCA, etc.), the stored first dimension data is integrated to obtain a new data structure, namely, "second dimension data".

[0024] Perform feature extraction calculations on the second-dimensional data, including operations such as selecting important features and reducing dimensionality.

[0025] The goal of feature extraction is to find key features from the data that can represent the essential attributes of the data, thereby simplifying the subsequent analysis process.

[0026] The result obtained after feature extraction and calculation is called "dimensional feature". Dimensional features are used to describe the quality status of products in various aspects. Step S3: According to the preset association rules, the first dimension data obtained in the first step and the dimension features obtained in the second step are subjected to association analysis.

[0027] Through association analysis, the relationship between different data points is established to form an association network.

[0028] Use the established associations to perform anomaly detection on dimensional features and identify data points that do not conform to the expected pattern. Anomaly detection can be achieved through statistical methods and machine learning algorithms (such as isolation forests, autoencoders, etc.).

[0029] The information obtained after anomaly detection is called "anomaly information".

[0030] Step S4: The previously recorded historical first dimension data is obtained from the storage database. These data may be the quality inspection data of the same type of products in different time periods.

[0031] Based on historical data, the association relationship is used to perform trend analysis on dimension features.

[0032] Trend analysis includes time series analysis, moving average, exponential smoothing and other methods, which are used to discover the changing trends of data over time.

[0033] The dimensional features of the current batch are compared with the trend of the first historical dimensional data, and the degree of difference between the dimensional features and the first historical dimensional data is calculated.

[0034] The higher the degree of compliance, the closer the quality of the current batch of products is to the historical average or expected standard; otherwise, there may be quality problems.

[0035] Step S5: Combine the abnormal information and the degree of compliance to conduct a comprehensive analysis. The purpose of the analysis is to determine whether there are quality problems in the current batch of products and the severity of the problems.

[0036] Based on the results of the comprehensive analysis, a report containing all relevant information is formed, namely "comprehensive information". Finally, the judgment result of the first dimension data is generated based on the comprehensive information to determine whether the batch of products is qualified.

[0037] The judgment results include qualified, unqualified, need for further inspection and other conclusions, and can be accompanied by specific improvement suggestions or action plans.

[0038] The present invention provides an intelligent method for judging the authenticity of quality inspection data. The original quality inspection data of the product is collected in real time by the detection equipment deployed on the production line, and multi-dimensional processing is performed on it. The multi-dimensional information can be effectively used, and the problem of information omission caused by focusing only on the single first-dimensional data in the traditional method is overcome. By integrating multiple first-dimensional data through a preset matrix decomposition rule, and combining the feature extraction technology to obtain the dimensional features, and then using the preset association rules for abnormal detection, potential quality problems can be quickly and accurately identified, and the sensitivity and accuracy of abnormal detection are improved. By comparing the trend changes of the first historical dimension data and the current dimension features, the conformity of the data can be obtained, which is helpful to predict the quality problems that may occur in the future and provide a basis for taking preventive measures in advance. Comprehensive analysis of abnormal information and conformity is performed to generate the judgment result of the first-dimensional data, which can help quality inspectors make decisions faster and improve work efficiency. By performing multi-dimensional analysis and integration of data, not only the utilization rate of data is improved, but also more hidden valuable information can be excavated, which helps to optimize the production process and reduce resource waste.

[0039] In one embodiment, original quality inspection data of products are collected by detection equipment deployed on a production line, the original quality inspection data is multi-dimensionally processed to obtain first dimensional data, and the first dimensional data is stored in a storage database, including: The original quality inspection data of the products is collected through various inspection equipment deployed on the production line, and the original quality inspection data is classified and processed to obtain the original type data; Perform multidimensional processing on the original type data according to the preset multidimensional rules to obtain original multidimensional data, and perform standardization processing on the original multidimensional data to obtain first dimensional data; The first dimension data is stored in a storage database.

[0040] This embodiment collects the original quality inspection data of the product in real time through various detection equipment deployed on the production line, and classifies and processes it to obtain the original type data, ensuring the comprehensiveness and accuracy of the data. The original type data is multi-dimensionally processed according to the preset multi-dimensional rules, which can make full use of various types of detection data and overcome the problem of information omission caused by single-dimensional data processing that may exist in traditional methods. The original multi-dimensional data is standardized, and the first-dimensional data obtained is consistent and comparable, which is convenient for subsequent data analysis and storage.

[0041] In one embodiment, the first dimensional data is integrated according to a preset matrix decomposition rule to obtain second dimensional data, and feature extraction calculation is performed on the second dimensional data to obtain corresponding dimensional features, including: Perform matrix conversion on the first dimension data to obtain a first dimension data matrix, perform non-negative matrix decomposition on the first dimension data matrix to obtain a base matrix and a coefficient matrix; Reconstruct the first dimension data according to the basis matrix and the coefficient matrix to obtain reconstructed data; The reconstructed data is subjected to difference calculation according to the preset standard first dimension data to obtain the reconstruction error; The first dimension data is screened according to the reconstruction error to obtain screened first dimension data; feature extraction is performed on the screened first dimension data to obtain size dimension features, material composition dimension features, surface defect dimension features, functional test dimension features and time series dimension features; Construct feature vectors of size dimension features, material composition dimension features, surface defect dimension features, functional test dimension features and time series dimension features to obtain feature vector matrix, calculate feature weights of the feature vector matrix to obtain weighted feature data; Based on the weighted feature data, comprehensive feature extraction is performed on size dimension features, material composition dimension features, surface defect dimension features, functional test dimension features and time series dimension features to obtain dimensional features.

[0042] This embodiment uses matrix decomposition technology (such as non-negative matrix decomposition NMF) to reduce the dimension of the first dimension data, which can effectively extract key information from the data, reduce computational complexity, and improve data processing capabilities. By calculating the difference between the reconstructed data and the standard first dimension data, the reconstruction error is obtained, which helps to evaluate the data quality and the degree of model fitting, thereby screening out higher quality data for subsequent analysis. The multiple dimensional features (such as size, material composition, surface defects, functional tests, and time series features) obtained by feature extraction calculation can comprehensively reflect the product quality characteristics from different angles. The feature weights are calculated for the feature vector matrix to obtain weighted feature data, and different weights can be given according to the importance of different features, making the decision basis more scientific and reasonable. Comprehensive feature extraction not only contains the information of a single feature, but also captures the correlation between different features, enhancing the ability of pattern recognition and anomaly detection. By comprehensively considering multi-dimensional features, the detection accuracy can be significantly improved, and the processing efficiency is also improved because the data is effectively reduced in dimension.

[0043] In one embodiment, the first dimension data and the dimension feature are associated and analyzed according to a preset association rule to obtain an association relationship, and an anomaly detection is performed on the dimension feature according to the association relationship to obtain anomaly information, including: The correlation coefficient matrix between the first dimension data and the dimension features is calculated to obtain the initial correlation strength table. The adaptive threshold screening method is applied to the initial correlation strength table to screen out feature pairs with strong correlation, which are the core correlation pairs.

[0044] A multi-level association network is constructed based on core association pairs to form a hierarchical association structure. A topological analysis is performed on the hierarchical association structure to determine the key nodes (target nodes) in the network and the connection paths (association paths) between them.

[0045] The importance of dimension features is evaluated according to the target node and associated path to obtain the feature weight vector.

[0046] The preset anomaly detection rule set is set in a targeted manner according to the feature weight vector to generate a dynamic detection criterion suitable for the current scenario. Anomaly detection is performed on the dimensional features according to the dynamic detection criterion to obtain the original anomaly index.

[0047] The density clustering method is used to group the original abnormal indicators and obtain the abnormal category matrix. The abnormal category matrix is ​​used to screen and decompose the original abnormal indicators and extract the abnormal trend characteristics.

[0048] Based on the abnormal trend characteristics and hierarchical association structure, anomaly propagation simulation is performed to form abnormal impact characteristics. The abnormal impact characteristics are evaluated according to the preset evaluation rules to obtain the final abnormal level.

[0049] An exception description is generated according to the final exception level and the hierarchical association structure to obtain exception information.

[0050] This embodiment more accurately identifies which features are associated with abnormal behavior through in-depth analysis of the relationship between the first dimension data and the dimensional features, thereby improving the accuracy of anomaly detection. The setting of dynamic detection criteria enables the anomaly detection rules to be adjusted according to the importance of the features and the changes in the actual data, thereby improving the flexibility and adaptability of the detection rules. The use of anomaly propagation simulation makes the impact path of the anomaly visual, which helps to understand how the anomaly propagates in the system and facilitates subsequent problem location and resolution. The construction of a multi-level association network can reveal the complex relationship between features of different dimensions, thereby helping analysts understand the inherent connection of data from multiple perspectives. The application of density clustering methods can automatically classify anomaly indicators into different categories, which helps to quickly identify anomalies of similar types and facilitates classification processing. By extracting abnormal trend features, valuable information can be extracted from large amounts of data more efficiently, reducing the workload of manual analysis. The existence of preset evaluation rules ensures the consistency and standardization of anomaly evaluation and avoids deviations caused by subjective judgment.

[0051] In one embodiment, first historical dimension data is obtained from a storage database, and trend comparison of dimension features is performed based on the association relationship according to the first historical dimension data to obtain a degree of conformity, including: Extracting first historical dimension data matching the time span of the dimension feature from a storage database; The first historical dimension data is decomposed into time series to obtain three components: periodicity, seasonality and randomness. According to the preset feature mapping rules, feature mapping is performed on the dimensional characteristics and the three components: periodicity, seasonality and randomness to obtain a multi-dimensional mapping feature matrix. The feature mapping rules are functions defined based on domain knowledge.

[0052] The feature importance of the multidimensional mapping feature matrix is ​​evaluated based on the association relationship to obtain the feature weight vector.

[0053] Using the predefined adaptive time window, the feature weight vector is dynamically adjusted to obtain a time-varying weight sequence. The correlation between the time-varying weight sequence and the dimensional feature is analyzed according to the preset nonlinear trend comparison algorithm to obtain the correlation index. Based on a preset adaptive threshold mechanism, the correlation index is dynamically adjusted to obtain an adjustment index, and the adjustment index is hierarchically evaluated according to a predefined multi-level evaluation standard to obtain the degree of compliance, wherein the adaptive threshold mechanism dynamically adjusts the threshold according to the actual distribution of the data, and the multi-level evaluation standard includes multiple thresholds and corresponding grade divisions.

[0054] This embodiment decomposes the historical data into time series to gain a deeper understanding of the periodicity, seasonality and randomness characteristics of the data, providing richer information for subsequent analysis. By mapping the dimensional features to the different components of the historical data, a closer association can be established to improve the accuracy and effectiveness of the analysis. The feature weight vector obtained by the feature importance assessment can determine which features are most critical for the analysis, thereby improving the pertinence and efficiency of the analysis. Dynamic adjustment of feature weights using an adaptive time window can better reflect the importance of features that change over time and improve the adaptability and accuracy of the analysis. The correlation index obtained through correlation analysis can quantify the similarity between historical data and current data, which helps to identify potential trends and patterns. The use of a nonlinear trend comparison algorithm can more accurately capture the complex relationship between data and improve the accuracy of the analysis. By dynamically adjusting the correlation index, data fluctuations can be handled more flexibly and the robustness of the analysis can be improved.

[0055] In one embodiment, the abnormal information and the compliance are comprehensively analyzed to obtain comprehensive information, and a judgment result of the first dimension data is generated based on the comprehensive information, including: The abnormal information is weighted to obtain the abnormal weight coefficient, the compliance is numerically standardized to obtain the standardized compliance, and a comprehensive evaluation is performed based on the abnormal weight coefficient and the standardized compliance to obtain comprehensive information.

[0056] Based on the preset evaluation threshold, the comprehensive information is graded to obtain preliminary classification results, and the preliminary classification results are intelligently analyzed to obtain optimized judgment parameters.

[0057] The historical judgment data is extracted from the storage database, and the preliminary classification results are corrected using the optimized judgment parameters to obtain the corrected results.

[0058] According to the correction result, a judgment report is generated, wherein the judgment report includes authenticity assessment and credibility analysis. Based on the judgment report, the first dimension data is feature identified to obtain the final judgment result.

[0059] This embodiment can more accurately reflect the importance of abnormal events by assigning different weights to abnormal information, and avoid the excessive influence of a single abnormal point on the overall evaluation results. Standardization can eliminate the influence of different dimensions and magnitudes, so that the conformity of different attributes can be fairly compared in the comprehensive evaluation. Combining the abnormal weight coefficient with the standardized conformity can comprehensively consider the abnormal degree and normal performance of the data, and improve the accuracy and reliability of the comprehensive evaluation. Based on the preset evaluation threshold, the comprehensive information is graded, and the data categories that need to be focused on can be quickly screened out. Using the historical judgment data stored in the database can provide a reference for the judgment of the current data and enhance the consistency and stability of the judgment results. By optimizing the judgment parameters to correct the preliminary classification results, the erroneous judgment caused by model deviation or data changes can be corrected. The generated judgment report contains an evaluation of the authenticity of the data and an analysis of the credibility, which provides a reliable basis for subsequent decision-making.

[0060] Reference Figure 2 As shown, the present invention also provides an intelligent quality inspection data authenticity judgment device, which is applied to any of the above-mentioned intelligent quality inspection data authenticity judgment methods, including: The acquisition module is used to collect original quality inspection data of the product through the detection equipment deployed on the production line, perform multi-dimensional processing on the original quality inspection data to obtain first dimensional data, and store the first dimensional data in the storage database; An analysis module, the analysis module is used to integrate the first dimensional data according to a preset matrix decomposition rule to obtain second dimensional data, perform feature extraction calculation on the second dimensional data, and obtain corresponding dimensional features; An association module is used to perform association analysis on the first dimension data and the dimension features according to a preset association rule to obtain an association relationship, and perform anomaly detection on the dimension features according to the association relationship to obtain anomaly information; A processing module, the processing module is used to obtain the first historical dimension data from the storage database, and compare the trend of the dimension feature according to the first historical dimension data based on the association relationship to obtain the conformity; The control module is used to conduct a comprehensive analysis on the abnormal information and the compliance, obtain the comprehensive information, and generate the judgment result of the first dimension data based on the comprehensive information.

[0061] The present invention provides an intelligent quality inspection data authenticity judgment device, which collects the original quality inspection data of the product in real time through the detection equipment deployed on the production line, and performs multi-dimensional processing on it, which can effectively utilize multi-dimensional information, and overcome the problem of information omission caused by focusing only on a single first-dimensional data in the traditional method. By integrating multiple first-dimensional data through a preset matrix decomposition rule, and combining the feature extraction technology to obtain dimensional features, and then using the preset association rules for abnormal detection, potential quality problems can be quickly and accurately identified, and the sensitivity and accuracy of abnormal detection are improved. By comparing the trend changes of the first historical dimension data and the current dimension features, the conformity of the data can be obtained, which is helpful to predict the quality problems that may occur in the future and provides a basis for taking preventive measures in advance. Comprehensive analysis of abnormal information and conformity is performed to generate the judgment result of the first-dimensional data, which can help quality inspectors make decisions faster and improve work efficiency. By performing multi-dimensional analysis and integration of data, not only the utilization rate of data is improved, but also more hidden valuable information can be excavated, which helps to optimize the production process and reduce resource waste.

[0062] Reference Figure 3 As shown, the present invention also provides an intelligent quality inspection data authenticity judgment system, comprising: Memory, used to store programs; The processor is used to execute the program to implement each step of any one of the above-mentioned intelligent methods for judging the authenticity of quality inspection data.

[0063] In this embodiment, the processor and the memory may be connected via a bus or other means. The memory may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a read-only memory, a flash memory, a hard disk, or a solid-state drive. The processor may be a general-purpose processor, such as a central processing unit, a digital signal processor, an application-specific integrated circuit, or one or more integrated circuits configured to implement the embodiments of the present invention.

[0064] The present invention also provides a storage medium storing computer instructions, wherein the computer instructions are used to enable a computer to execute any of the above methods.

[0065] It should be noted that technicians in the relevant technical field can clearly understand that for the convenience and conciseness of description, the specific working process of the system and each module described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0066] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. An intelligent method for judging the authenticity of quality inspection data, characterized in that: include: Collecting original quality inspection data of the product through the detection equipment deployed on the production line, performing multi-dimensional processing on the original quality inspection data to obtain first dimensional data, and storing the first dimensional data in a storage database; Integrate the first dimensional data according to a preset matrix decomposition rule to obtain second dimensional data, perform feature extraction calculation on the second dimensional data to obtain corresponding dimensional features; Performing association analysis on the first dimensional data and the dimensional features according to a preset association rule to obtain an association relationship, and performing anomaly detection on the dimensional features according to the association relationship to obtain anomaly information; Acquire first historical dimension data from the storage database, and compare the trend of the dimension feature according to the first historical dimension data based on the association relationship to obtain a degree of conformity; The abnormal information and the conformity are comprehensively analyzed to obtain comprehensive information, and a judgment result of the first dimensional data is generated based on the comprehensive information.

2. The intelligent method for judging the authenticity of quality inspection data according to claim 1 is characterized in that: The method of collecting original quality inspection data of the product through the detection equipment deployed on the production line, performing multi-dimensional processing on the original quality inspection data to obtain first dimensional data, and storing the first dimensional data in the storage database includes: Collect the original quality inspection data of the product through various inspection devices deployed on the production line, and classify and process the original quality inspection data to obtain original type data; Performing multidimensional processing on the original type data according to a preset multidimensional rule to obtain original multidimensional data, and performing standardization processing on the original multidimensional data to obtain the first dimensional data; The first dimensional data is stored in a storage database.

3. The intelligent method for judging the authenticity of quality inspection data according to claim 1 is characterized in that: The step of integrating the first dimensional data according to a preset matrix decomposition rule to obtain second dimensional data, and performing feature extraction calculation on the second dimensional data to obtain corresponding dimensional features includes: Performing matrix conversion on the first dimensional data to obtain a first dimensional data matrix, and performing non-negative matrix decomposition on the first dimensional data matrix to obtain a base matrix and a coefficient matrix; Reconstruct the first dimensional data according to the basis matrix and the coefficient matrix to obtain reconstructed data; The reconstructed data is subjected to difference calculation according to the preset standard first dimension data to obtain the reconstruction error; The first dimensional data is screened according to the reconstruction error to obtain screened first dimensional data; feature extraction is performed on the screened first dimensional data to obtain size dimensional features, material composition dimensional features, surface defect dimensional features, functional test dimensional features and time series dimensional features; Construct feature vectors of size dimension features, material composition dimension features, surface defect dimension features, functional test dimension features and time series dimension features to obtain feature vector matrix, calculate feature weights of the feature vector matrix to obtain weighted feature data; Based on the weighted feature data, comprehensive feature extraction is performed on size dimensional features, material composition dimensional features, surface defect dimensional features, functional test dimensional features and time series dimensional features to obtain the dimensional features.

4. The intelligent method for judging the authenticity of quality inspection data according to claim 1 is characterized in that: The associating analysis is performed on the first dimensional data and the dimensional features according to a preset association rule to obtain an association relationship, and anomaly detection is performed on the dimensional features according to the association relationship to obtain anomaly information, including: Calculating a correlation coefficient matrix between the first dimensional data and the dimensional features to obtain an initial correlation strength table, applying an adaptive threshold screening method to the initial correlation strength table to obtain core correlation pairs; Based on the core association pairs, a multi-level association network is constructed to obtain a hierarchical association structure; a topological analysis is performed on the hierarchical association structure to obtain target nodes and association paths; Evaluate the importance of the dimension features according to the target node and the associated path to obtain a feature weight vector; Directingly setting a preset anomaly detection rule set according to the feature weight vector to obtain a dynamic detection criterion, and performing anomaly detection on the dimensional feature according to the dynamic detection criterion to obtain an original anomaly indicator; The original abnormal indicators are grouped and constructed by using a density clustering method to obtain an abnormal category matrix, and the abnormal category matrix is ​​used to screen and decompose the original abnormal indicators to extract abnormal trend characteristics; Based on the abnormal trend characteristics and hierarchical association structure, anomaly propagation simulation is performed to form abnormal impact characteristics. The abnormal impact characteristics are evaluated according to the preset evaluation rules to obtain the final abnormal level. An exception description is generated according to the final exception level and the hierarchical association structure to obtain the exception information.

5. The intelligent method for judging the authenticity of quality inspection data according to claim 1 is characterized in that: The acquiring the first historical dimension data from the storage database, and performing trend comparison on the dimension feature according to the first historical dimension data based on the association relationship to obtain the degree of conformity includes: Extracting first historical dimension data matching the time span of the dimension feature from the storage database; Performing time series decomposition on the first historical dimension data to obtain three components of periodicity, seasonality and randomness, and performing feature mapping on the dimension features and the three components of periodicity, seasonality and randomness according to a preset feature mapping rule to obtain a multi-dimensional mapping feature matrix; Based on the association relationship, feature importance evaluation is performed on the multidimensional mapping feature matrix to obtain a feature weight vector; Using a predefined adaptive time window, dynamically adjust the feature weight vector to obtain a time-varying weight sequence, and perform a correlation analysis on the time-varying weight sequence and the dimensional feature according to a preset nonlinear trend comparison algorithm to obtain a correlation index; The correlation index is dynamically adjusted based on a preset adaptive threshold mechanism to obtain an adjustment index, and the adjustment index is hierarchically evaluated according to a predefined multi-level evaluation standard to obtain the degree of compliance.

6. The intelligent method for judging the authenticity of quality inspection data according to claim 1 is characterized in that: The comprehensively analyzing the abnormal information and the conformity to obtain comprehensive information, and generating a judgment result of the first dimensional data according to the comprehensive information, includes: Performing weight calculation on the abnormal information to obtain an abnormal weight coefficient, performing numerical standardization on the compliance to obtain a standardized compliance, and performing a comprehensive evaluation based on the abnormal weight coefficient and the standardized compliance to obtain comprehensive information; Based on the preset evaluation threshold, the comprehensive information is graded to obtain preliminary classification results, and the preliminary classification results are intelligently analyzed to obtain optimized judgment parameters; Extracting historical judgment data from the storage database, and correcting the preliminary classification result using optimized judgment parameters to obtain a correction result; A judgment report is generated according to the correction result, wherein the judgment report includes authenticity assessment and credibility analysis. Based on the judgment report, feature identification is performed on the first dimensional data to obtain the final judgment result.

7. An intelligent device for judging the authenticity of quality inspection data, characterized in that: The intelligent method for judging the authenticity of quality inspection data applied to any one of claims 1 to 6 above comprises: A collection module, the collection module is used to collect original quality inspection data of the product through the detection equipment deployed on the production line, perform multi-dimensional processing on the original quality inspection data to obtain first dimensional data, and store the first dimensional data in a storage database; An analysis module, the analysis module is used to integrate the first dimensional data according to a preset matrix decomposition rule to obtain second dimensional data, perform feature extraction calculation on the second dimensional data, and obtain corresponding dimensional features; An association module, the association module is used to perform association analysis on the first dimensional data and the dimensional features according to a preset association rule to obtain an association relationship, and perform anomaly detection on the dimensional features according to the association relationship to obtain anomaly information; A processing module, the processing module is used to obtain the first historical dimension data from the storage database, and perform trend comparison on the dimension feature according to the first historical dimension data based on the association relationship to obtain a degree of conformity; A control module is used to perform a comprehensive analysis on the abnormal information and the compliance to obtain comprehensive information, and generate a judgment result of the first dimensional data based on the comprehensive information.

8. An intelligent quality inspection data authenticity judgment system, characterized in that: include: Memory, used to store programs; The processor is used to execute the program to implement each step of an intelligent method for determining the authenticity of quality inspection data as described in any one of claims 1 to 6.

9. A storage medium, characterized in that: Computer instructions are stored, and the computer instructions are used to make a computer execute the method according to any one of claims 1 to 6.

Images