MES system quality management traceability and early warning method

Abstract

The present invention discloses a MES system quality management traceability and early warning method, which belongs to the field of manufacturing information technology, and specifically includes: collecting production process data in real time through an intelligent sensor array, and using edge computing for preprocessing and storage; the system has built-in error prevention and error prevention rules, and combines machine learning algorithms to conduct in-depth analysis of production process data, automatically detect and handle errors or abnormal situations, trigger an alarm system and start predictive maintenance; the MES system can automatically adjust production parameters, optimize the production process, and record complete data to form a product production process database; when product quality problems occur, the problem is quickly located through traceability and simulation functions, improvement measures are formulated and automatically updated to the system, so as to achieve continuous optimization of the production process.

Description

Technical Field

[0001] The invention belongs to the technical field of manufacturing informationization, and specifically is a MES system quality management tracing and early warning method. Background Art

[0002] The MES system, or manufacturing execution system, is a production information management system for the workshop execution layer of manufacturing enterprises. The system aims to provide production managers with visual monitoring and scheduling methods through real-time collection, analysis and processing of production site data, thereby ensuring the smooth execution of production plans and improving production efficiency and quality.

[0003] The production process based on the MES system generates a large amount of quality data. These quality data involve various aspects such as personnel, equipment, parts, inspection, quality, process, manufacturing and management. Due to the variety of data types, obvious interrelationships and numerous sources, traditional quality management relies on the experience of experts, and the quality inspection results are mainly based on data records and feedback on unqualified items. There is little systematic statistics and analysis of quality inspection data in each inspection link, which cannot provide preventive guidance for the production process. Therefore, how to effectively control quality data and improve management efficiency is a technical problem that needs to be solved.

[0004] For example, the patent with publication number CN111626548A discloses a quality management method and device based on an MES system, including: collecting original quality data based on a preset collection frequency and collection type; determining quality feature data based on the result of preprocessing the original quality data, the preprocessing including data extraction, data conversion and data loading; obtaining multiple quality pattern data sets based on the matching results of the quality feature data and preset quality factors, the preset quality factors are factors affecting product quality determined based on historical production data; determining effective quality pattern data based on preset quality pattern selection rules corresponding to each of the quality pattern data sets, and performing quality management through the effective quality pattern data, thereby realizing comprehensive management of quality data and improving management efficiency.

[0005] The above existing technologies all have the following problems: 1) Lack of real-time and dynamic performance; 2) Insufficient prediction and maintenance capabilities; 3) Limitations in adaptive and optimization capabilities. Summary of the invention

[0006] In view of the shortcomings of the existing technology, the present invention proposes a quality management traceability and early warning method for the MES system, which collects production process data in real time through an intelligent sensor array, and uses edge computing for preprocessing and storage; the system has built-in error prevention and error prevention rules, and combines machine learning algorithms to conduct in-depth analysis of production process data, automatically detect and handle errors or abnormal situations, trigger the alarm system and start predictive maintenance; the MES system can automatically adjust production parameters, optimize the production process, and record complete data to form a product production process database; when product quality problems occur, the traceability and simulation functions are used to quickly locate the problems, formulate improvement measures and automatically update them to the system, so as to achieve continuous optimization of the production process.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] MES system quality management traceability and early warning methods include:

[0009] Step S1: The MES system collects production process data in real time through an intelligent sensor array, and performs edge computing preprocessing and database storage;

[0010] Step S2: Process and analyze the pre-processed production process data using machine learning algorithms and data analysis methods, and perform predictive maintenance based on the data analysis results. At the same time, based on the predictive maintenance results, if any error or abnormality is detected, the alarm system is automatically triggered and the operation of the next process is automatically prohibited;

[0011] Step S3: Based on the detected error information, the intelligent optimization algorithm is used to automatically adjust the production parameters, correct the problems in the current production process and optimize the production process. After the processing is completed, the production is restarted or the next process is carried out. At the same time, the adjustment process and results are autonomously learned to optimize the adjustment strategy;

[0012] Step S4: Record the data and results of each production link in the MES system to form a complete product production process database. When a product has a quality problem, quickly locate the link and cause of the problem through the traceability function, and use the digital twin algorithm to simulate the scenario where the problem occurs. According to the traceability and simulation results, automatically generate improvement measures, and automatically update the improvement measures to the MES system;

[0013] The specific steps of step S2 include:

[0014] S2.1: According to the production plan and process requirements, identify the control points in the production process, set error-proofing and error-proofing rules for each control point, and convert the error-proofing and error-proofing rules into execution logic;

[0015] S2.2: Collect historical production process data and perform preprocessing. Use the PCA algorithm to extract features from the preprocessed historical production process data. At the same time, perform feature engineering and fuse the feature engineering data with the feature extraction data to obtain a feature data set.

[0016] S2.3: Use the improved isolation forest algorithm to train a machine learning-based anomaly prediction model based on the feature data set, deploy the trained anomaly prediction model to the production environment, and use the trained anomaly prediction model to perform predictive analysis on the production process to identify equipment failures or production problems. Based on the prediction results, develop a maintenance plan and update the MES system.

[0017] Specifically, the specific steps of the improved isolation forest algorithm in S2.3 include:

[0018] S2.31: Collect feature datasets , Represents the mth feature data, sets the subsample size, number of trees, and maximum height of the isolation forest, where y k , k ∈ [ 1 , m ] represents the kth feature data, and m represents the number of feature data;

[0019] S2.32: For each tree, randomly select a specified number of sample points from the feature data set as subsamples, put them into the root node of the tree, and randomly generate a cutting point e in the current root node data;

[0020] S2.33: Generate a hyperplane at the cutting point e to divide the current root node data space into two subspaces: , then put the data in the left root node, if , then put the data in the right root node, y e is the cutting point position data;

[0021] S2.34: Repeat S2.32-S2.33 to construct a preset number of isolated trees and form an isolation forest;

[0022] The specific steps of the improved isolation forest algorithm in S2.3 also include:

[0023] S2.35: For each feature data, calculate the number of edges contained in the path from the root node to the leaf node of the feature data, and obtain the path length of the feature data in each tree ;

[0024] S2.36: Based on path length , the weighted mean method is used to calculate the average path length of feature data in all trees ,in, express The feature weight vector of express The vector of path lengths in each tree, represents the mean function;

[0025] The specific steps of the improved isolation forest algorithm in S2.3 also include:

[0026] S2.37: Use the anomaly score formula to calculate the anomaly score of the feature data. The formula is:

[0027] ;

[0028] in, Representing feature data The anomaly score, represents the normalization factor, g represents the maximum height of the tree;

[0029] S2.38: Set the threshold h, if , then the feature data point is identified as an abnormality or equipment failure.

[0030] Specifically, the specific steps of step S2 also include:

[0031] S2.4: Based on the updated MES system, real-time monitoring of production process data is performed to detect whether the set error prevention and error prevention rules are met. If not, the unsatisfactory production process data is collected and sorted in ascending order to obtain the data set to be tested. , Represents the nth data to be detected, and uses the improved quartile algorithm to detect anomalies in the data set to be detected. The formula is:

[0032] ;

[0033] Among them, IQR represents the interquartile range, n represents the number of data in the data set to be tested, x j , j ∈ [ 1 , n ] represents the jth data to be detected, represents the skew factor of the data set to be tested, Represents the discrete factor of the data set to be tested, represents the scaling factor of the dataset to be tested, represents the first quartile, represents the third quartile;

[0034] S2.5: If , it means there is no abnormal situation;

[0035] like ,or , then the data to be detected is abnormal data, which automatically triggers the alarm system and prohibits the operation of the next process.

[0036] Specifically, the formulas for the error prevention and error protection rules in S2.1 are:

[0037] ;

[0038] in, Indicates the current temperature. , express The minimum and maximum values ​​of represents the i-th operator, Indicates the device permissions corresponding to the i-th operator, Indicates the corresponding relationship, Indicates the operation priority from large to small, and l indicates the type of priority level to be set.

[0039] Specifically, the specific steps of converting the error prevention and error prevention rules into execution logic in S2.1 include:

[0040] S2.11: Match the error prevention and error proofing rules with each control point one by one, and fine-tune the error prevention and error proofing rules during the matching process. For each control point, formulate the execution logic of trigger condition-operation sequence-operation standard-error handling mechanism;

[0041] S2.12: Write operational instructions based on execution logic, and implement and monitor execution logic during actual production.

[0042] Compared with the prior art, the present invention has the following beneficial effects:

[0043] 1. The present invention proposes a MES system quality management traceability and early warning method. By collecting production process data in real time and processing it quickly, the MES system can quickly respond to changes in production, reduce production interruption time, and thus improve production efficiency; use machine learning algorithms and data analysis methods to perform predictive maintenance, timely discover and resolve potential equipment failures, reduce unplanned downtime, and reduce maintenance costs; automatically adjust production parameters based on intelligent optimization algorithms, correct problems in the production process, optimize the production process, and improve product quality and consistency.

[0044] 2. The present invention proposes a MES system quality management traceability and early warning method. When an error or abnormal situation is detected, the alarm system is automatically triggered and the operation of the next process is prohibited to prevent the accident from expanding and ensure production safety. The MES system records the data and results of each production link to form a complete product production process database, which is convenient for management personnel to trace and analyze and improve management efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a schematic diagram of the MES system quality management traceability and early warning method of the present invention;

[0046] Figure 2 This is a work flow diagram of the MES system quality management traceability and early warning method of the present invention;

[0047] Figure 3 The flowchart of the improved isolation forest algorithm for the MES system quality management traceability and early warning method of the present invention;

[0048] Figure 4 The present invention provides a flowchart for realizing the abnormal prediction model of the MES system quality management traceability and early warning method. DETAILED DESCRIPTION

[0049] Example 1

[0050] See also Figure 1-Figure 2 , an embodiment provided by the present invention: a MES system quality management traceability and early warning method, comprising the following steps:

[0051] Step S1: The MES system collects production process data in real time through an intelligent sensor array, and performs edge computing preprocessing and database storage;

[0052] Furthermore, the specific steps of step S1 include:

[0053] S1.1: Connect the equipment in the production process to the MES system through wired or wireless means to ensure smooth communication between the equipment and the MES system. The smart sensor array is installed on the equipment to detect the operating status and production data of the equipment in real time;

[0054] S1.2: The intelligent sensor array is used to collect the production process data of the equipment in real time, including the physical quantities of temperature, pressure, speed, and position, as well as the logical quantities of the equipment's on / off status and production count;

[0055] S1.3: Use wireless communication to transmit the collected data to the server of the MES system through the network to ensure the timeliness and accuracy of the data;

[0056] S1.4: Before the data reaches the MES system server, use edge computing devices to perform preliminary data processing and analysis. The edge computing devices perform data cleaning, data compression, and simple data analysis tasks to reduce the processing burden on the server.

[0057] S1.5: Store the pre-processed data into the database of the MES system to form a production database.

[0058] Step S2: Process and analyze the pre-processed production process data using machine learning algorithms and data analysis methods, and perform predictive maintenance based on the data analysis results. At the same time, based on the predictive maintenance results, if any error or abnormality is detected, the alarm system is automatically triggered and the operation of the next process is automatically prohibited;

[0059] Step S3: Based on the detected error information, the intelligent optimization algorithm is used to automatically adjust the production parameters, correct the problems in the current production process and optimize the production process. After the processing is completed, the production is restarted or the next process is carried out. At the same time, the adjustment process and results are autonomously learned to optimize the adjustment strategy;

[0060] In the present invention, the intelligent optimization algorithm adopts a genetic algorithm, which can automatically adjust the parameters in the production process according to the detected error information, find the optimal parameter configuration, thereby correcting the problems in the current production process and improving production efficiency and quality. Among them, the genetic algorithm is the prior art content in this field and is not the creative solution of this application, so it will not be elaborated here.

[0061] Step S4: Record the data and results of each production link in the MES system to form a complete product production process database. When a product has a quality problem, the traceability function can be used to quickly locate the link and cause of the problem, and the digital twin algorithm can be used to simulate the scenario where the problem occurs. According to the traceability and simulation results, improvement measures are automatically generated and automatically updated to the MES system.

[0062] Example 2

[0063] See also Figure 3-Figure 4 In this embodiment, the specific steps of step S2 include:

[0064] S2.1: According to the production plan and process requirements, identify the control points in the production process, set error-proofing and error-proofing rules for each control point, and convert the error-proofing and error-proofing rules into execution logic;

[0065] S2.2: Collect historical production process data and perform preprocessing, use the PCA algorithm to extract features from the preprocessed historical production process data, and at the same time, perform feature engineering, merge the feature engineering data with the feature extraction data, and obtain a feature data set, wherein the PCA algorithm feature extraction is the prior art content in this field, is not the inventive solution of the present application, and is not elaborated here;

[0066] S2.3: Use the improved isolation forest algorithm to train a machine learning-based anomaly prediction model based on the feature data set, deploy the trained anomaly prediction model to the production environment, and use the trained anomaly prediction model to perform predictive analysis on the production process to identify equipment failures or production problems. Based on the prediction results, formulate maintenance plans and update the MES system.

[0067] The maintenance plan includes: 1) inspection and diagnosis; 2) repair and replacement, such as replacing damaged parts, adjusting parameter settings or fixing system vulnerabilities; 3) spare parts management, which is responsible for preparing spare parts in advance to reduce downtime caused by waiting for spare parts.

[0068] The specific steps of step S2 also include:

[0069] S2.4: Based on the updated MES system, real-time monitoring of production process data is performed to detect whether the set error prevention and error prevention rules are met. If not, the unsatisfactory production process data is collected and sorted in ascending order to obtain the data set to be tested. , Represents the nth data to be detected, and uses the improved quartile algorithm to detect anomalies in the data set to be detected. The formula is:

[0070] ;

[0071] Among them, IQR represents the interquartile range, n represents the number of data in the data set to be tested, x j , j ∈ [ 1 , n ] represents the jth data to be detected, Represents the skew factor of the data set to be tested, which is used to adjust the IQR according to the skewness of the data. If the data is skewed to the right, that is, positively skewed, then , thereby increasing the IQR. If it is skewed to the left, that is, negatively skewed, then , thereby reducing the IQR, Represents the discrete factor of the data set to be tested, which is used to adjust the IQR according to the degree of dispersion of the data. If the degree of dispersion of the data is high, then , thereby increasing the IQR; if the dispersion is lower, then , Represents the scaling factor of the data set to be tested, which is used to adjust the scale of the IQR as a whole. represents the first quartile, represents the third quartile;

[0072] Furthermore, if is an integer, then is the value of the position; if is a decimal, then ,in, represents the first quartile of the jth data to be tested, Indicates The first quartile of the data to be tested.

[0073] S2.5: If , it means there is no abnormal situation;

[0074] like ,or , then the data to be detected is abnormal data, which automatically triggers the alarm system and prohibits the operation of the next process.

[0075] The formula for error proofing and error proofing rules in S2.1 is:

[0076] ;

[0077] in, Indicates the current temperature. , express The minimum and maximum values ​​of represents the i-th operator, Indicates the device permissions corresponding to the i-th operator, Indicates the corresponding relationship, Indicates the operation priority from large to small, and l indicates the type of priority level to be set.

[0078] The specific steps in S2.1 to convert error-proofing and error-proofing rules into execution logic include:

[0079] S2.11: Match the error prevention and error proofing rules with each control point one by one, and fine-tune the error prevention and error proofing rules during the matching process. For each control point, formulate the execution logic of trigger condition-operation sequence-operation standard-error handling mechanism;

[0080] Furthermore, 1) trigger conditions: clarify under what circumstances error prevention or error prevention measures need to be implemented; 2) operation sequence: specify which operations need to be performed in what order under the trigger conditions; 3) operation standards: specify the specific requirements for each operation, including operation steps and control parameters; 4) set up error handling mechanism: if an error occurs, how to deal with it, including stopping production, alarm notification, and corrective measures.

[0081] S2.12: Write operational instructions based on execution logic, and implement and monitor execution logic during actual production.

[0082] The specific steps of the improved isolation forest algorithm in S2.3 include:

[0083] S2.31: Collect feature datasets , Represents the mth feature data, sets the subsample size, number of trees, and maximum height of the isolation forest, where y k , k ∈ [ 1 , m ] represents the kth feature data, m represents the number of feature data, and in the present invention, the isolation forest parameters are set, including the subsample size is 256, the number of trees is 100, and the maximum height is 8;

[0084] S2.32: For each tree, randomly select a specified number of sample points from the feature data set as subsamples, put them into the root node of the tree, and randomly generate a cut point e in the current root node data, where the cut point is generated between the maximum and minimum values ​​of the specified dimension in the current node data;

[0085] S2.33: Generate a hyperplane at the cutting point e to divide the current root node data space into two subspaces: , then put the data in the left root node, if , then put the data in the right root node, y e is the cutting point position data;

[0086] S2.34: Repeat S2.32-S2.33 to construct a preset number of isolated trees and form an isolation forest.

[0087] S2.35: For each feature data, calculate the number of edges contained in the path from the root node to the leaf node of the feature data, and obtain the path length of the feature data in each tree ;

[0088] Furthermore, the specific steps of path length include:

[0089] Starting from the root node, the count is 0;

[0090] Traverse the tree and move down along the decision path of the feature data points;

[0091] For each node in the tree, if the node is a split point based on a certain feature and the feature data point meets the split condition of the node, move to a child node of the node and add 1 to the path length count;

[0092] The above process is repeated until the leaf node containing the feature data point is reached, that is, the count of the path length is the path length of the feature data point from the root node to the leaf node.

[0093] S2.36: Based on path length , the weighted mean method is used to calculate the average path length of feature data in all trees ,in, express The feature weight vector of express The vector of path lengths in each tree, represents the mean function.

[0094] S2.37: Use the anomaly score formula to calculate the anomaly score of the feature data. The formula is:

[0095] ;

[0096] in, Representing feature data The anomaly score, represents the normalization factor, g represents the maximum height of the tree;

[0097] S2.38: Set the threshold h, if , then the feature data point is identified as an abnormality or equipment failure.

[0098] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation modes, which are merely illustrative rather than restrictive. Under the guidance of the present invention, ordinary technicians in the field may also change, modify, replace and modify the above-mentioned embodiments without departing from the purpose and scope of protection of the present invention, and all of these are within the protection of the present invention.

Claims

1. MES system quality management traceability and early warning method, characterized by: include: Step S1: The MES system collects production process data in real time through an intelligent sensor array, and performs edge computing preprocessing and database storage; Step S2: Process and analyze the pre-processed production process data using machine learning algorithms and data analysis methods, and perform predictive maintenance based on the data analysis results. At the same time, based on the predictive maintenance results, if any error or abnormality is detected, the alarm system is automatically triggered and the operation of the next process is automatically prohibited; Step S3: Based on the detected error information, the intelligent optimization algorithm is used to automatically adjust the production parameters, correct the problems in the current production process and optimize the production process. After the processing is completed, the production is restarted or the next process is carried out. At the same time, the adjustment process and results are autonomously learned to optimize the adjustment strategy; Step S4: Record the data and results of each production link in the MES system to form a complete product production process database. When a product has a quality problem, quickly locate the link and cause of the problem through the traceability function, and use the digital twin algorithm to simulate the scenario where the problem occurs. According to the traceability and simulation results, automatically generate improvement measures, and automatically update the improvement measures to the MES system; The specific steps of step S2 include: S2.1: According to the production plan and process requirements, identify the control points in the production process, set error-proofing and error-proofing rules for each control point, and convert the error-proofing and error-proofing rules into execution logic; S2.2: Collect historical production process data and perform preprocessing. Use the PCA algorithm to extract features from the preprocessed historical production process data. At the same time, perform feature engineering and fuse the feature engineering data with the feature extraction data to obtain a feature data set. S2.3: Use the improved isolation forest algorithm to train a machine learning-based anomaly prediction model based on the feature data set, deploy the trained anomaly prediction model to the production environment, and use the trained anomaly prediction model to perform predictive analysis on the production process to identify equipment failures or production problems. Based on the prediction results, develop a maintenance plan and update the MES system. The specific steps of the improved isolation forest algorithm in S2.3 include: S2.31: Collect feature datasets , Represents the mth feature data, sets the subsample size, number of trees, and maximum height of the isolation forest, where represents the kth feature data, and m represents the number of feature data; S2.32: For each tree, randomly select a specified number of sample points from the feature data set as subsamples, put them into the root node of the tree, and randomly generate a cutting point e in the current root node data; S2.33: Generate a hyperplane at the cutting point e to divide the current root node data space into two subspaces: , then put the data in the left root node, if , then put the data in the right root node, y e is the cutting point position data; S2.34: Repeat S2.32-S2.33 to construct a preset number of isolated trees and form an isolation forest; The specific steps of the improved isolation forest algorithm in S2.3 also include: S2.35: For each feature data, calculate the number of edges contained in the path from the root node to the leaf node of the feature data, and obtain the path length of the feature data in each tree ; S2.36: Based on path length , the weighted mean method is used to calculate the average path length of feature data in all trees ,in, express The feature weight vector of express The vector of path lengths in each tree, represents the mean function; The specific steps of the improved isolation forest algorithm in S2.3 also include: S2.37: Use the anomaly score formula to calculate the anomaly score of the feature data. The formula is: ； in, Representing feature data The anomaly score, represents the normalization factor, g represents the maximum height of the tree; S2.38: Set the threshold h, if , the characteristic data is identified as abnormal or equipment failure.

2. The MES system quality management traceability and early warning method according to claim 1, characterized in that: The specific steps of step S2 also include: S2.4: Based on the updated MES system, real-time monitoring of production process data is performed to detect whether the set error prevention and error prevention rules are met. If not, the unsatisfactory production process data is collected and sorted in ascending order to obtain the data set to be tested. , Represents the nth data to be detected, and uses the improved quartile algorithm to detect anomalies in the data set to be detected. The formula is: ； Among them, IQR represents the interquartile range, n represents the number of data in the data set to be tested, represents the jth data to be detected, represents the skew factor of the data set to be tested, Represents the discrete factor of the data set to be tested, represents the scaling factor of the dataset to be tested, represents the first quartile, represents the third quartile; S2.5: If , it means there is no abnormal situation; like ,or , then the data to be detected is abnormal data, which automatically triggers the alarm system and prohibits the operation of the next process.

3. The MES system quality management traceability and early warning method according to claim 2, characterized in that: The formula for error prevention and error proofing rules in S2.1 is: ； in, Indicates the current temperature. , express The minimum and maximum values ​​of represents the i-th operator, Indicates the device permissions corresponding to the i-th operator, Indicates the corresponding relationship, Indicates the operation priority from large to small, and l indicates the type of priority level to be set.

4. The MES system quality management traceability and early warning method according to claim 3, characterized in that: The specific steps of converting error prevention and error prevention rules into execution logic in S2.1 include: S2.11: Match the error prevention and error proofing rules with each control point one by one, and fine-tune the error prevention and error proofing rules during the matching process. For each control point, formulate the execution logic of trigger condition-operation sequence-operation standard-error handling mechanism; S2.12: Write operational instructions based on execution logic, and implement and monitor execution logic during actual production.

Images