A method and system for monitoring quality of numerical control machining based on statistical process control
By adopting CNC machining quality monitoring method based on statistical process control during CNC machining, using CNN-SVM and PSO-SVR models for intelligent pattern recognition and abnormal traceability, the problem that traditional SPC technology cannot achieve real-time quality control is solved, and the realization of the realization of the Real-time monitoring and automated adjustment of CNC processing process improve production stability and product quality.
Patent Information
- Application Number
- CN202411230778.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-04
- Publication Date
- 2024-11-29
- Estimated Expiration
- 2044-09-04
AI Technical Summary
Traditional statistical process control (SPC) technology has problems such as offline detection, inability to conduct online detection, timely diagnosis of abnormal causes and quality adjustments during CNC processing, resulting in the inability to realize real-time quality control and information sharing, and the inability to meet the modern quality engineering Product quality control requirements.
A CNC machining quality monitoring method based on statistical process control is proposed, including online data acquisition, tool repair and processing quality correlation analysis, quality control analysis and abnormal traceability adjustment, and intelligent pattern recognition and PSO-SVR model are used for intelligent pattern recognition and PSO-SVR model. Exception traceability, realize automated tool-compensation adaptive compensation and closed-loop dynamic control of quality.
Real-time quality monitoring and automated adjustment of the CNC machining process are realized, the stability of the production process and product quality are improved, and the processing quality is improved through secondary quality warning and optimal tool compensation feedback.
Smart Images

Figure CN118735365B_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the technical field of quality monitoring, and in particular to a numerical control machining quality monitoring method based on statistical process control, a numerical control machining quality monitoring system based on statistical process control, and electronic equipment. Background Art
[0002] The monitoring and control of machining processes is becoming increasingly important. In current industrial practice, quality is ensured at two different stages of the product engineering cycle.
[0003] First, the Taguchi method is used in the product design stage, aiming to improve the quality and stability of the product through design optimization, which is done before the product is manufactured.
[0004] Secondly, statistical process control methods are used to check product quality during the inspection phase after product manufacturing. Traditional SPC technology uses standard Shewhart control charts to monitor whether the process is in a stable state and to warn of process abnormalities.
[0005] The above two-stage quality control method has been widely used and achieved good economic benefits.
[0006] However, although SPC technology is quite mature, there are still some problems in the practical application of traditional SPC technology: First, offline detection is usually used, which has poor real-time performance and cannot be used for online detection; second, when the control chart is abnormal, the cause of the abnormality cannot be diagnosed. Third, quality problems cannot be adjusted and improved in a timely manner, and the purpose of timely feedback and adjustment cannot be achieved.
[0007] In addition, CNC machining has been widely used in the manufacturing industry due to its automation and high efficiency. As the complexity of machined parts increases, the control of the machining process becomes more and more difficult. The integration of machining and testing is an effective way to control the machining process and improve the machining quality.
[0008] Traditional SPC technology is a post-process detection, which cannot achieve information sharing well, nor can it automatically analyze and process according to the expected plan, and can no longer meet the requirements of modern quality engineering for product quality control. In the CNC machining process, the workpiece quality is closely related to the tool compensation value, but the relationship is not clear, and it is not possible to form a good guiding knowledge in the machining process. Summary of the invention
[0009] In order to solve the above problems, the present application proposes a CNC machining quality monitoring method based on statistical process control, a CNC machining quality monitoring system based on statistical process control, and an electronic device.
[0010] On the one hand, the present application proposes a CNC machining quality monitoring method based on statistical process control, comprising the following steps:
[0011] S1. Online data collection: obtain the quality data in the current CNC machining process, collect the quality data, and select the control chart corresponding to the CNC machining control according to the characteristics of the quality data;
[0012] S2. Analysis of the correlation between tool compensation and machining quality: Apply the stability criterion to make a stability judgment on the selected control chart to determine whether the current CNC machining process is stable:
[0013] If yes, go to the next step;
[0014] Otherwise, determine whether it is out of control again:
[0015] If so, it is not the first out-of-control, and a strong out-of-control alarm prompt is issued, and the abnormal diagnosis knowledge base is used to perform abnormal diagnosis, and the CNC machining process is re-entered after the abnormal factors of the process are eliminated;
[0016] On the contrary, it is the first time that the machine loses control, and a weak out-of-control alarm is given. The quality data of the current time series is recorded in the tool compensation and machining quality association database, and the tool compensation adaptive compensation is performed on the current CNC machining process.
[0017] S3. Quality control analysis: Conduct process capability analysis on the CNC machining process that is currently in a stable state to determine whether the process capability is sufficient:
[0018] If yes, the control chart is used to continue monitoring, and intelligent pattern recognition is performed on the control chart based on the CNN-SVM model to determine whether there is an abnormal pattern in the control chart pattern: if yes, proceed to the next step; otherwise, the control chart is used to continue monitoring;
[0019] On the contrary, the abnormal factors in the process are eliminated, and the CNC machining process is re-entered after the abnormal factors in the process are eliminated;
[0020] S4. Abnormal tracing and adjustment: Based on the abnormal mode parameter estimation model: PSO-SVR, the abnormal information under the abnormal mode is intelligently evaluated and the adjustment measures for the abnormal information are output; the system eliminates the process abnormal factors and re-enters the CNC machining process based on the adjustment measures of the abnormal information.
[0021] As an optional implementation scheme of the present application, optionally, in step S1, the obtaining of quality data in the current CNC machining process and the collection of quality data include:
[0022] Collect quality data of the current CNC machining process in real time and import it into the preset data classification model;
[0023] The quality data is identified by the data classification model, and the data type of the current quality data is determined according to the identified feature data:
[0024] If it is new data, configure the timed delegation parameters and the first cache ID corresponding to it in the cache database;
[0025] If the data is of a fault / failure event, an irregular delegation parameter and a corresponding second cache ID in the cache database are configured for it.
[0026] As an optional implementation scheme of the present application, optionally, in step S1, the control diagram corresponding to the numerical control machining control is selected according to the characteristics of the quality data, including:
[0027] Identify the data characteristics of the current quality data:
[0028] If the current quality data is metrological data, the first control chart corresponding to the numerical control machining control is selected according to the sample size;
[0029] If the current quality data is count data, the current quality data is judged according to the number of defects and the number of unqualified objects of the standard object, and the second control chart corresponding to the CNC machining control is selected.
[0030] As an optional implementation scheme of the present application, optionally, in step S2, recording the quality data of the current time series to a tool compensation and processing quality association database, and performing tool compensation adaptive compensation on the current CNC processing process, includes:
[0031] Obtain the processing quality data and tool compensation data of the current time series;
[0032] Preprocess the processing quality data and tool compensation data;
[0033] Based on the Mean-shift model, cluster analysis is performed on the processing quality data and tool compensation data to obtain cluster categories;
[0034] Divide the quality intervals according to cluster categories;
[0035] Establish an Apriori model to explore the relationship between tool compensation value and processing size;
[0036] Tool compensation feedback is performed based on the correlation between the tool compensation value and the processing size.
[0037] As an optional implementation scheme of the present application, optionally, in step S3, intelligent pattern recognition is performed on the control chart based on the CNN-SVM model, including:
[0038] Importing the current control graph into a preset CNN-SVM model, and using the CNN-SVM model to perform feature extraction on the control graph;
[0039] Classify according to the extracted features, identify and output the current control chart mode;
[0040] The CNN-SVM model is generated by performing CNN-SVM classification training on historical samples under several different control chart modes, and the historical sample data includes historical sample feature images of several control charts and the historical sample feature data therein.
[0041] As an optional implementation scheme of the present application, optionally, in step S3, determining whether the control diagram mode has an abnormal mode includes:
[0042] Based on the CNN-SVM model, intelligent pattern recognition is performed on the real-time sample feature data in the control chart;
[0043] Determine whether the real-time sample feature data matches the historical sample feature data learned by the model:
[0044] If they match, it is determined that there is no abnormal pattern in the control chart pattern;
[0045] Otherwise, it is determined that there is an abnormal mode in the control chart mode, and the abnormal mode parameters are calculated to trace the abnormality.
[0046] As an optional implementation scheme of the present application, optionally, in step S3, determining whether the control diagram mode has an abnormal mode further includes:
[0047] If it is determined that there is no abnormal pattern in the control chart pattern, then processing will continue to obtain additional samples;
[0048] Calculate the feature data of the newly added samples and determine whether the relative change value of the feature data of the newly added samples exceeds the preset threshold:
[0049] If exceeded, retrain the CNN-SVM model;
[0050] Otherwise, the CNN-SVM model is maintained to continue to perform intelligent pattern recognition on the control chart.
[0051] As an optional implementation scheme of the present application, optionally, S4, abnormal source tracing adjustment: based on the abnormal mode parameter estimation model: PSO-SVR, intelligently evaluate the abnormal information under the abnormal mode, and output adjustment measures for the abnormal information, including:
[0052] Analyze the factors affecting CNC machining quality, determine the main factors affecting CNC machining quality, and use them as training sample data;
[0053] Read the training sample data, import it into the preset SVR model, and initialize the SVR model;
[0054] The optimal parameters of the SVR model are iteratively calculated through the PSO algorithm, and the final abnormal pattern parameter estimation model is constructed based on the optimal parameters: the PSO-SVR model;
[0055] Construct confidence intervals for each anomaly cause;
[0056] The control chart pattern abnormal data is imported into the abnormal pattern parameter estimation model: PSO-SVR model, and based on the confidence interval, various abnormal causes in the abnormal information are ranked by abnormality, and the best adjustment measures for the abnormal information are evaluated and output.
[0057] On the other hand, the present application proposes a CNC machining quality monitoring system based on statistical process control, which is used to implement a CNC machining quality monitoring method based on statistical process control, including:
[0058] General module, used to: provide general services for CNC machining;
[0059] The online data acquisition module is used to: obtain the quality data in the current CNC machining process, collect the quality data, and select the control chart corresponding to the CNC machining control according to the characteristics of the quality data;
[0060] Tool compensation and processing quality correlation analysis module is used to: apply the stability judgment criteria to make stability judgment on the selected control chart and judge whether the current CNC machining process is stable:
[0061] If yes, go to the next step;
[0062] Otherwise, determine whether it is out of control again:
[0063] If so, it is not the first out-of-control, and a strong out-of-control alarm prompt is issued, and the abnormal diagnosis knowledge base is used to perform abnormal diagnosis, and the CNC machining process is re-entered after the abnormal factors of the process are eliminated;
[0064] On the contrary, it is the first time that the machine loses control, and a weak out-of-control alarm is given. The quality data of the current time series is recorded in the tool compensation and machining quality association database, and the tool compensation adaptive compensation is performed on the current CNC machining process.
[0065] The quality control analysis module is used to: perform process capability analysis on the CNC machining process that is currently in a stable state to determine whether the process capability is sufficient:
[0066] If yes, the control chart is used to continue monitoring, and intelligent pattern recognition is performed on the control chart based on the CNN-SVM model to determine whether there is an abnormal pattern in the control chart pattern: if yes, proceed to the next step; otherwise, the control chart is used to continue monitoring;
[0067] On the contrary, the abnormal factors in the process are eliminated, and the CNC machining process is re-entered after the abnormal factors in the process are eliminated;
[0068] The abnormality tracing and adjustment module is used to: based on the abnormal mode parameter estimation model: PSO-SVR, intelligently evaluate the abnormal information under the abnormal mode and output adjustment measures for the abnormal information; based on the adjustment measures of the abnormal information, the system eliminates the process abnormal factors and re-enters the CNC machining process.
[0069] In another aspect, the present application further provides an electronic device, comprising:
[0070] processor;
[0071] a memory for storing processor-executable instructions;
[0072] Wherein, the processor is configured to implement the CNC machining quality monitoring method based on statistical process control when executing the executable instructions.
[0073] Technical effects of the present invention:
[0074] This application uses a secondary quality warning and establishes a tool compensation and processing quality association model, which can automatically call the optimal tool compensation value adjustment in the CNC machining program. The abnormal mode of the control chart of the machining process is intelligently identified through the CNN-SVM online update recognition method. When an abnormality occurs, the control chart abnormal mode information and process status information are used as input, and the confidence of the abnormal cause is used as output. The PSO-SVR model of each abnormal cause is trained to achieve abnormal traceability of the machining process. It is used to improve the stability of the production process and the quality of the product. The present invention establishes a complete set of quality data collection, analysis, monitoring, diagnosis, and tool compensation-related processing quality online detection system to achieve closed-loop dynamic quality control and achieve optimal tool compensation value feedback to improve machining quality.
[0075] Further features and aspects of the present disclosure will become apparent from the following detailed description of exemplary embodiments with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS
[0076] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate exemplary embodiments, features, and aspects of the disclosure and, together with the description, serve to explain the principles of the disclosure.
[0077] Figure 1 It is a flow chart of the CNC machining quality monitoring method based on statistical process control;
[0078] Figure 2 It is a selection flow chart of control chart;
[0079] Figure 3 It is a flow chart of the correlation analysis between tool compensation and processing quality;
[0080] Figure 4 There are eight control chart modes;
[0081] Figure 5 It is the CNN-SVM model structure;
[0082] Figure 6 This is the schematic diagram of SVM online update;
[0083] Figure 7 It is based on the PSO-SVR anomaly tracing reasoning flow chart;
[0084] Figure 8 It is a factor affecting the quality of CNC machining process;
[0085] Fig. 9 CNC machining quality monitoring system module based on statistical process control. DETAILED DESCRIPTION
[0086] Various exemplary embodiments, features and aspects of the present disclosure will be described in detail below with reference to the accompanying drawings. The same reference numerals in the accompanying drawings represent elements with the same or similar functions. Although various aspects of the embodiments are shown in the accompanying drawings, the drawings are not necessarily drawn to scale unless otherwise specified.
[0087] The word “exemplary” is used exclusively herein to mean “serving as an example, example, or illustration.” Any embodiment described herein as “exemplary” is not necessarily to be construed as preferred or advantageous over other embodiments.
[0088] In addition, in order to better illustrate the present disclosure, numerous specific details are given in the following specific embodiments. It should be understood by those skilled in the art that the present disclosure can also be implemented without certain specific details. In some examples, means, components and circuits well known to those skilled in the art are not described in detail in order to highlight the main purpose of the present disclosure.
[0089] Example 1
[0090] This application mainly integrates the CNC machining quality optimization method and system of statistical process control to improve the stability of the production process and product quality. It realizes dynamic closed-loop quality control, realizes optimal tool compensation value feedback to improve machining quality, and establishes a complete set of quality data collection, analysis, monitoring, diagnosis, and tool compensation-related machining quality online detection system.
[0091] like Figure 1As shown, on the one hand, the present application proposes a CNC machining quality monitoring method based on statistical process control, comprising the following steps:
[0092] S1. Online data collection: obtain the quality data in the current CNC machining process, collect the quality data, and select the control chart corresponding to the CNC machining control according to the characteristics of the quality data;
[0093] S2. Analysis of the correlation between tool compensation and machining quality: Apply the stability criterion to make a stability judgment on the selected control chart to determine whether the current CNC machining process is stable:
[0094] If yes, go to the next step;
[0095] Otherwise, determine whether it is out of control again:
[0096] If so, it is not the first out-of-control, and a strong out-of-control alarm prompt is issued, and the abnormal diagnosis knowledge base is used to perform abnormal diagnosis, and the CNC machining process is re-entered after the abnormal factors of the process are eliminated;
[0097] On the contrary, it is the first time that the machine loses control, and a weak out-of-control alarm is given. The quality data of the current time series is recorded in the tool compensation and machining quality association database, and the tool compensation adaptive compensation is performed on the current CNC machining process.
[0098] S3. Quality control analysis: Conduct process capability analysis on the CNC machining process that is currently in a stable state to determine whether the process capability is sufficient:
[0099] If yes, the control chart is used to continue monitoring, and intelligent pattern recognition is performed on the control chart based on the CNN-SVM model to determine whether there is an abnormal pattern in the control chart pattern: if yes, proceed to the next step; otherwise, the control chart is used to continue monitoring;
[0100] On the contrary, the abnormal factors in the process are eliminated, and the CNC machining process is re-entered after the abnormal factors in the process are eliminated;
[0101] S4. Abnormal tracing and adjustment: Based on the abnormal mode parameter estimation model: PSO-SVR, the abnormal information under the abnormal mode is intelligently evaluated and the adjustment measures for the abnormal information are output; the system eliminates the process abnormal factors and re-enters the CNC machining process based on the adjustment measures of the abnormal information.
[0102] The main application descriptions of the above steps are as follows:
[0103] S1. Online data acquisition: The quality control data of the current manufacturing process is acquired in real time through the online data acquisition module. The most appropriate control chart is selected according to the characteristics of the quality data. The measurement value uses the measurement control chart, and the count value uses the count control chart. Secondly, select the appropriate control chart according to the size of the subgroup capacity.
[0104] The range, mean, standard deviation, statistical information and parameters required by the statistical process control algorithm of the target quality data are calculated.
[0105] S2. Analysis of the correlation between tool compensation and processing quality: Draw a control chart for analysis, and apply the stability criterion to determine whether the process is in a state of statistical control. If it is in an unstable state (i.e., out-of-control state), a weak out-of-control alarm prompt will be issued when it loses control for the first time, and the data of the current time series will be recorded in the tool compensation and workpiece quality correlation database. According to the association rules between workpiece quality and tool compensation, a correlation analysis model between the tool compensation value and the corresponding workpiece quality data is constructed. In the CNC machining stage, the register tool compensation value adjustment is automatically called through the machining program to achieve adaptive compensation machining. Then, the data at the initial time point is eliminated one by one and the data at the new time point is obtained to determine whether the process is in a state of statistical control again. If it is still in an unstable state (i.e., out-of-control state), a strong out-of-control alarm prompt will be issued, and other causes of the abnormality will be found and the process will be investigated.
[0106] S3. Quality control analysis: When the production process is in a state of statistical control, the process capability index is calculated and evaluated. If the process capability of the process fails to meet the requirements of the enterprise or customers, the process is readjusted to improve the process capability. If the requirements are met, the control chart is used to monitor the quality of daily production.
[0107] By using the CNN-SVM model to perform intelligent pattern recognition (recognize control chart features, including chart features and data feature values, etc.), obtain the processing process statistical parameters and set the process parameter change threshold, when the parameter change is greater than the threshold, re-learn and train the SVM to obtain a new SVM model, and then the online update of the SVM can be realized. If the control chart pattern is normal, the process continues; if the control chart pattern is abnormal, the abnormal cause diagnosis stage is entered to find the abnormal cause, and the data is continuously input into the corresponding abnormal pattern parameter estimation model to achieve the current requirements of fast and accurate processing process abnormality tracing.
[0108] S4. Abnormal tracing and adjustment: The abnormal tracing interface is divided into three parts: abnormal information list, abnormal tracing rule list, and process adjustment list. When an abnormality occurs in the control chart, the abnormal information list automatically displays the category and parameters of the corresponding control chart mode; the abnormal tracing rule list corresponds to the confidence level obtained based on the PSO-SVR model (abnormal mode parameter estimation) for each abnormal cause; the process adjustment list displays the adjustment measures for the abnormal cause.
[0109] The implementation principle of the present invention will be further described below.
[0110] As an optional implementation scheme of the present application, optionally, in step S1, the obtaining of quality data in the current CNC machining process and the collection of quality data include:
[0111] Collect quality data of the current CNC machining process in real time and import it into the preset data classification model;
[0112] The quality data is identified by the data classification model, and the data type of the current quality data is determined according to the identified feature data:
[0113] If it is new data, configure the timed delegation parameters and the first cache ID corresponding to it in the cache database;
[0114] If the data is of a fault / failure event, an irregular delegation parameter and a corresponding second cache ID in the cache database are configured for it.
[0115] During the CNC machining process, new quality data will be generated; and if there is an abnormality in the quality data of the previous stage, data of failure / unqualified events will be generated. Therefore, in order to facilitate the classification, identification and storage of quality data and the classification processing of CNC machining quality data, this embodiment sets a data classification model for identifying each data type and assigning corresponding entrusted transmission methods and cache address IDs to it, so as to facilitate orderly data cache processing.
[0116] Specifically, after data processing, the information is transmitted to the database through delegation triggering. There are two forms of delegation: one is scheduled delegation, that is, sending new quality data to the database regularly; the other is irregular delegation, the occurrence of failure events and defective product events is irregular, and the data of such irregular events is sent to the cache database through triggering.
[0117] It can provide diversified data collection methods to meet the needs of the company's various production equipment with different degrees of automation and achieve docking with different enterprise systems.
[0118] During data collection, in order to prevent data overflow in the acquisition card buffer area and interruption of collection, the data in the buffer area needs to be read regularly according to the sampling frequency.
[0119] The steps to build a data classification model usually include the following stages:
[0120] 1. Data collection: Collect data sets for training and testing models, including different types of historical data, as well as their historical transmission patterns and historical storage information.
[0121] 2. Data preprocessing: clean data, handle missing values, and standardize or normalize data.
[0122] 3. Feature engineering: Select or construct features that are helpful for classification tasks (corresponding to transmission features and storage features of different data types, IDs, etc.).
[0123] 4. Model selection: Select an appropriate classification algorithm based on the nature of the problem, such as decision tree, random forest, support vector machine, etc.
[0124] 5. Training model: Use the training data set to train the selected classification model. Classification models include logistic regression, decision tree, random forest, support vector machine, naive Bayes, K nearest neighbor, neural network, etc.
[0125] 6. Model evaluation: Evaluate the performance of the model through cross-validation or using an independent test set.
[0126] 7. Parameter tuning: Adjust model parameters based on the evaluation results to improve the accuracy and generalization ability of the model.
[0127] 8. Model deployment: Deploy the trained model to the production environment for actual data classification tasks.
[0128] The solution of using the data classification model to identify data types and automatically assign corresponding transmission modes and cache addresses can be:
[0129] 1. Data Reception: The system receives data from different sources.
[0130] 2. Data classification: Input the received data into the trained classification model.
[0131] 3. Type identification: type label of model output data.
[0132] 4. Transmission mode decision: Based on the data type, the system automatically selects the most appropriate transmission mode, such as fast transmission for data with high real-time requirements and low-cost transmission for non-real-time data.
[0133] 5. Cache address allocation: Based on the data type and transmission method, the system automatically allocates the corresponding cache address to ensure that data can be stored and accessed efficiently.
[0134] 6. Execute transmission and storage: Execute data transmission and storage operations according to the decision results.
[0135] 7. Monitoring and optimization: Continuously monitor the performance of data transmission and storage, and make necessary adjustments and optimizations based on feedback.
[0136] As an optional implementation scheme of the present application, optionally, in step S1, the control diagram corresponding to the numerical control machining control is selected according to the characteristics of the quality data, including:
[0137] Identify the data characteristics of the current quality data:
[0138] If the current quality data is metrological data, the first control chart corresponding to the numerical control machining control is selected according to the sample size;
[0139] If the current quality data is count data, the current quality data is judged according to the number of defects and the number of unqualified objects of the standard object, and the second control chart corresponding to the CNC machining control is selected.
[0140] See also Figure 2 (The figure shows the types of control charts. Each symbol represents a different type of control chart (either built-in or set by the administrator): It is a single value-moving range control chart. is the mean-standard deviation control chart, is the mean-range control chart, xR is the median-range control chart, NP is the number of defective products control chart, P is the rate of defective products control chart, C is the number of defects control chart, and U is the number of defects per unit control chart), select the control chart according to the selection principle of the control chart, and calculate the range, mean, standard deviation, statistical information of the target quality data and the parameters required by the statistical process control algorithm.
[0141] Draw a control chart for analysis and apply the stability criteria to determine whether the process is in a state of statistical control. Eight control chart modes (specifically, when one of the following eight variation modes appears, the process can be determined to be out of control: (1) one point is more than three standard deviations away from the center line; (2) nine consecutive points are on the same side of the center line; (3) six consecutive points increase or decrease; (4) 14 consecutive points alternate up and down; (5) two of three consecutive points are more than two standard deviations away from the center line; (6) four of five consecutive points are more than one standard deviation away from the center line; (7) all fifteen consecutive points are within one standard deviation of the center line; (8) eight consecutive points are more than one standard deviation away from the center line).
[0142] If the process is out of control, it will be judged whether it is out of control again. When it is out of control for the first time, a weak out of control alarm will be issued and the current time series data will be recorded in the tool compensation and workpiece quality association database. According to the established tool compensation value and the corresponding processing quality association analysis model; in the CNC processing stage, the tool compensation value of the register is automatically called through the processing program to adjust, and adaptive compensation processing is realized;
[0143] If the system loses control again, a strong out-of-control alarm will be issued, and the abnormal cause will be matched through the abnormal tracing adjustment module, the process will be adjusted and maintained, and the abnormal factors in the process will be eliminated.
[0144] As an optional implementation scheme of the present application, optionally, in step S2, recording the quality data of the current time series to a tool compensation and processing quality association database, and performing tool compensation adaptive compensation on the current CNC processing process, includes:
[0145] Obtain the processing quality data and tool compensation data of the current time series;
[0146] Carry out data preprocessing on machining quality data and tool compensation data;
[0147] Based on the Mean-shift model, cluster analysis is performed on the processing quality data and tool compensation data to obtain cluster categories;
[0148] Divide the quality intervals according to cluster categories;
[0149] Establish an Apriori model to explore the relationship between tool compensation value and processing size;
[0150] Tool compensation feedback is performed based on the correlation between the tool compensation value and the processing size.
[0151] like Figure 3 The flow chart of the tool compensation and processing quality correlation analysis is shown in the figure. First, the processing quality data and tool compensation data are preprocessed, and then the data is clustered based on the Mean-shift model, and the quality intervals are divided according to the clustering categories. The Apriori model is established to mine the correlation between the tool compensation value and the processing size, and finally the tool compensation feedback is performed.
[0152] First, select a suitable feature space and determine the bandwidth parameter. Then, initialize a set of candidate cluster center points. Next, for each data point, calculate its distance to all candidate center points and find the nearest center point. After that, update the position of each candidate center point according to the density distribution of the data points, so that it moves to the area with higher density of surrounding data points. Repeat the above process until the position of the center point no longer changes significantly or reaches the preset number of iterations. Finally, according to the position of the center point, assign the data point to the nearest cluster center to complete the clustering.
[0153] When clustering the processing quality data and tool compensation data, other factors can also be considered:
[0154] 1. Evaluate clustering effect:
[0155] - The quality of clustering results can usually be evaluated through indicators such as Silhouette Coefficient and Calinski-Harabasz Index.
[0156] - Check whether the clustering results meet the expected business or research needs.
[0157] 2. Handling outliers:
[0158] - The clustering process may find some points that are significantly different from most of the data points. These may be outliers or noise.
[0159] - You can consider whether to retain these outliers or handle them appropriately (such as deletion, marking, etc.) according to specific needs.
[0160] 3. Optimization parameters:
[0161] - Bandwidth is a critical parameter in Mean-shift clustering, which directly affects the clustering results.
[0162] - The bandwidth parameters can be optimized through cross-validation, grid search and other methods to obtain better clustering effect.
[0163] 4. Explain the clustering results:
[0164] - Conduct in-depth analysis of clustering results to understand the data characteristics, distribution and other information within each cluster.
[0165] - Based on the clustering results, put forward corresponding business recommendations or research hypotheses.
[0166] 5. Visualize clustering results:
[0167] - Use visualization tools such as scatter plots and heat maps to intuitively present clustering results.
[0168] - Through visualization, you can more clearly see the relationship between different clusters, the distribution within the clusters and other information.
[0169] 6. Apply clustering results:
[0170] - Apply clustering results to actual business or research, such as customer segmentation, market positioning, risk assessment, etc.
[0171] - Based on the clustering results, formulate corresponding strategies or plans to optimize business operations or improve research efficiency.
[0172] In order to establish the Apriori model to mine the relationship between tool compensation value and processing size, the following steps need to be followed:
[0173] 1. Data collection: First, collect transaction data including tool compensation values and processing dimensions. Make sure the data format is suitable for association rule mining.
[0174] 2. Data preprocessing: clean the data, handle missing values and outliers. Convert the data into a format suitable for the Apriori algorithm, usually by converting the data into a series of transaction records, each of which contains a set of items (for example, tool compensation value and processing size for each processing task).
[0175] 3. Determine parameters: Set the minimum support and minimum confidence threshold of the Apriori algorithm. These parameters will determine which rules are considered strong association rules.
[0176] 4. Apply Apriori algorithm: Use data mining tools or programming languages (such as Python's mlxtend library) to run the Apriori algorithm. The algorithm will find frequent itemsets based on minimum support.
[0177] 5. Analyze the results: Extract association rules from the frequent itemsets and evaluate the strength of these rules based on the minimum confidence. Analyze these rules to understand the correlation between tool compensation value and machining size.
[0178] 6. Interpretation and application: Convert the mined association rules into practical applications, such as adjusting tool compensation values to optimize processing dimensions, or using them to predict processing results.
[0179] Ensure that the business context and knowledge of the actual processing process are considered during the analysis process to ensure that the mined rules have practical significance and application value.
[0180] As an optional implementation scheme of the present application, optionally, in step S3, intelligent pattern recognition is performed on the control chart based on the CNN-SVM model, including:
[0181] Importing the current control graph into a preset CNN-SVM model, and using the CNN-SVM model to perform feature extraction on the control graph;
[0182] Classify according to the extracted features, identify and output the current control chart mode;
[0183] The CNN-SVM model is generated by performing CNN-SVM classification training on historical samples under several different control chart modes, and the historical sample data includes historical sample feature images of several control charts and the historical sample feature data therein.
[0184] The CNN-SVM model is generated by CNN-SVM classification training through historical samples under several different control chart modes. The historical samples can be collected and prepared by the administrator.
[0185] When the production process is in a state of statistical control, the process capability index is calculated and evaluated. If the process capability of the process fails to meet the requirements of the enterprise or customers, the process is readjusted to improve the process capability. If the requirements are met, the control chart is used to monitor the quality of daily production.
[0186] The control chart patterns are identified based on the CNN-SVM model. Specifically, the control chart patterns are divided into 8 types, such as Figure 4 As shown (8 identified control chart modes): (1) normal mode NOR; (2) upward step mode US; (3) downward step mode DS; (4) upward trend mode UT; (5) downward trend mode DT; (6) cycle mode CYC; (7) layered mode STR; (8) system mode SYS.
[0187] The established CNN-SVM model includes input layer, feature extraction layer, classification layer and output layer. The feature extraction layer contains three convolutional layers and three pooling layers. The classification layer uses SVM as the classifier. The CNN-SVM model structure is as follows: Figure 5 In SVM, the choice of kernel function parameters will have a great impact on classification accuracy. Choosing appropriate parameters can effectively improve the performance of SVM.
[0188] Preferably, the whale optimization algorithm can be used to optimize the parameters of the SVM to avoid the randomness and uncertainty caused by manual selection of parameters and improve the generalization and adaptability of the SVM.
[0189] By obtaining the statistical parameters of the machining process and setting the threshold of process parameter changes, when the parameter change is greater than the threshold, the SVM is retrained to obtain a new SVM model, thus realizing the online update of the SVM. Figure 6 As shown, specifically, firstly, the initial mean can be calculated based on historical samples , initial standard deviation (Threshold). As the process progresses, when the control chart pattern is abnormal, calculate the mean of the additional samples , Standard Deviation . The mean of the sample , Standard Deviation The relative change is compared with the set threshold. When the relative change is greater than the set threshold, the SVM model is retrained to obtain a new SVM model. When the relative change is less than the set threshold, the initial SVM model is used to identify the abnormal pattern.
[0190] As an optional implementation scheme of the present application, optionally, in step S3, determining whether the control diagram mode has an abnormal mode includes:
[0191] Based on the CNN-SVM model, intelligent pattern recognition is performed on the real-time sample feature data in the control chart;
[0192] Determine whether the real-time sample feature data matches the historical sample feature data learned by the model:
[0193] If they match, it is determined that there is no abnormal pattern in the control chart pattern;
[0194] Otherwise, it is determined that there is an abnormal mode in the control chart mode, and the abnormal mode parameters are calculated to trace the abnormality.
[0195] If the control chart pattern is normal, the process continues; if the control chart pattern is abnormal, the abnormal cause diagnosis stage is entered to find the abnormal cause, and the data is continued to be input into the corresponding abnormal pattern parameter estimation model to achieve the current requirements of fast and accurate processing process abnormality tracing.
[0196] As an optional implementation scheme of the present application, optionally, in step S3, determining whether the control diagram mode has an abnormal mode further includes:
[0197] If it is determined that there is no abnormal pattern in the control chart pattern, then processing will continue to obtain additional samples;
[0198] Calculate the feature data of the newly added samples and determine whether the relative change value of the feature data of the newly added samples exceeds the preset threshold:
[0199] If exceeded, retrain the CNN-SVM model;
[0200] Otherwise, the CNN-SVM model is maintained to continue to perform intelligent pattern recognition on the control chart.
[0201] As an optional implementation scheme of the present application, optionally, S4, abnormal source tracing adjustment: based on the abnormal mode parameter estimation model: PSO-SVR, intelligently evaluate the abnormal information under the abnormal mode, and output adjustment measures for the abnormal information, including:
[0202] Analyze the factors affecting CNC machining quality, determine the main factors affecting CNC machining quality, and use them as training sample data;
[0203] Read the training sample data, import it into the preset SVR model, and initialize the SVR model;
[0204] The optimal parameters of the SVR model are iteratively calculated through the PSO algorithm, and the final abnormal pattern parameter estimation model is constructed based on the optimal parameters: the PSO-SVR model;
[0205] Construct confidence intervals for each anomaly cause;
[0206] The control chart pattern abnormal data is imported into the abnormal pattern parameter estimation model: PSO-SVR model, and based on the confidence interval, various abnormal causes in the abnormal information are ranked by abnormality, and the best adjustment measures for the abnormal information are evaluated and output.
[0207] Specifically, the PSO-SVR anomaly tracing reasoning flowchart is as follows: Figure 7 As shown in the figure, the process is described as follows: (1) Determine the main quality influencing factors in combination with the CNC machining production environment and production process. (2) Read the training sample data and establish the initial SVR model. (3) Iterate the optimal parameters through the PSO algorithm, calculate the fitness of each particle, and input the position vector of each particle as a hyperparameter into the SVR model. (4) Update the individual optimal value and the population optimal value. (5) Determine whether the termination condition is met. If the termination condition is met, the particle position vector corresponding to the minimum fitness of the population is used as the optimal parameter of the SVR model. If the termination condition is not met, return to step 2. (6) Construct the final PSO-SVR model. (7) Construct a confidence interval for each common abnormal cause. (8) Input the abnormal data of the control chart model, sort the abnormality degree of various abnormal causes, and finally make process maintenance adjustments based on the evaluation results (prioritize the sorting results to recommend adjustment measures, and the system database pre-stores adjustment measures corresponding to different abnormality degrees (the characteristic value pairing method can be used to match the adjustment measures with the corresponding abnormality degrees). According to the abnormality degree sorting, the corresponding adjustment measures are recommended. The smaller the abnormality degree ranking, the milder the adjustment measures are recommended (mild, for example, its characteristic value is level III, corresponding to the third level of abnormality degree sorting)).
[0208] like Figure 8 The following are common influencing factors in actual CNC machining, which can be further classified. Preferably, binary coding can be used layer by layer to make abnormality tracing faster and more accurate.
[0209] Obviously, those skilled in the art should understand that the implementation of all or part of the processes in the above-mentioned embodiments can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the above-mentioned embodiments. Those skilled in the art can understand that the implementation of all or part of the processes in the above-mentioned embodiments can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the above-mentioned embodiments. Among them, the storage medium can be a disk, an optical disk, a read-only memory (ROM), a random access memory (RAM), a flash memory (Flash Memory), a hard disk (Hard Disk Drive, abbreviated as: HDD) or a solid-state drive (SSD), etc.; the storage medium can also include a combination of the above-mentioned types of memory.
[0210] Example 2
[0211] Based on the implementation principle of Example 1, on the other hand, the present application proposes a CNC machining quality monitoring system based on statistical process control, which is used to implement a CNC machining quality monitoring method based on statistical process control described in Example 1, including:
[0212] General module, used to: provide general services for CNC machining;
[0213] The online data acquisition module is used to: obtain the quality data in the current CNC machining process, collect the quality data, and select the control chart corresponding to the CNC machining control according to the characteristics of the quality data;
[0214] Tool compensation and processing quality correlation analysis module is used to: apply the stability judgment criteria to make stability judgment on the selected control chart and judge whether the current CNC machining process is stable:
[0215] If yes, go to the next step;
[0216] Otherwise, determine whether it is out of control again:
[0217] If so, it is not the first out-of-control, and a strong out-of-control alarm prompt is issued, and the abnormal diagnosis knowledge base is used to perform abnormal diagnosis, and the CNC machining process is re-entered after the abnormal factors of the process are eliminated;
[0218] On the contrary, it is the first time that the machine loses control, and a weak out-of-control alarm is given. The quality data of the current time series is recorded in the tool compensation and machining quality association database, and the tool compensation adaptive compensation is performed on the current CNC machining process.
[0219] The quality control analysis module is used to: perform process capability analysis on the CNC machining process that is currently in a stable state to determine whether the process capability is sufficient:
[0220] If yes, the control chart is used to continue monitoring, and intelligent pattern recognition is performed on the control chart based on the CNN-SVM model to determine whether there is an abnormal pattern in the control chart pattern: if yes, proceed to the next step; otherwise, the control chart is used to continue monitoring;
[0221] On the contrary, the abnormal factors in the process are eliminated, and the CNC machining process is re-entered after the abnormal factors in the process are eliminated;
[0222] The abnormality tracing and adjustment module is used to: based on the abnormal mode parameter estimation model: PSO-SVR, intelligently evaluate the abnormal information under the abnormal mode and output adjustment measures for the abnormal information; based on the adjustment measures of the abnormal information, the system eliminates the process abnormal factors and re-enters the CNC machining process.
[0223] The general module includes general settings such as user login, system initialization, technical parameter management, data backup, etc., realizes user login information verification, database configuration, sets and manages product information in the system, and saves it through database or configuration files.
[0224] The online data acquisition module is used to collect all real-time quality data that require control. After the collected quality data is processed, the information is transmitted to the database through a delegated trigger.
[0225] Tool compensation and processing quality correlation analysis module, the tool compensation and processing quality correlation analysis module establishes a correlation analysis model based on tool compensation value data and workpiece size or workpiece size range data, finds the optimal tool compensation value combination of each size in the processing process, and plans the processing process. According to the current workpiece size data, the optimal tool compensation value can be matched and fed back to the CNC machine tool. The tool compensation value adjustment of the register is automatically called through the processing program to realize product adaptive compensation processing.
[0226] The quality control analysis module includes the creation and analysis of SPC projects, drawing of various charts, real-time monitoring of the processing process, intelligent pattern recognition, saving and printing of charts, and visual display of test results. According to the technical parameters and product information parameters set by the administrator in the general module, statistical process control is performed on the processing process required by the user. The optimal control chart is automatically selected according to the control chart selection principle, and the corresponding SPC project analysis chart is drawn to monitor in real time whether there are any abnormalities in the processing process. A secondary quality warning is set. When the control is out of control for the first time, a weak out-of-control quality alarm is issued, and the abnormal data is stored in the tool compensation and workpiece quality association database for tool compensation adaptive compensation; when the control is out of control again, a strong out-of-control alarm is issued, and the CNN-SVM online update classifier is called for intelligent identification. The abnormal data will be marked and enter the abnormal traceability adjustment module to find the cause of the abnormality and eliminate the process abnormality factors.
[0227] Abnormal tracing adjustment module, the abnormal tracing adjustment module contains the corresponding abnormal causes and the abnormal diagnosis knowledge base based on the PSO-SVR model. The abnormal tracing interface is divided into three parts: abnormal information list, abnormal tracing rule list, and process adjustment list. When an abnormality occurs in the control chart, the abnormal information list automatically displays the category and parameters of the corresponding control chart mode; the abnormal tracing rule list corresponds to the confidence obtained by the PSO-SVR model for each abnormal cause; the process adjustment list displays the adjustment measures for the abnormal cause.
[0228] The functions and interactions of the above modules should be understood and implemented in conjunction with Example 1, and will not be described in detail in this example.
[0229] The modules or steps of the present invention described above can be implemented by a general-purpose computing system, they can be concentrated on a single computing system, or distributed on a network composed of multiple computing systems, and optionally, they can be implemented by a program code executable by a computing system, so that they can be stored in a storage system and executed by the computing system, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.
[0230] Example 3
[0231] Furthermore, in another aspect, the present application also proposes an electronic device for a CNC machining quality monitoring method based on statistical process control, comprising:
[0232] processor;
[0233] a memory for storing processor-executable instructions;
[0234] Wherein, the processor is configured to implement a CNC machining quality monitoring method based on statistical process control as described in Example 2 when executing the executable instructions.
[0235] The electronic device of the embodiment of the present disclosure includes a processor and a memory for storing processor executable instructions. The processor is configured to implement a numerical control machining quality monitoring method based on statistical process control described in the above embodiment 2 when executing the executable instructions.
[0236] Here, it should be noted that the number of processors can be one or more. At the same time, the electronic device of the embodiment of the present disclosure may also include an input system and an output system. Among them, the processor, memory, input system and output system may be connected through a bus or in other ways, which are not specifically limited here.
[0237] The memory, as a computer-readable storage medium, can be used to store software programs, computer executable programs and various modules, such as the program or module corresponding to the CNC machining quality monitoring method based on statistical process control in the embodiment of the present disclosure. The processor executes various functional applications and data processing of the electronic device by running the software programs or modules stored in the memory.
[0238] The input system can be used to receive input numbers or signals. The signal can be a key signal related to user settings and function control of the device / terminal / server. The output system can include display devices such as display screens.
[0239] The embodiments of the present disclosure have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or technical improvements in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.
Claims
1. A method for monitoring the quality of CNC machining based on statistical process control, characterized in that: The steps include: S1. Online data collection: obtain the quality data in the current CNC machining process, collect the quality data, and select the control chart corresponding to the CNC machining control according to the characteristics of the quality data; S2. Analysis of the correlation between tool compensation and machining quality: Apply the stability criterion to make a stability judgment on the selected control chart to determine whether the current CNC machining process is stable: If yes, go to the next step; Otherwise, determine whether it is out of control again: If so, it is not the first out-of-control, and a strong out-of-control alarm prompt is issued, and the abnormal diagnosis knowledge base is used to perform abnormal diagnosis, and the CNC machining process is re-entered after the abnormal factors of the process are eliminated; Otherwise, it is the first time that the machine loses control, and a weak out-of-control alarm is given. The quality data of the current time series is recorded in the tool compensation and machining quality association database, and the tool compensation adaptive compensation is performed on the current CNC machining process, including: Obtain the processing quality data and tool compensation data of the current time series; Carry out data preprocessing on machining quality data and tool compensation data; Based on the Mean-shift model, cluster analysis is performed on the processing quality data and tool compensation data to obtain cluster categories; Divide the quality intervals according to cluster categories; Establish an Apriori model to explore the relationship between tool compensation value and processing size; Provide tool compensation feedback based on the relationship between tool compensation value and processing size; S3. Quality control analysis: Conduct process capability analysis on the CNC machining process that is currently in a stable state to determine whether the process capability is sufficient: If yes, the control chart is used to continue monitoring, and intelligent pattern recognition is performed on the control chart based on the CNN-SVM model to determine whether there is an abnormal pattern in the control chart pattern: if yes, proceed to the next step; otherwise, the control chart is used to continue monitoring; On the contrary, the abnormal factors in the process are eliminated, and the CNC machining process is re-entered after the abnormal factors in the process are eliminated; S4. Abnormal tracing adjustment: Based on the abnormal mode parameter estimation model: PSO-SVR, the abnormal information under the abnormal mode is intelligently evaluated and the adjustment measures for the abnormal information are output; the system eliminates the process abnormal factors and re-enters the CNC machining process based on the adjustment measures of the abnormal information, including: Analyze the factors affecting CNC machining quality, determine the main factors affecting CNC machining quality, and use them as training sample data; Read the training sample data, import it into the preset SVR model, and initialize the SVR model; The optimal parameters of the SVR model are iteratively calculated through the PSO algorithm, and the final abnormal pattern parameter estimation model is constructed based on the optimal parameters: the PSO-SVR model; Construct confidence intervals for each anomaly cause; The control chart pattern abnormal data is imported into the abnormal pattern parameter estimation model: PSO-SVR model, and based on the confidence interval, various abnormal causes in the abnormal information are ranked by abnormality, and the best adjustment measures for the abnormal information are evaluated and output.
2. A method for monitoring the quality of numerical control machining based on statistical process control according to claim 1, characterized in that: In step S1, the quality data in the current numerical control machining process is obtained and quality data collection is performed, including: Collect quality data of the current CNC machining process in real time and import it into the preset data classification model; The quality data is identified by the data classification model, and the data type of the current quality data is determined according to the identified feature data: If it is new data, configure the timed delegation parameters and the first cache ID corresponding to it in the cache database; If the data is of a fault / failure event, an irregular delegation parameter and a corresponding second cache ID in the cache database are configured for it.
3. The method for monitoring the quality of numerical control machining based on statistical process control according to claim 1, characterized in that: In step S1, the control diagram corresponding to the numerical control machining control is selected according to the characteristics of the quality data, including: Identify the data characteristics of the current quality data: If the current quality data is metrological data, the first control chart corresponding to the numerical control machining control is selected according to the sample size; If the current quality data is count data, the current quality data is judged according to the number of defects and the number of unqualified objects of the standard object, and the second control chart corresponding to the CNC machining control is selected.
4. The method for monitoring the quality of numerical control machining based on statistical process control according to claim 1, characterized in that: In step S3, intelligent pattern recognition is performed on the control chart based on the CNN-SVM model, including: Importing the current control graph into a preset CNN-SVM model, and using the CNN-SVM model to perform feature extraction on the control graph; Classify according to the extracted features, identify and output the current control chart mode; The CNN-SVM model is generated by performing CNN-SVM classification training on historical samples under several different control chart modes, and the historical sample data includes historical sample feature images of several control charts and the historical sample feature data therein.
5. The method for monitoring the quality of numerical control machining based on statistical process control according to claim 4, characterized in that: In step S3, judging whether there is an abnormal mode in the control diagram mode includes: Based on the CNN-SVM model, intelligent pattern recognition is performed on the real-time sample feature data in the control chart; Determine whether the real-time sample feature data matches the historical sample feature data learned by the model: If they match, it is determined that there is no abnormal pattern in the control chart pattern; Otherwise, it is determined that there is an abnormal mode in the control chart mode, and the abnormal mode parameters are calculated to trace the abnormality.
6. A method for monitoring the quality of numerical control machining based on statistical process control according to claim 5, characterized in that: In step S3, judging whether there is an abnormal mode in the control diagram mode also includes: If it is determined that there is no abnormal pattern in the control chart pattern, then processing will continue to obtain additional samples; Calculate the feature data of the newly added samples and determine whether the relative change value of the feature data of the newly added samples exceeds the preset threshold: If exceeded, retrain the CNN-SVM model; Otherwise, the CNN-SVM model is maintained to continue to perform intelligent pattern recognition on the control chart.
7. A CNC machining quality monitoring system based on statistical process control, used to implement a CNC machining quality monitoring method based on statistical process control as described in any one of claims 1 to 6, characterized in that: include: General module, used to: provide general services for CNC machining; The online data acquisition module is used to: obtain the quality data in the current CNC machining process, collect the quality data, and select the control chart corresponding to the CNC machining control according to the characteristics of the quality data; Tool compensation and processing quality correlation analysis module is used to: apply the stability judgment criteria to make stability judgment on the selected control chart and judge whether the current CNC machining process is stable: If yes, go to the next step; Otherwise, determine whether it is out of control again: If so, it is not the first out-of-control, and a strong out-of-control alarm prompt is issued, and the abnormal diagnosis knowledge base is used to perform abnormal diagnosis, and the CNC machining process is re-entered after the abnormal factors of the process are eliminated; Otherwise, it is the first time that the machine loses control, and a weak out-of-control alarm is given. The quality data of the current time series is recorded in the tool compensation and processing quality association database, and tool compensation adaptive compensation is performed on the current CNC machining process, including: obtaining the processing quality data and tool compensation data of the current time series; performing data preprocessing on the processing quality data and tool compensation data; performing cluster analysis on the processing quality data and tool compensation data based on the Mean-shift model to obtain cluster categories; dividing the quality intervals according to the cluster categories; establishing an Apriori model to explore the correlation between the tool compensation value and the processing size; and performing tool compensation feedback according to the correlation between the tool compensation value and the processing size. The quality control analysis module is used to: perform process capability analysis on the CNC machining process that is currently in a stable state to determine whether the process capability is sufficient: If yes, the control chart is used to continue monitoring, and intelligent pattern recognition is performed on the control chart based on the CNN-SVM model to determine whether there is an abnormal pattern in the control chart pattern: if yes, proceed to the next step; otherwise, the control chart is used to continue monitoring; On the contrary, the abnormal factors in the process are eliminated, and the CNC machining process is re-entered after the abnormal factors in the process are eliminated; The abnormality tracing adjustment module is used for: based on the abnormal mode parameter estimation model: PSO-SVR, intelligently evaluating the abnormal information under the abnormal mode and outputting the adjustment measures for the abnormal information; based on the adjustment measures of the abnormal information, the system eliminates the process abnormal factors and re-enters the CNC machining process, including: analyzing the factors affecting the quality of CNC machining, determining the main factors affecting the quality of CNC machining, and using them as training sample data; reading the training sample data, and importing it into the preset SVR model to initialize the SVR model; iteratively calculating the optimal parameters of the SVR model through the PSO algorithm, and constructing the final abnormal mode parameter estimation model: PSO-SVR model based on the optimal parameters; constructing a confidence interval for each abnormal cause; importing the control chart mode abnormal data into the abnormal mode parameter estimation model: PSO-SVR model, and based on the confidence interval, sorting the abnormality of various abnormal causes in the abnormal information, and evaluating and outputting the best adjustment measures for the abnormal information.
8. An electronic device, characterized in that include: processor; a memory for storing processor-executable instructions; Wherein, the processor is configured to implement a CNC machining quality monitoring method based on statistical process control as described in any one of claims 1 to 6 when executing the executable instructions.
Citation Information
Patent Citations
Quality control chart pattern recognition method based on improved genetic algorithm optimization
CN109919130A
Power utilization potential safety hazard intelligent diagnosis method based on knowledge graph
CN113505241A
Cited By
Numerical control machining quality monitoring method and system based on statistical process control
CN120370859A
