Numerical control machining size error prediction method

Abstract

The invention discloses a numerical control machining size error prediction method which is characterized by comprising the following steps: collecting various historical data in the numerical control machining process of aviation parts, and preprocessing the historical data; carrying out feature analysis on the preprocessed data, screening out features related to the numerical control machining of the aviation parts, obtaining a feature data set, and analyzing the change trend of each feature; on the basis of an analysis result, a size error prediction model is constructed in combination with a machine learning algorithm, and the size error of the aviation part in the numerical control machining process is predicted; combining a prediction result of the size error prediction model and feature data of the feature data set to obtain a size error trend coefficient, and analyzing a part size error change trend of the production environment; the constructed size error prediction model is applied to the actual production process, key parameters in the machining process are monitored in real time, and the size error state of aviation part machining is predicted; according to the prediction result of the size error state, the abnormal condition of the size error is judged, whether an early warning signal is sent out or not is judged, and corresponding adjustment measures are taken. According to the technical scheme, the problems that various interference factors exist in the actual production environment in an existing numerical control machining mode, size errors of the aviation parts in the numerical control machining process can be caused, and then the machining quality of the aviation numerical control parts is affected are solved.

Description

Technical Field

[0001] The present invention relates to the field of, but is not limited to, CNC parts processing technology, and in particular to a CNC processing dimension error prediction method. Background Art

[0002] The production and processing of aviation CNC parts is a key link in the aviation manufacturing industry. It has the characteristics of small batches and diversification. With the continuous development of the manufacturing industry, the requirements for the processing accuracy and product quality of aviation CNC parts are getting higher and higher. As an important pillar of modern manufacturing, CNC processing technology has a direct impact on the quality and performance of products. Therefore, by predicting the dimensional error in the CNC processing process and taking corresponding measures to control and compensate, the processing accuracy can be significantly improved to ensure the consistency and stability of the production quality of aviation CNC parts.

[0003] The existing CNC machining methods have various interference factors in the actual production environment, which will cause dimensional errors in the CNC machining process of aviation parts, and thus affect the machining quality of aviation CNC parts. Summary of the invention

[0004] Purpose of the invention: In order to solve the above-mentioned technical problems, an embodiment of the present invention provides a method for predicting dimensional errors in CNC machining to solve the problem that various interference factors exist in the existing CNC machining methods in the actual production environment, which will cause dimensional errors in the CNC machining process of aviation parts and further affect the machining quality of aviation CNC parts.

[0005] Technical solution of the present invention: An embodiment of the present invention provides a method for predicting dimensional errors in numerical control machining, comprising:

[0006] Step 1, collecting various historical data in the CNC machining process of aviation parts, and preprocessing the collected historical data;

[0007] Step 2: Perform feature analysis on various types of preprocessed historical data, screen out features related to CNC machining of aviation parts, obtain feature data sets, and analyze the change trends of each feature;

[0008] Step 3: Based on the analysis results of the feature data set, a dimensional error prediction model is constructed in combination with a machine learning algorithm to predict the dimensional error of the CNC machining process of aviation parts;

[0009] Step 4, combining the prediction results of the dimensional error prediction model and the characteristic data of the characteristic data set to obtain the dimensional error trend coefficient and analyze the dimensional error change trend of the parts in the production environment;

[0010] Step 5, applying the constructed dimensional error prediction model to the actual production process, monitoring the key parameters in the processing process in real time, and predicting the dimensional error state of aviation parts processing;

[0011] Step 6: According to the prediction result of the dimensional error state, determine the abnormal situation of the dimensional error, determine whether to issue an early warning signal, and implement corresponding adjustment measures.

[0012] Optionally, in the above-mentioned method for predicting dimensional errors in numerical control machining, in step 1, the process of collecting and preprocessing historical data of the numerical control machining process of aviation parts includes:

[0013] Step 1.1, clarify the data types that need to be collected for CNC machining of aviation parts, including: machining parameters, equipment characteristic values ​​and part dimensional error data;

[0014] Step 1.2, using the data acquisition function of the CNC machine tool, read the machine tool operation data through the interface of the CNC control system, collect the historical data of the processing parameters and equipment characteristic values ​​of the preset length evaluation period, and use the measuring tool to measure the size of the aviation parts, compare it with the design size, and calculate the part size error data;

[0015] Step 1.3, preprocessing the collected historical data, the preprocessing operation includes data cleaning and data conversion;

[0016] In step 1.4, the preprocessed processing parameters, equipment characteristic values, and part dimensional error data are matched and integrated according to the processing batches to form a unified data set.

[0017] Optionally, in the above-mentioned method for predicting dimensional errors in numerical control machining, in step 2, the analysis process of the change trend of each feature is as follows:

[0018] Step 2.1, statistically analyzing the preprocessed historical data, extracting the CNC machining feature data of aviation parts from the machining parameters, equipment characteristic values ​​and part size error data, and obtaining the feature data set by comprehensively screening the feature data;

[0019] Among them, for the processing parameter data, the characteristic data of spindle speed, cutting speed, feed speed and cutting depth are extracted; for the equipment characteristic value data, the characteristic data of machine tool vibration, spindle current and lubricating oil temperature are extracted; for the part size error data, the characteristic data of size deviation, size deviation change rate, shape deviation and position deviation are extracted;

[0020] Step 2.2, based on each characteristic data of the processing parameters in the characteristic data set and in combination with the reference value of each characteristic data of the processing parameters, the process trend index is calculated to analyze the change trend of the processing parameters within the preset length evaluation time period;

[0021] Step 2.3, based on each characteristic data of the equipment characteristic value in the characteristic data set and in combination with the reference value of each characteristic data of the equipment characteristic value, the mechanical property trend index is calculated to analyze the change trend of the equipment characteristic value within the preset length evaluation time period;

[0022] Step 2.4, based on the characteristic data of the part dimensional error in the characteristic data set and combined with the baseline value of each characteristic data of the part dimensional error, calculate the part error trend index and analyze the change trend of the part dimensional error within the preset length evaluation time period.

[0023] Optionally, in the above-mentioned method for predicting dimensional errors in numerical control machining, the process trend index is calculated as follows:

[0024]

[0025] Among them, ATI is the process trend index, SP is the current value of the spindle speed, BP is the reference value of the cutting speed, LP is the current value of the feed speed, RP is the reference value of the cutting depth, and α is the coefficient of the cutting depth, which is used to adjust the influence of the cutting depth on the index;

[0026] The calculation method of the mechanical property trend index is:

[0027]

[0028] Among them, MTI is the mechanical property trend index, V m is the current machine tool vibration value, V 0 is the reference value of machine tool vibration, V max is the maximum allowable value of machine tool vibration, I s is the current spindle current value, I 0 is the reference value of the spindle current, I max is the maximum allowable value of the spindle current, T oil is the current lubricating oil temperature, T 0 is the reference value of lubricating oil temperature, T max is the maximum allowable value of lubricating oil temperature;

[0029] The calculation method of the part error trend index is:

[0030]

[0031] in,

[0032] Among them, PTI is the part error trend index, F(d, r, f) is the deviation influence function, d s is the current size deviation, d 0 is the reference value of the dimensional deviation, d max is the maximum allowable value of dimensional deviation, r s is the current size deviation change rate, r 0 is the reference value of the dimensional deviation change rate, r max is the maximum allowable value of the dimensional deviation change rate, f s is the current shape deviation, f 0 is the reference value of shape deviation, f max is the maximum allowable value of shape deviation, p s is the current position deviation, p 0 is the reference value of position deviation, p max is the maximum allowable value of position deviation.

[0033] Optionally, in the above-mentioned method for predicting dimensional errors in numerical control machining, the process of constructing the dimensional error prediction model in step 3 includes:

[0034] Step 3.1, randomly divide the entire feature data set into a training set and a test set. For the size error prediction problem, a linear regression model is selected for model training;

[0035] Step 3.2, using the training set data to learn and adjust the parameters of the selected linear regression model, calculate the model parameters to minimize the error between the predicted value and the actual value, and then build a size error prediction model;

[0036] In step 3.3, the trained dimensional error prediction model is evaluated using the test set data, the difference between the predicted value and the actual value is compared, and the dimensional error prediction model is adjusted and optimized according to the evaluation results, so as to predict the dimensional error of the CNC machining process of aviation parts.

[0037] Optionally, in the above-mentioned method for predicting dimensional errors in numerical control machining, in step 4, the process of analyzing the dimensional error variation trend of parts in the production environment is as follows:

[0038] Step 4.1, compare the prediction results of the size error prediction model with the actual size error data in the feature data set one by one, calculate the prediction error, and analyze the distribution and size of the prediction error;

[0039] Step 4.2, combining the prediction results of the dimensional error prediction model and the relevant data of the characteristic data set, calculating the process trend index, mechanical property trend index and part error trend index, and analyzing the change trends of processing parameters, equipment characteristic values ​​and part dimensional errors;

[0040] Step 4.3, combine the process trend index, mechanical property trend index and part error trend index to obtain a comprehensive dimensional error trend coefficient, analyze the time series change of the dimensional error trend coefficient, determine its size and change direction, and clarify the change trend of the dimensional error of parts in the production environment;

[0041] Step 4.4, based on the changing trend of the dimension error trend coefficient, perform anomaly detection on the production environment. If the dimension error trend coefficient exceeds the normal range, it is judged that there is an abnormality in the production environment.

[0042] Optionally, in the above-mentioned method for predicting dimensional errors in numerical control machining, the dimensional error trend coefficient in step 4.3 is calculated as follows:

[0043]

[0044] Among them, ECI is the dimensional error trend coefficient, ATI is the process trend index, which reflects the changing trend of processing parameters, MTI is the mechanical property trend index, which reflects the changing trend of equipment characteristic values, PTI is the part error trend index, which reflects the changing trend of part dimensional error, ∈ is a positive number, close to 0, which is used to avoid division by 0, and the value range of ECI is between 0 and 1.

[0045] Optionally, in the above-mentioned method for predicting dimensional errors in numerical control machining, the process of predicting the dimensional error state of aviation parts in step 5 includes:

[0046] Step 5.1, save the trained dimension error prediction model in a loadable format, deploy it to the server in the production environment, and integrate the dimension error prediction model in the CNC control system;

[0047] Step 5.2, configure sensors and data acquisition modules in the CNC control system to monitor the key parameter data in the processing process in real time, send the key parameter data monitored in real time to the data server through the data interface of the CNC control system, develop a data preprocessing module in the data server, and perform preprocessing operations such as cleaning, format conversion, and normalization on the collected key parameter data;

[0048] Step 5.3, the preprocessed key parameter data is used as input into the dimensional error prediction model;

[0049] Step 5.4, call the dimensional error prediction model through the interface of the CNC control system, pass the input data to the dimensional error prediction model for processing, calculate and output the dimensional error state prediction results of aviation parts processing based on the input data, and output the prediction results to the operator through the interface of the CNC control system.

[0050] Optionally, in the above-mentioned method for predicting dimensional errors in numerical control machining, the process of issuing the warning signal in step 6 is:

[0051] Step 6.1, obtain the prediction result of the dimensional error state from the CNC control system, set the warning threshold of the dimensional error according to the product quality standard and historical data, compare the prediction result with the preset warning threshold, and determine whether the prediction result exceeds the allowable error range;

[0052] Step 6.2: If the prediction result exceeds the allowable error range, it is judged as a dimensional error abnormality, and the cause of the abnormality is analyzed;

[0053] Step 6.3, combining the dimension error trend coefficient and the deviation degree of the prediction result, the dimension error anomaly level is divided into slight anomaly, medium anomaly, and severe anomaly, and the corresponding deviation thresholds are matched for different levels of dimension error anomaly. Different levels of dimension error anomaly correspond to different warning signals and adjustment measures;

[0054] Step 6.4, according to the level of dimensional error anomaly, set the corresponding warning signal rules, a slight anomaly will issue a yellow warning, a moderate anomaly will issue an orange warning, and a severe anomaly will issue a red warning. According to the set warning rules, a warning signal will be issued through the interface of the CNC control system, and corresponding countermeasures will be taken;

[0055] Step 6.5, integrate the data and feedback from the actual processing process, optimize the dimensional error prediction model, and continuously monitor the dimensional error status during the processing after implementing the adjustment measures.

[0056] Optionally, in the above-mentioned method for predicting dimensional errors in numerical control machining, a plurality of dimensional error abnormality levels correspond to a plurality of deviation thresholds, wherein the deviation thresholds include an upper threshold and a lower threshold;

[0057] The multiple dimensional error abnormality levels and the multiple deviation thresholds satisfy the following relationship:

[0058] Mildly abnormal ECI L ≤ECI<1;

[0059] Moderately abnormal ECI M ≤ECI <ECI L ;

[0060] Severe abnormality 0 <ECI<ECI M ;

[0061] Among them, ECI is the dimensional error tendency coefficient, ECI L is the lower threshold corresponding to a slight anomaly and the upper threshold corresponding to a moderate anomaly, ECI Mis the lower threshold corresponding to moderate anomalies and the upper threshold corresponding to severe anomalies, ECI L =0.9, ECI M =0.7.

[0062] Beneficial effects of the present invention: The embodiment of the present invention provides a method for predicting dimensional errors in numerical control machining. Compared with the prior art, the following technical advances have been achieved by using the method for predicting dimensional errors in numerical control machining:

[0063] First, the CNC machining dimensional error prediction method provided by the present invention obtains data during the machining process and uses the dimensional error prediction model for analysis to discover and warn of potential dimensional errors, which not only improves the accuracy of the dimensions of aviation parts, but also significantly enhances the stability of product quality. During the production process, even in the face of slight changes in machining parameters or equipment wear, it can respond quickly and guide operators to adjust machining conditions, thereby avoiding the accumulation and amplification of dimensional errors.

[0064] Second, the CNC machining dimensional error prediction method provided by the present invention not only reduces the waste of raw materials but also reduces production costs by discovering potential problems in advance and avoiding rework and scrapping due to unqualified dimensions. At the same time, it can also promptly discover equipment failures or signs of wear, thereby performing maintenance and replacement in advance, avoiding production interruptions and delays due to equipment failures, and greatly reducing the risks and uncertainties in the production process. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] The accompanying drawings are used to provide a further understanding of the technical solution of the present invention and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the technical solution of the present invention and do not constitute a limitation on the technical solution of the present invention.

[0066] Figure 1 A flowchart of a method for predicting dimensional errors in numerical control machining provided by an embodiment of the present invention;

[0067] Figure 2 Flow chart of the analysis of the changing trend of each dimension error characteristic in the embodiment of the present invention;

[0068] Figure 3 Flow chart of the analysis of the variation trend of the dimensional error of parts in the production environment according to the embodiment of the present invention;

[0069] Figure 4 A flowchart of predicting the machining dimension error state of aviation parts in an embodiment of the present invention;

[0070] Figure 5 The figure is a flow chart of issuing a warning signal in an embodiment of the present invention. DETAILED DESCRIPTION

[0071] In order to make the purpose, technical solution and advantages of the present invention more clear, the embodiments of the present invention will be described in detail with reference to the accompanying drawings. It should be noted that the embodiments and features in the embodiments of the present application can be combined with each other arbitrarily without conflict.

[0072] The above background technology has explained that the existing CNC machining methods have various interference factors in the actual production environment, which will cause dimensional errors in the CNC machining process of aviation parts, thereby affecting the machining quality of aviation CNC parts.

[0073] Therefore, how to analyze the key parameters in the machining process of aviation parts and timely discover the changing trends and abnormal conditions of the production environment is the problem to be solved by the present invention. To this end, the embodiment of the present invention provides a CNC machining dimensional error prediction method, which acquires data in the machining process and uses a dimensional error prediction model for analysis to discover and warn of potential dimensional errors, thereby not only improving the accuracy of the dimensions of aviation parts, but also significantly enhancing the stability of product quality. During the production process, even in the face of slight changes in machining parameters or equipment wear, it can respond quickly and guide operators to adjust machining conditions, thereby avoiding the accumulation and amplification of dimensional errors.

[0074] The present invention provides the following specific embodiments which can be combined with each other. The same or similar concepts or processes may not be described in detail in some embodiments.

[0075] The implementation of the CNC machining dimension error prediction method provided by the present invention is described below through some embodiments.

[0076] Embodiment 1:

[0077] like Figure 1 , Figure 2 As shown, the present invention provides a method for predicting dimensional errors in numerical control machining, comprising the following steps:

[0078] Step 1, collecting various historical data in the CNC machining process of aviation parts, and preprocessing the collected historical data;

[0079] The implementation method of this step is as follows: first, clarify the data types of CNC machining of aviation parts that need to be collected, including: machining parameters, equipment characteristic values ​​and part size error data; second, use the data acquisition function of the CNC machine tool to read the machine tool operation data through the interface of the CNC control system, collect the historical data of machining parameters and equipment characteristic values ​​for a preset length evaluation period, and use measuring tools to measure the dimensions of aviation parts, compare them with the design dimensions, and calculate the part size error data; then, preprocess the collected historical data, and the preprocessing operation includes data cleaning and data conversion, wherein data cleaning is used to check whether there are duplicate records in the data set and delete them. For missing data, fill (such as using the mean, median, mode, etc.) or delete according to the specific situation, identify and process outliers in the data, and convert non-numeric data into numerical data for subsequent analysis and processing. For data of different dimensions, standardize them to eliminate the impact of the dimension on the data; finally, match and integrate the preprocessed machining parameters, equipment characteristic values ​​and part size error data according to the machining batch to form a unified data set.

[0080] Step 2: Perform feature analysis on various types of preprocessed historical data, screen out features related to CNC machining of aviation parts, obtain feature data sets, and analyze the change trends of each feature;

[0081] The implementation method of this step is: statistically analyze the pre-processed historical data, extract the CNC machining feature data of aviation parts from the machining parameters, equipment characteristic values ​​and part size error data, and obtain the feature data set through comprehensive screening of the feature data; wherein, for the machining parameter data, the feature data of the spindle speed, cutting speed, feed speed and cutting depth are extracted, wherein the spindle speed affects the linear speed of contact between the tool and the workpiece, thereby affecting the quality of the machined surface, the change in cutting speed directly affects the cutting force and cutting temperature, thereby affecting the dimensional accuracy of the part, the increase in feed speed will lead to an increase in cutting force, thereby affecting the dimensional stability of the part, and the increase in cutting depth will aggravate the friction and wear between the tool and the workpiece, thereby affecting the dimensional error of the part.

[0082] Among them, for the equipment characteristic value data, the characteristic data of machine tool vibration, spindle current and lubricating oil temperature are extracted. Among them, the magnitude of machine tool vibration directly affects the processing accuracy, especially high-frequency vibration may cause increased surface roughness of parts and increased dimensional errors. The change of spindle current reflects the load condition of the spindle motor and indirectly reflects the change of cutting force during the cutting process. The change of lubricating oil temperature will affect the lubrication effect and the thermal expansion of machine tool components, thereby affecting the dimensional accuracy of the parts.

[0083] Among them, for the part dimensional error data, the characteristic data of dimensional deviation, dimensional deviation change rate, shape deviation and position deviation are extracted. Among them, the dimensional deviation is the deviation between the actual size of the part and the designed size, which directly reflects the size of the dimensional error. The dimensional deviation change rate is the change rate of the dimensional deviation between adjacent workpieces, which reflects the stability of the machining process. The shape deviation is the difference between the shape of the part after machining and the designed shape. The position deviation is the deviation between the position of the part after machining and the designed position.

[0084] Based on each characteristic data of the processing parameters in the characteristic data set, combined with the reference value of each characteristic data of the processing parameters, the process trend index is calculated, and the change trend of the processing parameters within the preset length evaluation period is analyzed; based on each characteristic data of the equipment characteristic value in the characteristic data set, combined with the reference value of each characteristic data of the equipment characteristic value, the mechanical property trend index is calculated, and the change trend of the equipment characteristic value within the preset length evaluation period is analyzed; based on each characteristic data of the part dimensional error in the characteristic data set, combined with the reference value of each characteristic data of the part dimensional error, the part error trend index is calculated, and the change trend of the part dimensional error within the preset length evaluation period is analyzed;

[0085] In a specific implementation, the process trend index is calculated as follows:

[0086]

[0087] Among them, ATI is the process trend index, SP is the current value of the spindle speed, BP is the reference value of the cutting speed, LP is the current value of the feed speed, RP is the reference value of the cutting depth, and α is the coefficient of the cutting depth, which is used to adjust the influence of the cutting depth on the index.

[0088] It should be noted that when ATI is close to 1, it means that the processing parameters are very close to the reference values ​​and the processing process is in a well-controlled state. When ATI is close to 0, it means that the processing parameters are greatly different from the reference values ​​and may need to be adjusted to improve the processing quality. Measure the deviation between spindle speed and cutting speed and combine it with an exponential function Smooth the spindle speed and cutting speed deviations on the final index using the difference Measures the deviation of feed rate and cutting depth. Since the influence of feed rate and cutting depth is not as significant as that of spindle speed and cutting speed, the square difference Reduce the effect of feed rate and cutting depth on index.

[0089] In a specific implementation, the mechanical property trend index is calculated as follows:

[0090]

[0091] Among them, MTI is the mechanical property trend index, V m is the current machine tool vibration value, V 0 is the reference value of machine tool vibration, V max is the maximum allowable value of machine tool vibration, I s is the current spindle current value, I 0 is the reference value of the spindle current, I max is the maximum allowable value of the spindle current, T oil is the current lubricating oil temperature, T 0 is the reference value of lubricating oil temperature, T max It is the maximum allowable value of lubricating oil temperature.

[0092] It should be noted that when MTI is close to 1, it means that the equipment characteristic value is close to its benchmark value and the mechanical performance is in good condition. When MTI is close to 0, it means that the equipment characteristic value deviates from its benchmark value and there is a risk of performance degradation or failure. Measure of the deviation in vibration level, exponential function Effect of the smoothed vibration level deviation on the final index, using square terms for spindle current and lubricant temperature and It measures the degree of deviation from the maximum allowed value and uses the square term to amplify the impact of the deviation;

[0093] In a specific implementation, the calculation method of the part error trend index is:

[0094]

[0095] in,

[0096] Among them, PTI is the part error trend index, F(d, r, f) is the deviation influence function, d s is the current size deviation, d 0 is the reference value of the dimensional deviation, d max is the maximum allowable value of dimensional deviation, r s is the current size deviation change rate, r 0 is the reference value of the dimensional deviation change rate, r max is the maximum allowable value of the dimensional deviation change rate, f s is the current shape deviation, f is the reference value of the shape deviation, and f max is the maximum allowable value of shape deviation, p s is the current position deviation, p 0 is the reference value of position deviation, p max is the maximum allowable value of position deviation.

[0097] It should be noted that when PTI is close to 1, it means that the characteristic value of the part size error is close to its reference value and the part processing quality is in good condition. When PTI is close to 0, it means that the characteristic value of the part size error deviates from its reference value a lot and there may be processing quality problems.

[0098] Step 3: Based on the analysis results of the feature data set, a dimensional error prediction model is constructed in combination with a machine learning algorithm to predict the dimensional error of the CNC machining process of aviation parts;

[0099] The implementation process of this step is as follows: randomly divide the entire feature data set into a training set and a test set to ensure that the training set and the test set are consistent in data distribution to avoid introducing deviations. For the problem of dimensional error prediction, select a linear regression model for model training; use the training set data to learn and adjust the parameters of the selected linear regression model, calculate the model parameters to minimize the error between the predicted value and the actual value, and then build a dimensional error prediction model; use the test set data to evaluate the trained dimensional error prediction model, compare the difference between the predicted value and the actual value, evaluate the accuracy and generalization ability of the dimensional error prediction model, and adjust and optimize the dimensional error prediction model according to the evaluation results to improve the accuracy and generalization ability of the dimensional error prediction model, and then predict the dimensional error of the CNC machining process of aviation parts.

[0100] Step 4, combining the prediction results of the dimensional error prediction model with the relevant data of the feature data set, obtaining the dimensional error trend coefficient, and analyzing the dimensional error change trend of the parts in the production environment;

[0101] In the embodiment of the present invention, whether the production environment is stable and whether adjustment measures need to be taken are judged according to the size and change direction of the dimensional error trend coefficient.

[0102] Step 5: Apply the constructed dimensional error prediction model to the actual production process, monitor the key parameters of the processing process in real time, and predict the dimensional error status of aviation parts processing.

[0103] Step 6: According to the prediction result of the dimensional error state, determine the abnormal situation of the dimensional error, determine whether to issue an early warning signal, and implement corresponding adjustment measures.

[0104] Embodiment 2:

[0105] like Figures 3 to 5 As shown, on the basis of Example 1, in the numerical control machining dimension error prediction method provided in Example 2; preferably, as Figure 3 As shown, in step 4 of embodiment 2, the process of analyzing the variation trend of the dimensional error of parts in the production environment includes:

[0106] Compare the prediction results of the dimensional error prediction model with the actual dimensional error data in the feature data set one by one to ensure that the prediction results and actual data correspond in time series, so as to make accurate analysis, calculate the prediction error, and analyze the distribution and size of the prediction error; to evaluate the accuracy and reliability of the prediction model; combine the prediction results of the dimensional error prediction model with the relevant data of the feature data set, calculate the process trend index, mechanical property trend index and part error trend index, and analyze the changing trends of processing parameters, equipment characteristic values ​​and part dimensional errors; combine the process trend index, mechanical property trend index and part error trend index, Obtain a comprehensive dimensional error trend coefficient, analyze the time series changes of the dimensional error trend coefficient, determine its size and change direction, and clarify the change trend of the component dimensional error in the production environment; perform anomaly detection on the production environment according to the change trend of the dimensional error trend coefficient. If the dimensional error trend coefficient exceeds the normal range, it is judged that there is an abnormality in the production environment, and the stability of the production environment is evaluated. If the dimensional error trend coefficient remains relatively stable within a certain period of time and the change range is within an acceptable range, the production environment is considered to be stable. If the dimensional error trend coefficient fluctuates frequently or exceeds the normal range, adjustment measures need to be taken to optimize the production environment;

[0107] Furthermore, the calculation method of the dimensional error tendency coefficient is:

[0108] Among them, ECI is the dimensional error trend coefficient, ATI is the process trend index, which reflects the changing trend of processing parameters, MTI is the mechanical property trend index, which reflects the changing trend of equipment characteristic values, PTI is the part error trend index, which reflects the changing trend of part dimensional error, ∈ is a very small positive number, close to 0, which is used to avoid division by 0, and the value range of ECI is between 0 and 1.

[0109] It should be noted that when ECI is close to 1, it means that the three trend indexes are close to or equal to 1, that is, the production environment is stable and the dimensional error is controlled within an acceptable range. When ECI is close to 0, it means that at least one trend index is low, and there may be problems with processing parameters, equipment performance or part dimensional errors, which require further analysis and adjustment.

[0110] like Figure 4 As shown, in step 5 of this embodiment 2, the process of predicting the machining dimension error state of aviation parts includes:

[0111] The trained dimensional error prediction model is saved in a loadable format and deployed to the server in the production environment to ensure that the network connection between the server and the production CNC control system is stable. The dimensional error prediction model is integrated in the CNC control system to ensure that the CNC control system can access the model server in real time and send and receive data. Sensors and data acquisition modules are configured in the CNC control system to monitor the key parameter data (such as cutting speed, feed rate, cutting depth, spindle speed, tool wear, etc.) in the processing process in real time. The key parameter data monitored in real time are sent to the data server through the data interface of the CNC control system to ensure the integrity and accuracy of the data and avoid data loss or errors. A data preprocessing module is developed in the data server to perform preprocessing operations such as cleaning, format conversion, and normalization on the collected key parameter data. The preprocessed data should meet the input requirements of the dimensional error prediction model. The preprocessed key parameter data is used as input and input into the dimensional error prediction model. The dimensional error prediction model is called through the interface of the CNC control system, and the input data is passed to the dimensional error prediction model for processing. The dimensional error state prediction results of aviation parts processing are calculated and output according to the input data, and the prediction results are output to the operator through the interface of the CNC control system.

[0112] like Figure 5 As shown, in step 6 of this embodiment 2, the process of issuing the warning signal includes:

[0113] The prediction results of the dimensional error status are obtained from the CNC control system. According to the product quality standards and historical data, the warning threshold of the dimensional error is set. The prediction results are compared with the preset warning threshold to determine whether the prediction results exceed the allowable error range. If the prediction results exceed the allowable error range, it is determined that the dimensional error is abnormal, and then the causes of the abnormality are analyzed, such as improper processing parameter settings, equipment failure, material problems, etc. Combined with the dimensional error trend coefficient and the degree of deviation of the prediction results, the dimensional error abnormality level is divided into slight abnormality, medium abnormality, and severe abnormality, and the corresponding deviation thresholds are matched for different levels of dimensional error abnormality. Different levels of dimensional error abnormality correspond to different warning signals and adjustment measures. According to the level of dimensional error abnormality, the corresponding warning signal rules are set. Slight abnormality issues a yellow warning, medium abnormality issues an orange warning, and severe abnormality issues an orange warning. A red warning is issued, and according to the set warning rules, a warning signal is issued through the interface of the CNC control system to ensure that the operator can notice the warning signal in time and take corresponding countermeasures. According to the prediction results and abnormal causes, the processing parameters of cutting speed, feed rate, and cutting depth are adjusted to optimize the processing process to reduce dimensional errors. Regular inspection and maintenance of CNC equipment are carried out to ensure the accuracy and stability of the equipment and avoid dimensional errors caused by equipment failure. If the prediction results show that the dimensional error is caused by material or tool problems, the material or tool is replaced in time, and materials and tools with stable quality and excellent performance are selected to improve the processing accuracy. The data and feedback in the actual processing process are integrated to optimize the dimensional error prediction model. After the adjustment measures are implemented, the dimensional error status in the processing process is continuously monitored to ensure the effectiveness of the adjustment measures and discover new problems in time.

[0114] Further, the multiple dimension error anomaly levels correspond to multiple deviation thresholds, wherein the deviation threshold includes an upper threshold and a lower threshold;

[0115] Multiple dimension error anomaly levels and multiple deviation thresholds satisfy the following relationship:

[0116] Mildly abnormal ECI L ≤ECI<1;

[0117] Moderately abnormal ECI M ≤ECI <ECI L ;

[0118] Severe abnormality 0 <ECI<ECI M ;

[0119] Among them, ECI is the dimensional error tendency coefficient, ECI L is the lower threshold corresponding to a slight anomaly and the upper threshold corresponding to a moderate anomaly, ECI M is the lower threshold corresponding to moderate anomalies and the upper threshold corresponding to severe anomalies, ECIL =0.9, ECI M =0.7.

[0120] The embodiment of the present invention provides a method for predicting dimensional errors in numerical control machining. Compared with the prior art, the method for predicting dimensional errors in numerical control machining has achieved the following technical progress:

[0121] First, the CNC machining dimensional error prediction method provided by the present invention obtains data during the machining process and uses the dimensional error prediction model for analysis to discover and warn of potential dimensional errors, which not only improves the accuracy of the dimensions of aviation parts, but also significantly enhances the stability of product quality. During the production process, even in the face of slight changes in machining parameters or equipment wear, it can respond quickly and guide operators to adjust machining conditions, thereby avoiding the accumulation and amplification of dimensional errors.

[0122] Second, the CNC machining dimensional error prediction method provided by the present invention not only reduces the waste of raw materials but also reduces production costs by discovering potential problems in advance and avoiding rework and scrapping due to unqualified dimensions. At the same time, it can also promptly discover equipment failures or signs of wear, thereby performing maintenance and replacement in advance, avoiding production interruptions and delays due to equipment failures, and greatly reducing the risks and uncertainties in the production process.

[0123] Although the embodiments disclosed in the present invention are as above, the contents are only embodiments adopted to facilitate understanding of the present invention and are not intended to limit the present invention. Any technician in the field to which the present invention belongs can make any modifications and changes in the form and details of implementation without departing from the spirit and scope disclosed in the present invention, but the patent protection scope of the present invention shall still be subject to the scope defined in the attached claims.

Claims

1. A method for predicting dimensional errors in numerical control machining, characterized in that: include: Step 1, collecting various historical data in the CNC machining process of aviation parts, and preprocessing the collected historical data; Step 2: Perform feature analysis on various types of preprocessed historical data, screen out features related to CNC machining of aviation parts, obtain feature data sets, and analyze the change trends of each feature; Step 3: Based on the analysis results of the feature data set, a dimensional error prediction model is constructed in combination with a machine learning algorithm to predict the dimensional error of the CNC machining process of aviation parts; Step 4, combining the prediction results of the dimensional error prediction model and the characteristic data of the characteristic data set to obtain the dimensional error trend coefficient and analyze the dimensional error change trend of the parts in the production environment; Step 5, applying the constructed dimensional error prediction model to the actual production process, monitoring the key parameters in the processing process in real time, and predicting the dimensional error state of aviation parts processing; Step 6: According to the prediction result of the dimensional error state, determine the abnormal situation of the dimensional error, determine whether to issue an early warning signal, and implement corresponding adjustment measures.

2. A method for predicting dimensional errors in numerical control machining according to claim 1, characterized in that: In step 1, the collection and preprocessing of historical data of the CNC machining process of aviation parts includes: Step 1.1, clarify the data types that need to be collected for CNC machining of aviation parts, including: machining parameters, equipment characteristic values ​​and part dimensional error data; Step 1.2, using the data acquisition function of the CNC machine tool, read the machine tool operation data through the interface of the CNC control system, collect the historical data of the processing parameters and equipment characteristic values ​​of the preset length evaluation period, and use the measuring tool to measure the size of the aviation parts, compare it with the design size, and calculate the part size error data; Step 1.3, preprocessing the collected historical data, the preprocessing operation includes data cleaning and data conversion; In step 1.4, the preprocessed processing parameters, equipment characteristic values, and part dimensional error data are matched and integrated according to the processing batches to form a unified data set.

3. A method for predicting dimensional errors in numerical control machining according to claim 2, characterized in that: In step 2, the analysis process of each characteristic change trend is as follows: Step 2.1, statistically analyzing the preprocessed historical data, extracting the CNC machining feature data of aviation parts from the machining parameters, equipment characteristic values ​​and part size error data, and obtaining the feature data set by comprehensively screening the feature data; Among them, for the processing parameter data, the characteristic data of spindle speed, cutting speed, feed speed and cutting depth are extracted; for the equipment characteristic value data, the characteristic data of machine tool vibration, spindle current and lubricating oil temperature are extracted; for the part size error data, the characteristic data of size deviation, size deviation change rate, shape deviation and position deviation are extracted; Step 2.2, based on each characteristic data of the processing parameters in the characteristic data set and in combination with the reference value of each characteristic data of the processing parameters, the process trend index is calculated to analyze the change trend of the processing parameters within the preset length evaluation time period; Step 2.3, based on each characteristic data of the equipment characteristic value in the characteristic data set and in combination with the reference value of each characteristic data of the equipment characteristic value, the mechanical property trend index is calculated to analyze the change trend of the equipment characteristic value within the preset length evaluation time period; Step 2.4, based on the characteristic data of the part dimensional error in the characteristic data set and combined with the baseline value of each characteristic data of the part dimensional error, calculate the part error trend index and analyze the change trend of the part dimensional error within the preset length evaluation time period.

4. A method for predicting dimensional errors in numerical control machining according to claim 3, characterized in that: The process trend index is calculated as follows: Among them, ATI is the process trend index, SP is the current value of the spindle speed, BP is the reference value of the cutting speed, LP is the current value of the feed speed, RP is the reference value of the cutting depth, and α is the coefficient of the cutting depth, which is used to adjust the influence of the cutting depth on the index; The calculation method of the mechanical property trend index is: Among them, MTI is the mechanical property trend index, V m is the current machine vibration value, V0 is the reference value of machine vibration, V max is the maximum allowable value of machine tool vibration, I s is the current spindle current value, I0 is the reference value of the spindle current, I max is the maximum allowable value of the spindle current, T oil is the current lubricating oil temperature value, T0 is the reference value of lubricating oil temperature, T max is the maximum allowable value of lubricating oil temperature; The calculation method of the part error trend index is: in, Among them, PTI is the part error trend index, F(d, r, f) is the deviation influence function, d s is the current size deviation, d0 is the reference value of the size deviation, d max is the maximum allowable value of dimensional deviation, r s is the current size deviation change rate, r0 is the reference value of the size deviation change rate, r max is the maximum allowable value of the dimensional deviation change rate, f s is the current shape deviation, f is the reference value of the shape deviation, and f max is the maximum allowable value of shape deviation, p s is the current position deviation, p0 is the reference value of the position deviation, and p max is the maximum allowable value of position deviation.

5. A method for predicting dimensional errors in numerical control machining according to claim 4, characterized in that: The construction process of the size error prediction model in step 3 includes: Step 3.1, randomly divide the entire feature data set into a training set and a test set. For the size error prediction problem, a linear regression model is selected for model training; Step 3.2, using the training set data to learn and adjust the parameters of the selected linear regression model, calculate the model parameters to minimize the error between the predicted value and the actual value, and then build a size error prediction model; In step 3.3, the trained dimensional error prediction model is evaluated using the test set data, the difference between the predicted value and the actual value is compared, and the dimensional error prediction model is adjusted and optimized according to the evaluation results, so as to predict the dimensional error of the CNC machining process of aviation parts.

6. A method for predicting dimensional errors in numerical control machining according to claim 5, characterized in that: In step 4, the analysis process of the variation trend of the dimensional error of the parts in the production environment is as follows: Step 4.1, compare the prediction results of the size error prediction model with the actual size error data in the feature data set one by one, calculate the prediction error, and analyze the distribution and size of the prediction error; Step 4.2, combining the prediction results of the dimensional error prediction model and the relevant data of the characteristic data set, calculating the process trend index, mechanical property trend index and part error trend index, and analyzing the change trends of processing parameters, equipment characteristic values ​​and part dimensional errors; Step 4.3, combine the process trend index, mechanical property trend index and part error trend index to obtain a comprehensive dimensional error trend coefficient, analyze the time series change of the dimensional error trend coefficient, determine its size and change direction, and clarify the change trend of the dimensional error of parts in the production environment; Step 4.4, based on the changing trend of the dimension error trend coefficient, perform anomaly detection on the production environment. If the dimension error trend coefficient exceeds the normal range, it is judged that there is an abnormality in the production environment.

7. A method for predicting dimensional errors in numerical control machining according to claim 6, characterized in that: The calculation method of the dimensional error tendency coefficient in step 4.3 is: Among them, ECI is the dimensional error trend coefficient, ATI is the process trend index, MTI is the mechanical property trend index, PTI is the part error trend index, ∈ is a positive number, close to 0, and the value range of ECI is between 0 and 1.

8. A method for predicting dimensional errors in numerical control machining according to claim 7, characterized in that: The process of predicting the machining dimension error state of aviation parts in step 5 includes: Step 5.1, save the trained dimension error prediction model in a loadable format, deploy it to the server in the production environment, and integrate the dimension error prediction model in the CNC control system; Step 5.2, configure sensors and data acquisition modules in the CNC control system to monitor the key parameter data in the processing process in real time, send the key parameter data monitored in real time to the data server through the data interface of the CNC control system, develop a data preprocessing module in the data server, and perform preprocessing operations such as cleaning, format conversion, and normalization on the collected key parameter data; Step 5.3, the preprocessed key parameter data is used as input into the dimensional error prediction model; Step 5.4, call the dimensional error prediction model through the interface of the CNC control system, pass the input data to the dimensional error prediction model for processing, calculate and output the dimensional error state prediction results of aviation parts processing based on the input data, and output the prediction results to the operator through the interface of the CNC control system.

9. A method for predicting dimensional errors in numerical control machining according to claim 8, characterized in that: The process of issuing the warning signal in step 6 is as follows: Step 6.1, obtain the prediction result of the dimensional error state from the CNC control system, set the warning threshold of the dimensional error according to the product quality standard and historical data, compare the prediction result with the preset warning threshold, and determine whether the prediction result exceeds the allowable error range; Step 6.2: If the prediction result exceeds the allowable error range, it is judged as a dimensional error abnormality, and the cause of the abnormality is analyzed; Step 6.3, combining the dimension error trend coefficient and the deviation degree of the prediction result, the dimension error anomaly level is divided into slight anomaly, medium anomaly, and severe anomaly, and the corresponding deviation thresholds are matched for different levels of dimension error anomaly. Different levels of dimension error anomaly correspond to different warning signals and adjustment measures; Step 6.4, according to the level of dimensional error anomaly, set the corresponding warning signal rules, a slight anomaly will issue a yellow warning, a moderate anomaly will issue an orange warning, and a severe anomaly will issue a red warning. According to the set warning rules, a warning signal will be issued through the interface of the CNC control system, and corresponding countermeasures will be taken; Step 6.5, integrate the data and feedback from the actual processing process, optimize the dimensional error prediction model, and continuously monitor the dimensional error status during the processing after implementing the adjustment measures.

10. A method for predicting dimensional errors in numerical control machining according to claim 9, characterized in that: A plurality of the dimension error abnormality levels correspond to a plurality of the deviation thresholds, wherein the deviation thresholds include an upper threshold and a lower threshold; The multiple dimensional error abnormality levels and the multiple deviation thresholds satisfy the following relationship: Mildly abnormal ECI L ≤ECI<1; Moderately abnormal ECI M ≤ECI <ECI L ; Severe abnormality 0 <ECI<ECI M ; Among them, ECI is the dimensional error tendency coefficient, ECI L is the lower threshold corresponding to a slight anomaly and the upper threshold corresponding to a moderate anomaly, ECI M is the lower threshold corresponding to moderate anomalies and the upper threshold corresponding to severe anomalies, ECI L =0.9, ECI M =0.7.

Images