Processing process abnormity monitoring method based on signal consistency analysis
By collecting and analyzing the time and frequency domain characteristics of vibration signals on the machine tool and calculating comprehensive consistency indicators, the model complexity and subjectivity problems of abnormal monitoring of machine tool processing in the existing technology are solved, and more efficient abnormal monitoring is achieved and assessment.
Patent Information
- Application Number
- CN202410942456.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-15
- Publication Date
- 2024-11-08
AI Technical Summary
In the monitoring of abnormal abnormalities in machine tool processing, the existing technology has problems such as many model parameters and difficulty in verification, low applicability and high cost. There are large subjective factors in the evaluation of comprehensive information based on multi-components, and fewer fault samples lead to difficulty in evaluating abnormalities.
By using vibration sensors installed on the machine tool to collect data, extract time-domain and frequency-domain features, draw a three-dimensional curve chart, calculate the feature matrix and index weights, comprehensive consistency indicators are obtained, which are used to judge whether there are abnormalities in the processing process.
It realizes more efficient monitoring of machine tool processing abnormalities under small samples, improves the accuracy and efficiency of signal consistency analysis, reduces subjective factors and reduces costs.
Smart Images

Figure CN118915615A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the application field of machine tool monitoring technology, and in particular to a machining process abnormality monitoring method based on signal consistency analysis. Background Art
[0002] With the continuous improvement of CNC technology and the improvement of the accuracy and stability of CNC machine tools, CNC machine tools have been widely used in aerospace, automobile, wind power, nuclear power, high-tech and other industries. As a key component of the manufacturing industry, CNC machine tools, on the one hand, ensure their safe and stable operation, which can not only create a large amount of material wealth for the national economy, but also produce huge social benefits; on the other hand, if a certain link in the machine tool fails, it will interrupt the production process at the least, and cause a catastrophic accident at the worst. Abnormal monitoring of the machining process is the basis of machine tool body monitoring and tool wear monitoring. Therefore, abnormal monitoring of the machining process of CNC machine tools, by collecting various sensor data on the equipment and using data analysis technology, can timely discover abnormal conditions of the equipment, which has important practical significance for daily maintenance and early safety warning. The abnormal monitoring information of the machining process mainly includes vibration, cutting force, temperature, current, voltage, power signal and machining process of the machine tool operation process. Effective monitoring and processing of operation data can realize the transparency of the machine tool operation process. In turn, it can meet the needs of improving machine tool operation efficiency, tool wear monitoring, machine tool health warning, fault diagnosis, energy saving and consumption reduction.
[0003] In the prior art, the abnormality monitoring strategy of the machine tool processing process mainly includes the evaluation strategy based on the model and the evaluation strategy based on the comprehensive information of multiple components, which specifically include the following technical solutions. (1): By establishing the torque model of the tool spindle motor, the abnormality monitoring function of the machine tool based on the working current is realized. (2): The online monitoring of the rotating machinery spindle fault based on the virtual instrument realizes the abnormality monitoring of the processing process. Specifically, by collecting the speed, vibration, temperature and other signals of the processing process, the feature extraction parameters are used as the model input, and the fuzzy C-means clustering algorithm is combined to realize the fault warning of the machine tool. (3): Based on the state acquisition monitoring and health assessment of typical CNC machine tools, the corresponding data is obtained. Specifically, the vibration signal is used to collect the signal of the key components of the machine tool, the spindle, X-axis bearing, Y-axis nut, and Z-axis bearing, and the weight of each component is determined according to the expert experience. The health status of the machine tool is comprehensively evaluated using these four components. (5): In view of the problem that the health status of the machine tool hydraulic system is difficult to accurately evaluate, the abnormality evaluation is carried out by combining gray clustering with fuzzy comprehensive evaluation. Although the above technologies can evaluate abnormal conditions relatively accurately, it is extremely difficult to develop a reliable physical model because too many parameters are required to establish the model, and the model verification is difficult, the applicability is low, and the cost is high. In addition, when evaluating based on the comprehensive information of multiple components, the combined weighting method is always combined. The combined weighting method is based on expert scoring and has a large number of subjective factors. The traditional similarity evaluation method relies on fault samples. When calculating the similarity, it is necessary to normalize the healthy samples and the faulty samples and convert them into a health index. However, there are fewer faulty samples in machine tools, and it is more difficult to perform abnormal evaluation.
[0004] In actual production, after the design and installation of the machine tool is completed, it is often necessary to judge the accuracy of the assembly, analyze the signals collected from different machine tools under the same processing conditions, and analyze whether there will be abnormal impacts or even failures during the stable operation of the machine tool. When optimizing the installation position of the sensor measuring point, whether the signals collected by sensors at different positions are consistent also requires a signal analysis method. Therefore, based on the above, it is particularly necessary to provide a technology that can monitor the abnormalities of the machine tool processing process in real time. Summary of the invention
[0005] In order to overcome the existing abnormal monitoring technology for machine tool processing processes, due to technical limitations and the drawbacks described in the background technology, the present invention provides a method for collecting data using an installed vibration sensor, extracting time domain and frequency domain features of the collected multiple processing signals, and drawing a three-dimensional curve graph, normalizing the obtained feature matrix to obtain a standard matrix, calculating the feature covariance matrix and indicator weights, and obtaining a comprehensive consistency index for each processing signal, judging whether the processing process is abnormal based on the consistency information, and compared with the existing technology, a processing process abnormality monitoring method based on signal consistency analysis can obtain abnormal monitoring data for machine tool processing processes relatively more effectively.
[0006] The technical solution adopted by the present invention to solve its technical problem is:
[0007] The method for monitoring abnormality in a machining process based on signal consistency analysis is characterized in that it includes the following processes: S1: using a vibration sensor installed on a machine tool workbench to collect vibration signals generated during multiple machine tool tool machining processes; S2: extracting time domain and frequency domain features of the collected vibration signals, and the extracted features include mean, effective value, variance, maximum value, peak-to-peak value, skewness, kurtosis, waveform factor, pulse factor, mean square frequency, center of gravity frequency, and spectrum kurtosis, a total of 12 features, which together constitute a feature set F = {f j},j=1,2,......,12; S3: draw the extracted feature set into a radar chart for comparative analysis; S4: analyze the stability index of the data extracted signal obtained in step S3 during the processing process, slide the window of the original signal, unify the number of windows, and calculate the various feature indicators in each window, so as to avoid the situation where the signal length is different due to different feed speeds, which leads to the situation where the number of sliding windows is not uniform, and it is convenient to compare the features at the same processing position under different processing conditions; S5: draw a three-dimensional line comparison chart for the data obtained in step S4, and compare the features calculated by the signal sliding window; S6: normalize the data feature matrix calculated in step S5 to make it dimensionless and obtain a standard matrix; S7: calculate the correlation coefficient matrix of each feature for the data obtained in step S6; S8: for the data obtained in step S7, calculate the index weight of each feature based on the obtained correlation coefficient matrix, and obtain the index weight W={w j}, j = 1, 2, ..., 12, through this calculation, redundant information is removed and the difference information between each signal is extracted; S9: Based on the data of step S8, the comprehensive consistency evaluation result is calculated to obtain the comprehensive consistency index B = {B i}, i = 1, 2, 3, ... m, where B iIt is the result of the comprehensive consistency index of the i-th processing signal. The comprehensive consistency index reflects the overall difference information of each index between the signals. The smaller the difference between the two values, the better the consistency of the signal. S10: The data obtained in steps S1-S9 is used for the abnormal monitoring system software of the processing process in the PC to monitor and warn the abnormalities in the machining process in real time.
[0008] Furthermore, in step S1, the vibration signal is marked as, D = {X i}, i = 1, 2, ..., N, where X i The signal representing the i-th processing process, a total of N signals are collected.
[0009] Furthermore, in step S6, the calculation formula applied is as follows: Among them, f i,j is the jth feature calculated for the i-th processed signal, F i,j is the result of its normalization, with a total of m processed signals, and m = 128 in this case.
[0010] Furthermore, in step S7, the calculation formula applied is as follows: In the formula, and is the average of the kth and lth standardized features, r k,l is the correlation coefficient between the kth and lth features.
[0011] Furthermore, in step S8, the calculation formula applied is as follows: In the formula represents the amount of information contained in the jth feature, σ j represents the standard deviation of the jth standardized feature, is the average value of the jth standardized feature, w j represents the weight of the jth feature.
[0012] Furthermore, in step S9, the calculation formula applied is as follows: i =[w 1 ,w 2 ,w 1 ,...,w 12 ][F i,1 ,F i,2 ,F i,3 ,...,F i,12 ] T .
[0013] Compared with the prior art, the present invention has the following beneficial effects: in the application of the present invention, it is mainly used for abnormal monitoring of machine tool processing under small samples, and the installed vibration sensor is used to collect data. First, the time domain and frequency domain features of the collected multiple processing signals are extracted, and then the characteristic change curve of the processing process is calculated by sliding the signal, and the obtained characteristic matrix is normalized to obtain a standard matrix; then, the characteristic covariance matrix and the indicator weight are calculated; finally, the comprehensive consistency index of each processing signal is obtained, and whether the processing process is abnormal is judged based on the consistency information. Compared with the prior art, the abnormal monitoring data of the machine tool processing process can be obtained relatively more effectively. BRIEF DESCRIPTION OF THE DRAWINGS
[0014] Figure 1 It is a schematic diagram of the machine tool structure and sensor installation location.
[0015] Figure 2 It is a process framework diagram of the present invention.
[0016] Figure 3 It is a schematic diagram of repeatedly processing signals.
[0017] Figure 4 It is a comparison chart of sliding window indicators.
[0018] Figure 5 It is a single indicator comparison radar chart.
[0019] Figure 6 It is a comparison chart of comprehensive consistency indicators. DETAILED DESCRIPTION
[0020] The present invention takes the same processing process of two machine tools of different models as an example for explanation, and the machine tools are numbered as machine tool 1 and machine tool 2. The structural schematic diagram of the two machine tools is shown in FIG. Figure 1 As shown in Figure 1, the installation position of the vibration sensor used to collect monitoring data is as follows: Figure 1 As shown in the figure, the sampling frequency is 4000Hz. The two machine tools use eight different spindle speeds, and the faster the speed, the faster the feed speed. The speeds are 1000RPM, 1500RPM, 2000RPM, 2500RPM, 3000RPM, 4000RPM, 5000RPM, and 7000RPM. Collect multiple repeated processing signals at each speed and intercept the processing segment signal, such as Figure 3 shown.
[0021] Figure 2 As shown in the figure, the abnormal monitoring method of machining process based on signal consistency analysis, step 1, using the vibration sensor installed on the machine tool workbench, collects the vibration signal D = {X i}, i = 1, 2, ..., N; where X iThe signal representing the i-th processing process is collected in total N; in this embodiment, 8 repeated processing signals are collected in each case, totaling 128 processing process signals (this step provides data for step 2). Step 2, extract the time domain and frequency domain features of the collected vibration signal, and the extracted features include mean, effective value, variance, maximum value, peak-to-peak value, skewness, kurtosis, waveform factor, pulse factor, mean square frequency, center of gravity frequency, and spectrum kurtosis, a total of 12 features, which together constitute the feature set F = {f j}, j = 1, 2, ..., 12 (this step provides data for step 3 and step 6). Step 3, draw the features extracted in step 2 into a radar chart for comparative analysis; Figure 5 As shown, it is a comparison chart of one feature of multiple indicators in the same processing process under 8 rotational speeds of the two machine tools. It can be seen that the change consistency of this indicator is good, and there is no obvious abnormality (the purpose of this step is to calculate the characteristics of each indicator in the entire processing section and make an intuitive comparison). Step four, analyze the stability indicators of the extracted signal during the processing process, slide the window of the original signal, unify the number of windows, and calculate each characteristic indicator in each window, so as to avoid the situation where the signal length is different due to different feed speeds, and then the number of sliding windows is not uniform, which is convenient for comparing the features at the same processing position under different processing conditions (the purpose of this step is to calculate the changes of each indicator in the processing process and provide data for step five). Step five, draw a three-dimensional line comparison chart to compare the features calculated by the signal sliding window; in this embodiment, the number of sliding windows is fixed to 70, such as Figure 4 The figure shows the comparison of the changes of a certain indicator during 8 repeated processing (the purpose of this step is to intuitively reflect the consistency changes in each processing process by drawing the changes in the indicators during the processing. The comprehensive consistency index calculated at the end can be used to reflect whether the machine tool processing has any abnormalities).
[0022] Figure 2 As shown, step six, according to formula (1) and (2), the characteristic matrix of the signal is normalized to make it dimensionless and obtain a standard matrix; where f i,j is the jth feature calculated for the i-th processed signal, F i,j It is the result of normalization, with a total of m processed signals, and in this embodiment, m=128 (the function of this step is to normalize different features, unify the dimensions and magnitudes of different features, and facilitate subsequent comparison and weighting of different features).
[0023]
[0024] Step 7: Calculate the correlation coefficient of each feature according to formula (3) and (4) and obtain the correlation coefficient matrix; where: and is the average of the kth and lth standardized features, r k,l is the correlation coefficient between the kth and lth features (this step is to calculate the correlation between the features and provide a basis for calculating the weights of each feature in step eight).
[0025]
[0026] Step 8: Based on the obtained correlation coefficient matrix, the index weight of each feature is calculated by formula (5) and (6), and the index weight W = {w j},j=1,2,...,12,(where represents the amount of information contained in the jth feature, σ j represents the standard deviation of the jth standardized feature, is the average value of the jth standardized feature, w j Represents the weight of the jth feature). This calculation can remove redundant information to a certain extent and extract the difference information between each signal (the role of this step is to calculate It can reflect the size of the information contained in the jth feature. The larger its value is, the more difference information it contains. The corresponding weight calculated according to formula (6) is also larger. The weight of each feature is calculated through this step, which provides support for the calculation of the next comprehensive index).
[0027]
[0028] Figure 2 As shown, in step nine, the comprehensive consistency evaluation result is calculated by formula (7) to obtain B = {B i}, i = 1, 2, 3, ..., m, where B i is the result of the comprehensive consistency index of the i-th processed signal. The comprehensive consistency index reflects the overall difference information of each index between the signals. The smaller the difference between the two values, the better the consistency of the signal. Figure 6 The figure shows the average comprehensive consistency index of 8 repeated processing signals of the two machine tools at 1000RPM and 1500RPM, indicating that the signal consistency of the two machine tools is good at 1500RPM.
[0029] B i =[w 1 ,w 2 ,w 1 ,...,w 12 ][F i,1 ,F i,2 ,F i,3 ,...,F i,12 ] T (7)
[0030] Step 10: The data obtained from steps 1 to 9 are used in the abnormality monitoring system software of the machining process in the PC to monitor and warn of abnormalities in the machining process in real time.
[0031] The present invention is mainly used for abnormal monitoring of machine tool processing under small samples. Through the above technical scheme, firstly, the vibration sensor installed at the machine tool is used to collect data, and the time domain and frequency domain features of the collected multiple processing signals are extracted; then, the characteristic changes in the processing process are calculated by sliding the signal, and a three-dimensional curve graph is drawn, and the obtained characteristic matrix is normalized to obtain a standard matrix; then, the characteristic covariance matrix and the index weight are calculated; finally, the comprehensive consistency index of each processing signal is obtained, and whether the processing process is abnormal is judged according to the consistency information (the signal consistency is intuitively compared and analyzed). In order to verify the effectiveness of the invented method, experimental verification is carried out by vibration signals collected from the actual machine tool processing process. In the present invention, since the data of different feed speeds can be processed with a unified sliding window number, compared with the case of a fixed sliding window width, the signal length of the same processing process under different feed speeds can be optimized, resulting in the situation that it is difficult to achieve effective comparison of the results, and it is convenient to compare the same processing position under different feed speeds.
[0032] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It is obvious to those skilled in the art that the present invention is limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or basic features of the present invention. Therefore, no matter from which point of view, the embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is defined by the appended claims rather than the above description, and it is intended that all changes that fall within the meaning and scope of the equivalent elements of the claims are included in the present invention.
[0033] In addition, it should be understood that although the present specification is described according to the implementation mode, the implementation mode does not only include an independent technical solution. This narrative method of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in the embodiments can also be appropriately combined to form other implementation modes that can be understood by those skilled in the art.
Claims
1. A method for monitoring abnormalities in a machining process based on signal consistency analysis, characterized in that: The method includes the following steps: S1: using a vibration sensor installed on a machine tool workbench to collect vibration signals generated during multiple machine tool tool processing; S2: extracting time domain and frequency domain features of the collected vibration signals, and the extracted features include mean, effective value, variance, maximum value, peak-to-peak value, skewness, kurtosis, waveform factor, pulse factor, mean square frequency, center of gravity frequency, and spectrum kurtosis, a total of 12 features, which together constitute a feature set F = {f j },j=1,2,......,12; S3: draw the extracted feature set into a radar chart for comparative analysis; S4: analyze the stability index of the data extracted signal obtained in step S3 during the processing process, slide the window of the original signal, unify the number of windows, and calculate the various feature indicators in each window, so as to avoid the situation where the signal length is different due to different feed speeds, which leads to the situation where the number of sliding windows is not uniform, and it is convenient to compare the features at the same processing position under different processing conditions; S5: draw a three-dimensional line comparison chart for the data obtained in step S4, and compare the features calculated by the signal sliding window; S6: normalize the data feature matrix calculated in step S5 to make it dimensionless and obtain a standard matrix; S7: calculate the correlation coefficient matrix of each feature for the data obtained in step S6; S8: for the data obtained in step S7, calculate the index weight of each feature based on the obtained correlation coefficient matrix, and obtain the index weight W={w j }, j = 1, 2, ..., 12, through this calculation, redundant information is removed and the difference information between each signal is extracted; S9: Based on the data of step S8, the comprehensive consistency evaluation result is calculated to obtain the comprehensive consistency index B = {B i }, i = 1, 2, 3, ... m, where B i It is the result of the comprehensive consistency index of the i-th processing signal. The comprehensive consistency index reflects the overall difference information of each index between the signals. The smaller the difference between the two values, the better the consistency of the signal. S10: The data obtained in steps S1-S9 is used for the abnormal monitoring system software of the processing process in the PC to monitor and warn the abnormalities in the machining process in real time.
2. The method for monitoring abnormality in a machining process based on signal consistency analysis according to claim 1, characterized in that: In step S1, the vibration signal is marked as, D = {X i }, i = 1, 2, ..., N, where X i The signal representing the i-th processing process, a total of N signals are collected.
3. The method for monitoring abnormality in a machining process based on signal consistency analysis according to claim 1, characterized in that: In step S6, the calculation formula applied is as follows, Among them, f i,j is the jth feature calculated for the i-th processed signal, F i,j is the result of its normalization, with a total of m processed signals, and m = 128 in this case.
4. The method for monitoring abnormality in a machining process based on signal consistency analysis according to claim 1, characterized in that: In step S7, the calculation formula applied is as follows: In the formula, and is the average of the kth and lth standardized features, r k,l is the correlation coefficient between the kth and lth features.
5. The method for monitoring abnormality in a machining process based on signal consistency analysis according to claim 1, characterized in that: In step S8, the calculation formula applied is as follows, In the formula represents the amount of information contained in the jth feature, σ j represents the standard deviation of the jth standardized feature, is the average value of the jth standardized feature, w j represents the weight of the jth feature.
6. The method for monitoring abnormality in a machining process based on signal consistency analysis according to claim 1, characterized in that: In step S9, the calculation formula used is as follows: i =[w1,w2,w1,...,w 12 ][F i,1 ,F i,2 ,F i,3 ,...,F i,12 ] T .
Citation Information
Cited By
Machine tool health degree ranking feature analysis method and system based on acoustic emission technology
CN121919524A
Underwater propulsion system assembly consistency multi-dimension quantitative evaluation method
CN122689128A
