Method for identifying and detecting abnormal machining of screw tap based on smoothing prior method

Through the tap processing abnormality identification and detection method based on the smooth prior method, the problems of complex signal acquisition and noise interference in the actual production environment are solved, and the accurate identification and detection of tap faults is realized, which improves the production efficiency and signal processing stability.

CN119089236APending Publication Date: 2024-12-06YOUJI TECH (SHANGHAI) CO LTD

Patent Information

Application Number
CN202410579905.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-05-11
Publication Date
2024-12-06

AI Technical Summary

Technical Problem

The prior art is difficult to effectively identify abnormal situations during tap processing in actual production environments, mainly due to complex signal acquisition, large noise interference, and traditional methods are difficult to accurately describe tap processing characteristics, resulting in low accuracy and efficiency of fault detection.

Method used

The tap processing abnormality recognition detection method based on the smooth prior method is adopted, and the real-time vibration signal is collected through the acceleration sensor, and the nonlinear trend of the signal is extracted using the sliding window envelope correlation algorithm and the smooth prior method, and the signal is intercepted adaptively to reduce the noise influence. Achieve accurate judgment of the tap processing status.

Benefits of technology

It improves the accuracy and efficiency of fault detection, can effectively identify tap faults in actual production, reduce production costs, provide time information on signal changes, and enhances the stability and applicability of signal processing.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119089236A_ABST
    Figure CN119089236A_ABST
Patent Text Reader

Abstract

A screw tap machining anomaly recognition and detection method based on a smoothing prior method comprises the following steps that S1, vibration signals of a screw tap are collected; s2, single-section machining process signals are intercepted and aligned from the initial vibration signals; s3, obtaining a trend signal # imgabs0 # of the processing signal and a trend signal # imgabs1 # of the reference signal; S4, obtaining a plurality of groups of residual signals; and S5, the obtained data is applied to a screw tap software judgment system, and judgment of the machining state of the screw tap is achieved by delimiting a threshold value of the kurtosis of the obtained residual signals. According to the invention, the acquisition signal truncation process can be adjusted according to the signal characteristics, the characteristics and modes of fault signals can be captured more accurately, the signals are filtered in the time domain, the screw tap processing change process is extracted, the expression of the signals on key time nodes is displayed more visually, the influence of noise on the signal processing result is reduced, and the processing accuracy is improved. Time information of signal change can be provided for subsequent result analysis, and favorable technical support is provided for effective identification of screw tap faults.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of drilling tool fault diagnosis, in particular to a method for identifying and detecting abnormalities in tap machining based on a smoothing priori method. Background Art

[0002] my country is making great strides from "Made in China" to "Smart Manufacturing in China". In intelligent manufacturing systems, signal processing plays an important role as a powerful tool for mechanical equipment fault diagnosis. By collecting real-time signals of mechanical equipment operation, such as vibration signals, acoustic emission signals, etc., and analyzing them using signal processing technology under the action of relevant software units, it is possible to identify possible faults in mechanical equipment production, such as bearing wear, gear tooth breakage, tool edge breakage, etc. This mechanical equipment fault diagnosis method based on signal processing has the advantages of high efficiency, accuracy, and strong robustness. It can help companies detect equipment problems in a timely manner, avoid production line shutdowns and equipment damage, thereby improving production efficiency and reducing maintenance costs. Therefore, the role of signal processing methods in promoting the development of intelligent manufacturing systems cannot be ignored.

[0003] As the teeth of industrial mother machines, tool systems are one of the most important core components of various machine tools. Their health status directly affects processing quality, machine tool life and production efficiency. Therefore, timely and accurate understanding of the operating status of tools is of great significance to extending tool life and ensuring safe and stable operation of machine tools. As a tool for hole processing of various metal and non-metal materials, taps are an indispensable part of the tool system. The existing technology has carried out a lot of research on the health status monitoring method of taps, mainly including the following technical solutions. (1) Extract signal features through variational mode decomposition and continuous wavelet transform, and then combine convolutional neural network to perform health diagnosis of tools; (2) After decomposing acceleration and force signals into multiple modal components through empirical mode decomposition, select effective components to recombine new signals, and then extract features from the signals and use support vector machine to judge the degree of tool wear; (3) Reconstruct the signal through wavelet packet decomposition, extract the signal feature vector, use singular value decomposition to select the singular value of the signal as the feature vector again, and finally combine the obtained feature vectors into one feature vector and input it into the support vector machine for tool wear identification. Although the existing technical solutions have made certain progress in the research methods of health status monitoring of tools including taps, their application is generally limited to experimental environments and cannot be effectively applied to actual production. In actual production sites, the processing flow of taps is complex, the processing environment and working conditions have great interference, and the complete signal analysis of tap processing is difficult. Specifically, on the one hand, during the signal acquisition process, due to the short processing time and complex processing actions of taps, it is difficult for on-site workers to intercept the processing signals of the complete cycle of taps, which greatly increases the difficulty of the subsequent software system signal analysis of the separated signals; on the other hand, traditional time domain and frequency domain signal indicators are difficult to accurately describe the overall processing characteristics of taps, and the decomposed signals obtained by traditional frequency domain signal decomposition methods have poor interpretability for tap processing, and require the pre-assumption of the trend item type of the signal, the selection of basis functions or the number of layers of the decomposition results. Therefore, how to effectively and efficiently intercept signals for tap tools with complex processing actions, and accurately identify the important nodes and change trends of their responses during the processing process, has very important academic research significance and practical application value for the accurate identification of tap faults by related software systems. Summary of the invention

[0004] In order to overcome the drawback of the prior art that, due to technical limitations, it is impossible to effectively identify abnormalities in tap processing, the present invention, under the joint action of relevant technical solutions, on the one hand, solves the problem of artificially or mechanically performing fixed-length signal collection during the signal collection process when the tap is working, so that the truncation process of the collected signal can be adjusted according to the signal characteristics, and the characteristics and patterns of the fault signal can be captured more accurately, thereby improving the accuracy of fault detection. On the other hand, the signal is filtered in the time domain to extract the change process of tap processing, which more intuitively displays the performance of the signal at key time nodes, reduces the impact of noise on the signal processing results, and can provide time information of signal changes for subsequent result analysis, which provides favorable technical support for the effective identification of tap faults. It is a tap processing abnormality identification and detection method based on the smoothing prior method.

[0005] The technical solution adopted by the present invention to solve its technical problem is:

[0006] A method for detecting and identifying abnormal tap machining based on a smoothing prior method, characterized in that it includes the following processes: S1: using an acceleration sensor installed on a machine tool turret to collect real-time vibration signals X and Y of the tap during normal machining and abnormal machining, wherein X is the entire normal machining signal and Y is the entire abnormal machining signal; S2: manually intercepting a complete normal machining process of the tap as a reference signal, denoted as RS, and then calculating the envelope correlation coefficient by using the reference signal and the entire real-time signal collected in step 1 in a moving sliding window manner, then using the number of machining times n performed during the signal acquisition process as prior knowledge, dividing the obtained envelope spectrum coefficients into equal intervals, finding the maximum point positions ind1, ind2...indn in each segment as the initial position of signal interception, and finally intercepting a signal of the same length as the reference signal, thereby completing the purpose of intercepting and aligning the single-segment machining process signal from the initial vibration signal; S3: multiple single-segment machining signal data sets X intercepted in step S2 n As well as the reference signal RS, the nonlinear trend of the signal is extracted using the smoothing priori method to obtain the trend signal of the processed signal And the trend signal of the benchmark signal S4: Perform residual processing on the trend signal of the reference signal obtained in step S3, together with the trend signals of other normal processing signals and the trend signals of abnormal processing signals, to obtain multiple groups of residual signals; S5: Apply the data obtained in steps S1-S4 to the tap software judgment system, and determine the tap processing status by setting a threshold for the kurtosis of the obtained residual signal.

[0007] Furthermore, in step S1, the real-time vibration signal of each processing process needs to be collected for a certain length of time to ensure that multiple processing processes are included.

[0008] Furthermore, in step S2, the entire vibration signal collected in step S1 is firstly analyzed using the formula The Hilbert transform of the complex signal is obtained, and the formula is used Let the original signal be the real part and the complex signal be the imaginary part, perform envelope demodulation calculation to obtain the envelope signal of the original signal, and then use the formula X i =H X (i:len+i-1) is used to perform moving window interception, and the formula is used The intercepted signal and the reference signal are cross-correlated, where X i is the signal intercepted from the envelope signal of the original signal starting from the i-th point, H X is the envelope signal of the X signal, H RS is the envelope signal of the reference signal, corr i is the mutual correlation value calculated between the intercepted signal starting from the ith point and the reference signal;

[0009] Furthermore, in step S3, for the reference signal, the reference signal is set to include two parts, such as the formula RS=RS stat +RS trend As shown, where RS stat is the stationary term of the reference signal RS, RS trend is a nonlinear trend term, where RS trend The formula RS trend =Hθ+v, where H is the observation matrix, θ is the regression parameter, and v is the observation error. The nonlinear trend extraction of the signal is converted into the problem of optimizing the regression parameter θ value through the expression of the formula.

[0010] Furthermore, in step S3, the regularized least squares method is used to estimate the regression parameters, and the specific formula used is as follows: In the formula, λ is the regularization parameter, D d is the discrete form of the d-th order differential operator, That is the nonlinear trend signal of the required reference signal.

[0011] Furthermore, in S5, when the kurtosis exceeds the defined threshold, it indicates that the machining state of the tap has changed. A two-way feedback is formed between the tool state at each shutdown and the kurtosis threshold, thereby achieving accurate judgment of the machining state of the tap.

[0012] Compared with the prior art, the present invention has the following beneficial effects: (1) The present invention can adaptively intercept a complete single-segment processing signal from the entire collected signal, and use a smoothing prior method to extract a nonlinear trend signal, thereby achieving the purpose of judging the processing state of the tap. (2) The sliding window envelope correlation algorithm used in the present invention calculates the correlation between signals according to the overall characteristics of the signal, which can better solve the noise and interference problems in the actual signal, and the algorithm complexity is relatively low, so it is more suitable for signals collected in real processing, saving the time cost of the production process and laying a solid foundation for subsequent signal processing; (3) The smoothing prior method used in the present invention can effectively deal with the problem of abnormal values ​​in the signal, making the extraction of key time nodes in the tap processing process more stable and reliable, and can also adjust parameters according to specific signal characteristics. The algorithm has strong applicability. In addition, the algorithm has low complexity, which provides a new idea for realizing the nonlinear trend extraction of real-time signals. When applied to actual production, it can provide favorable technical support for effectively identifying tap faults. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0014] Figure 1 It is a structural schematic diagram of a tapping machine tool of the present invention.

[0015] Figure 2 It is a schematic diagram of the normal state of the tap and the damaged state of the tap of the present invention.

[0016] Figure 3 It is a flowchart of the present invention.

[0017] Figure 4 It is a schematic diagram of a whole vibration signal including five sections of complete processing according to the present invention.

[0018] Figure 5 It is a schematic diagram of the whole-segment collected processing vibration signal and the reference signal of the present invention.

[0019] Figure 6 It is a schematic diagram of the sliding window correlation coefficient between the reference signal and the entire signal of the present invention.

[0020] Figure 7 It is a schematic diagram of the vibration signal intercepted in the normal processing state of the present invention.

[0021] Figure 8 It is a schematic diagram of the vibration signal intercepted in the damaged processing state of the present invention.

[0022] Fig. 9 It is a trend signal schematic diagram of the normal processing vibration signal of the present invention.

[0023] Fig.10It is a trend signal schematic diagram of the damage processing signal of the present invention.

[0024] Fig.11 It is a trend signal schematic diagram of the reference signal of the present invention.

[0025] Fig.12 It is a signal schematic diagram of residual processing between normal processing of the present invention.

[0026] Fig.13 It is a schematic diagram of signal comparison obtained by residual processing of the present invention. DETAILED DESCRIPTION

[0027] Figure 3 As shown in FIG. 1 , a method for detecting and identifying abnormal tap machining based on a smooth prior method is shown. Step 1: using an acceleration sensor installed on a tool holder, collects real-time vibration signals X and Y of the tap during normal machining and abnormal machining, where X is the entire normal machining signal and Y is the entire abnormal machining signal; the specific sampling frequency is 4000 Hz, the tap speed is 3500 rpm, and the real-time vibration signal of each machining process of the tap needs to be collected for a certain length of time to ensure that multiple machining processes are included; the tapping machine tool structure is as follows: Figure 1 As shown, the normal tap and the tap damage state are as follows Figure 2 As shown; this step provides raw data for subsequent processing steps.

[0028] Figure 3 As shown, step 2: first, for the entire vibration signal collected in step 1, use the Hilbert transform of formula (1) to obtain a complex signal, and use formula (2) to make the original signal as the real part and the complex signal as the imaginary part, perform envelope demodulation calculation, and then obtain the envelope signal of the original signal.

[0029]

[0030]

[0031] Then, the envelope signal is intercepted by a moving sliding window using formula (3), and the intercepted signal is cross-correlated with the reference signal using formula (4).

[0032] X i = H X (i:len+i-1) (3)

[0033]

[0034] Among them, X i is the signal intercepted from the envelope signal of the original signal starting from the i-th point, H X is the envelope signal of the X signal, H RSis the envelope signal of the reference signal, corr i is the correlation value calculated between the intercepted signal starting from the i-th point and the reference signal. Then, the envelope spectrum coefficients obtained are divided into equal intervals using the number of processing times n performed during the signal acquisition process as prior knowledge, and the positions of the maximum points ind1, ind2…indn in each segment are found as the initial positions for signal interception. Finally, a signal of the same length as the reference signal is intercepted to achieve the purpose of intercepting and aligning a single-segment processing process signal from the initial vibration signal. The entire segment of the collected processing vibration signal and the reference signal are as follows: Figure 4 , 5 The sliding window correlation coefficient between the reference signal and the entire signal is shown as Figure 6 The signals intercepted under different processing conditions are shown in Figure 7 , 8 The function of this step is to adaptively truncate the entire segment of data in step 1 according to the working conditions, and provide aligned single-segment complete processing data for the subsequent steps.

[0035] Figure 3 As shown, step three: multiple single-segment processing signal data sets X intercepted in step two n As well as the reference signal RS, the nonlinear trend of the signal is extracted using the smoothing priori method to obtain the trend signal of the processed signal And the trend signal of the benchmark signal Specifically, according to the trend signals of different processing states obtained by the prior method, the trend signals of the benchmark are as follows: Fig. 9 , 10 , as shown in 11. Taking the reference signal as an example, it is assumed that the reference signal includes two parts, as shown in formula (5):

[0036] RS=RS stat +RS trend (5)

[0037] In the formula, RS stat is the stationary term of the reference signal RS, RS trend is a nonlinear trend term. trend It can be expressed by formula (6):

[0038] RS trend =Hθ+v (6)

[0039] Where H is the observation matrix, θ is the regression parameter, and v is the observation error. By expressing in formula (6), the nonlinear trend extraction of the signal is converted into the problem of optimizing the value of the regression parameter θ. In the present invention, the regularized least squares method is used to estimate the regression parameter, as shown in formulas (7) and (8):

[0040]

[0041]

[0042] Where λ is the regularization parameter; D d It is the discrete form expression of the d-th order differential operator; This is the nonlinear trend signal of the desired reference signal. Based on the data intercepted in step 2, this step further extracts the nonlinear trend of the tap signal to provide a nonlinear trend signal for the subsequent steps.

[0043] Figure 3 As shown, step 4: perform residual processing on the trend signal of the reference signal and the trend signals of other normal processing signals and the trend signals of abnormal processing signals to obtain multiple groups of residual signals. Fig.12 , 13 As shown; this step provides a new indicator for monitoring the tap processing process after the above steps, so as to achieve the purpose of tool impact protection. Step 5: The data obtained in steps S1-S4 are applied to the tap software judgment system. The kurtosis threshold of the residual signal obtained is set according to the vibration signal of the normal tap processing. When the kurtosis exceeds the set threshold, it means that the processing state of the tap has changed. The machine is immediately shut down on site and nearby professionals are called for investigation and verification; the tool state and the kurtosis threshold are used to form a two-way feedback at each shutdown to achieve accurate judgment of the tap processing state.

[0044] The present invention solves the problem of artificially or mechanically collecting signals for a fixed time during the signal collection process when the tap is working through the combined effect of all the above technical solutions. It can adaptively intercept a complete single-segment processing signal from the entire collected signal, and use a smoothing prior method to extract a nonlinear trend signal, so as to more accurately capture the characteristics and patterns of the fault signal and more intuitively display the performance of the signal at key time nodes. The sliding window envelope correlation algorithm used specifically calculates the correlation between signals according to the overall characteristics of the signals, which can better solve the noise and interference problems in the actual signals and reduce the impact of noise on the signal processing results. The algorithm complexity is relatively low, and it can provide the time information of signal changes for the subsequent result analysis to achieve the purpose of judging the tap processing status, thereby improving the accuracy of fault detection. Therefore, it is more suitable for signals collected in real processing, saving the time cost of the production process and laying a solid foundation for subsequent signal processing; the smoothing prior method used can effectively deal with the problem of outliers in the signal, making the extraction of key time nodes in the tap processing process more stable and reliable; and the parameters can also be adjusted according to the specific signal characteristics, and the algorithm has strong applicability; in addition, the algorithm has low complexity, which provides a new idea for realizing the nonlinear trend extraction of real-time signals. When applied to actual production, it can provide favorable technical support for the effective identification of tap faults.

[0045] It will be apparent to those skilled in the art that the present invention is limited to the details of the exemplary embodiments described above and that the present invention can be implemented in other specific forms without departing from the spirit or essential features of the present invention. Therefore, the embodiments should be considered exemplary and non-restrictive in all respects, and the scope of the present invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations that fall within the meaning and scope of the equivalent elements of the claims be included in the present invention.

[0046] 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 detecting abnormality in tap machining based on a smooth prior method, characterized in that: The method comprises the following processes: S1: using an acceleration sensor installed on a machine tool turret, collecting real-time vibration signals X and Y of a tap in a normal machining process and an abnormal machining process, wherein X is a whole section of a normal machining signal and Y is a whole section of an abnormal machining signal; S2: manually intercepting a whole section of a normal machining process of a tap as a reference signal, recorded as RS, and then calculating an envelope correlation coefficient by using a moving sliding window method with the reference signal and the whole section of the real-time signal collected in step 1, and then using the number of machining times n performed in the signal acquisition process as prior knowledge, dividing the obtained envelope spectrum coefficients into equal intervals, finding the maximum point positions ind1, ind2, ..., indn in each section as the initial positions of signal interception, and finally intercepting a signal of the same length as the reference signal, thereby completing the purpose of intercepting and aligning a single-segment machining process signal from the initial vibration signal; S3: performing a plurality of single-segment machining signal data sets X intercepted in step S2 n As well as the reference signal RS, the nonlinear trend of the signal is extracted using the smoothing priori method to obtain the trend signal of the processed signal And the trend signal of the benchmark signal S4: Perform residual processing on the trend signal of the reference signal obtained in step S3, together with the trend signals of other normal processing signals and the trend signals of abnormal processing signals, to obtain multiple groups of residual signals; S5: Apply the data obtained in steps S1-S4 to the tap software judgment system, and determine the tap processing status by setting a threshold for the kurtosis of the obtained residual signal.

2. The method for detecting and identifying abnormality in tap machining based on the smoothing prior method according to claim 1, characterized in that: In step S1, the real-time vibration signal of each processing process needs to be collected for a certain length of time to ensure that multiple processing processes are included.

3. The method for detecting and identifying abnormality in tap machining based on the smoothing prior method according to claim 1, characterized in that: In step S2, firstly, the entire vibration signal collected in step S1 is analyzed using the formula The Hilbert transform of the complex signal is obtained, and the formula is used Let the original signal be the real part and the complex signal be the imaginary part, perform envelope demodulation calculation to obtain the envelope signal of the original signal, and then use the formula X i =H X (i:len+i-1) is used to perform moving window interception, and the formula is used The intercepted signal and the reference signal are cross-correlated, where X i is the signal intercepted from the envelope signal of the original signal starting from the i-th point, H X is the envelope signal of the X signal, H RS is the envelope signal of the reference signal, corr i The mutual correlation value calculated between the intercepted signal starting from the i-th point and the reference signal.

4. The method for detecting abnormality in tap machining based on the smoothing prior method according to claim 1, characterized in that: In step S3, for the reference signal, the reference signal is set to include two parts, such as the formula RS = RS stat +RS trend As shown in the formula, RS stat is the stationary term of the reference signal RS, RS trend is a nonlinear trend term, where RS trend The formula RS trend =Hθ+v, where H is the observation matrix, θ is the regression parameter, and v is the observation error. The nonlinear trend extraction of the signal is converted into the problem of optimizing the regression parameter θ value through the expression of the formula.

5. The method for detecting and identifying abnormality in tap machining based on the smoothing prior method according to claim 4, characterized in that: In step S3, the regularized least squares method is used to estimate the regression parameters. The specific formula used is as follows: and In the formula, λ is the regularization parameter, D d is the discrete form of the d-th order differential operator, That is the nonlinear trend signal of the required reference signal.

6. The method for detecting and identifying abnormality in tap machining based on the smoothing prior method according to claim 1, characterized in that: In S5, when the kurtosis exceeds the designated threshold, it means that the processing state of the tap has changed. The tool state at each shutdown and the kurtosis threshold form a two-way feedback, thereby achieving accurate judgment of the tap processing state.

Citation Information

Patent Citations

  • Machine tool screw error signal adaption trend detection method

    CN102059585A

  • Tool damage online and in-situ detection system in clean cutting environment and method

    CN110340733A

  • Physiological electric signal entropy analysis method based on de-trending terms

    CN110490257A

  • Bearing early fault diagnosis method based on EDM and spectral kurtosis

    CN111521400A

  • Safety apparatus for faucet and electric range comprising the same

    KR102903990B1

Cited By

  • Thread precision monitoring method and system for pipe fitting machined by automatic threading machine

    CN121199253A