Mechanical turntable damage monitoring method and system based on combination of acoustic emission technology and SVM (Support Vector Machine) damage classification

Abstract

The invention discloses a mechanical turntable damage monitoring method and system based on combination of an acoustic emission technology and SVM damage classification, and belongs to the technical field of mechanical indexing turntable damage detection.The method comprises the steps that an acoustic emission sensor is installed on a mechanical turntable to be detected; acquiring an acoustic emission waveform acquired by an acoustic emission sensor in real time, and extracting an AE event based on the original waveform; calculating acoustic emission characteristic parameters of each AE event, and constructing a data set; aE signals of different damages in different areas of the mechanical turntable are collected, and acoustic emission characteristic parameters of AE events are jointly used as a training set; taking the characteristic parameters corresponding to the AE signals of different damages of the same area detected subsequently as a test set; inputting the training set and the test set into an SVM (Support Vector Machine) to obtain a result; and monitoring the damage condition of the turntable by using the accuracy of the SVM according to the result of the SVM. According to the invention, damage dynamic monitoring can be carried out on the mechanical indexing table in operation, and fault prevention can be carried out in advance.

Description

Technical Field

[0001] The present invention relates to the technical field of damage detection of mechanical indexing turntables, and specifically to a mechanical turntable damage monitoring method and system based on acoustic emission technology combined with SVM damage classification. Background Technique

[0002] During long-term continuous operation of mechanical turntables on industrial production lines and due to inadequate daily maintenance, the main shaft is prone to wear and even fracture. Therefore, in order to avoid the problem that the main shaft of the mechanical turntable affects product quality during industrial production due to the above situations, and even causes major safety accidents due to severe damage, it is of great significance to monitor the damage of the main shaft of the mechanical turntable.

[0003] Traditional detection methods for rotating machinery mainly include vibration detection and temperature detection. Among them, the vibration detection method analyzes and processes the vibration signals of the equipment to extract the normal and fault characteristics of the signals. However, vibration detection requires installing sensors on the components to be inspected, is suitable for detecting known fault types, and is not sensitive enough to early material damage. The temperature-based detection method monitors the temperature changes of the equipment. When the equipment suddenly produces abnormal temperature changes during operation, it means that the equipment may have a fault. This detection method is only applicable to some fault types and is less sensitive to certain mechanical problems. Both of these methods are insensitive to equipment faults and cannot effectively solve the problem of online detection. Therefore, a non-destructive detection method that can dynamically detect mechanical indexing turntables is needed.

[0004] Acoustic Emission (AE), also known as stress wave emission, is a phenomenon in which materials are deformed or fractured under the action of external or internal forces, and energy is released in the form of elastic waves. The AE technology can not only dynamically monitor the stress changes inside and on the surface of materials, but also identify early damage to various materials such as metals, rocks, and composite materials. In recent years, the application of acoustic emission technology in the damage detection of industrial equipment such as bearings, speed reducers, and rotating machinery has become more and more extensive. Summary of the Invention

[0005] The technical solution of the present invention provides a solution significantly different from the prior art for the technical problem of the overly single solution of the prior art. It mainly provides a mechanical turntable damage monitoring method and system based on acoustic emission technology combined with SVM damage classification to solve the technical problem of the inability of the traditional detection method to achieve online dynamic monitoring proposed in the above background technique.

[0006] The technical solution adopted by the present invention to solve the above technical problems is as follows:

[0007] A mechanical turntable damage monitoring method based on acoustic emission technology combined with SVM damage classification includes the following steps:

[0008] S1. Install an acoustic emission sensor on the mechanical turntable to be detected;

[0009] S2. Obtain the acoustic emission waveform collected by the acoustic emission sensor in real time, and extract AE events based on the original waveform;

[0010] S3. Calculate the acoustic emission characteristic parameters of each AE event and construct a data set;

[0011] S4. Collect the AE signals of different damages in different regions of the mechanical turntable, calculate the acoustic emission characteristic parameters of each AE event, and jointly form a training set for different damages;

[0012] S5. Calculate the acoustic emission characteristic parameters corresponding to the AE signals of different damages in the same region detected subsequently, and use them as a test set;

[0013] S6. Input the above training set and test set into the SVM to obtain the result;

[0014] S7. According to the result of the SVM, use the accuracy rate of the SVM to monitor the damage condition of the turntable.

[0015] Further, in step S2, AE events are identified based on a preset threshold voltage and duration.

[0016] Further, in step S2, first screen the data points in the original waveform that exceed the threshold voltage, and determine the positions of the data points where the interval between two adjacent data points is greater than the preset duration; then record the start and end points of each AE event and count the duration of the AE event.

[0017] Further, in step S3, the acoustic emission characteristic parameters include amplitude, ring count, margin, kurtosis, and center frequency.

[0018] Further, in step S3, the calculation formula for the acoustic emission characteristic parameters is:

[0019] AM = max{|x i |}

[0020]

[0021] Mgn = X max / X R

[0022]

[0023] Among them, X i represents the time-domain voltage signal, AM represents the amplitude, X R represents the root mean square amplitude, Mgn represents the margin index, X maxThe maximum value of the signal is denoted as, the standard deviation is denoted as σ, the mean value of the signal is denoted as u, the kurtosis index is denoted as Kurt, and the center frequency is denoted as FC. f represents the frequency, and P(f) represents the power spectral density of the signal.

[0024] Further, in step S3, for the i-th AE event, the data set is X i =[AM i , RC i , Kurt i , Mgn i , FC i , where i = 1, 2, …, m; among them, AM represents the amplitude, RC represents the ring count, Mgn represents the margin, Kurt represents the kurtosis, FC represents the center frequency, and m represents the total number of extracted AE events.

[0025] Further, in step S7, count the number of samples that should be in normal condition but are predicted as slightly damaged and the number of samples that should be slightly damaged but are predicted as severely damaged, and combine with the total number of samples in the test set to calculate the accuracy rate.

[0026] The present invention also provides a mechanical turntable damage monitoring system based on acoustic emission technology combined with SVM damage classification, which is used to implement the steps of the above-mentioned mechanical turntable damage monitoring method based on acoustic emission technology combined with SVM damage classification. The damage monitoring system includes an acoustic emission signal acquisition module, an acoustic emission signal processing module, and a damage monitoring module; where:

[0027] The acoustic emission acquisition module includes an acoustic emission sensor, which is used to collect the AE signals released during the operation of the mechanical turntable to be detected and convert the sound signals into electrical signals;

[0028] The acoustic emission signal processing module includes a preamplifier and an acoustic emission instrument. The output end of the acoustic emission sensor is connected to the input end of the preamplifier through a signal cable; the acoustic emission instrument is used to filter the signal, perform analog-to-digital conversion, and generate a waveform file for transmission to the host;

[0029] The damage monitoring module includes a host, which is used to obtain the original acoustic emission waveform file and perform monitoring according to the steps of the mechanical turntable damage monitoring method based on acoustic emission technology combined with SVM damage classification.

[0030] Further, the damage monitoring module is also provided with a signal lamp alarm device, which is connected to the host and displays different damage degrees of the mechanical turntable to be detected through different states of the signal lamp.

[0031] Further, the acoustic emission sensor is a piezoelectric ceramic sensor, and the bandwidth frequency range is 50 kHz - 400 kHz, and the resonance frequency is 150 kHz; the amplification factor of the preamplifier is 40 dB.

[0032] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0033] (1) Based on the acoustic emission technology, the present invention collects AE signals of different damages in different regions of the mechanical turntable, takes the corresponding acoustic emission characteristic parameters as the training set, and takes the acoustic emission characteristic parameters corresponding to the AE signals of different damages at each subsequent detection as the test set. Then, the training set and the test set are input into the SVM. Through the SVM classification results, the accuracy of the SVM classification can be obtained by comparing the predicted value and the true value. When the accuracy is significantly reduced, it indicates that the damage of the mechanical turntable is relatively serious and timely maintenance is required. Compared with the traditional method, the present invention can perform dynamic damage monitoring on the running mechanical indexing turntable online, timely monitor the time and degree of equipment damage, and thus can prevent faults in advance.

[0034] (2) The system provided by the present invention is simply deployed and has a wide range of application scenarios. The acoustic emission sensor is easy to install and has high detection sensitivity. It can be installed at different detection parts according to requirements and is applicable to a variety of scenarios: new machine detection, running detection, and shutdown detection. It can detect early damage and has high accuracy.

[0035] The present invention will be explained and described in detail below in conjunction with the accompanying drawings and specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 is a flowchart of the damage monitoring method provided by the present invention;

[0037] Figure 2 is a structural block diagram of the mechanical turntable damage monitoring system in the present invention;

[0038] FIG. 3 is a time domain diagram corresponding to a single complete rotation cycle and different damages in the embodiment;

[0039] Figure 4 is a result diagram obtained by inputting the characteristic parameters of the second acoustic emission experiment in the embodiment into the SVM;

[0040] Figure 5 is a result diagram obtained by inputting the characteristic parameters of the third acoustic emission experiment in the embodiment into the SVM. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0041] To facilitate the understanding of the present invention, the present invention will be described more comprehensively below with reference to the relevant accompanying drawings. Several embodiments of the present invention are given in the drawings. However, the present invention can be implemented in different forms and is not limited to the embodiments described in the text. On the contrary, these embodiments are provided to make the disclosure of the present invention more thorough and comprehensive.

[0042] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the technical field to which this invention belongs. The terms used in the description of the present invention herein are for the purpose of describing specific embodiments only and are not intended to limit the present invention. The term "and / or" as used herein includes any and all combinations of one or more of the related listed items.

[0043] Please refer specifically to the attached Figure 2 , a mechanical turntable damage monitoring system based on acoustic emission technology combined with SVM damage classification, comprising: an acoustic emission signal acquisition module, an acoustic emission signal processing module, and a damage monitoring module. Among them:

[0044] The acoustic emission acquisition module is completed by an acoustic emission sensor. The acoustic emission sensor is installed on the mechanical turntable to be detected, used to collect the AE signals released during its operation, and convert the sound signals into electrical signals;

[0045] The acoustic emission signal processing module is composed of a preamplifier and an acoustic emission instrument. The output end of the acoustic emission sensor is connected to the input end of the preamplifier through a signal cable. Since the output of the acoustic emission signal after being converted into an electrical signal is often small. After such weak signals are transmitted over a long distance, it may be impossible to distinguish useful signals from noise signals. Therefore, a preamplifier is set to improve the signal-to-noise ratio. The acoustic emission instrument can filter the signal, perform analog-to-digital conversion, generate a waveform file and transmit it to the host;

[0046] The damage monitoring module transmits the acquired original acoustic emission waveform file to the data analysis software MATLAB on the host, and then monitors according to the steps of the mechanical turntable damage monitoring method based on acoustic emission technology combined with SVM damage classification.

[0047] The above system further includes a signal lamp alarm device, which is connected to the host.

[0048] (1) Calculate the accuracy rate of damage classification. The change in the SVM damage classification accuracy rate is not significant, that is, when the normal damage becomes minor damage ≤ 10%, and when the normal and minor damage becomes severe damage ≤ 5%, the signal lamp does not light up.

[0049] (2) When the accuracy rate decreases significantly, and when there are more than 10% of the originally normal condition signals that become minor damage signals, and ≤ 5% become severe damage signals, the signal lamp flashes.

[0050] (3) When the accuracy rate decreases significantly, and when there are more than 10% of the originally normal condition signals that become minor damage signals, and > 5% become severe damage signals, the signal lamp is always on.

[0051] A mechanical turntable damage monitoring method based on acoustic emission technology combined with SVM damage classification using the above system includes the following steps:

[0052] S1. Install at least one acoustic emission sensor on the mechanical turntable to be detected;

[0053] S2. Obtain the acoustic emission waveforms collected by the acoustic emission sensor in real time, and extract AE events based on the original waveforms. Set the threshold voltage (h) and duration (T) in the data analysis software; screen the data points in the original waveforms that exceed the threshold voltage, and determine the positions of the data points where the interval between two adjacent impacts is greater than the duration T; if all the intervals of the found data points are less than the set duration, it is regarded as an AE event. Then record the start and end points of each AE event, and count the duration of the AE events. For AE events with too short a duration, they can be deleted based on a preset duration (analyze the extracted AE events, and when AE events with unsatisfactory and non-standard waveform distributions are found, reset the preset duration for secondary screening).

[0054] Among them, the threshold voltage refers to the preset voltage value, and only when the amplitude of the AE signal exceeds this voltage value can it be detected; the duration refers to the duration of the AE event; an impact refers to any signal that exceeds the threshold and causes the system channel to collect data; the preset duration refers to the duration of the ideal AE event for secondary screening.

[0055] S3. Calculate and count the acoustic emission characteristic parameters of each AE event, including five indicators: amplitude (AM), ring count (RC), margin (Mgn), kurtosis (Kurt), and center frequency (FC).

[0056] Among them, assume that a set of discrete data obtained by sampling a certain signal x(t) is x1, x2, …, x i , …x N , N is the signal length, σ is the standard deviation, and the formula is:

[0057] AM = max{|x i |}

[0058]

[0059] Mgn = X max / X R

[0060]

[0061] Among them, X i represents the time-domain voltage signal, AM represents the amplitude, X R represents the root mean square amplitude, Mgn represents the margin index, X max represents the signal maximum value, σ represents the standard deviation, u represents the signal mean, Kurt represents the kurtosis index, FC represents the center frequency. f represents the frequency, and P(f) represents the power spectral density of the signal.

[0062] For the i-th AE event, construct a data set X composed of these five indicators i =[AM i , RC i , Kurt i , Mgn i , FC i , where i = 1, 2, …, m (m represents the total number of extracted AE events);

[0063] S4. Conduct multiple experiments, collect AE signals of different damages in different regions of the mechanical turntable, and calculate the above five characteristic parameters to form a training set for different damages;

[0064] S5. Calculate the acoustic emission characteristic parameters corresponding to the AE signals of different damages in the same regions detected subsequently, and use them as a test set;

[0065] S6. Input the above training set and test set into SVM (Support Vector Machine) to obtain the results;

[0066] S7. According to the results of SVM, use the accuracy rate of SVM to monitor the damage situation of the turntable through the accuracy rates of the normal state and minor damage.

[0067] The calculation method of the accuracy rate is as follows: Count the number of samples A that should be in the normal condition but are predicted as minor damage and the number of samples B that should be in the minor damage but are predicted as severe damage. Combining with the total number of samples C in the test set, the calculation formula for the total accuracy rate is: (C - B - A) / C * 100%.

[0068] If there are a large number of situations in a certain monitoring where the normal condition should be but becomes minor damage and the minor damage should be but becomes severe damage, resulting in a significant decrease in the classification accuracy rate, it is considered that the mechanical turntable needs to be maintained in a timely manner.

[0069] Example: Apply the technical method proposed by the present invention to the mechanical turntable on the rear floor line of a certain automobile company, with the model EDX1370. The turntable to be detected is composed of a turntable plane, bearings, a housing, a main shaft, a main shaft cam, a main shaft guide rail, a motor, etc.; there are three working states in one working cycle: rotating 180° counterclockwise, waiting for shutdown, and rotating 180° clockwise; the mechanical turntable completes the transmission operation of automobile welding parts by rotating clockwise and counterclockwise.

[0070] The acoustic emission instrument uses the DS5-16C acoustic emission instrument produced by Beijing Ruandao. This instrument uses a USB3.0 interface, its sampling rate can reach 10M, and it has 16 acquisition channels. Among them, the signal sampling frequency of each channel is set to 2.5MHz, and the amplification factor of the preamplifier is set to 40dB. The acoustic emission sensor is an RS-2A piezoelectric ceramic sensor, the sensor bandwidth frequency range is 50kHz - 400kHz, and the resonance frequency is 150kHz. The acoustic emission sensor is installed at the main shaft of the mechanical turntable.

[0071] 1. Conduct an acoustic emission experiment on the above mechanical turntable. The time-domain diagram results corresponding to a single complete rotation cycle and different damages are shown in Figure 3:

[0072] As can be seen from Figure 3(a), there are many burst signals of different degrees in the whole AE signal. Through the comparative study of the machine and the signal, it is found that the image intercepted from the slight damage stage is shown in Figure 3(c). Compared with the normal condition AE signal in Figure 3(b), the image of slight damage shows continuous and relatively high-amplitude AE signals. And the AE signal of serious damage to the turntable in Figure 3(d) shows sudden and particularly obvious amplitude-changing AE signals.

[0073] 2. Conduct an acoustic emission experiment on the above mechanical turntable at three different time periods (with a difference of 3 months between each adjacent two):

[0074] Collect the AE signals during the reverse rotation of 180° and forward rotation of 180° of the mechanical indexing turntable. Set the threshold voltage h = 200mV and the duration T = 500uS in the MATLAB analysis software, extract the AE events, and calculate 5 indexes including amplitude (AM), ring count (RC), margin (Mgn), kurtosis (Kurt), and center frequency (FC) in the AE events.

[0075] Use the characteristic parameters corresponding to different damage positions in the first experiment as the training set; use the characteristic parameters calculated at the same damage positions in the latter two experiments as the test set. Input the training set and the test set into the SVM at the same time to get the results as Figure 4 and Figure 5 shown.

[0076] As Figure 4 can be seen, using the data of the first experiment as the training set, the prediction accuracy for the second experiment is 91.7%; further analyze the specific situation of the prediction results, as Figure 4 shown, among which there is 1 group of samples originally in the normal state that are mispredicted as slight damage, and at the same time, 4 groups of samples originally with slight damage are predicted as serious damage. As Figure 5As shown, using the data from the first experiment as the training set, the accuracy of the prediction results for the third experiment is 85%. Specifically, 2 groups of samples that were originally in a normal state were predicted to have minor damage, and 7 groups of samples that were originally slightly damaged were predicted to have severe damage, resulting in a decrease in the accuracy of the prediction results to 85%. Comparing Figure 4 and Figure 5 the classification results, it can be clearly seen that over time, the damage to the mechanical turntable shows an obvious trend of worsening.

[0077] Compared with traditional methods, the method provided by the present invention can achieve real-time monitoring of the damage situation, and thus can carry out fault prevention in advance. Then the present invention can more simply monitor the damage of the mechanical turntable.

[0078] The above has made an exemplary description of the present invention in conjunction with the accompanying drawings. Obviously, the specific implementation of the present invention is not limited by the above methods. As long as such non-substantial improvements are made using the method concept and technical solution of the present invention, or the concept and technical solution of the present invention are directly applied to other occasions without improvement, they are all within the protection scope of the present invention.

Claims

1. A mechanical turntable damage monitoring method based on acoustic emission technology combined with SVM damage classification, characterized by: The steps include: S1. Install the acoustic emission sensor on the mechanical turntable to be tested; S2, acquiring the acoustic emission waveform collected by the acoustic emission sensor in real time, and extracting AE events based on the original waveform; S3, calculating the acoustic emission characteristic parameters of each AE event and constructing a data set; S4, collecting AE signals of different damages in different areas of the mechanical turntable, and calculating the acoustic emission characteristic parameters of each AE event to form a training set of different damages; S5, calculating acoustic emission characteristic parameters corresponding to AE signals of different damages detected in the same area subsequently, and using them as a test set; S6. Input the above training set and test set into SVM to obtain the result; S7. Based on the results of SVM, use the accuracy of SVM to monitor the damage of the turntable.

2. The mechanical turntable damage monitoring method based on acoustic emission technology combined with SVM damage classification according to claim 1 is characterized by: In step S2, an AE event is identified based on a preset threshold voltage and duration.

3. The mechanical turntable damage monitoring method based on acoustic emission technology combined with SVM damage classification according to claim 2 is characterized in that: In step S2, firstly, the data points exceeding the threshold voltage in the original waveform are screened, and the positions of the data points whose interval between two adjacent data points is greater than the preset duration are determined; then, the starting point and the end point of each AE event are recorded, and the duration of the AE event is counted.

4. The mechanical turntable damage monitoring method based on acoustic emission technology combined with SVM damage classification according to claim 1 is characterized in that: In step S3, the acoustic emission characteristic parameters include amplitude, ringing count, margin, kurtosis and center of gravity frequency.

5. The mechanical turntable damage monitoring method based on acoustic emission technology combined with SVM damage classification according to claim 4 is characterized in that: In step S3, the calculation formula of the acoustic emission characteristic parameter is: AM=max{|x i |} Mgn=X max / X R Among them, X i represents the time domain voltage signal, AM represents the amplitude, X R represents the square root amplitude, Mgn represents the margin index, X max represents the maximum value of the signal, σ represents the standard deviation, u represents the mean value of the signal, Kurt represents the kurtosis index, FC represents the center of gravity frequency, f represents the frequency, and P(f) represents the power spectral density of the signal.

6. The mechanical turntable damage monitoring method based on acoustic emission technology combined with SVM damage classification according to claim 5 is characterized in that: In step S3, for the i-th AE event, the data set is X i =[AM i ,RC i ,Kurt i ,Mgn i ,FC i ],i=1,2,…,m; where AM represents amplitude, RC represents ringing count, Mgn represents margin, Kurt represents kurtosis, FC represents center of gravity frequency, and m represents the total number of extracted AE events.

7. The mechanical turntable damage monitoring method based on acoustic emission technology combined with SVM damage classification according to claim 1 is characterized in that: In step S7, the number of samples that should be in normal condition but are predicted to be slightly injured and the number of samples that should be slightly injured but are predicted to be severely injured are counted, and the accuracy is calculated in combination with the total number of samples in the test set.

8. A mechanical turntable damage monitoring system based on acoustic emission technology combined with SVM damage classification, characterized by: The damage monitoring system for implementing the method for mechanical turntable damage monitoring based on acoustic emission technology combined with SVM damage classification according to any one of claims 1 to 7 comprises an acoustic emission signal acquisition module, an acoustic emission signal processing module, and a damage monitoring module; wherein: The acoustic emission acquisition module includes an acoustic emission sensor, which is used to collect AE signals released during the operation of the mechanical turntable to be detected and convert the sound signals into electrical signals; The acoustic emission signal processing module includes a preamplifier and an acoustic emission instrument. The output end of the acoustic emission sensor is connected to the input end of the preamplifier through a signal cable; the acoustic emission instrument is used to filter and convert the signal into analog and digital form, and generate a waveform file to transmit to the host; The damage monitoring module includes a host, which is used to obtain the original acoustic emission waveform file and perform monitoring according to the steps of the mechanical turntable damage monitoring method based on acoustic emission technology combined with SVM damage classification.

9. The mechanical turntable damage monitoring system based on acoustic emission technology combined with SVM damage classification according to claim 8 is characterized in that: The damage monitoring module is also provided with a signal light alarm device, which is connected to the host and displays different damage degrees of the mechanical turntable to be detected through different states of the signal light.

10. The mechanical turntable damage monitoring system based on acoustic emission technology combined with SVM damage classification according to claim 8 or 9, characterized in that: The acoustic emission sensor is a piezoelectric ceramic sensor with a bandwidth frequency range of 50kHz-400kHz and a resonance frequency of 150kHz; the preamplifier has an amplification factor of 40dB.

Images