CN115758083A - A Fault Diagnosis Method for Motor Bearings Based on Fusion of Time Domain and Time-Frequency Domain - Google Patents

Abstract

The invention relates to the field of motor bearing fault diagnosis methods, and discloses a motor bearing fault diagnosis method based on time domain and time-frequency domain fusion, which comprises the following steps: carrying out data acquisition and processing on the time domain index and the time-frequency domain index of the rolling bearing by using a vibration signal acquisition device; performing VMD decomposition on the time-frequency domain indexes to obtain optimal parameters, calculating the kurtosis of each IMF component after VMD decomposition, and reconstructing a vibration signal; performing SVD (singular value decomposition) to obtain a singular value matrix according to each IMF component reconstruction matrix, and selecting the largest singular value of each IMF component to form a fault characteristic vector; fusing the extracted time domain indexes and the characteristic vector of the VMD-SVD to form a composite characteristic vector of multi-dimensional information; and inputting the composite feature vector into a support vector machine for training and testing, and finally realizing the judgment and diagnosis of the fault type. The invention provides an algorithm for integrating time domain indexes and VMD-SVD decomposition characteristics on a time-frequency domain, and the accuracy of fault diagnosis of a rolling bearing of a motor is improved.

Description

A Fault Diagnosis Method for Motor Bearings Based on Fusion of Time Domain and Time-Frequency Domain

technical field

The invention belongs to the field of motor bearing fault diagnosis methods, in particular to a motor bearing fault diagnosis method based on fusion of time domain and time-frequency domain.

Background technique

There are various types of motors in my country's power plants, and these motors play an important role in water supply, lubrication, and drainage in power plants. However, after the motor has been in operation for a long time, the unit equipment has a series of problems such as aging parts, difficulty in collecting information, and difficulty in finding faults in time. These problems will cause certain safety hazards. The stable operation of the motor also affects the stable operation of the generator set.

Like other fault diagnosis systems, the fault diagnosis of rolling bearings is to realize fault diagnosis by analyzing and processing the signals generated by it. The fault diagnosis process includes signal acquisition, feature extraction, and fault identification. Rolling bearing fault diagnosis adopts the following four methods according to different principles:

(1) Temperature diagnosis method

In the fault diagnosis of rolling bearings, the temperature diagnosis method is the earliest application. It mainly monitors the bearing temperature and judges whether the bearing is faulty according to the temperature change of the bearing. However, it is greatly affected by the external environment and can only be used as an auxiliary diagnosis. method.

(2) Acoustic diagnosis method

Acoustic diagnosis method is acoustic emission technology. This technology judges the fault by sound. At present, most of the bearings are made of metal materials, so when the bearing is running in a fault state, some special noise will be generated. This method requires experience. Rich engineers monitor the noise from time to time and judge the fault according to the noise change, but this method has some disadvantages that it is impossible to get the specific location of the bearing failure, and the acoustic signal is easily affected by the bearing structure and position.

(3) Oil film resistance diagnosis method

Oil film resistance diagnosis method, which is used to diagnose the resistance value generated by the oil film in the rolling bearing. When the rolling bearing fails, the oil film layer between them will be destroyed, which will cause the change of the resistance value, and then diagnose the bearing according to the change of the resistance value Is there a failure. In the face of bearing faults caused by wear and corrosion, this method has a good effect, but this technical method has high requirements for bearing sealing, and it cannot be widely used in practice at present.

(4) Vibration signal diagnosis method

Vibration signal diagnosis method, the vibration signal can truly reflect the running state of the bearing. Scholars at home and abroad have conducted a lot of research on this. This method first collects the vibration signal of the bearing, and then extracts the fault characteristics of the signal through a series of analysis methods, and finally diagnoses . A series of fault diagnosis systems for rolling bearings appearing on the market are based on the principle of vibration. On this basis, the patent of the present invention proposes a diagnostic method based on vibration signals to realize the fault diagnosis of rolling bearings.

At present, the mainstream method is to use pattern recognition algorithm to intelligently diagnose rolling bearings. According to the difference of recognition principles, fault pattern recognition can be divided into two categories. The first category is supervised pattern recognition. Commonly used are support vector machines and neural networks. and deep learning etc. The principle of these classifiers is to build a classification model through learning. First, learn the known samples to form a specific mathematical classification model, which can identify and judge unknown samples; the second type is unsupervised pattern recognition, which It is completely different from supervised pattern recognition. It does not need to learn and train samples before pattern recognition. It performs pattern recognition by finding similarities between input samples. Commonly used unsupervised pattern recognition has fuzzy clustering. method. Yao Liguo and others successfully applied the K-means clustering identification method to the fault diagnosis of rolling bearings. This method successfully solved the local solution problem in the clustering algorithm by using simulated annealing, and then collected the fault signals of rolling bearings by constructing fault experiments to carry out diagnosis, experiments have shown that the method works well. Xiang Ling and others also used the fuzzy clustering method in the pattern recognition of bearing faults. They started with the vibration signal and decomposed the vibration signal generated during the operation of the rolling bearing by EMD. After the EMD decomposition, the corresponding modal components were generated. The approximate entropy of the state is used as a feature input into the FCM clustering for classification and recognition. The good recognition rate proves the feasibility of this method. Neural network is more and more widely used in rolling bearing fault diagnosis. Its core idea is to simulate the human brain neuron network to construct a mathematical classification model based on empirical knowledge learning and obtained through machine learning. Zhang Shuqing and others proposed a diagnostic method based on BP neural network. This method first decomposes the bearing vibration signal into a series of modal components through EEMD, and then uses the approximate entropy of each modal component as a feature, and then uses BP neural network to analyze The characteristic samples are used for training, and the fault signal is diagnosed and identified after the training is completed. The experimental results verify the effectiveness of the method.

Support vector machine is a classifier based on statistics, and its basic principle is to find the optimal hyperplane that can distinguish them in different types of data samples. Compared with neural network, support vector machine has the advantage that it is especially suitable for small sample classification, and it will not appear the common problems of neural network such as overfitting and underfitting. Zhang Peipeng proposed a combination of time domain features Support vector machine fault diagnosis method, the experiment proves that the method is effective. Wang Jindong and others proposed a fault diagnosis method based on EMD and support vector machine. Firstly, EMD is used to decompose the original signal of bearing, and the information entropy of each modal component is used as the parameter input of support vector machine for training. The sample analysis method has a certain validity.

And the above-mentioned method has the following defects:

The temperature diagnosis method is greatly affected by the external environment and can only be used as an auxiliary diagnosis method; the disadvantage of the acoustic diagnosis method is that it is impossible to obtain the specific location of the bearing failure, and the acoustic signal is easily affected by the bearing structure and position; oil film resistance The diagnostic method has high requirements on the bearing seal and cannot be widely used in practice at present; the fault recognition rate of the EMD method is not high.

Contents of the invention

The purpose of the present invention is to provide a motor bearing fault diagnosis method based on fusion of time domain and time-frequency domain, so as to solve the technical problem of low accuracy existing in existing fault diagnosis methods.

In order to solve the problems of the technologies described above, the specific technical solutions of the present invention are as follows:

In some embodiments of the present application, a motor bearing fault diagnosis method based on fusion of time domain and time-frequency domain is provided, comprising the following steps:

S1. Use the vibration signal collector to collect and process data on the time-domain indicators and time-frequency domain indicators of the rolling bearing;

S2. Set the VMD decomposition parameters. Within the set parameter range, perform VMD decomposition on the time-frequency domain index to obtain the optimal parameters, calculate the kurtosis of each IMF component after VMD decomposition, and reconstruct the vibration signal;

S3. Reconstruct the matrix according to each IMF component, perform SVD decomposition to obtain a singular value matrix, and select the largest singular value of each IMF component to form a fault feature vector;

S4. Fusing the extracted time domain index and the feature vector extracted by VMD-SVD decomposition to form a composite feature vector of multi-dimensional information;

S5. Input the composite feature vector into the support vector machine for training and testing, and finally realize the discrimination and diagnosis of the fault type.

Preferably, in one embodiment of the motor bearing fault diagnosis method based on time domain and time-frequency domain fusion, the time domain indicators in the step S1 include: kurtosis, peak factor, pulse factor and margin factor.

Preferably, in one embodiment of the motor bearing fault diagnosis method based on time domain and time-frequency domain fusion, the specific process of step S2 is as follows:

S21. Perform parameter initialization on the time-frequency domain index through the PSO optimization algorithm;

S22. Using the VMD decomposition algorithm to decompose the filtered vibration signal into multiple modal components, and calculate the kurtosis coefficient value of each modal;

S23, setting the current number of iterations is not n, the preset maximum number of iterations N; judging whether the current number of iterations n is greater than or equal to the preset maximum number of iterations N, if so, then enter step S24, otherwise, make n= n+1, and return to step S22;

S24, save the optimal parameter and perform VMD decomposition on the optimal parameter, decompose into multiple modal components, calculate the kurtosis coefficient value of each modal component, and select the modal component corresponding to the maximum kurtosis coefficient value for signal synthesis ;

S25. After filtering the composite signal obtained in step S24, perform envelope demodulation to generate an envelope spectrum;

S26. Analyze and extract the fault feature vector according to the envelope spectrum obtained in step S25.

Preferably, in one embodiment of the motor bearing fault diagnosis method based on time domain and time-frequency domain fusion, the parameters in the step S2 are the quadratic penalty factor α and the number K of modal components.

Preferably, in one embodiment of the motor bearing fault diagnosis method based on time domain and time-frequency domain fusion, it also includes: determining the working state of the motor bearing according to the collected vibration signal, and then performing different adjust.

Preferably, in one embodiment of the motor bearing fault diagnosis method based on time domain and time-frequency domain fusion, the determination of the working state of the motor bearing according to the collected vibration signal is specifically: combining the collected vibration signal with The preset vibration signal is compared based on the comparison result to determine the current working state of the motor bearing and transmit it to the fault management terminal; wherein the working state includes a normal state and a fault state; when the working state is normal, If the working state is failure, it is necessary to judge the current failure level according to the number of failures and carry out early warning processing at the same time, and the failure types include: inner ring failure, outer ring failure and rolling element failure.

Preferably, in one embodiment of the motor bearing fault diagnosis method based on time domain and time-frequency domain fusion, judging the current fault level according to the number of faults and performing early warning processing specifically includes:

Preset the preset fault state degree matrix A0, set A0=(A1, A2, A3), where A1 is the first preset fault state degree, A2 is the second preset fault state degree, and H3 is the third Preset fault state degree, where A1<A2<A3;

Preset the preset warning level matrix B0, set B0=(B1, B2, B3), where B1 is the first preset warning level, B2 is the second preset warning level, and B3 is the third preset warning level , and B1<B2<B3; set the warning level G according to the relationship between the leakage degree H and each preset leakage degree: when A<A1, select the first preset warning level B1 as the warning level B; when A1≤A<A2, select the second preset warning level B2 as the warning level B; when A2≤A<A3, select the third preset warning level B3 as the warning level B.

It can be seen through the above-mentioned technical solution that, compared with the prior art, the beneficial effects of the present invention are:

In this scheme, a vibration signal collector based on ZYNQ is designed to realize the collection and processing of vibration data of rolling bearings; at the PL end, a signal conditioning circuit, AD acquisition and control circuit, FIR digital filter, and system clock frequency are designed, and at the PS The terminal completes data storage and Ethernet data transmission. In this design, acceleration sensors are used to collect vibration signals of rolling bearings at different measuring points, and after preprocessing, they are transmitted to the PC terminal through Ethernet, and the PC terminal performs fault identification on the operating status of the rolling bearings of the unit;

In this scheme, the variational mode decomposition (VMD) is used to decompose and process the bearing vibration signal. This method has a solid theoretical foundation and strong robustness, and is suitable for dealing with nonlinear and non-stationary signals such as rolling bearing vibration signals. This method can adaptively select the two important parameters of VMD decomposition layer number k and penalty factor according to the signal characteristics, and then complete the signal decomposition. Finally, the validity of the method is verified by using the fault simulation signal of the rolling bearing;

This program uses the support vector machine algorithm to realize the fault diagnosis of rolling bearings. The eigenvalues extracted by VMD-SVD decomposition and the indicators in the time domain are fused to form a composite feature matrix of multi-dimensional information, and the features are input to the support vector machine for training. The application effect of the method proposed in the present invention is demonstrated, and the method is used to carry out diagnostic tests on the bearing fault data collected in the experiment, and a good diagnostic effect is obtained, which verifies the effectiveness of the present invention.

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only It is an embodiment of the present invention, and those skilled in the art can also obtain other drawings according to the provided drawings without creative work.

Fig. 1 is the flow chart of feature extraction provided by the embodiment of the present invention;

FIG. 2 is a flow chart of parameter adaptive VMD feature extraction provided by an embodiment of the present invention;

Fig. 3 is the recognition result diagram of the time-domain feature fusion SVM-VMD feature sample in SVM provided by the embodiment of the present invention

Fig. 4 is the recognition result figure of time domain characteristic sample in SVM;

Fig. 5 is a diagram of the recognition results of time-domain feature fusion SVM-EMD feature samples in SVM.

Detailed ways

The specific implementation manners of the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

In the description of this application, it is to be understood that the terms "center", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", The orientations or positional relationships indicated by "top", "bottom", "inner", "outer", etc. are based on the orientations or positional relationships shown in the drawings, and are only for the convenience of describing the application and simplifying the description, rather than indicating or implying References to devices or elements must have a particular orientation, be constructed, and operate in a particular orientation and therefore should not be construed as limiting the application.

The terms "first" and "second" are used for descriptive purposes only, and cannot be understood as indicating or implying relative importance or implicitly specifying the quantity of indicated technical features. Thus, a feature defined as "first" and "second" may explicitly or implicitly include one or more of these features. In the description of the present application, unless otherwise specified, "plurality" means two or more.

In the description of this application, it should be noted that unless otherwise specified and limited, the terms "installation", "connection", and "connection" should be understood in a broad sense, for example, it can be a fixed connection or a detachable connection. Connected, or integrally connected; it can be mechanically connected or electrically connected; it can be directly connected or indirectly connected through an intermediary, and it can be the internal communication of two components. Those of ordinary skill in the art can understand the specific meanings of the above terms in this application in specific situations.

In order to better understand the purpose, structure and function of the present invention, the present invention will be described in further detail below in conjunction with the accompanying drawings.

Referring to Figures 1-2, according to some embodiments of the present application, in some embodiments of the present application, a motor bearing fault diagnosis method based on time domain and time-frequency domain fusion is provided, including the following steps:

S1. Use the vibration signal collector to collect and process data on the time-domain indicators and time-frequency domain indicators of the rolling bearing;

S2. Set the VMD decomposition parameters. Within the set parameter range, perform VMD decomposition on the time-frequency domain index to obtain the optimal parameters, calculate the kurtosis of each IMF component after VMD decomposition, and reconstruct the vibration signal;

S3. Reconstruct the matrix according to each IMF component, perform SVD decomposition to obtain a singular value matrix, and select the largest singular value of each IMF component to form a fault feature vector;

S4. Fusing the extracted time domain index and the feature vector extracted by VMD-SVD decomposition to form a composite feature vector of multi-dimensional information;

S5. Input the composite feature vector into the support vector machine for training and testing, and finally realize the discrimination and diagnosis of the fault type.

Specifically, the vibration signal collector includes: some acceleration sensors, some signal conditioners, AD acquisition cards, PL end, PS end and PC end; wherein, the interface of the acceleration sensor, the signal conditioner and the AD acquisition card are connected in sequence; It is used to amplify and filter the collected vibration electrical signal, convert it into a digital signal and transmit it to the acquisition control module; the PL end is the logic end, including: several FIFD buffers, FIR filter module, acquisition control module, Internet storage and expansion MIO pin; store and transmit the collected data by receiving; the PS end is the processing end, completes the data storage, and transmits the data to the PC end for display through the PC end.

Through the above-mentioned technical scheme, the technical effects that the present application can achieve are:

The vibration signal collector based on ZYNQ of the present invention realizes the collection and processing of the vibration data of rolling bearings, and designs including signal conditioning circuit, AD collection and control circuit, FIR digital filtering and system clock frequency at the PL end, and completes the process at the PS end Data storage and Ethernet data transmission, etc.; in this design, acceleration sensors are used to collect vibration signals of rolling bearings at different measuring points, and after preprocessing, they are transmitted to the PC terminal through Ethernet, and the PC terminal performs fault identification on the operating status of the rolling bearings of the unit;

The invention combines the time domain index with the VMD-SVD decomposition feature in the time-frequency domain, and the verification of experimental data shows that the method improves the classification accuracy.

In another preferred embodiment of the present application, referring to the accompanying drawing 2, the specific process of the step S2 is:

S21. Perform parameter initialization on the time-frequency domain index through the PSO optimization algorithm;

S22. Using the VMD decomposition algorithm to decompose the filtered vibration signal into multiple modal components, and calculate the kurtosis coefficient value of each modal;

S23, setting the current number of iterations is not n, the preset maximum number of iterations N; judging whether the current number of iterations n is greater than or equal to the preset maximum number of iterations N, if so, then enter step S24, otherwise, make n= n+1, and return to step S22;

S24, save the optimal parameter and perform VMD decomposition on the optimal parameter, decompose into multiple modal components, calculate the kurtosis coefficient value of each modal component, and select the modal component corresponding to the maximum kurtosis coefficient value for signal synthesis ;

S25. After filtering the composite signal obtained in step S24, perform envelope demodulation to generate an envelope spectrum;

S26. Analyze and extract the fault feature vector according to the envelope spectrum obtained in step S25.

Specifically, the parameters in the step S2 are the quadratic penalty factor α and the number of modal components.

Through the above-mentioned technical scheme, the technical effects that the present application can achieve are:

The VMD algorithm is used to decompose the short-circuit protection current signal of the converter valve. Compared with the traditional non-stationary signal processing method EMD and its improved algorithm, it has a solid theoretical foundation and better decomposition effect;

The disclosure combines VMD and SVM, and further improves the accuracy of fault diagnosis of rolling bearings of motors by virtue of VMD algorithm's good ability to divide similar frequency components and SVM learning and training ability.

In another preferred embodiment of the present application, it further includes: determining the working state of the motor bearing according to the collected vibration signal, and then performing different adjustments in different working states.

Specifically, the determination of the working state of the motor bearing according to the collected vibration signal is specifically: comparing the collected vibration signal with the preset vibration signal and based on the comparison result, determining the current working state of the motor bearing and transmitting it to Fault management terminal; wherein, the working state includes a normal state and a fault state; when the working state is normal, no adjustment is performed; when the working state is a fault, it is necessary to judge the current fault level according to the number of faults At the same time, early warning processing is carried out, and the fault types include: inner ring fault, outer ring fault and rolling element fault.

Specifically, judging the current fault level according to the number of faults and performing early warning processing specifically includes:

Preset the preset fault state degree matrix A0, set A0=(A1, A2, A3), where A1 is the first preset fault state degree, A2 is the second preset fault state degree, and H3 is the third Preset fault state degree, where A1<A2<A3;

Preset the preset warning level matrix B0, set B0=(B1, B2, B3), where B1 is the first preset warning level, B2 is the second preset warning level, and B3 is the third preset warning level , and B1<B2<B3; set the warning level G according to the relationship between the leakage degree H and each preset leakage degree: when A<A1, select the first preset warning level B1 as the warning level B; when A1≤A<A2, select the second preset warning level B2 as the warning level B; when A2≤A<A3, select the third preset warning level B3 as the warning level B.

Through the above-mentioned technical scheme, the technical effects that the present application can achieve are:

By presetting the fault state degree matrix and the early warning level, different early warning levels can be selected according to different fault degrees, so that the fault can be adjusted conveniently.

The technical effect of the present invention is further stated below by several embodiments:

Experimental results of fault diagnosis based on SVM:

Example 1:

The test sample is divided into feature samples that are fused with time-domain indicators and SVD-VMD features. The above-mentioned test samples are input into the trained SVM classification model for diagnosis. As shown in Figure 3, only the time-domain feature samples can identify 96% of the results in SVM.

Comparative example 1:

The test samples are divided into time-domain feature samples (kurtosis, peak factor, impulse factor, margin factor), and the above test samples are input into the trained SVM classification model for diagnosis. As shown in Figure 4, only the time-domain feature samples are in the SVM The middle recognition result is 89%.

Comparative example 2:

The test sample is divided into feature samples fused with time-domain indicators and SVD-EMD features. The above-mentioned test samples are input into the trained SVM classification model for diagnosis. As shown in Figure 5, only the time-domain feature samples can identify 92% of the results in SVM.

From the above, it can be seen that the present invention combines the time domain index with the VMD-SVD decomposition feature in the time-frequency domain, and the experimental data verification shows that the recognition result is 96%, which greatly improves the classification accuracy.

In the present invention:

Variational mode decomposition (VMD) is a signal adaptive decomposition method proposed by UCLA scholars Dragomiretskiy and Zosso in 2014. As an improved empirical mode decomposition method, VMD has a solid mathematical theoretical foundation, noise Robustness and signal separation performance have also been greatly improved. However, VMD decomposition parameters such as the number of modal components and frequency bandwidth control parameters of modal components have a significant impact on its decomposition results.

Support Vector Machine (SVM) is a kind of generalized linear classifier (generalized linear classifier) that performs binary classification on data according to supervised learning, and its decision boundary is the maximum margin hyperplane (maximum- margin). The SVM integrated classifier is used for training and testing. The primary learner is obtained from the initial data set training, and the output of the primary learner is used as a sample input feature to generate a new training set for secondary learner training; SVM classification is selected. The primary learner is the primary learner, and the secondary learner adopts the learning-based AdaBoost multi-category ensemble learning algorithm. The weighted majority voting method is used to integrate the individual SVM classifiers generated in each cycle to obtain a new strong classifier, which will be trained The classification model of is used for rolling bearing fault classification to obtain a significant improvement in classification accuracy.

Empirical mode decomposition (EMD) is a new method for dealing with non-stationary signals proposed by Dr. Nordene.Huang, a Chinese scientist of NASA in 1998 - an important component of the Hilbert-Huang transform part. The time-frequency analysis method based on EMD is not only suitable for the analysis of nonlinear and non-stationary signals, but also suitable for the analysis of linear and stationary signals, and the analysis of linear and stationary signals is better than other time-frequency analysis methods. The physical meaning of the signal.

Those skilled in the art should understand that the embodiments of the present application may be provided as methods, systems, or computer program products. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.

The present application is described with reference to flowcharts and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the present application. It should be understood that each procedure and/or block in the flowchart and/or block diagram, and a combination of procedures and/or blocks in the flowchart and/or block diagram can be realized by computer program instructions. These computer program instructions may be provided to a general purpose computer, special purpose computer, embedded processor, or processor of other programmable data processing equipment to produce a machine such that the instructions executed by the processor of the computer or other programmable data processing equipment produce a An apparatus for realizing the functions specified in one or more procedures of the flowchart and/or one or more blocks of the block diagram.

These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing apparatus to operate in a specific manner, such that the instructions stored in the computer-readable memory produce an article of manufacture comprising instruction means, the instructions The device realizes the function specified in one or more procedures of the flowchart and/or one or more blocks of the block diagram.

These computer program instructions can also be loaded onto a computer or other programmable data processing device, causing a series of operational steps to be performed on the computer or other programmable device to produce a computer-implemented process, thereby The instructions provide steps for implementing the functions specified in the flow chart or blocks of the flowchart and/or the block or blocks of the block diagrams.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: the present invention can still be Any modifications or equivalent replacements that do not depart from the spirit and scope of the present invention shall fall within the protection scope of the claims of the present invention.

Claims (7)

1. A motor bearing fault diagnosis method based on time domain and time-frequency domain fusion is characterized by comprising the following steps:

s1, carrying out data acquisition and processing on time domain indexes and time-frequency domain indexes of a rolling bearing by using a vibration signal acquisition device;

s2, setting VMD decomposition parameters, carrying out VMD decomposition on the time-frequency domain indexes in the set parameter range to obtain optimal parameters, calculating the kurtosis of each IMF component after VMD decomposition, and reconstructing a vibration signal;

s3, according to the IMF component reconstruction matrix, carrying out SVD to obtain a singular value matrix, and selecting the singular value with the largest IMF component to form a fault feature vector;

s4, fusing the extracted time domain indexes and the characteristic vectors extracted by the VMD-SVD decomposition to form a composite characteristic vector of multi-dimensional information;

and S5, inputting the composite feature vector into a support vector machine for training and testing, and finally realizing the judgment and diagnosis of the fault type.

2. The method for diagnosing the fault of the motor bearing based on the fusion of the time domain and the time-frequency domain according to claim 1, wherein the time domain index in the step S1 comprises the following steps: kurtosis, peak factor, impulse factor, and margin factor.

3. The method for diagnosing the fault of the motor bearing based on the fusion of the time domain and the time-frequency domain as claimed in claim 1, wherein the specific process of the step S2 is as follows:

s21, carrying out parameter initialization on the time-frequency domain indexes through a PSO optimization algorithm;

s22, decomposing the filtered vibration signal into a plurality of modal components by utilizing a VMD decomposition algorithm, and calculating a kurtosis coefficient value of each mode;

s23, setting the current iteration times to be not N and the preset maximum iteration times to be N; judging whether the current iteration number N is greater than or equal to a preset maximum iteration number N, if so, entering a step S24, otherwise, enabling N = N +1, and returning to the step S22;

s24, storing the optimal parameters, performing VMD decomposition on the optimal parameters, decomposing the optimal parameters into a plurality of modal components, calculating the kurtosis coefficient value of each modal component, and selecting the modal component corresponding to the maximum kurtosis coefficient value to perform signal synthesis;

s25, after filtering processing is carried out on the synthesized signal obtained in the step S24, envelope demodulation is carried out to generate an envelope spectrum;

and S26, analyzing according to the envelope spectrum obtained in the step S25 and extracting a fault feature vector.

4. The method for diagnosing the fault of the motor bearing based on the fusion of the time domain and the time-frequency domain as claimed in claim 3, wherein the parameters in the step S2 are a secondary penalty factor α and a number K of modal components.

5. The method for diagnosing the fault of the motor bearing based on the fusion of the time domain and the time-frequency domain as claimed in claim 1, wherein the method further comprises the following steps: and determining the working state of the motor bearing according to the acquired vibration signal, and then carrying out different adjustments according to different working states.

6. The method for diagnosing the fault of the motor bearing based on the fusion of the time domain and the time-frequency domain according to claim 5, wherein the determining of the working state of the motor bearing according to the collected vibration signal specifically comprises: comparing the collected vibration signal with a preset vibration signal, determining the current working state of the motor bearing based on a comparison result, and transmitting the working state to a fault management terminal; wherein the working state comprises a normal state and a fault state; when the working state is normal, the adjustment is not carried out; when the working state is a fault, the current fault level needs to be judged according to the number of the faults, and the early warning processing is carried out at the same time, and the fault type comprises the following steps: inner ring failure, outer ring failure, and rolling element failure.

7. The method for diagnosing the fault of the motor bearing based on the fusion of the time domain and the time-frequency domain as claimed in claim 6, wherein the step of judging the current fault level according to the number of the faults and performing early warning processing at the same time specifically comprises the steps of:

presetting a preset fault state degree matrix A0, and setting A0= (A1, A2, A3), wherein A1 is a first preset fault state degree, A2 is a second preset fault state degree, and H3 is a third preset fault state degree, wherein A1 is more than A2 and less than A3;

presetting a preset early warning level matrix B0, and setting B0= (B1, B2 and B3), wherein B1 is a first preset early warning level, B2 is a second preset early warning level, B3 is a third preset early warning level, and B1 is greater than B2 and is greater than B3; setting an early warning grade G according to the relation between the leakage degree H and each preset leakage degree: when A is smaller than A1, selecting the first preset early warning grade B1 as an early warning grade B; when A1 is more than or equal to A and less than A2, selecting the second preset early warning level B2 as an early warning level B; and when A2 is not less than A and is less than A3, selecting the third preset early warning grade B3 as an early warning grade B.

Images