Bearing fault diagnosis method and device for multi-sensor driven flexible support tensor machine

Abstract

This application discloses a bearing fault diagnosis method and device using a multi-sensor driven flexible support tensor machine. This method decomposes multi-sensor signals and sequentially extracts time-domain characteristic parameters to design a "frequency band component-statistical parameter-multi-sensor" third-order characteristic tensor; introduces flexible factors and permutation factors to construct a flexible support tensor machine for training and testing the third-order characteristic tensor; and finally, inputs the third-order characteristic tensor into the flexible support tensor machine intelligent fault diagnosis model for bearing fault diagnosis. Bearing fault diagnosis experiments have verified that this method can fully utilize the high-order correlation information of multi-sensor signals, improve the model's robustness and diagnostic accuracy, and has significant effects in intelligent fault diagnosis of rotating machinery, surpassing existing technologies.

Description

Technical Field

[0001] The present application belongs to the field of fault diagnosis technology, and specifically relates to a bearing fault diagnosis method, device and electronic equipment for a multi-sensor driven flexible support tensor machine. Background Art

[0002] As rotating machinery becomes increasingly integrated and intelligent, the degree of coupling between components increases. It's difficult to represent the operating status of rotating machinery with a single sensor alone. Furthermore, the intelligent diagnostic results obtained from a single sensor are unstable and uncertain. Therefore, when different types of equipment faults occur, it's difficult to identify them with just a single status signal. More status signals are often required to reflect the operating status. Therefore, it's crucial to explore intelligent fault diagnosis methods for rotating machinery based on multi-source signals.

[0003] Signal fusion is currently the main research method for multi-source signal fault diagnosis. However, signal fusion essentially relies on the eigenvectors of multi-source signals for intelligent diagnosis, but ignores the spatial structure and related intrinsic information of multi-source signals. In contrast, the eigentensor of multi-source signals can contain richer data structure information.

[0004] The tensor convex hull, like the convex hull of a vector space, is an underestimation of tensor samples and is susceptible to interference from outliers. Therefore, improving the tensor convex hull model to make the corresponding multi-classifier more robust is of great practical significance for improving the recognition accuracy of intelligent diagnosis of multiple fault states in rotating machinery. Summary of the Invention

[0005] The purpose of the embodiments of the present application is to provide a bearing fault diagnosis method, device and electronic device for a multi-sensor driven flexible support tensor machine, which is inspired by the affine hull and introduces a flexibility factor on the convex hull model to increase its elasticity and make the model looser to solve the problem of under-estimation of the tensor convex hull; at the same time, a substitution factor is introduced to reduce the interference of outliers on the model, thereby increasing the robustness of the model, thereby solving the technical problems involved in the background technology.

[0006] In order to solve the above technical problems, this application is implemented as follows:

[0007] In a first aspect, embodiments of the present application provide a bearing fault diagnosis method and apparatus for a multi-sensor driven flexible support tensor machine, comprising:

[0008] S1. Collect multi-source signals of bearings in different health states to obtain multi-source signals of samples corresponding to each health state;

[0009] S2. Obtaining a third-order feature tensor of each sample from the multi-source signal of each sample collected according to a multi-source signal tensor feature construction method;

[0010] S3. Based on the convex hull geometry model, the flexible factor and permutation factor are introduced to create a flexible support tensor machine model;

[0011] S4. Randomly select a small number of third-order feature tensors from each health state as tensor training samples to train the flexible support tensor machine model to obtain an intelligent fault diagnosis decision function;

[0012] S5. The remaining samples of each health state constitute test samples to test the feasibility of the trained intelligent fault diagnosis decision function.

[0013] As an improvement of the present application, in step S1, multi-source signals are collected by sound sensors and vibration sensors.

[0014] As an improvement of the present application, in step S2, the third-order feature tensor is a frequency band component-statistical parameter-multi-sensor third-order feature tensor.

[0015] As an improvement of the present application, step S2 specifically includes:

[0016] For each sample of the multi-source signal collected, each signal in the multi-source signal is decomposed into M components using ensemble empirical mode decomposition;

[0017] Extract statistical characteristic parameters, extract H time domain parameters from the component components obtained by decomposing each signal, where the characteristic of each signal in the sample is expressed as F = [f1, f2, ..., f H ] T ∈R M×H ;

[0018] Get the feature tensor. The feature of each sample can be expressed as a third-order feature tensor Where I1 = M, I2 = H, I3 = P, corresponding to the frequency band component M, statistical parameter H and the number of sensors P respectively;

[0019] The frequency band component, statistical parameter and multi-sensor third-order feature tensor of each sample are formed.

[0020] As an improvement of the present application, the time domain parameters include mean, variance, root mean square, peak-to-peak value, skewness, and kurtosis.

[0021] As an improvement of the present application, step S3 specifically includes:

[0022] S31. Define the convex hull geometric model of the tensor space:

[0023] where β i is the corresponding tensor sample x iThe combination coefficient of , the tensor convex hull is a minimum convex set containing l tensor samples;

[0024] S32. Introduce flexibility factor and permutation factor to define a new geometric model of tensor space

[0025] Where λ is a flexibility factor that can make the geometric model looser, and μ is a displacement factor that can improve the robustness of the geometric model.

[0026] As an improvement of the present application, in step S4, the tensor training sample can be expressed as The corresponding label is y i ∈γ={1,-1}.

[0027] As an improvement of the present application, step S5 specifically includes:

[0028] S41. Use Tucker decomposition to decompose the tensor training sample and obtain Where G is the core tensor; A k The factor matrix represented by ; Represents the column vectors of each factor matrix;

[0029] Then select a suitable kernel function to embed the kernel function of Tucker decomposition, which is expressed by the following formula:

[0030]

[0031] S42. Construct a maximum margin optimization model for linear and nonlinear space, expressed by the following formula:

[0032] Linear Maximum Margin Optimization Model

[0033]

[0034] The nonlinear maximum margin optimization model is

[0035]

[0036] S43. Solve the QP problem using the standard algorithm and obtain the optimal solution: and nearest neighbor and

[0037] S44. Get the weight tensor ω * and intercept b * :

[0038]

[0039] According to the linear equation of the hyperplane, ω T x + b = 0, we get:

[0040]

[0041] S45. Finally, the decision function f(x) is calculated:

[0042] Decision function for linear models

[0043]

[0044] Decision Function for Nonlinear Models

[0045]

[0046] In a second aspect, embodiments of the present application provide a bearing fault diagnosis method and apparatus for a multi-sensor driven flexible support tensor machine, including:

[0047] A multi-source signal acquisition module is used to collect multi-source signals under different health conditions of the bearing and obtain multi-source signals of samples corresponding to each health condition;

[0048] A feature tensor acquisition module is used to obtain a third-order feature tensor of each sample according to a multi-source signal tensor feature construction method for the multi-source signal of each sample collected;

[0049] A tensor machine model building module is used to introduce flexibility factors and permutation factors based on the convex hull geometry model to build a flexible support tensor machine model;

[0050] The decision function acquisition module is used to randomly select a small number of third-order feature tensors from each health state as tensor training samples to train the flexible support tensor machine model and obtain the intelligent fault diagnosis decision function;

[0051] The testing module is used to form test samples from the remaining samples of each health state to test the feasibility of the trained intelligent fault diagnosis decision function.

[0052] In a third aspect, an embodiment of the present application provides an electronic device comprising a processor, a memory, and a program or instruction stored in the memory and executable on the processor. When the program or instruction is executed by the processor, the steps of the method for intelligent fault diagnosis of bearings based on multi-sensor signal fusion and flexible support tensor machine are implemented.

[0053] In a fourth aspect, an embodiment of the present application provides a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, the steps of the method described in the first aspect are implemented.

[0054] In the embodiment of the present application, the characteristic tensor representation of multi-source signals is adopted, which can contain richer data structure information, and solves the problem that the current diagnosis based on multi-source signals is mainly concentrated in signal fusion, but ignores the spatial structure and related intrinsic information of the multi-source signals; the flexible support tensor machine is adopted to effectively identify and diagnose multiple health states of bearings, and can achieve better diagnostic accuracy; when outliers are mixed in the training sample set, the algorithm can be made more robust by introducing a permutation factor, so that the best diagnostic results can be achieved while maintaining the relative stability of the diagnostic results. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 This is a flow chart of a bearing fault diagnosis method for a multi-sensor driven flexible support tensor machine provided in an embodiment of the present application;

[0056] Figure 2 This is a structural framework diagram of a bearing fault diagnosis device for a multi-sensor driven flexible support tensor machine provided in an embodiment of the present application;

[0057] Figure 3 This is a structural framework diagram of an electronic device provided in an embodiment of the present application;

[0058] Figure 4 This is a structural diagram of the bearing fault simulation experimental platform provided in Example 1 of the present application;

[0059] Figure 5 These are three fault mode diagrams for a bearing with a damage degree of 0.2 mm provided in Example 1 of the present application;

[0060] Figure 6 It is the original vibration acceleration time domain signal diagram of ten types of bearing fault states in the target domain of the present invention;

[0061] Figure 7 It is the original sound time domain signal diagram of ten types of bearing fault states in the target domain of the present invention. DETAILED DESCRIPTION

[0062] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0063] The terms "first," "second," and the like in the specification and claims of this application are used to distinguish similar objects, and are not used to describe a specific order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate, so that the embodiments of this application can be implemented in an order other than that illustrated or described herein, and that the objects distinguished by "first," "second," and the like are generally of the same type, and do not limit the number of objects; for example, the first object can be one or more. In addition, the term "and / or" in the specification and claims refers to at least one of the connected objects, and the character " / " generally indicates that the objects connected are in an "or" relationship.

[0064] The following, in conjunction with the accompanying drawings, describes in detail the bearing intelligent fault diagnosis method based on multi-sensor signal fusion and flexible support tensor machine provided in the embodiment of the present application through specific embodiments and their application scenarios.

[0065] See Figure 1 , is a bearing fault diagnosis method and apparatus for a multi-sensor driven flexible support tensor machine provided in an embodiment of the present application, comprising:

[0066] S1. Collect multi-source signals of bearings in different health states to obtain multi-source signals of samples corresponding to each health state;

[0067] It should be noted that multi-source signals are collected through sound sensors and vibration sensors.

[0068] S2. Obtaining a third-order feature tensor of each sample from the multi-source signal of each sample collected according to a multi-source signal tensor feature construction method;

[0069] It should be noted that the third-order characteristic tensor is a frequency band component-statistical parameter-multi-sensor third-order characteristic tensor.

[0070] S3. Based on the convex hull geometry model, the flexible factor and permutation factor are introduced to create a flexible support tensor machine model;

[0071] S4. Randomly select a small number of third-order feature tensors from each health state as tensor training samples to train the flexible support tensor machine model to obtain an intelligent fault diagnosis decision function;

[0072] S5. The remaining samples of each health state constitute test samples to test the feasibility of the trained intelligent fault diagnosis decision function.

[0073] Step S2 specifically includes:

[0074] For each sample of the multi-source signal collected, each signal in the multi-source signal is decomposed into M components using ensemble empirical mode decomposition;

[0075] Extract statistical characteristic parameters, extract H time domain parameters from the component components obtained by decomposing each signal, where the characteristic of each signal in the sample is expressed as F = [f1, f2, ..., f H ] T ∈R M×H ;

[0076] Get the feature tensor. The feature of each sample can be expressed as a third-order feature tensor Where I1 = M, I2 = H, I3 = P, corresponding to the frequency band component M, statistical parameter H and the number of sensors P respectively;

[0077] The frequency band component, statistical parameter and multi-sensor third-order feature tensor of each sample are formed.

[0078] Specifically, the time domain parameters include but are not limited to mean, variance, root mean square, peak-to-peak value, skewness, and kurtosis.

[0079] Step S3 specifically includes:

[0080] S31. Define the convex hull geometric model of the tensor space:

[0081] where β i is the corresponding tensor sample x i The combination coefficient of , the tensor convex hull is a minimum convex set containing l tensor samples;

[0082] S32. Introduce flexibility factor and permutation factor to define a new geometric model of tensor space

[0083] Where λ is a flexibility factor that can make the geometric model looser, and μ is a displacement factor that can improve the robustness of the geometric model.

[0084] As an improvement of the present application, in step S4, the tensor training sample can be expressed as

[0085] Among them, y i ∈γ={1,-1},i=1,…,l.

[0086] As an improvement of the present application, step S5 specifically includes:

[0087] S41. Use Tucker decomposition to decompose the tensor training sample and obtain Where G is the core tensor; A k The factor matrix represented by ; Represents the column vectors of each factor matrix;

[0088] Then select a suitable kernel function to embed the kernel function of Tucker decomposition, which is expressed by the following formula:

[0089]

[0090] S42. Construct a maximum margin optimization model for linear and nonlinear space, expressed by the following formula:

[0091] Linear Maximum Margin Optimization Model

[0092]

[0093] The nonlinear maximum margin optimization model is

[0094]

[0095] S43. Solve the QP problem using the standard algorithm and obtain the optimal solution: and nearest neighbor and

[0096] S44. Get the weight tensor ω * and intercept b * :

[0097]

[0098] According to the linear equation of the hyperplane, ω T x + b = 0, we get:

[0099]

[0100] S45. Finally, the decision function f(x) is calculated:

[0101] Decision function for linear models

[0102]

[0103] Decision Function for Nonlinear Models

[0104]

[0105] See Figure 2 An embodiment of the present application provides a bearing fault diagnosis method and device for a multi-sensor driven flexible support tensor machine, including a multi-source signal acquisition module 10, a feature tensor acquisition module 20, a tensor machine model building module 30, a decision function acquisition module 40 and a verification module 50.

[0106] The multi-source signal acquisition module 10 is used to acquire multi-source signals under different health conditions of the bearing, and obtain multi-source signals of samples corresponding to each health condition. The multi-source signals are acquired by sound sensors and vibration sensors.

[0107] The feature tensor acquisition module 20 is used to obtain a third-order feature tensor of each sample according to a multi-source signal tensor feature construction method for the multi-source signal of each sample acquired.

[0108] The tensor machine model building module 30 is used to introduce flexibility factors and permutation factors to build a flexible support tensor machine model based on the convex hull geometric model.

[0109] The decision function acquisition module 40 is used to randomly select a small number of third-order feature tensors of samples from each health state as tensor training samples to train the flexible support tensor machine model to obtain an intelligent fault diagnosis decision function.

[0110] The testing module 50 is used to form test samples from the remaining samples of each health state to test the feasibility of the trained intelligent fault diagnosis decision function.

[0111] The device in the embodiments of the present application may also be a component, integrated circuit, or chip in a terminal. The device may be a mobile electronic device or a non-mobile electronic device. For example, the mobile electronic device may be a mobile phone, a tablet computer, a laptop computer, a PDA, an in-vehicle electronic device, a wearable device, an ultra-mobile personal computer (UMPC), a netbook, or a personal digital assistant (PDA), etc. The non-mobile electronic device may be a server, a network attached storage (NAS), a personal computer (PC), a television (TV), a teller machine, or a self-service machine, etc., and the embodiments of the present application do not specifically limit this.

[0112] The device in the embodiment of the present application may be a device having an operating system. The operating system may be an Android operating system, an iOS operating system, or other possible operating systems, which are not specifically limited in the embodiment of the present application.

[0113] The display device provided by the embodiment of the present application can achieve Figure 1 To avoid repetition, the various processes implemented in the method embodiment are not described here.

[0114] Optional, such as Figure 3As shown, an embodiment of the present application also provides an electronic device 100, which includes a processor 101, a memory 102, and a program or instruction stored in the memory 102 and executable on the processor 101. When the program or instruction is executed by the processor 101, each process of the above-mentioned embodiment of the intelligent fault diagnosis method for bearings based on multi-sensor signal fusion and flexible support tensor machine is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0115] It should be noted that the first electronic device in the embodiment of the present application includes the mobile electronic device and the non-mobile electronic device mentioned above.

[0116] An embodiment of the present application also provides a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, the various processes of the above-mentioned embodiment of the bearing intelligent fault diagnosis method based on multi-sensor signal fusion and flexible support tensor machine are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0117] The processor is the processor in the electronic device described in the above embodiment. The readable storage medium includes a computer-readable storage medium, such as a computer read-only memory (ROM), random access memory (RAM), a magnetic disk, or an optical disk.

[0118] An embodiment of the present application further provides a chip, which includes a processor and a communication interface, wherein the communication interface is coupled to the processor, and the processor is used to run programs or instructions to implement the various processes of the above-mentioned embodiment of the bearing intelligent fault diagnosis method based on multi-sensor signal fusion and flexible support tensor machine, and can achieve the same technical effect. To avoid repetition, it will not be repeated here.

[0119] It should be understood that the chip mentioned in the embodiments of the present application can also be called a system-level chip, a system chip, a chip system or a system-on-chip chip, etc.

[0120] The following is a specific example 1 of a bearing fault diagnosis method and device for a multi-sensor driven flexible support tensor machine provided in an embodiment of the present application.

[0121] Example 1

[0122] The selected bearing fault simulation experimental platform is as follows Figure 4As shown, it mainly consists of a drive motor 1, a bearing seat 2, a rotor 3, a test bearing 4, a gearbox 5, and a brake 6. In addition, an acceleration sensor (not shown) and an acoustic sensor (not shown) are also installed on the test bearing 4. The sampling frequency of the sensor is 25.6kHz, and the data acquisition system is LMS. At three speeds (1500, 1800, and 2000rpm), the acceleration signals of the normal bearing state (NC), outer ring fault (OF), inner ring fault (IF), and roller fault (RF) were collected in sequence. Each fault type was set to three different damage levels of 0.2, 0.4, and 0.6mm. Figure 5 Shown are three failure modes of a bearing (0.2 mm).

[0123] This case study includes samples within the distribution and three types of samples outside the distribution. Dataset A0, collected at 1500 rpm, contains ten types of samples: normal (NC), outer race fault 0.2 mm (OF1), outer race fault 0.4 mm (OF2), outer race fault 0.6 mm (OF3), inner race fault 0.2 mm (IF1), inner race fault 0.4 mm (IF2), inner race fault 0.6 mm (IF3), roller fault 0.2 mm (RF1), roller fault 0.4 mm (RF2), and roller fault 0.6 mm (RF3). Unlike dataset A0, datasets A1 and A2 are single-channel signals, while dataset A3 is a two-channel spliced ​​signal.

[0124] Table 1 Comparison of bearing fault diagnosis results using different methods

[0125]

[0126]

[0127] Table 1 shows the comparison results of the method of the present invention (Flexible replaceable support tensor machine, FRSTM) and intelligent recognition methods such as support vector machine (SVM), maximum margin classification based on Flexible Convex Hulls (MMC-FCH), extreme gradient boosting (XGBoost), and random forest (RF) after ten independent runs.

[0128] Table 2 Diagnosis results of different methods under different numbers of outliers

[0129]

[0130] Table 2 shows the comparison results of the method of the present invention with intelligent recognition methods such as SVM, MMC-FCH, XGBoost and RF after ten independent runs of experiments with 30 training samples and 10 test samples for each category of outliers.

[0131] The input for the method of the present invention is tensor data. The input for the support vector machine is data A1, A2, and A3. The input for the maximum margin elastic convex hull classification is A3, the input for the extreme gradient boosting is A3, and the input for the random forest is A3. Finally, the final diagnosis results of FRSTM and its comparison method under different training samples are shown in Table 1, and the final diagnosis results after adding outliers are shown in Table 2. A specific analysis of the average recognition results of the method of the present invention over 10 runs shows that FRSTM has the highest accuracy.

[0132] Refer again Figure 6 As shown, the content of the present invention can be mainly divided into four parts. The first part is to collect the simulated time domain vibration signals under different health states of the bearing fault simulation experimental platform as sample data sets for extraction of multi-source signal feature tensors, and then divide them into training samples and test samples respectively; the second part is to introduce an elasticity factor and a shift factor based on the convex hull model to obtain a looser estimation model in view of the under-estimation problem of the convex hull geometric model of the tensor space; the third part is to randomly select the third-order feature tensors of some samples from each state as training samples, and use the training samples to train the FRSTM model to obtain its decision function; the fourth part is to use the remaining samples of each state as test samples, and use the decision function of the FRSTM model trained in the previous step to test the test samples.

[0133] Refer again Figure 7 Figure 2 shows the raw time-domain vibration signals of various bearing fault conditions tested experimentally, including four different bearing fault states: normal state, inner race fault, rolling element fault, and outer race fault. The sampling frequency was 25.6 kHz, the sampling time was 16 seconds, and the number of sampling points was 537,600. The horizontal axis in the figure represents time in seconds, and the vertical axis represents amplitude in meters per second. Figure 2 shows the raw time-domain sound signals of various bearing fault conditions tested experimentally, including four different bearing fault states: normal state, inner race fault, rolling element fault, and outer race fault. The sampling frequency was 25.6 kHz, the sampling time was 16 seconds, and the number of sampling points was 537,600. The horizontal axis in the figure represents time in seconds, and the vertical axis represents amplitude in meters per second. Points 1 to 163,840 were selected to divide the training and test samples, with each sample containing 1,024 points, and no overlap between the two samples, for a total of 400 samples.

[0134] It should be noted that, in this article, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the statement "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device comprising the element. In addition, it should be noted that the scope of the methods and devices in the embodiments of the present application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in the opposite order according to the functions involved. For example, the described method may be performed in an order different from that described, and various steps may also be added, omitted, or combined. In addition, the features described with reference to certain examples may be combined in other examples.

[0135] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus the necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for enabling a terminal (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0136] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of this application, ordinary technicians in this field can also make many forms without departing from the purpose of this application and the scope of protection of the claims, all of which are within the protection of this application.

Claims

1. A bearing fault diagnosis method for a multi-sensor signal driven flexible support tensor machine, characterized in that: include: S1. Collect multi-source signals of bearings in different health states to obtain multi-source signals of samples corresponding to each health state; S2. Obtaining a third-order feature tensor of each sample from the multi-source signal of each sample collected according to a multi-source signal tensor feature construction method; S3. Based on the convex hull geometry model, the flexible factor and permutation factor are introduced to create a flexible support tensor machine model; S4. Randomly select a small number of third-order feature tensors from each health state as tensor training samples to train the flexible support tensor machine model to obtain an intelligent fault diagnosis decision function; S5. The remaining samples of each health state constitute test samples to test the feasibility of the trained intelligent fault diagnosis decision function.

2. The method according to claim 1, characterized in that In step S1, multi-source signals are collected by sound sensors and vibration sensors.

3. The method according to claim 1, characterized in that In step S2, the third-order characteristic tensor is a frequency band component-statistical parameter-multi-sensor third-order characteristic tensor.

4. The method according to claim 3, characterized in that Step S2 specifically includes: For each sample of the multi-source signal collected, each signal in the multi-source signal is decomposed into M components using ensemble empirical mode decomposition; Extract statistical characteristic parameters and extract H time domain parameters from the component components decomposed from each signal, where the characteristic of each signal in the sample is expressed as ; Get the feature tensor. The feature of each sample can be expressed as a third-order feature tensor ,in , corresponding to the frequency band component M, statistical parameter H and number of sensors P respectively; The frequency band component, statistical parameter and multi-sensor third-order feature tensor of each sample are formed.

5. The method according to claim 4, characterized in that The time domain parameters include mean, variance, root mean square, peak-to-peak value, skewness, and kurtosis.

6. The method according to claim 4, characterized in that Step S3 specifically includes: S31. Define the convex hull geometric model of the tensor space: ; in is the corresponding tensor sample The combination coefficient of the tensor is the convex hull containing A minimal convex set of tensor samples; S32. Introduce flexibility factor and permutation factor to define a new geometric model of tensor space ; in is the flexibility factor that can make the geometric model looser, is a permutation factor that can improve the robustness of the geometric model.

7. The method according to claim 6, characterized in that In step S4, the tensor training sample can be expressed as , ; The corresponding label is .

8. The method according to claim 7, characterized in that Step S5 specifically includes: S41. Use Tucker decomposition to decompose the tensor training sample and obtain , where G is the kernel tensor; The factor matrix represented by ; Represents the column vectors of each factor matrix; Then select a suitable kernel function to embed the kernel function of Tucker decomposition, which is expressed by the following formula: ； S42. Construct a maximum margin optimization model for linear and nonlinear space, expressed by the following formula: Linear Maximum Margin Optimization Model ； The nonlinear maximum margin optimization model is ； S43. Solve the QP problem using the standard algorithm and obtain the optimal solution: and nearest neighbor and ; S44. Get weight tensor and intercept : ； According to the linear equation of the hyperplane, get: ； S45. Finally, calculate the decision function : Decision function for linear models ； Decision Function for Nonlinear Models 。 9. A bearing intelligent fault diagnosis device based on multi-sensor signal fusion and flexible support tensor machine, characterized in that: include: A multi-source signal acquisition module is used to collect multi-source signals under different health conditions of the bearing and obtain multi-source signals of samples corresponding to each health condition; A feature tensor acquisition module is used to obtain a third-order feature tensor of each sample according to a multi-source signal tensor feature construction method for the multi-source signal of each sample collected; A tensor machine model building module is used to introduce flexibility factors and permutation factors based on the convex hull geometry model to build a flexible support tensor machine model; The decision function acquisition module is used to randomly select a small number of third-order feature tensors from each health state as tensor training samples to train the flexible support tensor machine model and obtain the intelligent fault diagnosis decision function; The testing module is used to form test samples from the remaining samples of each health state to test the feasibility of the trained intelligent fault diagnosis decision function.

10. An electronic device, characterized in that: It comprises a processor, a memory, and a program or instruction stored in the memory and executable on the processor, wherein when the program or instruction is executed by the processor, the steps of the method for intelligent fault diagnosis of bearings using multi-sensor signal fusion and flexible support tensor machine are implemented as described in any one of claims 1 to 8.

Images