Bearing fault diagnosis method and system based on multi-scale and multi-direction Log-Gabor modulation convolution

Abstract

The invention relates to the technical field of mechanical fault diagnosis, in particular to a bearing fault diagnosis method and system based on multi-scale and multi-direction Log-Gabor modulation convolution, and the method comprises the steps: obtaining first vibration signal data of a rolling bearing test bed, and carrying out the first preprocessing of the first vibration signal data, and obtaining second vibration signal data; presetting a first bearing fault detection model and inputting the second vibration signal data into the first bearing fault detection model; and carrying out bearing fault intelligent diagnosis according to the output of the first bearing fault detection model. A multi-scale and multi-direction Log-Gabor filter is adopted to modulate a convolution kernel, features of different scales and directions are extracted from vibration signals, the comprehensiveness of feature extraction is improved, a feature fusion mechanism based on energy entropy is introduced, adaptive weighting of the features of different scales and directions is achieved, and the problem of feature redundancy is solved; finally, high diagnosis accuracy can still be kept under the complex working conditions and the strong noise environment, and the method is easy to operate, high in applicability and capable of being directly applied to industrial sites.

Description

Technical Field

[0001] The present invention relates to the technical field of mechanical fault diagnosis, and in particular to a bearing fault diagnosis method and system based on multi-scale and multi-directional Log-Gabor modulation convolution. Background Art

[0002] Rotating machinery is one of the most important equipment in the industrial field. However, due to the complex and harsh working conditions of rotating machinery, various faults often occur, causing economic losses and even threatening personal safety. Therefore, it is very necessary to accurately diagnose the faults of rotating machinery. With the development of science and technology, the vibration signals of rotating machinery can be easily collected by sensors in industry. The fault of rotating machinery will generate a pulse with a short duration in the vibration signal, which can be used for fault diagnosis. In addition, the vibration signals collected from rotating machines are usually composed of many vibration components, such as the meshing of gears and the rotation of bearings. Therefore, there are often some non-stationary and nonlinear components in the vibration signal. Therefore, how to extract fault features from vibration signals under strong background noise is a challenge.

[0003] Signal processing is one of the most effective tools for bearing fault diagnosis, but it is always time-consuming and requires a lot of manual intervention. With the advancement of computer technology, deep learning has achieved certain success in mechanical equipment fault diagnosis, thanks to its powerful feature learning ability. However, in some strong noise environments, it is often difficult to obtain satisfactory results using deep learning models. Log-Gabor filters can extract features of different scales and directions from vibration signals by designing different scales, bandwidths and other parameters. Therefore, it is very meaningful to design multi-scale and multi-directional Log-Gabor filters to modulate convolution kernels to construct deep learning networks and extract fault features from vibration signals. However, since the features of different scales and directions extracted by multi-scale and multi-directional Log-Gabor filters may be redundant, the model's ability to recognize the differences between features is insufficient, which may cause the model to fall into a local minimum during the training process. Therefore, it is necessary to design a feature fusion method to guide the model to focus on key features. The feature fusion method based on energy entropy can highlight important features and effectively adjust the model's attention to features of different scales and directions according to the training process. Therefore, a deep learning network is constructed based on the idea of ​​Log-Gabor filter modulated convolution kernel, and the feature fusion based on energy entropy is used to guide the model to focus on important features. This is very meaningful for achieving high-accuracy rolling bearing fault diagnosis under various fault types and damage degrees. Summary of the invention

[0004] In view of the problems existing in the prior art, the present invention is proposed.

[0005] Therefore, the problem to be solved by the present invention is how to solve the problem that the working conditions of rotating machinery are complex and harsh, various faults often occur, and how to extract fault characteristics from vibration signals under strong background noise.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, an embodiment of the present invention provides a bearing fault diagnosis method based on multi-scale multi-directional Log-Gabor modulation convolution, which includes obtaining first vibration signal data of a rolling bearing test bench, and performing a first preprocessing on the first vibration signal data to obtain second vibration signal data;

[0008] Presetting a first bearing fault detection model, and inputting the second vibration signal data into the first bearing fault detection model;

[0009] Intelligent bearing fault diagnosis is performed based on the output of the first bearing fault detection model.

[0010] As a preferred solution of the bearing fault diagnosis method based on multi-scale multi-directional Log-Gabor modulation convolution described in the present invention, wherein: the first vibration signal data includes: first-class vibration signal data, second-class vibration signal data, and third-class vibration signal data;

[0011] The first vibration signal data are all vibration signal data collected by the rolling bearing test bench under preset working conditions.

[0012] As a preferred solution of the bearing fault diagnosis method based on multi-scale and multi-directional Log-Gabor modulation convolution described in the present invention, wherein: the first preprocessing includes signal sampling preprocessing and data enhancement preprocessing;

[0013] The signal sampling preprocessing includes dividing the first vibration signal data according to a preset sampling length;

[0014] The data enhancement preprocessing is used to expand the first vibration signal data after the signal sampling preprocessing.

[0015] As a preferred solution of the bearing fault diagnosis method based on multi-scale and multi-directional Log-Gabor modulated convolution described in the present invention, wherein: the first bearing fault detection model includes: a multi-scale and multi-directional Log-Gabor filter module, which is used to modulate the convolution kernel through multiple Log-Gabor filters of different scales and directions;

[0016] Feature fusion module, used to fuse features of different scales and directions based on energy entropy;

[0017] The classifier module is used to classify the fault type based on the fused features.

[0018] As a preferred solution of the bearing fault diagnosis method based on multi-scale multi-directional Log-Gabor modulated convolution described in the present invention, the filtering process of the multi-scale multi-directional Log-Gabor filter module includes:

[0019] Inputting the second vibration signal data into a plurality of Log-Gabor filters of different scales and directions for feature extraction;

[0020] Calculate the energy entropy of the extracted features to obtain the weight coefficient of each feature;

[0021] Different features are weightedly fused according to the weight coefficient.

[0022] As a preferred solution of the bearing fault diagnosis method based on multi-scale and multi-directional Log-Gabor modulation convolution described in the present invention, the feature fusion module includes:

[0023] An energy calculation unit, used to calculate the energy value of each channel feature;

[0024] A weight coefficient generating unit, used for calculating a weight coefficient of feature fusion based on the energy value;

[0025] The feature weighting unit is used to perform weighted fusion on the features according to the weight coefficient.

[0026] As a preferred solution of the bearing fault diagnosis method based on multi-scale and multi-directional Log-Gabor modulated convolution described in the present invention, the multi-scale and multi-directional Log-Gabor filter module includes filters of 4 scales and 4 directions, constituting 16 Log-Gabor filters of different scales and directions.

[0027] In a second aspect, an embodiment of the present invention provides a bearing fault diagnosis system based on multi-scale multi-directional Log-Gabor modulation convolution, which includes a signal acquisition module, which acquires first vibration signal data of a rolling bearing test bench, and performs a first preprocessing on the first vibration signal data to obtain second vibration signal data;

[0028] A model operation module, presetting a first bearing fault detection model, and inputting the second vibration signal data into the first bearing fault detection model;

[0029] The result output module performs intelligent diagnosis of bearing faults according to the output of the first bearing fault detection model.

[0030] In a third aspect, an embodiment of the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program instructions are executed by the processor, the steps of the bearing fault diagnosis method based on multi-scale and multi-directional Log-Gabor modulated convolution as described in the first aspect of the present invention are implemented.

[0031] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program instructions are executed by a processor, the steps of the bearing fault diagnosis method based on multi-scale and multi-directional Log-Gabor modulated convolution as described in the first aspect of the present invention are implemented.

[0032] The beneficial effects of the present invention are as follows: by setting up three acceleration sensors to collect multi-channel vibration signals and dividing the samples according to the length of 2048, the vibration characteristics of different positions of the bearing are fully captured, the problem of incomplete information collection by a single sensor is overcome, and a high-quality data foundation is provided for subsequent feature extraction. By constructing a Log-Gabor filter with 16 different parameters in 4 scales and 4 directions, a multi-scale and multi-directional feature extraction network is formed, which effectively overcomes the limitation that a single scale or direction filter is difficult to cope with complex fault characteristics. By introducing a feature fusion mechanism based on energy entropy, adaptive weighting of features of different scales and directions is achieved, and the problem of poor model training effect caused by feature redundancy is solved, so that the model can automatically focus on the most discriminative fault features, and the accuracy and reliability of diagnosis are improved. By adopting a classifier structure with two fully connected layers and combining the random gradient descent method to update the network parameters, an efficient mapping of features to fault types is achieved, overfitting problems caused by overly complex network structures are avoided, and the generalization ability of the model in practical applications is guaranteed. Compared with the traditional single feature extraction method, the present invention can maintain a high diagnostic accuracy in a strong noise environment through the synergy of multi-scale and multi-directional features. The improvement in anti-interference ability exceeds expectations.

[0033] The feature fusion mechanism based on energy entropy not only improves the diagnostic accuracy, but also significantly reduces the model training time. This improvement in training efficiency is an additional advantage discovered unexpectedly. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0035] Figure 1This is a flow chart of the bearing fault diagnosis method based on multi-scale and multi-directional Log-Gabor modulation convolution;

[0036] Figure 2 Computer equipment diagram for bearing fault diagnosis method based on multi-scale and multi-directional Log-Gabor modulation convolution;

[0037] Figure 3 This is the multi-scale and multi-directional Log-Gabor network structure diagram of the bearing fault diagnosis method based on multi-scale and multi-directional Log-Gabor modulated convolution. DETAILED DESCRIPTION

[0038] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0039] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0040] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or selective embodiment that is mutually exclusive with other embodiments.

[0041] Example 1

[0042] Reference Figure 1-2 , which is the first embodiment of the present invention, provides a bearing fault diagnosis method based on multi-scale and multi-directional Log-Gabor modulation convolution, including:

[0043] S100: Acquire first vibration signal data of a rolling bearing test bench, and perform first preprocessing on the first vibration signal data to obtain second vibration signal data;

[0044] Preferably, a rolling bearing test bench is built to collect real vibration signal data of rolling bearings under different speed conditions; a rolling bearing test bench is established to simulate and collect data on various fault types and damage degrees that may be encountered in real scenarios, such as inner ring fault 0.2mm, outer ring fault 0.4mm, etc.

[0045] Specifically, the rolling bearing test bench consists of a drive motor, rolling bearing, gearbox, accelerometer and signal acquisition system. The accelerometer is used to collect vibration signals with a sampling frequency of 20 khz. Three acceleration sensors are installed on the gearbox to collect multi-channel vibration signals. A total of seven health conditions of bearings with different fault types and damage degrees are considered.

[0046] In this embodiment, the vibration signal is divided into data sets according to a single sample length of 2048.

[0047] S101: The first vibration signal data includes: first-type vibration signal data, second-type vibration signal data, and third-type vibration signal data;

[0048] The first vibration signal data are all vibration signal data collected by the rolling bearing test bench under preset working conditions.

[0049] Preferably, a multi-scale and multi-directional Log-Gabor filter is constructed to modulate the convolution kernel to extract the vibration signal features; the convolution kernel is modulated using the multi-scale and multi-directional characteristics of the Log-Gabor filter to achieve the extraction of features of different scales and directions from the vibration signal.

[0050] Preferably, the collected real vibration signal data of rolling bearings under different speed conditions and the corresponding label inputs are used to construct a multi-scale and multi-orientation Log-Gabor network (MMLGNet) to extract features of different scales and directions;

[0051] Specifically, the number of scales and the number of directions of the Log-Gabor filter are both set to 4.

[0052] In this embodiment, the classification error is used as the loss function in the model training stage.

[0053] S102: The first preprocessing includes signal sampling preprocessing and data enhancement preprocessing;

[0054] The signal sampling preprocessing includes dividing the first vibration signal data according to a preset sampling length;

[0055] The data enhancement preprocessing is used to expand the first vibration signal data after the signal sampling preprocessing.

[0056] S200: Preset a first bearing fault detection model, and input the second vibration signal data into the first bearing fault detection model;

[0057] S201: The first bearing fault detection model includes: a multi-scale and multi-directional Log-Gabor filter module, which is used to modulate the convolution kernel through multiple Log-Gabor filters of different scales and directions;

[0058] Feature fusion module, used to fuse features of different scales and directions based on energy entropy;

[0059] The classifier module is used to classify the fault type based on the fused features.

[0060] S202: The filtering process of the multi-scale multi-directional Log-Gabor filter module includes:

[0061] Inputting the second vibration signal data into a plurality of Log-Gabor filters of different scales and directions for feature extraction;

[0062] Calculate the energy entropy of the extracted features to obtain the weight coefficient of each feature;

[0063] Different features are weightedly fused according to the weight coefficient.

[0064] Preferably, a feature fusion mechanism based on energy entropy is introduced to guide the model to pay attention to the characteristics of features of different scales and directions and highlight more important features;

[0065] During the offline training stage, in order to make the model focus on key features, a feature fusion mechanism based on energy entropy is introduced to highlight the key information in features of different scales and directions.

[0066] In this embodiment, the stochastic gradient descent method is used to update the network weight parameters.

[0067] S203: The feature fusion module includes:

[0068] An energy calculation unit, used to calculate the energy value of each channel feature;

[0069] A weight coefficient generating unit, used for calculating a weight coefficient of feature fusion based on an energy value;

[0070] The feature weighting unit is used to perform weighted fusion of features according to weight coefficients.

[0071] S300: Perform intelligent bearing fault diagnosis according to the output of the first bearing fault detection model.

[0072] S301: The multi-scale and multi-directional Log-Gabor filter module includes filters of 4 scales and 4 directions, forming 16 Log-Gabor filters of different scales and directions.

[0073] Preferably, the test set is input into the trained model for online testing to achieve rolling bearing fault diagnosis under real working conditions.

[0074] Figure 3 MMLGNet network structure diagram of an embodiment of the present invention.

[0075] like Figure 3 As shown, in this embodiment, the MMLGNet network structure includes a multi-scale and multi-directional Log-Gabor filter, a classifier, and a feature fusion module based on energy entropy.

[0076] Specifically, the filters of 4 scales and 4 directions of the multi-scale and multi-directional Log-Gabor filter module are combined to form 16 Log-Gabor filters of different scales and directions. The classifier consists of two fully connected layers. The feature fusion module based on energy entropy uses the energy entropy of features of different scales and directions to guide the model to focus on important features. The specific feature extraction process is:

[0077] Step T1: Use a multi-scale and multi-directional Log-Gabor filter to modulate the convolution kernel. The modulation process is as follows:

[0078]

[0079] Where r and θ are the radial coordinate and angular coordinate of the Log-Gabor filter in polar coordinates, s and o are the indicators of scale and direction (s = 0, 1, nscales-1; o = 0, 1, norientations-1). and are used to set the radial and angular bandwidth of the filter, respectively, 0 (s) is the scale-varying central spatial frequency, and λ(s) is the scale-varying wavelength.

[0080] Step T2: After modulating the convolution kernel, the input data is used to extract features of different scales and directions. The feature fusion method based on energy entropy is used to guide the model to focus on key features. The feature fusion process based on energy entropy is as follows:

[0081]

[0082] f MMLG =W c ·f c #(5)

[0083] Where E i is the energy of the i-th channel feature, n is the length of the feature, X ik The amplitude of each discrete point of each vibration signal, W is the energy entropy, W c and f care the energy entropy and extracted features of the cth channel, respectively, MMLG It is the feature after energy entropy weighted fusion.

[0084] The test set is input into the trained model for online testing to realize rolling bearing fault diagnosis under real working conditions.

[0085] The workflow of defect feature recognition of the fault diagnosis model is steps S100 to S300.

[0086] Furthermore, this embodiment also provides a bearing fault diagnosis system based on multi-scale multi-directional Log-Gabor modulation convolution, comprising:

[0087] A signal acquisition module, which acquires first vibration signal data of the rolling bearing test bench and performs a first preprocessing on the first vibration signal data to obtain second vibration signal data;

[0088] A model operation module presets a first bearing fault detection model and inputs the second vibration signal data into the first bearing fault detection model;

[0089] The result output module performs intelligent diagnosis of bearing faults according to the output of the first bearing fault detection model.

[0090] In summary, this embodiment provides a bearing fault diagnosis method based on multi-scale multi-directional Log-Gabor modulated convolution, which is used for rolling bearing fault diagnosis, including the following steps: building a rolling bearing test bench to collect real vibration signal data of rolling bearings under different speed conditions; constructing a multi-scale multi-directional Log-Gabor filter for modulating the convolution kernel to extract vibration signal features; inputting the collected real vibration signal data of rolling bearings under different speed conditions and the corresponding labels into the constructed MMLGNet to extract features of different scales and directions; introducing a feature fusion mechanism based on energy entropy to guide the model to pay attention to the characteristics of features of different scales and directions and highlight more important features; inputting the test set into the trained model for online testing to realize rolling bearing fault diagnosis under real working conditions. The rolling bearing fault diagnosis method of the present invention can effectively perform mechanical fault diagnosis on fault samples under various fault types and damage degrees.

[0091] Furthermore, the bearing fault diagnosis method based on multi-scale and multi-directional Log-Gabor modulated convolution provided in this embodiment makes full use of the convolution kernels modulated by Log-Gabor filters of different scales and directions, so that the model can extract features containing rich scale and direction information from the vibration signal.

[0092] The bearing fault diagnosis method based on multi-scale and multi-directional Log-Gabor modulated convolution provided in this embodiment introduces a feature fusion method based on energy entropy, so that the model focuses on key features, thereby achieving a higher fault diagnosis accuracy.

[0093] Example 2

[0094] Reference Figure 1 - Figure 3 , which is the second embodiment of the present invention, provides a bearing fault diagnosis method based on multi-scale and multi-directional Log-Gabor modulation convolution. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0095] In order to verify the practical application effect of the method of the present invention, a three-month field test was carried out on a rolling mill production line in a large steel plant. The main drive bearing of the 1450mm hot rolling mill of the plant was selected as the test object. The bearing model is SKF23196CAK / W33, and the operating speed range is 300-1200rpm. During the test, vibration data under seven working conditions were collected, including normal working conditions, inner ring fault (two damage degrees of 0.2mm and 0.4mm), outer ring fault (two damage degrees of 0.2mm and 0.4mm), rolling element fault (0.3mm) and compound fault.

[0096] The test device includes: PCB356A16 triaxial acceleration sensor (sensitivity 100mV / g), NI9234 data acquisition card (sampling rate 25.6kHz) and self-developed data acquisition software. The three acceleration sensors are installed in the radial, axial and horizontal directions of the bearing seat respectively. The acquisition time is 8 hours for each working condition, and a set of data is recorded every 30 minutes, and each set of data is 10 seconds long.

[0097] The sliding window method was used in the data preprocessing stage, with the window length set to 2048 points and an overlap rate of 50%, and a total of 28,000 training samples were obtained. In order to enhance the representativeness of the data, various working conditions such as equipment start and stop, load changes, etc. were deliberately covered during the sampling process. By adding Gaussian white noise (with signal-to-noise ratios of 20dB, 15dB, and 10dB, respectively), the training set size was further expanded to 84,000 groups.

[0098] In the feature extraction stage, a multi-scale and multi-directional Log-Gabor filter bank based on logarithmic polar coordinates was designed. The center frequency is distributed from 0.05π to 0.4π in a geometric series (4 scales), and the directional angles are evenly distributed in the range of 0° to 180° (4 directions). The radial bandwidth and angular bandwidth of the filter are set to 0.65 and π / 6, respectively. These parameters are optimized by the grid search method. To ensure the stability of feature extraction, the frequency response of each filter is normalized.

[0099] In the network training phase, an adaptive feature fusion strategy based on energy entropy is adopted. First, the energy distribution of 16 filter channels is calculated, and then the weights of each channel feature are dynamically adjusted based on the normalized energy entropy value. The training adopts the small batch stochastic gradient descent method, and the initial learning rate is set to 0.001, decaying by 10% every 50 epochs, and a total of 200 epochs are trained. To prevent overfitting, dropout (ratio 0.5) and L2 regularization (coefficient 0.0001) are introduced.

[0100] Performance tests were conducted under different working conditions and noise levels, and compared with four mainstream methods: traditional GaborNet, MLFNet, SCNet, and ResNet. To ensure fairness in the comparison, all methods used the same training and test sets and ran in the same hardware environment (Intel Xeon E5-2680 v4 CPU and NVIDIA Tesla V100 GPU). The test results are shown in the following table:

[0101] The results obtained by the bearing fault diagnosis method based on multi-scale multi-directional Log-Gabor modulation convolution in this embodiment are compared with the results obtained by other methods as shown in Table 1:

[0102] Table 1 Comparison of fault diagnosis results obtained by each method

[0103] method 1 2 3 4 5 average MMLGNet 99.54 99.92 99.62 99.46 99.09 99.53 GaborNet 98.33 98.1 97.38 98.89 97.06 97.95 MLFNet 98.1 98.41 98.57 98.33 96.75 98.03 SCNet 81.83 81.13 85.08 79.84 84.13 82.40 ResNet 87.31 87.46 89.76 88.81 89.19 88.51

[0104] From the results in Table 1, it can be seen that the average accuracy of the fault diagnosis model using MMLGNet in this embodiment reaches 99.53%, which is significantly higher than classic methods such as SCNet and ResNet, and has certain advantages over the latest methods such as GaborNet and MLFNet. This shows that the bearing fault diagnosis method based on multi-scale multi-directional Log-Gabor modulated convolution in this embodiment can effectively perform mechanical fault diagnosis on fault samples under various fault types and damage degrees, and has high application value.

[0105] In-depth analysis of the comparative experimental results shows that the method of the present invention (MMLGNet) exhibits significant advantages under various working conditions. Under normal working conditions, the average recognition accuracy reaches 99.54%, which is 1.21 percentage points higher than the second best GaborNet. This experiment uses a five-fold crossover, and it is obvious that MMLGNet has better stability under different folds, especially under the fifth fold condition, the method can still maintain an accuracy of 99.09%, while the performance of traditional methods has dropped significantly, such as SCNet and ResNet The accuracy rate dropped to 84.13% and 89.19% respectively.

[0106] This excellent performance is mainly due to the following innovations: First, the multi-scale and multi-directional Log-Gabor filter provides richer feature expression capabilities. Through experiments, it was found that when processing bearing inner ring faults, the high-frequency scale (0.4π) filter contributed about 45% of the discriminant information, while for outer ring faults, the contribution of the medium-frequency scale (0.2π) reached more than 50%. This shows that filters of different scales do capture the characteristic frequency bands of different types of faults.

[0107] Secondly, the feature fusion mechanism based on energy entropy significantly improves the model's adaptability. Experimental data shows that in the case of a sudden load change, the mechanism can automatically adjust the feature weights within 0.5 seconds, allowing the model to quickly adapt to new working conditions. This solves the problem that traditional methods need to be retrained when working conditions change.

[0108] Finally, in terms of computational efficiency, the average single diagnosis time of this method is 15ms, which meets the requirements of real-time monitoring. Moreover, the model storage space only requires 12MB, which is much smaller than the 45MB of ResNet, making it more suitable for the deployment of edge computing devices. These data fully demonstrate the feasibility and advancement of this invention in engineering practice.

[0109] Example 3

[0110] This embodiment also provides a computer device, which is applicable to a bearing fault diagnosis method based on multi-scale and multi-directional Log-Gabor modulation convolution, and includes a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement a forced oscillation detection and positioning method for a distribution network as proposed in the above embodiment.

[0111] This embodiment further provides a storage medium on which a computer program is stored. When the program is executed by a processor, a forced oscillation detection and positioning method for a distribution network is implemented as proposed in the above embodiment.

[0112] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0113] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0114] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0115] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.

[0116] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used to implement: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0117] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A bearing fault diagnosis method based on multi-scale and multi-directional Log-Gabor modulated convolution, characterized in that: The method comprises acquiring first vibration signal data of a rolling bearing test bench, and performing a first preprocessing on the first vibration signal data to obtain second vibration signal data; Presetting a first bearing fault detection model, and inputting the second vibration signal data into the first bearing fault detection model; Intelligent bearing fault diagnosis is performed based on the output of the first bearing fault detection model.

2. The bearing fault diagnosis method based on multi-scale and multi-directional Log-Gabor modulation convolution according to claim 1 is characterized in that: The first vibration signal data includes: first-type vibration signal data, second-type vibration signal data, and third-type vibration signal data; The first vibration signal data are all vibration signal data collected by the rolling bearing test bench under preset working conditions.

3. The bearing fault diagnosis method based on multi-scale and multi-directional Log-Gabor modulation convolution according to claim 2 is characterized in that: The first preprocessing includes signal sampling preprocessing and data enhancement preprocessing; The signal sampling preprocessing includes dividing the first vibration signal data according to a preset sampling length; The data enhancement preprocessing is used to expand the first vibration signal data after the signal sampling preprocessing.

4. The bearing fault diagnosis method based on multi-scale and multi-directional Log-Gabor modulation convolution according to claim 3 is characterized in that: The first bearing fault detection model includes: a multi-scale and multi-directional Log-Gabor filter module, which is used to modulate the convolution kernel through multiple Log-Gabor filters of different scales and directions; Feature fusion module, used to fuse features of different scales and directions based on energy entropy; The classifier module is used to classify the fault type based on the fused features.

5. The bearing fault diagnosis method based on multi-scale and multi-directional Log-Gabor modulation convolution according to claim 4 is characterized in that: The filtering process of the multi-scale multi-directional Log-Gabor filter module includes: Inputting the second vibration signal data into a plurality of Log-Gabor filters of different scales and directions for feature extraction; Calculate the energy entropy of the extracted features to obtain the weight coefficient of each feature; Different features are weightedly fused according to the weight coefficient.

6. The bearing fault diagnosis method based on multi-scale and multi-directional Log-Gabor modulation convolution according to claim 5, characterized in that: The feature fusion module includes: An energy calculation unit, used to calculate the energy value of each channel feature; A weight coefficient generating unit, used for calculating a weight coefficient of feature fusion based on the energy value; The feature weighting unit is used to perform weighted fusion on the features according to the weight coefficient.

7. The bearing fault diagnosis method based on multi-scale and multi-directional Log-Gabor modulation convolution according to claim 6, characterized in that: The multi-scale and multi-directional Log-Gabor filter module includes filters of 4 scales and 4 directions, forming 16 Log-Gabor filters of different scales and directions.

8. A bearing fault diagnosis system based on multi-scale multi-directional Log-Gabor modulation convolution, based on the bearing fault diagnosis method based on multi-scale multi-directional Log-Gabor modulation convolution according to any one of claims 1 to 7, characterized in that: Also includes, A signal acquisition module, which acquires first vibration signal data of the rolling bearing test bench and performs a first preprocessing on the first vibration signal data to obtain second vibration signal data; A model operation module, presetting a first bearing fault detection model, and inputting the second vibration signal data into the first bearing fault detection model; The result output module performs intelligent diagnosis of bearing faults according to the output of the first bearing fault detection model.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the bearing fault diagnosis method based on multi-scale and multi-directional Log-Gabor modulation convolution described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the bearing fault diagnosis method based on multi-scale and multi-directional Log-Gabor modulation convolution described in any one of claims 1 to 7 are implemented.

Images