Rolling bearing fault diagnosis method based on Schwang optimization algorithm parameter optimization variational mode decomposition

Abstract

The invention discloses a rolling bearing fault diagnosis method and a rolling bearing fault diagnosis system based on parameter optimization variational mode decomposition of a South goose optimization algorithm, and belongs to the technical field of rotating machinery fault diagnosis. The method comprises the following steps: firstly, carrying out noise reduction processing on an acquired bearing vibration signal by adopting a wavelet threshold value; then introducing a Schwander optimization algorithm, and carrying out global optimization on key parameters (a modal decomposition number K and a penalty factor alpha) of the VMD by taking the minimum envelope entropy as a fitness function; performing VMD decomposition on the signal by using the optimized parameters, and screening out an intrinsic mode function (IMF) component with the minimum envelope entropy; and converting the IMF component into a grey-scale map, and inputting the grey-scale map into a Darknet-53 deep learning model for fault type identification. Experimental results show that the method can effectively extract fault features, the diagnosis accuracy is as high as 99.17%, and the method has the advantages of high convergence rate, high global optimization capability, good anti-noise performance and the like, and is suitable for intelligent fault diagnosis of the rolling bearing under complex working conditions.

Description

Technical Field

[0001] This invention relates to the field of mechanical equipment fault diagnosis technology, and in particular to a method and system for rolling bearing fault diagnosis based on wavelet threshold denoising, Snow Goose Optimization Algorithm (SGA) optimization of variational mode decomposition (VMD) parameters, and combined with a deep learning model. Background Technology

[0002] Rolling bearings are critical components in rotating machinery, and their operating status directly affects the safety and stability of the entire machine. Due to their complex working environment, fault characteristics are often masked by noise, making it difficult for traditional signal processing methods to accurately extract fault features. In recent years, Variational Mode Decomposition (VMD) has been widely used in fault diagnosis due to its excellent mode separation performance; however, its decomposition effect is heavily dependent on the selection of the penalty factor α and the number of modes K. In existing technologies, swarm intelligence optimization algorithms (such as sparrow search algorithm and locust optimization algorithm) are often used to optimize VMD parameters, but these methods suffer from problems such as being prone to getting trapped in local optima and slow convergence speed, leading to inaccurate parameter selection and thus affecting the accuracy and efficiency of fault diagnosis. Summary of the Invention

[0003] This invention aims to provide a method and system for diagnosing rolling bearing faults by combining wavelet thresholding denoising, Snow Goose Optimization (SGA) algorithm to optimize VMD parameters, and image classification using a deep learning model, thereby addressing the problems of poor parameter optimization and low diagnostic accuracy in existing technologies. To achieve the above objectives, this invention provides the following technical solution:

[0004] Step 1: Signal Acquisition and Preprocessing

[0005] Vibration signals of rolling bearings are collected, and the original signals are denoised using the wavelet thresholding method to obtain the denoised real signals.

[0006] Step 2: VMD Parameter Optimization

[0007] Using the denoised signal as input, the Snow Goose Optimization (SGA) algorithm is used to globally optimize the two key parameters of VMD—the mode decomposition number K and the penalty factor α; the minimum envelope entropy is used as the fitness function to select the optimal parameter combination.

[0008] Step 3: Signal Decomposition and IMF Screening

[0009] The signal is decomposed using the optimized K and α parameters to obtain several intrinsic mode functions (IMFs), and the IMF with the smallest envelope entropy is selected as the optimal component.

[0010] Step 4: Grayscale image conversion

[0011] The optimal IMF component is segmented into segments of fixed length, and a weighted average method is used to convert it into a two-dimensional grayscale image, preserving the temporal characteristics and energy distribution of the signal.

[0012] Step 5: Fault Classification

[0013] Grayscale images are input into the Darknet-53 deep learning model for training and classification to identify the fault types of rolling bearings.

[0014] The above-described solution of the present invention has the following beneficial effects:

[0015] Significant noise reduction effect: By using wavelet thresholding, environmental noise is effectively removed and signal quality is improved.

[0016] Precise parameter optimization: The introduction of the Xueyan optimization algorithm has the advantages of strong global search capability, fast convergence speed, and strong ability to escape local optima, which significantly improves the accuracy of VMD parameter selection.

[0017] High diagnostic accuracy: Combining a deep learning image classification model, it achieves efficient identification of fault features, with experimental verification showing a fault identification rate of up to 99.17%. Attached Figure Description

[0018] Figure 1 Fault Diagnosis Method Flowchart Figure 2 Darknet-53 model structure diagram Figure 3 Unprocessed confusion matrix diagram

[0019] Figure 4 Wavelet threshold-VMD confusion matrix diagram Figure 5 Wavelet thresholding-SGA-VMD confusion matrix diagram Figure 6 : Bearing vibration data acquisition system diagram Detailed Implementation

[0020] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings:

[0021] Implementation Case:

[0022] Step 1: Collect the vibration signal of the rolling bearing, perform 5-level wavelet decomposition using the db4 wavelet basis, and perform noise reduction using a hard threshold function.

[0023] Step 2: Initialize the Snow Goose optimization algorithm population, set the search space as K∈[2,10], α∈[100,2000], and use the minimum envelope entropy as the fitness function.

[0024] Step 3: After SGA optimization, the optimal parameter combination is obtained (e.g., K=4, α=100), and VMD decomposition is performed.

[0025] Step 4: Select the IMF2 component with the smallest envelope entropy and convert it to a grayscale image.

[0026] Step 5: Input the grayscale image into the Darknet-53 model for training, and finally achieve accurate identification of different fault types with an accuracy rate of 99.17%.

Claims

1. A method for diagnosing rolling bearing faults based on variational mode decomposition of parameters using the Xueyan optimization algorithm, characterized in that, include: Step 1: Acquire the raw vibration signal of the rolling bearing; Step 2: Perform wavelet threshold denoising on the original vibration signal to obtain a denoised signal; Step 3: Use the Snow Goose optimization algorithm to globally optimize the mode decomposition number K and penalty factor α of variational mode decomposition, with the minimum envelope entropy of the fitness function as the optimization objective, and output the optimal parameter combination [K, α]. Step 4: Use the optimal parameter combination to perform variational mode decomposition on the denoised signal to obtain several intrinsic mode function (IMF) components, calculate the envelope entropy of each IMF component, and select the IMF component with the smallest envelope entropy as the optimal IMF component. Step 5: Divide the optimal IMF component into segments of fixed length and convert them into a two-dimensional grayscale image using a weighted average method; Step 6: Input the grayscale image into the Darknet-53 deep learning model for training and classification to achieve intelligent identification of rolling bearing fault types.

2. The method according to claim 1, characterized in that: In step 2, the wavelet thresholding noise reduction process uses the db4 wavelet basis for multi-level decomposition and combines it with a hard threshold function to process the wavelet coefficients in order to preserve the effective features of the signal and suppress noise interference.

3. The method according to claim 1, characterized in that: The Xueyan optimization algorithm in step 3 includes a three-stage structure: (1) Initialization phase: Randomly generate a population of snow geese in the search space, with each snow goose representing a candidate parameter combination [K,α]; (2) V-shaped exploration stage: Simulate the V-shaped migration behavior of snow geese and conduct a global search through speed update and position adjustment mechanisms; (3) Linear development stage: When the population angle exceeds the set threshold, switch to linear flight mode, introduce random perturbation and Brownian motion mechanism to enhance the ability to escape local optima.

4. The method according to claim 1, characterized in that: The calculation of minimum envelope entropy in step 4 is based on the envelope signal of the IMF component. The smaller the entropy value, the higher the signal sparsity and the more obvious the fault characteristics.

5. The method according to claim 1, characterized in that: In step 5, during the grayscale image conversion process, the grayscale value is calculated using the following weighted average method: R, G, and B are analog color channel values, generated by amplitude mapping of IMF components.

6. The method according to claim 1, characterized in that: In step 6, the Darknet-53 model automatically extracts fault features from grayscale images through a convolutional neural network, enabling efficient classification and identification of different fault types in rolling bearings.

7. A rolling bearing fault diagnosis system based on variational mode decomposition of parameters using the Xueyan optimization algorithm, characterized in that, include: (1) Signal acquisition module, used to acquire the original vibration signal of the rolling bearing; (2) Noise reduction module, used to perform wavelet threshold noise reduction on the original vibration signal; (3) Parameter optimization module, used to optimize the parameters K and α of variational mode decomposition based on the Xueyan optimization algorithm; (4) Signal decomposition module, used to perform VMD decomposition on the signal according to the optimal parameters and select the optimal IMF component; (5) Image conversion module, used to convert the optimal IMF components into grayscale images; (6) Fault classification module, used to identify fault types in grayscale images using the Darknet-53 model.

8. The system according to claim 7, characterized in that: In step (3), the Xueyan optimization algorithm integrated in the parameter optimization module has global search and local development capabilities, and can quickly converge and avoid getting trapped in local optimal solutions.

9. The system according to claim 7, characterized in that: In step (6), the identification results output by the fault classification module include the normal state of the rolling bearing, inner ring fault, outer ring fault and rolling element fault type, with an identification accuracy of not less than 99%.

Images