10 March 2022 in Mathematical Problems in Engineering
Traditional prognostics and health management (PHM) methods for fault detection require complex signal processing and manual fault feature extraction, and the accuracy is low. To address these problems, a fault diagnosis method of planetary gearbox based on deep belief networks (DBNs) is proposed. Firstly, the vibration signals of the planetary gearbox are collected and analyzed in the time domain and the frequency domain. Then, the DBN model and optimal parameters are determined to meet the task requirements. Finally, the vibration data is divided into training set and test set and input into the DBN model, which can realize the automatic feature extraction and fault recognition of vibration signals. The results show that the identification accuracy reaches 97% under five working conditions of planetary gearbox.
6 December 2022 in Shock and Vibration
Traditional fault diagnosis methods require complex signal processing and expert experience, and the accuracy of fault identification is low. To solve these problems, a fault diagnosis method based on an improved convolutional neural network (CNN) is proposed. Based on the traditional CNN model, a variety of convolution stride modes were added to extract features of different scales of signals and expand the feature dimension. Firstly, the vibration signals were collected and grouped. Then, the data were divided into a training set and a test set and input into improved CNN for feature extraction and model training to realize fault identification. The proposed model achieved a classification accuracy of 99.3% when testing the vibration data of the armored vehicle. Finally, the proposed model was used to classify different fault types of planetary gearboxes. The gradient-weighted class activation mapping (Grad-CAM) method was used to visualize the classification weight of samples. The results showed that the classification accuracy reaches 98% under various working conditions of the planetary gearbox.
17 April 2020 in IEEE Access
The vibration-based analysis is an effective technique for planetary gearbox fault diagnosis, but a difficult task is how to accurately identify fault features from noisy vibration signals. In this paper, a nonconvex wavelet thresholding total variation (WATV) denoising method is proposed for planetary gearbox fault diagnosis, which combines wavelet-domain sparsity and total variation (TV) regularization. The TV regularization algorithm is employed to modify the retained wavelet coefficients so that the occurrence of oscillations caused by wavelet thresholding is suppressed. To overcome the underestimation shortcoming of L1-norm regularization, nonconvex penalty function regularization is used to strongly promote the sparsity of estimation while guaranteeing that the global optimal solutions are obtained even though the objective function is a nonconvex function. Then, the split augmented Lagrangian shrinkage (SALSA) method is developed to solve the nonconvex WATV denoising problem. Two experimental studies are performed to verify the performance and effectiveness of the proposed method. Comparisons with the soft thresholding and basis pursuit denoising (BPD) methods show that the proposed method can accurately estimate the fault features from vibration signals, which means that the proposed method is an effective and promising tool for planetary gearbox fault diagnosis.
20 June 2020 in Shock and Vibration
The use of the convolutional neural network for fault diagnosis has been a common method of research in recent years. Since this method can automatically extract fault features, it has played a good role in some research studies. However, this method has a clear drawback that the signals will be significantly affected by working conditions and sample size, and it is difficult to improve diagnostic accuracy by directly learning faults, regardless of working conditions. It is therefore a research orientation worthy of a diagnosis of high precision defect in various working conditions. In this article, using a fine-grained classification algorithm, the operating conditions of the object system are considered an approximate classification. A specific failure in different working conditions is considered a beautiful classification. Samples of different faults in different working conditions are learned uniformly and the common characteristics are extracted from the convolutional network so that different faults of different working conditions can simultaneously be identified on the basis of the entire sample. Experimental results show that the method effectively uses the set of samples of the working conditions of the variables to obtain the dual recognition of defects and specific working conditions and the accuracy of the recognition is significantly higher than the method of learning regardless of working conditions.