A review on deep learning in planetary gearbox health state recognition: methods, applications, and dataset publication

Abstract

Planetary gearboxes have various merits in mechanical transmission, but their complex structure and intricate operation modes bring large challenges in terms of fault diagnosis. Deep learning has attracted increasing attention in intelligent fault diagnosis and has been successfully adopted for planetary gearbox fault diagnosis, avoiding the difficulty in manually analyzing complex fault features with signal processing methods. This paper presents a comprehensive review of deep learning-based planetary gearbox health state recognition. First, the challenges caused by the complex vibration characteristics of planetary gearboxes in fault diagnosis are analyzed. Second, according to the popularity of deep learning in planetary gearbox fault diagnosis, we briefly introduce six mainstream algorithms, i.e., autoencoder (AE), deep Boltzmann machine (DBM), convolutional neural network (CNN), transformer, generative adversarial network (GAN), and graph neural network (GNN), and some variants of them. Then, the applications of these methods to planetary gearbox fault diagnosis are reviewed. Finally, the research prospects and challenges in this research are discussed. According to the challenges, a dataset is introduced in this paper to facilitate future investigations. We expect that this paper can provide new graduate students, institutions and companies with a preliminary understanding of methods used in this field. The dataset can be downloaded from https://github.com/Liudd-BJUT/WT-planetary-gearbox-dataset.

A TLOT train gearbox fault diagnosis method based on ridge extraction under variable speed conditions

“…With the incredible development of deep learning in recent years, intelligent diagnosis is also effective making it af research hotspot [31][32][33]. Han et al [34] proposed a semisupervised fault diagnosis approach specifically designed for wind turbines, aiming to train unlabeled samples effectively.…”

DDAGCN: an unsupervised cross-domain identification method for tie rod bolt loosening in a rod-fastening rotor system under different working conditions

“…Multi-channel CNN was used to extract the data of multiple monitoring sensors simultaneously by Junior et al [39], which could improve the feature extraction capability of the network and realize the accurate diagnosis of the motor. Lui et al [40] classified CNN into four categories: 1D-CNN, 2D-CNN, Fusion CNN, and Residual Network (ResNet), and summarized the application of CNN in planetary gear box fault diagnosis. Guo et al [41] generated fault data through a generative adversarial network to expand the fault dataset and help CNN extract a large amount of feature information, thus achieving fault classification.…”

“…The improved model can extract weighted map features of different scales and enhance diagnostic efficiency. Lui et al [40] reviewed the application of GNN as a model feature extraction module in planetary gear box fault diagnosis. Li et al [47] established a GNN model by using the observation mechanism of reinforcement learning, and updated the reinforcement learning parameters with feedback diagnosis accuracy, so as to find the optimal diagnosis network structure.…”

Rotating machinery fault diagnosis based on optimized Hilbert curve images and a novel bi-channel CNN with attention mechanism

“…Traditional ML methods [5][6][7][8] involve manual extraction and selection of features [9], which make it challenging to obtain effective fault features for complex equipment fault recognition. In contrast, deep learning (DL) methods can adaptively mine the features from the vibration signals [10][11][12]. Hence, they have been widely investigated in mechanical fault diagnosis, and became a hot topic recently.…”