Fault diagnosis of shearer rocker gear based on depth residual network

Shearer is one of the core equipment of coal mining face. Rocker arm is the key power component of shearer cutting coal seam. It bears shearer cutting load and nonlinear internal excitation of rocker arm transmission system, which is the weak link of shearer. The environment of coal mining face is bad, the gear of shearer rocker transmission system is prone to failure, which not only affects the efficiency of coal mining, causes economic losses of enterprises, but also causes casualties. Because of the characteristics of rocker transmission system, such as long transmission chain, many types of gears and strong environmental noise, it brings great challenges to gear fault diagnosis.

Domestic scholars have carried out a lot of research on Shearer fault diagnosis technology. Fu Jiacai and others combined wavelet analysis and neural network method to effectively detect the occurrence time of the rocker arm fault signal, and accurately diagnose the severity of the fault. D. Zhang et al. Installed two acceleration sensors on the surface of the gear box of the shearer transmission system. Through the power spectrum analysis of the vibration signal obtained, the possibility of failure of the gear box was predicted. Z. The kernel Fisher discriminant (vmd-srkfd) method of variable model decomposition spectrum regression optimization proposed by Li et al. Realizes the detection of gear mixed fault in the transmission system of shearer. Du Yuhui and others put forward a fault diagnosis method of shearer gearbox based on partial least squares regression to solve the problem of poor lubrication or abnormal wear during the operation of shearer gearbox. The experiment shows that this method can accurately judge the wear state of gears. Ren Zhong et al used particle swarm optimization algorithm to optimize the relevant parameters of support vector machine, and used the optimized support vector machine classifier to diagnose the fault of planetary gear reducer in shearer cutting part in real time. Hao Shangqing et al. Used blind source separation algorithm to predict the bearing fault of shearer rocker arm. The algorithm can separate the fault signal well, and the recognition rate is high. Most of the above methods are realized through a small amount of small-scale data analysis. In the face of multiple working conditions alternation of shearers and unknown fault information of massive signals, there are some problems such as one-sided analysis results, poor accuracy and low efficiency.

In recent years, the application of deep learning in fault diagnosis develops rapidly. Compared with traditional fault diagnosis methods, deep learning has breakthrough advantages. It can get rid of the dependence on a large number of signal processing technology and diagnosis experience, avoid the uncertainty caused by manual feature extraction, complete the adaptive feature extraction, and significantly improve the accuracy of fault recognition. The representative deep learning models include deep belief network (DBN), sparse auto encoder (SAE) and convolutional neural network (CNN), CNN has been successfully applied in the field of fault diagnosis. Most of the above researches are based on shallow deep learning model. With the increase of the number of network layers, the weights between layers can not be updated, which easily leads to the accuracy unchanged or decreased. In order to overcome this defect, K. he proposed the deep residual network (RESNET) model in 2015, and achieved good results in the field of fault diagnosis. M. Zhao et al. Designed a RESNET with dynamic weighted wavelet coefficients, which took a series of wavelet packet coefficients of different frequency bands as input, adaptively adjusted the weight of wavelet packet coefficients through dynamic weighting layer, found out a group of recognition features, and finally realized the fault diagnosis of planetary gearbox. Wang Jiugen et al. Used the residual network model to diagnose the fault of RV Reducer, and verified the effectiveness of the model through experiments.

Aiming at the problem that the traditional fault diagnosis method of shearer rocker gear can not extract features independently, which leads to the poor accuracy and efficiency of gear fault diagnosis, a fault diagnosis model of shearer rocker gear based on RESNET is constructed, and the experimental verification and comparative analysis are carried out.

Scroll to Top