Research on mixed-fault diagnosis of mine-used belt conveyor gearbox

Abstract

In view of problem that fault diagnosis methods of mine-used belt conveyor gearbox based on vibration signal analysis are not easy to process mixed-fault signals, a new mixed-fault diagnosis method of mine-used belt conveyor gearbox based on self-organizing map network was proposed. The standard multi-fault samples of mine-used belt conveyor gearbox are pre-processed by wavelet threshold denoising method incorporating Shannon entropy，Gaussian mixture distribution model is established for the standard multi-fault samples after pre-treatment, and the expectation-maximization algorithm is used to estimate the parameters of the model to obtain corresponding feature vectors which are input into the self-organizing map network. At last, the fault signals of different fault types are clustered and identified by self-organizing map network to determine the fault category. The test results show that the method can effectively diagnose the fault type of mine-used belt conveyor gearbox, and the overall accuracy of the diagnostic method is 88%, and the accuracy under six conditions is 100%. It provides a new method for gearbox fault diagnosis of mine electromechanical equipment.

Keywords

• mine-used belt conveyor
• gearbox
• vibration signal
• mixed-fault diagnosis
• expectation-maximization algorithm
• self-organizing map network
• gaussian mixture distribution model
• som