Research On Online Fault Diagnosis Method Of Mine Hoist Bearing Based On Deep Learning

The rolling bearing is a critical component of mine hoist equipment,and its health status affects the operation state of the whole hoist.Due to the complex structure and poor working environment of mineral hoists,it is challenging to realize the online fault diagnosis of rolling bearing.In recent years,deep learning algorithm has achieved rich research results in fault diagnosis.However,there is still a particular gap in applying this method to solve practical fault diagnosis problems.In order to solve the difficulties of the online fault diagnosis of mine hoist bearings,this paper: i)analyzes the fault mechanism of mine hoist rolling bearings,ii)presents a modified variational mode decomposition(VMD)algorithm to separate signal from strong noise,iii)improves the convolutional neural network(CNN)to diagnosis bearing fault,and iv)develop an online fault diagnosis system to validate the effectiveness of the presented algorithm.The main research contents of this paper are:Firstly,this paper analyzes the structure and mechanism of the mine hoist system to study the characteristics of vibration-based condition monitoring data of the mine host bearings.After studying the principle of a VMD algorithm,the paper combines the obtained characteristics to implement a VMD-based signal-to-noise separation,which combines the Pearson correlation coefficient index for accuracy enhancement.A numerical simulation verifies the signal-noise separation ability of the proposed VMD algorithm for the monitoring signals of the mine hoist system.Secondly,an improved fault diagnosis model based on CNN is constructed.An optimal parameter con Fig.uration for the improved CNN model is es Tab.lished(CNN-6,adding the Dropout layer and using a single intrinsic mode component as input).Based on the improved CNN fault diagnosis model,the impacts of convolution kernel number arrangement,activation function selection and training batch size on the diagnosis accuracy of the model network are analyzed.The diagnostic accuracy and robustness of the model under strong background noise interference and different load conditions are verified by open source test data.Finally,according to the design requirements of the mine hoist bearing fault diagnosis system,the overall design of the diagnosis system is completed,and the functional modules of the system are developed.Combined with the improved fault diagnosis method based on CNN,the Mine hoist bearing’s online fault diagnosis software system based on deep learning is built.The feasibility of the designed online fault diagnosis system is verified by inputting simulation test data.The online fault diagnosis and historical information management of the mine hoist bearing are realized,laying a technical foundation for the whole life cycle state monitoring of mine hoist.