Research On Fault Diagnosis Method Of Mine Main Ventilator Based On Data

Mine main ventilator,the main ancillary equipment in coal mines,its normal operation is an important guaranty for safe production,disaster relief ventilation and reasonable daily ventilation.However,once it is abnormal,huge economic losses may be caused,and security of underground workers may even be threatened.Therefore,It is very necessary to monitor the operation status of the ventilator and to carry out fault diagnosis in time.In this paper,with mine main ventilator as the research object,data acquisition system is designed based on PLC,MATLAB and IPC at first,then features extraction and fault diagnosis are carried out based on the extracted data.Extracting the features of the bearing vibration signals based on improved Hilbert-Huang transform(HHT).Original HHT has such disadvantages as modal aliasing,mean curve fitting,endpoint effects and so on.empirical modal decomposition(EMD)is very important in using HHT method,an adaptive local mean EMD algorithm is proposed in this paper.First,all the data between extreme points and zero are used as a local feature scale such that the obtained local mean is closer to the ideal mean.Then,the index of orthogonality is introduced to evaluate the correspondence between the extreme point order and the frequency component so as to realize the optimal order of the extreme points adaptively.It can be seen from the simulation that the improved algorithm can reduce the computational time and the number of iterations.The improved algorithm is used to decompose the ventilator vibration signals in time domain to obtain a series of intrinsic modal functions(IMF),and the method of energy entropy is used to extract the fault features in the last,which lays the data foundation for fault diagnosis.Extreme learning machine(ELM)algorithm is adopted to identify the fault types of mine main ventilator.Aimed at such defect that ELM algorithm selects the hidden layer input weights and thresholds randomly,particle swarm optimization(PSO)algorithm is employed to search for the optimal hidden layer node parameters.In order to overcome the shortcomings of long operation time and easily getting into the local optimum caused by introducing PSO algorithm,an algorithm named quantum-behaved particle swarm optimization(QPSO)is employed to optimize the hidden layer parameters.Also,in order to enhance the learning ability for unknown samples,the objective function of network training is to minimize experience risk and structural risk at the same time.Finally,the proposed algorithm is used for fault diagnosis of main ventilator and the simulation results show that the accuracy and timeliness of the algorithm can achieve satisfactory results when performing fault diagnosis.