Research On Health Monitoring Method For Steel Guide In Vertical Shaft

The mine hoisting system is responsible for the transportation of coal,equipment and personnel during the coal mining process.The working state of the mine hoisting system is directly related to the safe and efficient production of coal mines.The steel guide is a guarantee for the safe operation of the lifting container.Therefore,the research on the steel guide health monitoring is of great significance for the production of coal mines.Firstly,the structure of the vertical shaft lifting system is built by studying the working principle of the vertical shaft lifting system.The test set-up is composed of a motor,a reel,a lifting container,roller guide wheels and guide beams.In addition,there is a monitoring part composed of a PLC,a frequency converter,a touch screen,an encoder,and a sensor.The steel guide dynamic test equipment of the shaft hoisting system is designed to simulate three steel guide states which include: normal state,step excitation and connector gap excitation.Hence the vibration acceleration signal of the lifting container is collected.Secondly,the acquired acceleration signal of the lifting container is a non-stationary signal.Therefore,the wavelet packet and EMD-SVD are used to extract the fault signal characteristic parameters.The wavelet packet method is used to perform three-layer wavelet packet decomposition on the acceleration signals of different state lifting containers,and computed to obtain the energy values of eight frequency band signals as the fault characteristic parameters.EMD is used to decompose the acceleration signals of different state lifting containers;the first four natural modal functions are selected,and the singular value of each intrinsic mode function is calculated by SVD method as the fault characteristic parameters.Then,the signal fault feature parameters extracted by wavelet packet and EMD-SVD are taken as training samples and test samples of BP neural network and SVM respectively.The steel guide fault pattern recognition is realized by four methods based on wavelet packet and BP neural network,wavelet packet and SVM,EMD-SVD and BP neural network,and EMD-SVD and SVM.The success of the BP neural network and the average classification accuracy of the SVM are defined separately to evaluate the effectiveness of the pattern recognition.The simulation shows that the BP neural network can achieve the steel guide fault diagnosis,but considering the advantage of SVM in small samples,the SVM method is more reliable for pattern recognition.The analysis equally shows that the combination of EMD-SVD and SVM is better in fault identification.Finally,considering the K-CV method to optimize the SVM parameters c and g,the classification effect is not good and the optimization range is increased.Therefore,the parameters c and g in the SVM are optimized by GA.The optimized parameters c and g are brought into the SVM to recalculate the classification accuracy.The optimization results show that the average classification accuracy is improved by 7%.The analysis results show that the EMD-SVD as the fault feature parameter extraction method and the GA optimization SVM as the combination of fault pattern recognition can effectively classify the steel guide failure mode.Figure [36] table [18] reference [80]...