Fault Diagnosis Of S700K Switch Machine Based On Wavelet Transform And Improved Neural Network

In recent years,Chinese railways are developing in the direction of high-speed and heavy-load.The construction of high-speed railway is in full swing.As one of the basic signal equipment for high-speed railway operation,speed-up turnout plays a vital role in the safe operation of trains.S700 K Alternating Current(AC)electric switch machine is an important tool to control the action of speed-up turnout at present,and its operation status plays a vital role in whether the turnout can complete the conversion.From the practical application in the field,the judgement of the fault types of switch equipment mainly relies on the relevant technical personnel to look at the current or power curve of the switch machine collected by the microcomputer monitoring system.This method often has problems of long processing time and poor processing accuracy.In view of the above problems,it has important application value and practical significance to study the corresponding intelligent fault diagnosis method of turnout switch machine and to adapt to the rapid development direction of high-speed railway nowadays.Based on the analysis of domestic and foreign literature and references on fault diagnosis of switch machines,a fault diagnosis model of S700 K switch machines is proposed,which combines the wavelet transform and improved Back Propagation(BP)neural network.In this model,the power curve data of the switch machine collected by the microcomputer monitoring system is taken as the input of the wavelet decomposition.The feature of the signal is extracted by the wavelet decomposition,and the feature is taken as the input of the BP neural network optimized by the improved adaptive genetic algorithm to diagnose the fault of the S700 K switch machine.The main research work of the thesis is as follows.Firstly,according to the structure and working principle of S700 K AC electric switch machine,the relationship between its action force and action power is analyzed,and it is concluded that the change of operation power value of the switch machine can reflect the change of its operation state.Through consulting data and field fault records,eight common fault modes of S700 K switch machine are summarized,and the power curve and cause of each fault mode are listed.Then,the fault diagnosis model of S700 K switch machine is constructed.Haar wavelet decomposition is proposed to extract the input of the neural network,the Adaptive Genetic Algorithms(AGA)is used to optimize the weights and thresholds of the neural network,and the Adaptive Genetic Algorithms-Wavelet Neural Network(AGA-WNN)fault diagnosis model is constructed..After sampling the power curve data collected by the computer monitoring system,the characteristic quantity of the extracted signal is decomposed by Haar wavelet.The normalized characteristic quantity is used as the input of BP neural network.Anadaptive genetic algorithm which can change the crossover and mutation probability in the evolution process is used to optimize the connection weights and thresholds in the neural network.The network training and testing of the model are carried out by choosing the operation power curve in the microcomputer monitoring station machine database.The results show that the diagnostic model has high accuracy.Finally,the improved algorithm is used to optimize the fault diagnosis model of S700 K switch machine.Aiming at the weakness of the adaptive genetic wavelet neural network model in the three fault modes of "abnormal resistance in conversion process","turnout can not be converted" and "switch machine card gap",the Improved Adaptive Genetic Algorithm(IAGA)based on evolutionary algebraic attenuation factor is proposed to further enhance the operation ability of the algorithm and improve the diagnosis accuracy and speed of the system.The simulation results show that the improved adaptive genetic algorithm is better than the original algorithm in the optimization process and the optimal fitness value.The improved diagnosis model also has higher speed and accuracy,and can meet the needs of field operation...