Research On Fault Diagnosis Method Of Vehicle Data Transmission Equipment Based On Improved CS Algorithm

With the continuous development of China's high-speed railway,the importance of train operation efficiency has gradually increased.Train control on-board equipment is one of the key systems to ensure the efficiency of train operation,safe and comfortable driving.Once the on-board equipment is abnormal,it will affect the driving efficiency.In severe cases,it may cause a safety accident and endanger the safety of passengers.At present,the fault diagnosis of train control on-board equipment mainly depends on expert knowledge to achieve the purpose of fault diagnosis,and the recording and analysis of fault data phenomena are also manually completed by technical personnel.With the continuous improvement of mechanical complexity and precision,the structure of train control on-board equipment is becoming more and more complicated.The traditional fault intelligent diagnosis method relying on manual technology and experience has been far from fully able to meet its existing technical requirements.Therefore,we need to establish an intelligent diagnostic system that can safely,quickly and accurately find and eliminate faults in time.This thesis takes the on-board equipment of the CTCS3-300 T train control system as the research object.Based on the functions of the train control on-board equipment,the units with data transmission functions are collectively referred to as vehicle data transmission equipment.According to the characteristics of fault data,the corresponding relationship between fault characteristics and fault types is established,and the corresponding relationship between fault characteristics and fault types is reduced by using the rough set theory method based on the difference matrix to establish a fault decision table.Based on the in-depth study of the cuckoo search algorithm,this thesis proposes an improved cuckoo search algorithm based on Gaussian disturbance and Pareto's principle.In this thesis,the improved cuckoo search algorithm is combined with neural network algorithm and fault decision table to propose a fault diagnosis method for vehicle data transmission equipment based on the improved cuckoo search algorithm.The main research contents of the thesis are as follows:(1)Establishment of fault decision table: This thesis makes an in-depth analysis of the fault data characteristics of the vehicle data transmission equipment in the log file recorded by AElog and SDP log.According to the characteristics of these fault data,analyze and sort the data,summarize the fault characteristics and fault types of the vehicle data transmission equipment,and summarize the fault decision table of the vehicle data transmission equipment according to the actual case.Using the rough set(RS)theory based on discernibility matrix,discretization and dimensionality reduction of the summarized vehicle data transmissionequipment fault decision table are provided to provide data basis for the subsequent experimental verification part.(2)The establishment of the algorithm model: Through the deep learning research on the cuckoo search algorithm,the search execution speed in the later stage of the cuckoo search algorithm is slow,which is easy to make the search fall into the shortcomings of local optimization.The adaptive control strategy,Gaussian disturbance and Pareto's principle are used to improve the cuckoo search algorithm.The effectiveness of the improvement is verified by 7 benchmark functions.The improved cuckoo search algorithm is used to optimize the weight and threshold of the neural network algorithm,and a neural network algorithm model based on the improved cuckoo algorithm is established.Combined with the reduced fault decision table,an RS-GPICS-BP network model for vehicle data transmission equipment is established.(3)Experimental verification and analysis: In the mixed programming environment of MATLAB and VB,through data training and data testing,under the same conditions,the diagnostic accuracy rate,algorithm training speed and efficiency of network models such as RS-GPICS-BP,GPICS-BP,and CS-BP are evaluated.The indicators are compared and analyzed to verify the feasibility and accuracy of the RS-GPICS-BP network model.