The Fault Identification Algorithm Research Of Center Plate Bolts Based On Neural Network

Center Plate Bolts fault detection is one part of TFDS(Trouble of moving Freight car Detection System).The system detects fault through the computer.It is an important safeguard to ensure the safe running of trains. This paper researches for the fault detection algorithm of center plate bolts to effectively identify faultsof the bolts, but also can provide reference value for other fault of trains.The paper uses neural network approach to achieve fault identification of center plate bolts. Neural network has a better fitting ability to nonlinear system.It can master the inherent laws of the problem to be solved and then achieve intelligent fault identification through learning from the relationship between input and output. The optimization of neural network is an important decision factor. The genetic algorithm is choosed to optimize connection weights and neuron thresholds of neural network.It can achieve the global convergence of neural network parameters.And while the fitness computing is a large amount of computation,so master-slave parallel genetic algorithms is choosed to improve overall operating efficiency through the network communication by way of parallel computing processor.The neural network is used to identify fault of center plate bolts after training. The result shows that the accuracy of fault recognition rate can achieve 90%.It confirmed that neural network fault identification method meets the demand of the system.It can also adapt the environment change by learning. For other fault, neural network is also portable and has the meaning of popularization.