Study Of Diesel Engine Fault Diagnosis Technique Based On Neural Network

Based on.the analysis of the domestiic and foreign researeh about intelligent diagnosis technology and diesel engine fault diagnosis at the ceurrent status, a design proposal is given in this paper to carry on the diesel engine fault diagnosis through the neural network, and the fault diagnosis of fuel system and turbocharged system are mainly analysized.First, the contents and practical meanings of the subject have been discussed in this paper . After analyzing several popular diagnosis methods which are often used in the diesel engine fault diagnosis, the exist problems about fault diagnosis and possible development are pointed out, and neural network principle is describedThe simulated fault data of diesel fuel system and turbocharged system is obtained by collecting a lot of trial operation materials of diesel engine from documents;Taking these data as the input sample of neural network,the neural network is constructed and trained,and the simulated fault of diesel engine is diagnosised,so the diagnosis result is obtained.The neural network is a large scale of parallel nonlinear system,it has strong associative learning,self-orgnizing,self-adaptive and high nonlinear operation ability,so it has strong judge ability of identifying the causality of these complicated variables.The diesel engine fault is diagnosised and analysized by using the Radial Basis Function(RBF) neural network. RBF neural network is systematicly studied, the structure of three layers RBF neural network,the setting of network parameters and the choice of training pattern are detailed discussed, fault characteristic parameters are trained and identified by using K-means clustering algorithm, taking the characteristic parameters as the input neuron, its diagnosis result as the output neuron.The simulation experiment shows that the fault diagnosis result based on neural network is well consistent with measured values. As long as we choose enough typical initial fauly sample to train neural network, the network fault-tolerant and the stability are better.The method of fault pattern recognition based on neural network can fully use information feature,realize the mapping relation between input and output,get the acurate result.