The Study On Diesel Engine Fault Diagnosis Based On Information Fusion

Diesel engine is the power source of many equipments.It is widely used in many areas such as ships,electricity generation and mine field.It plays a very important role in national economy.As a typical power machine,diesel engine has complicated running conditions and internal structure.It has great significance to ensure a diesel engine run in safety and diagnose faults in time.There are many useful information when a diesel engine is running,so it is difficult for the fault diagnosis to be accurately performed by single sensor thus the information fusion among multiple sensors is required.The mainly work of this thesis are as follows:1.Research status of diesel engine fault diagnosis and the significance of adopting information fusion are discussed.The structure and common faults of diesel are studied.Choose the vibration signal and fuel oil pressure signal as monitoring information.2.Introduce the based structure,principles and algorithms of widely used BP neural network and RBF neural network.A modified chaos particle swarm optimization algorithm is proposed based on disadvantages of RBF.Optimize selection of the RBF hidden layer's central value,width and output layer's weight.Simulate fault diagnosis of these three methods by MATLAB.The results show that RBF neural network based on modified particle swarm algorithm has high precision diagnose,the effectiveness of this algorithm is validated.3.Combine the results of neural network with D-S evidence theory,with regard to vibration signal and fuel oil pressure signal,make decision level fusion diagnose and improve the reliability of the result.Proposed a relevant evidence combination method aimed at the independence of evidence combination and relevance of evidence,exclude the influence of reusing relevant information by evidence combination.Focused on the problem of conflicts among evidences of many sensors,the weighted combination method of conflict evidences is proposed.Use conflict information,exclude disturbance information of problematic sensor on the result,make the confusion result more reliable.4.The diesel engine fault diagnosis system is designed by MFC,with using ADO to load data from ACCESS database to diagnosis system,design a human-computer interaction interface and specific functions.