Fault Diagnosis Technology Of Marine Diesel Engine Based On Neural Network

For decades, shipping continues developing. Ship is the main transportation carrier ofthe shipping, diesel engine as the power plant of the entire ship takes absolutely importantposition in ship. The status of their work is directly related to the safe navigation of the ship.So the study on marine diesel engine fault diagnosis system is particularly important.The paper based on the research of existing fault diagnosis technology, and proposedan intelligent fault diagnosis method witch based on neural network. Firstly, a comparativeanalysis of BP and RBF network and simulation experiment of diesel engine fault diagnosisin MATLAB are given. The results show that BP neural network exists many problems,such as slow convergence, low accuracy, easy to fall into local minimum and other issues.Radial Basis Function neural network is superior to BP network in approximation capabilityclassification ability and learning speed. But the generalization ability of the network needsto be improved, the selection of center vector and width parameters has great influence onnetwork performance.Secondly, the article briefly describes the genetic algorithms and the application inneural network. Because of the standard genetic algorithm is prone to precocious, poorstability and other issues. To solve these deficiencies, the paper proposes a targetedimprovement, introducing a genetic algorithm according to the groups of individual fitnessadaptively changing the probability of crossover and transform to optimize the center vectorand width parameters of RBF, and using the Boolean vector optimize the network hiddennodes, to achieve the purpose of improving network fault recognition and diagnosticaccuracy.Thirdly, the fault diagnosis simulation experiments of marine diesel engine fuel systemtaken in MATLAB. The results shows that the performance of adaptive genetic RBFnetwork has been significantly improved, which used in marine diesel engine fault diagnosissystem has better effect in predictive failure. It also has good performers in diagnostic speed,convergence effect, diagnostic accuracy and network stability.Finally, using LabVIEW and MATLAB mixed programming technology, combined thewell interface design of LabVIEW with powerful math ability of MATLAB, and developeda fault diagnosis system of marine diesel engine. The practical application shows that thesystem has a good fault diagnosis effect and practical value.