Research Of Fault Diagnosis Expert System For MAN B&W Two-Stroke Diesel Engine

Over the last decade, marine diesel engines have taken an absolute dominance in the ship power plant because of its inherent advantages, especially in the field of civilian marine power devices. Meanwhile marine diesel engine parts’failure causes are complex and difficult to repair due to its numerous components and its internal logical structure is very complex. Due to limitations of operational capacity levels the engineer failed to exclude faults effectively and timely when they are in glitches stage. Then many accidents come up.Since the first expert system appeared in1965, during these decades the expert system have been used in various professional fields, and have achieved great economic benefit. Nowadays many expert systems have apply to the real ships for marine diesel engine fault diagnosis, however the complex structures and mechanisms of marine diesel engine, makes those expert systems long way to go before they can reach an ideal result. According to the survey, marine diesel engine expert systems in hand which have been used on real ships, remain in the stage that can only make some simple fault diagnosis or functional test. They can’t deal with the marine diesel engine faults which are a little bit complicated.In this paper, set the series of MAN B&W two-stroke diesel engine which is widely used currently for an example, on the basis of a mass of marine diesel engine failure data and sort out of these data, author select three mixed knowledge representation methods of the frames, production and Petri nets to dispose with those collected failures knowledge data and built the fault diagnosis expert knowledge base of this system. Then according to knowledge representation method mentioned above, author select a hybrid inference method combine with rule-based, model, case-based and a hybrid reasoning strategy combine with forward reasoning and backward reasoning, to finish the construction of the inference engine. This inference engine contains three reasoning model:unit failure reasoning, multivariate failure reasoning, and backward reasoning. Then use the Visual C++language to design and develop the program of expert knowledge base, inference engine, human-machine interface, and other expert system auxiliary modules, and then built a system level and non-on-line fault diagnosis expert system for marine diesel engines. Moreover through simulation test, the system has been proved its rationality, it could diagnosis the faults of diesel engines rapidly and accurately and the diagnosis is genuine and believable. In addition, in consideration of access mechanism of diesel engine parameter, author developed a data communication interface between the fault diagnosis expert system and the Diesel DMS simulation platform, by this way the system can obtain the real-time data of the running diesel engines, and provides technical support for an on-line fault diagnosis expert system.