Research On The Technologies Of Intellectualized Fault Diagnosis For The Diesel Engine Dynamotors Unit In Ships

A diesel engine dynamotors unit (DEDU), which is one of important parts of the power station in a ship, has a complex structure, so it is very difficult to diagnose its faults, and for the traditional fault diagnosis (FD) methods, it hard to make an accurate diagnosis. With the development of artificial intelligence, various intellectual FD technologies were applied widely in the domain of FD. In the paper, a technology of intellectualized fault diagnosis for DEDU in the ship was researched.Firstly, the various FD methods were analyzed, and the expert system and the fault tree theories were studied systemically for DEDU. For the expert system, the express of fault knowledge and the applications of possibility theory for FD were researched. Then the fault tree construction and its qualitative analysis methods were studied. Furthermore, a FD method that combined the fault tree diagnosis and the possibility theory was studied, which is some originality innovation of the expert system;Secondly, based on the DEDU structure and operation principle, the method to obtain fault knowledge and some representative faults were analyzed, and the fault structure was plotted. Then, DEDU were divided into several relevant subsystems, and several fault trees which set system faults as root nodes were established. According to those fault trees, the minimum cut set and qualitative-important degree were calculated and obtained. By analyzing the fault knowledge and the fault trees, an integrated intelligent FD system that associated possibility theory with fault tree and production rules to express fault knowledge were built for DEDU, which was very intelligent in FD, convenient to use, and easy to expand.Finally, the application of the integrated intelligent FD system for the DEDU in some unit in JiangYin city was introduced. And the software system was tested aboard and its effect on application was analyzed. The results show that the system can improve the diagnosis speed and reduce the rate of inaccurate diagnosis with strong universality. Especially, it can adapt itself to FD of complex systems, which has some important practical value in the engineering.