Research On Knowledge Mining Of Oil Monitoring Expert System

The development of expert system based on oil monitoring for the diesel engine is discussed in this paper. The key research point in this paper is the diagnosis knowledge mining and the establishment of knowledgebase.Knowledge is the kernel of an expert system. For acquisition of rules, the characters of diesel engine faults are discussed in this paper. The system in a diesel engine is divided into several sub system according to its structure, and the probable faults of each sub system are discussed. Those lead to the determination of diesel's fault modules. And then the applications of some general oil monitoring methods including physical chemistry analysis, ferrograph analysis, contamination degree analysis, etc, in fault diagnosis of diesel engine are analyzed, and also the relations between those methods and diesel engine's fault modules are discussed. These are the basis of the knowledge mining.On the basis of diesel engine's fault modules, the mining of knowledge rules about diagnosis of diesel engine are outlined in detail, including simple knowledge rules mining, cluster analysis and association rules mining. And the association rules mining is discussed in detail especially, including single level association rules and multilevel association rules. On the basis of rules mined, knowledgebase is established with CLIPS. The mining of association rules needs a great mount of data. When association rules mining is being done in this paper, some data about faults of ten more kinds of diesel engine is collected. The table structure of storing the data is given according to diesel engine's fault modules and characters of data mining. Finally the system Expert System Based on Oil Monitoring (ESBOM) is developed. Several functions are implemented in the system, such data management, data processing, fault diagnosis, etc. It is a function full system.The research result of this paper shows that association rules are of higherlevel than single rules. And they can be very helpful for people to do fault diagnosis and trend prediction. Although the research is about diesel engine), the research methods are also suitable for knowledge acquisition of different machine's diagnosis. It will be meaningful to the intelligentization of oil monitoring technique and the application of expert system in industry.