Non-Deterministic Reasoning Based Vehicle Fault Diagnosis Expert System

Fault diagnosis is a kind of technology which aims to diagnose the character and reason of faults by means of modern monitoring, testing and computer analysis, forecast the trend of faults and then take corresponding steps. It can help finding faults of equipments and avoiding loss. Researches on it make great sense to the society and have enormous economic value.Expert system is a kind of intelligent software system that has expert knowledge and experiences. It stores knowledge of human experts in computer, making computers think and solve problem like human experts.his paper use expert system technology in the vehicle fault diagnosis scenario, whose expert knowledge logically is a kind of causality representing the relation between faults and causes and has the feature of non-deterministic. Hereby, this paper used production representation, which is suitable for representing causality, to denote the expert knowledge, and used non-deterministic reasoning, which is appropriate for the non-deterministic knowledge, in the knowledge reasoning.Firstly, this paper gave a brief introduction of fault diagnosis and expert system technology, and then applied expert system technology on the vehicle fault diagnosis, designing a vehicle fault diagnosis expert system. Because vehicle has different fault rates during different use periods, the relation degree between fault phenomena and faults varies during different use periods. In allusion to that law and considering that different evidences of diagnosis rule have different significance, this paper process the reliability of diagnosis rule of non-deterministic reasoning in segments according the fault mileages, and bring in the dynamic weighting factor for diagnosis rule. At last, this paper gave the implementation of the vehicle fault diagnosis expert system, which contains three modules: knowledge module, diagnosis-reasoning module and auxiliary module, in which knowledge module and diagnosis-reasoning module are the main body of the whole system. Knowledge module consist of knowledge management sub-module, reliability computing sub-module and repair-record sub-module, and diagnosis-reasoning module is composed of reasoning sub-module and interpret sub-module.This system has been implemented on Windows platform, using the Borland C++ Builder IDE and the SQL Server 2000 DBMS, and has been used in the diagnosis of a type of tank in some army of PLA, helping that army save the defense appropriations and reflecting the feasibility and superiority of expert-system-aided fault diagnosis.