基于油液分析的汽车发动机摩擦系统故障诊断专家系统知识库的建立

Development of Oil Analysis Based Knowledge Base of Expert System to Diagnose Auto Engine Tribological System

采集汽车发动机润滑系统中摩擦副经历摩擦学行为的信息是诊断发动机摩擦系统故障的有效手段。油液分析从摩擦学系统的润滑剂和磨损物两方面获得摩擦副的润滑和磨损状态的信息。但常规的油液分析信息处理方法上的不足影响了油液分析技术在汽车行业的实际应用。开发基于油液分析的故障诊断专家系统,无疑会改善油液分析技术的诊断准确性和诊断成本。本文着重针对该诊断专家系统知识库的建立,提出了带有自学习功能的基于多元回归统计分析的规则获取方式和基于“规则架─规则体”的知识表示模式,并用PRO-LOG语言给出了汽油机换油指标的“规则架─规则体”表示模式的一个实例。

The collection of information about the tribological behaviour of friction pairs in auto engine lubrication system is the effective method to diagnose faults of engine tribological system. We get information about the lubrication and wear state of friction pairs by oil analysis which deals with the lubricant and wear particles in tribological system. But the insufficiency of traditional information process of oil analysis limits its practical applications on auto industry. It is no doubt , die development of oil analysis based expert system will improve the accurate of diagnostics and decrease the cost of diagnostics.This paper provides a multivariate regress statistics analysis based regulation acquisition methods with self learning function and a 'rule skeleton - rule body' based knowledge expression mode. At last, it provides a sample which is 'rule skeleton - rule body' based knowledge expression of gas engine oil change indicator with PROLOG language.