An expert system for diagnosing heavy-duty diesel engine faults

Abstract

Heavy-Duty Diesel Engines (HDDEs) support critical and high cost services, thus failure of such engines can have serious economic and health impacts. It is necessary that diagnosis is done during both preventive maintenance and when the engine has failed. Because of their complexity, HDDEs require high expertise for the diagnosis of their faults; such expertise is in many cases scarce, or just unavailable. Current computerized tools for diagnosing HDDEs are tied to particular manufacturer's products. In addition, most of them do not have the functionality that is required to assist inexperienced technicians to completely diagnose and repair HDDE faults, because most of the tools have only the capability to identify HDDE faults. These tools are not able to recommend corrective action. This paper presents an easy to use expert system for diagnosing HDDE faults that is based on Bayesian Network Technology. Using Bayesian Networks simplified the modeling of the complex process of diagnosing HDDEs. Moreover, it enabled us to capture the uncertainty associated with engine diagnosis, and to incorporate learning capabilities in the expert system.