An Automated Diagnostic Expert System for Diesel Engines

Abstract

Fault diagnosis of diesel engines can be a tedious time-consuming process, resulting in extended downtime, thus reducing productivity and increasing operational cost. This problem can be accentuated when experienced expert maintenance personnel are in short supply and also when the rate of development of new-generation engines using leading edge technology does not permit maintenance personnel to keep up with this change. An automated diagnostic system based on artificial intelligence criteria using mechanical signature analyses (MSA) of signals acquired from engine mounted sensors can overcome this problem by providing expert and consistent diagnostic advice. This paper describes the development and implementation of an automated diagnostic expert system for diesel engines. The system uses vibration signals together with oil pressure and temperature, crankcase pressures, exhaust gas temperature and pressure, exhaust emissions, manifold noise levels, inlet manifold pressure, fuel delivery pressure, and instantaneous engine speed to monitor and diagnose engine faults. State-of-the-art techniques used for signal processing to generate data required for effective diagnosis from raw signals acquired form the engine mounted sensors are described. Complexities of signal processing for diesel engines are discussed and solutions of a practical nature suggested. Signal analysis techniques relating to fault condition evaluation are also described.