Fault diagnosis and control of a diesel power generator using sliding mode techniques

The sliding mode concept, which is based on the notion of a variable structure system, has inherent advantages of robustness and performance specification and is here applied to the problems of fault diagnosis and control of the diesel engine. A real-time condition monitoring and fault diagnosis system and also control system using sliding mode techniques are described. The designs are assessed through tests on a particular Perkins diesel-electric power generator. A model-based approach incorporating a non-linear sliding mode observer scheme is proposed for fault monitoring. The diesel engine coolant system is considered. The system parameters are monitored using the concept of the equivalent injection signal which is required to maintain the sliding mode. The proposed diagnostic scheme is shown to be robust in estimating component parameters. The approach is applicable to many automotive engine approach problems and is cost effective as only low cost temperature sensors are involved in the implementation. The engine control strategy investigates both model-based and model-free sliding mode control techniques. The development of an engine model appropriate for speed control has been considered using closing-loop event-based system identification. Several control algorithms are proposed. Real-time speed control systems have been designed and implemented using Matlab/Simulink/dSPACE. The proposed model-based and model-free controllers show good tracking performance and disturbance rejection properties. The proposed model-free controller is shown to be an appropriate candidate for industrial control of the diesel engine system. The established gain-tuning algorithms allow non-experts to maintain and tune the resulting control schemes.