An integrated knowledge based approach for on-line condition monitoring & fault diagnostics of marine propulsion machinery

Sea has always baffled man due to inherent uncertainties involved and continues to do so despite his best efforts. Conventional monitoring techniques perform well however, their appeal fades when confronted with complex systems characterized by multiple criteria/decisions etc. An ''Artificial Intelligence'' based approach for monitoring the strongly interdependent propulsion machinery promises to be a viable solution. Optimum exploitation of propulsion and power generation machinery of warships require intelligent suggestions for usage pattern and incipient fault diagnosis to the operators. This paper projects an innovative approach which involves a conglomeration of various specialized techniques namely a shell resident Fault Diagnostic Knowledge Based System, an on-line non-linear parameter estimation propulsion independent technique for a class of marine diesel engines and marine gas turbines, knowledge elicited from vibration signatures and diesel engines combustion simulation in a prototype Knowledge Based System (KBS). The main objective of the KBS developed is to optimally exploit the ship borne machinery based on their ''condition'' and confidently predict incipient faults, besides using it as an onboard trainer. Introduction of KBS in the present ships as retrofits or as built-in system in future ships is justified due to continuing trend of rising cost of platform induction with reduced number of operators and ever increasing levels of reliability demanded of these platforms throughout their life cycle.