Application of pattern recognition with statistics for engine fault diagnosis

A combined method of pattern recognition and statistics was examined to diagnose the double faults of a double-shaft dividually-exhausting turbofan engine.The pattern recognition is based on the engine performance simulation model.Balance technique of engine was used during recognition process and the non-linear characteristic was considered.The result got by pattern recognition was treated statistically,compensating the deficiency in fault isolation decided purely by comparability.The right fault index is defined and calculated,proving that it is effective for double faults isolation.