Application of Intelligent Compensation to Engine Health Management System

Considering the prediction accuracy of health parameter in Engine Health Management (EHM) system used for aero-engine, a new compensation mechanism is proposed. The differences in the individuals of the same type engines and engine degradation after the use both will result in the modeling errors, while reducing the modeling errors is a key to improve the precision of fault diagnosis. A discrete series predictor based on multi-outputs least square support vector regression (LSSVR) is applied to the compensation of on-board self tuning model of aero-engine, and particle swarm optimization (PSO) is used to the kernels selection of multi-outputs LSSVR. The experimental results show that this compensation mechanism need not the model structure characteristics, compared with the other known approaches the aero-engine model has better generalization ability and higher prediction accuracy of health parameters.