In this paper, for the complex fault diagnosis problem of aero-engine with high dimensional nonlinearity, an intelligent fault diagnosis method of LSSVM based on damage repair and modified gray wolf optimization is proposed to characterize the performance deterioration trend with a damage accumulation prediction model under the interaction of multiple parameters, and to obtain the main health indicator parameters by damage repair method. The amount of model input information is increased. The improved Gray Wolf algorithm utilizes chaotic sequences to increase population diversity and uses adaptive inertia weights to achieve a good balance between global search capability and local exploitation capability to effectively optimize LSSVM parameters. The method proposed in this paper accurately diagnoses multiple types of engine faults, reduces the model learning time and accelerates the convergence process, providing an effective and feasible technical means for aircraft engine condition monitoring and fault diagnosis.
