In order to solve the questions of degenerate parameter estimation and performance recovery control of the engine components in supersonic condition, based on the improved Kalman filter, an approach of three-dimensional difference is proposed to achieve the parameter estimation. Performance recovery control increases the outer control loop based on the conventional inner speed control, and the outer control loop consists of thrust estimation model and outer controller. A thrust estimation model is designed based on the least squares support vector regression. The optimal inputs of model are selected through the thrust feature selection algorithm. The accuracy is improved greatly compared with the conventional non-selected thrust estimation model. Fuzzy PI controller is designed to achieve the purpose of degenerating engine performance recovery by adaptively adjusting the speed of the inner loop control. Finally, the numerical simulation of supersonic state verifies the effectiveness of the above programs.