Model-Based Intelligent Learning Control Method For Aeroengine

As the demand of aircraft engine performance has further increased,engine control way has gradually evolved from traditional control to intelligent control.In this paper,a turbofan engine is taken as the research object,and a model-based intelligent learning control method is developed for aeroengine.Inlet distortion of the engine is discussed,and the effects on the inlet,fan are then presented.The distortion index is introduced to depict the magnitude of inlet distortion.The total pressure recovery coefficient of inlet and fan map are corrected by the distortion index,which are utilized into the engine component level model.Thus,the influence of inlet distortion can be represented in the engine model,and it also be tested in simulation.The GS-NARMA-L2 and IL-NARMA-L2 intelligent speed control structures are designed to improve control quality due to the strong nonlinearity of aircraft engine.The neural network topology parameters are real time tuned in the NARMA-L2 model in the former method,while the iterative learning control strategy is combined to the NARMA-L2 control system in the latter one to fulfill the engine speed control.Furthermore,a pressure ratio control loop is added to the original speed control circuit to achieve multiple variables control.An intelligent multi-variable decentralized controller is designed by the combination of the proposed NARMA-L2 linear compensation strategy for aircraft engine based on the NARMA-L2 model.Simulation results show that the involved NARMA-L2 control methods provide fast response and low overshoot in the steady system in the flight envelope.A thrust degradation mitigation control method is proposed for aircraft engine multi-variable control based on neural network integrated with variable incremental LP optimization.It alleviates the engine thrust degeneration through that the inner loop controls high-pressure rotor speed and the engine pressure ratio,and the outer loop corrects the engine command signal.The inner loop NARMA-L2 speed controller is gained by neural network,and the engine command signal is calculated by the variable incremental LP optimization method in the outer command correction loop.Simulations were performed on a low-bypass-ratio turbofan engine,and the results show the satisfactory control performance of proposed methodology both under natural components degradation and inlet distortion.