Research On RBF Neural Network Sliding Mode Control And Optimization Algorithm For Aero-engine

The turbofan aero-engine is an extremely complex aerodynamic plant with complex structure,uncertainty and multi-objective constraints.How to design the aero-engine control system to meet the growing demand of aerodynamic power and how to obtain the optimal performance to improve the system quality while ensuring the engine work within the safety boundary are the focus and challenge of current researches.In this thesis,the nonlinear model of the turbofan aero-engine is established by using component-level modeling method,and the nonlinear model is linearized by the small disturbance method.However,if the sufficient thrust is only considered during the aero-engine design process,some variables may exceed the safety boundary which leads to a disaster to the engine components.In order to provide the necessary power to the aircraft,the engine is required to be in a constraint range,with no exceeding many boundaries and within the maximum boundary.Therefore,the Whale Optimization Algorithm is used to optimize the fuel flow for the engine transition state process.Firstly,the objective function and constraint function in every iteration point are discretized.Based on the WOA algorithm,the optimal law for each iteration point is found and combined all points to get the optimal laws of the entire process.Simulation results show that this method has good reliability and search ability in the engine system.To verify the superiority of the WOA algorithm,this thesis gives a comparison between the WOA algorithm and the classical Particle Swarm Optimization(PSO)algorithm.Under the optimal performance,the WOA algorithm has better convergence and accuracy than the PSO algorithm.In order to solve the time delay problem with the aero-engine control system,taking into consideration the influence of time delay,the mathematical model of the aero-engine control system is established in this paper.A sliding mode controller based on RBF neural network is designed,taking the advantage of the excellent robustness of sliding mode control.The power rate reaching law is adopted for the designed controller to suppress chattering.The RBF neural network is applied to provide a real-time estimation of the unknown disturbance and uncertainly of the control system.Finally,it can be seen from the simulation results that the controller designed in this thesis can effectively drive the plant to follow the reference model,and the closed-loop system also shows good static and dynamic performance.