RBF Neural Network Based Fast Converging Terminal Global Sliding Mode Control of Aeroengine

This paper focuses on the sliding mode control of aero-engine control system with system uncertainties and disturbances. A terminal sliding mode controller with variable exponential parameter has been proposed together with a global sliding surface. It has the advantages of both terminal global sliding control and global sliding mode control scheme, that is, both finite-time high speed convergence and global robustness can be realized. The exponential parameter, which decides the contradictory relation between convergence speed and chattering amplitude, has been optimized by adaptive PSO algorithm. Then to obtain the full-envelop controller, RBF neural network has been used to obtain the nonlinear mapping relation between different working conditions and the variable exponential parameter. Simulation results indicate that the proposed controller has great dynamic performance and can effectively suppress chattering.