ROBUST ADAPTIVE BANK-TO-TURN MISSILE AUTOPILOT DESIGN USING FUZZY CMAC NEURAL NETWORKS

Based on the fuzzy CMAC (FCMAC) neural networks (NNs), the theory of feedback linearization (FL), and the simplified bank to turn (BTT) missile control design model, a robust adaptive BTT missile autopilot design method is presented. First, based on the simplified BTT missile model for control design, a nonlinear feedback control law which depends on the accurate model of the controlled plant is obtained using the theory of FL. Secondly, based on the nominal BTT missile control design model, the FCMAC NNs are introduced to improve further the estimation accuracy of the BTT missile control design model in a online way, and a robustifying portion is included in the control law to suppress the effect of the NNs approximation errors on the missile system. A stability proof is given strictly in the sense of Lyapunov. Its shown that all the signals in the closed loop BTT missile system are uniformly ultimately bounded (UUB). The control law is valid throughout the entire flight envelope of the BTT missile and is fit for real time control due to the advantages of the FCMAC NNs. Simulation results have shown the rightness and effectiveness of the designed autopilot.