PD like fuzzy logic control of an autonomous underwater vehicle with the purpose of energy saving using ${H_\infty }$ robust filter and its optimized covariance matrices

Abstract

In this paper, new methods have been utilized to design and implement PD like fuzzy controllers for naval postgraduate school autonomous underwater vehicle. One of the main challenges for appropriate performance in the fuzzy controllers is to determine the optimal place for the membership functions for each inputs and outputs. Owing to the time variable dynamics, experimental knowledge or offline tuning with dynamic model cannot solve this problem correctly. The proposed method used ${H_\infty }$ robust filter with optimized covariance matrices for optimization of the membership functions of the designed controllers for heading and depth channels for the first time. Advantages of new investigated method are great convergence velocity and robustness against uncertainties of model and surrounding environment. In simulation results, the proposed method has been compared with extended Kalman filter (EKF), as another optimization approach. Results demonstrate that proposed method has impressive effects on tracking the desired path with dramatical decline in control efforts, which have crucial role in reducing energy consumption in practice, whereas the EKF simply leave its optimality when applying the real world conditions. Also, practical way for implementing designed controllers, multivariable regression analysis has been used. Statistical survey shows, fuzzy controllers can be easily substituted with obtained multivariable polynomials to be implemented in typical microcontrollers or other non-fuzzy hardware. As regarding to stability, passivity approach has been used to proof asymptotic stability of new designed controllers.