A recurrent neural network based fuzzy sliding mode control for 4-DOF ROV movements

A recurrent neural network (RNN) based fuzzy sliding mode control (RFSMC) method is proposed in this paper to achieve a stable movement for 4-degree freedom remotely operated vehicle (ROV) under the condition of ROV uncertainty and in the presence of external disturbance. The ROV model uncertainty term is estimated and compensated adaptively by training a RNN offline. Meanwhile, a fuzzy logic system (FLS) is employed as the switching term of the conventional sliding mode control (CSMC) in order to reduce the chattering phenomenon. By doing so, the external disturbance can be eliminated automatically during the movement of the ROV even though the upper bound of the external disturbance is unknown. The stability of the ROV with the presented RFSMC method is proven by Lyapunov stability analysis, and compared with the state-of-the-art methods including CSMC, double-loop sliding mode control (DSMC) and back propagation neural network based sliding mode control (BPSMC) method. The experiment results demonstrate that the proposed method has great robustness. The position error and trajectory error can rapidly converge to zero, and the chattering phenomenon can be suppressed consequently. • Neural network technology has been widely used in underwater vehicle, but most of them is radial basis function neural network or back propagation neural network. In this manuscript The ROV model uncertainty term is estimated and compensated adaptively by training a recurrent neural network offline. • A fuzzy logic system (FLS) is employed as the switching term of the conventional sliding mode control (CSMC) in order to reduce the chattering phenomenon. • The stability of the ROV with the presented method is proven by Lyapunov stability analysis, and compared with the other methods including CSMC, double-loop sliding mode control (DSMC) and back propagation neural network based sliding mode control (BPSMC) method.