Mohammad Javad Katouli - Amol Institute of Higher Education
Jalil Sadati - Faculty of Electrical Engineering, Babol Noshirvani University of Technology

In this study, a novel control scheme is presented for wing rock Phenomena. A sliding mode controller called twisting algorithm is combined with wavelet neural network (WNN) identifier to propone an adaptive control strategy. To train the proposed network, offline and online processes are considered. In offline training, initial weights of WNN identifier are obtained and this procedure helps the convergence rate of the network in online phase. Moreover, an identifier learns to predict one step ahead the dynamic of the plant and this characteristic helps the proposed control strategy to construct predictive-based control scheme. For offline training, a swarm optimization method called Bees algorithm is employed to gainoptimal weights of WNN. Also, the backpropagation algorithm due to its fast response and better performance in local search for optimization problems is considered for online training. To investigate the efficacy of the proposed scheme, this control strategy is compared with conventional method and the simulation results confirm the effectiveness of this strategy

wing rock, sliding mode, neural network,adaptive scheme