Control allocation (thrust allocation) is a very important issue in marine vessel applications and doing this task is really challenging in the presence of linear and nonlinear constraints on actuators and effectors. Usually, thrust allocation problems treated as an optimization problem. First, thrust and angle of each thrusters are optimized and then other goals such as minimizing the power, singularity avoidance, tear and wear minimization of thrusters are handled. In this thesis, a new approach is presented for thrust allocation based on the covariance matrix adaptation evolution strategy (CMA-ES) and it is implemented on a marine vessel with 7 azimuth thrusters to indicate appropriate performance of the proposed approach. Since the CMA-ES is unable to consider linear constraints, a term is defined and augmented in the cost function in order to handle linear constraints. The proposed approach determines actual allocated control effort with a little difference to command virtual control effort and with less consumed power respect to other approaches. In addition, it has a more convergence speed. Simulation results show the accuracy and efficiency of the proposed approach. Keywords: Control allocation, Thrust allocation, Dynamic positioning system, CMA-ES algorithm, Optimization, Thrusters
