Abstract: Aimed at nonlinear multi-input multi-output(MIMO) system,a predictive control strategy based on least squares support vector machines(LSSVMs) and chaos optimization is proposed.Predictive model is one of the three main factors of predictive control.Chaos-LSSVM algorithm based on chaos optimization is presented to obtain optimal LSSVM parameters and the model by iterative search in the feasible region.Online optimization is another essential factor.MSC-MPC(mutative scale chaos-model predictive control) algorithm based on mutative scale chaos optimization is developed,which can decide whether to reduce the search scope according to the size of control error,thus it can converge to the optimal solution rapidly.The algorithm is easy to compute and implement,and avoids the complicated derivation and inversion of other similar methods.The simulation results show that Chaos-LSSVM algorithm and MSC-MPC algorithm have good modeling and control performance,respectively.