Data-driven robust optimal control for nonlinear system with uncertain disturbances

Optimal control methods have attracted much attention for their promising performance in nonlinear systems. However, it is difficult to achieve satisfactory performance due to uncertain disturbances. To cope with this problem, a data-driven robust optimal control (DROC) method is proposed for uncertain nonlinear systems. The merits of the proposed DROC method are threefold: First, a data-driven evaluation strategy is introduced to capture the relationship between the approximating errors and the control variables. Then, the control performance indexes of nonlinear systems can be established within uncertain disturbances. Second, a multi-objective robust optimization algorithm is developed with a coevolution strategy. Then, robust optimal control laws can be obtained to improve the control performance. Third, the robust boundedness of DROC is discussed in theory. Then, the stability of the control systems can be guaranteed analytically. Finally, the effectiveness of DROC is illustrated with two multiple input multiple output second-order nonlinear systems. The optimal control performances are displayed in experiments to demonstrate the effectiveness of DROC.