Efficient predictive control algorithm for nonlinear discrete systems with persistent disturbances

An efficient model predictive control algorithm for nonlinear discrete systems with external persistent disturbances is characterized. The contribution of this article is two fold. First, we transfer the real-time nonlinear procedure into a nonlinear-quadratic regulator design for an off-line system by a novel successive approximation approach (SAA) at each control interval. Second, an adjoint vector is introduced to compensate the nonlinearities and the disturbances. Only the compensate term need to be calculated online in order to reduce the computation costs. This reduces the time in computing the feedforward and feedback optimal control law online. Simulations are conducted to demonstrate the real-time applicability of the approach proposed.