摘要：AbstractThis paper deals with computational delay in distributed nonlinear model predictive control. A fast, cooperative distributed model predictive control algorithm is proposed based on parametric sensitivity. The implementation strategy is divided into two different stages. In the background stage, the future state is estimated one step forward with the current state and input. The local MPC controllers perform distributed optimization based on the predicted state and iterate to obtain the nominal optimal solutions. In the online stage, all the controllers correct their nominal optimal inputs through parametric sensitivity. Specifically, each controller formulates its local sensitivity equation based on the state estimation error. In order to solve these linear equations in a distributed way, Jacobi iterative method is applied. The overall algorithm can provide fast control action. A case study is provided to show the promising performance of the proposed method.
