FPGA Implementation of a Recursive Rank One Updating Matrix Inversion Algorithm for Constrained MPC

The core computation load of a constrained model predictive control (MPC) problem with interior point method is to solve a linear equation during every iteration. According to the specific feature from problem formulation for constrained MPC, this paper proposed a recursive rank one updating matrix inversion algorithm for solving the linear equation. This algorithm is hardware implementation oriented. Its efficiency in implementing constrained MPC is verified by FPGA implementation results. More over, some important issues related to FPGA implementation of control algorithms are also investigated and discussed