A neural network approach to nonlinear model predictive control

Abstract

This paper proposes a neural network approach to nonlinear model predictive control (NMPC). The NMPC problem is formulated as a convex programming problem via Jacobain linearization. The unknown high-order term associated with the linearization is estimated by using a feedforward neural network via supervised learning. The convex optimization problem involved in MPC is solved by using a recurrent neural network. Simulation results are provided to demonstrate the performance of the approach.