Advanced-Multi-Step and Economically Oriented Nonlinear Model Predictive Control

This dissertation addresses two issues that arise in the field of Nonlinear Model Predictive Control (NMPC): computational delay and stability of economically oriented NMPC. NMPC has gained wide attention through the application of dynamic optimization. It has the ability to handle variable bounds and multi-input-multi-output systems. However, computational delay caused by large size of nonlinear programming (NLP) problems may lead to deterioration of controller performance and system stability. In this thesis we propose an advanced-multi-step formulation of NMPC (amsNMPC) based on NLP sensitivity. The basic idea of amsNMPC is to solve a background NLP problem in advance to get predictions of future manipulated variables. These are then updated online using NLP sensitivity when the actual states are obtained. This method could be applied to optimization problems whose solutions require multiple sampling times. We then analyze the nominal and robust stabilities of the two approaches. Two examples are studied to evaluate the performance of amsNMPC. The ultimate goal of any operation strategy for a process plant is to make profit. Traditionally this goal could be achieved by a two-layer Real-time Optimization (RTO) system, where the upper layer solves a steady state problem aiming at optimizing economic performance to get the optimal setpoints for the controlled variables in the layer below. The lower layer then keeps the controlled variables at their given setpoints using MPC/NMPC. However, there are some problems with this two-layer structure. One of the solutions is to combine these two layers and include the economic criterion directly into the cost function of the lower layer controller when an optimization-based controller such as MPC is used. This approach is often referred to as Economic MPC. The issue with Economic NMPC is that the controller may not be stable. In this dissertation we analyze its Lyapunov stability property and propose to stabilize it by adding quadratic regularization terms to the objective function, and we also provide a method to calculate the most appropriate weights on regularization terms to ensure the stability of Economic NMPC while achieving the best possible economic performance. Several challenging case studies are used to demonstrate these concepts.