Model Predictive Control: From Foundations to Advanced Topics

Model Predictive Control (MPC) is an optimisation-based control methodology that predicts future system behaviour and computes control actions by solving a constrained optimisation problem in real time. Originally developed for the process industries, it is now widely applied in robotics, autonomous vehicles, aerospace, renewable energy, and building climate control — wherever classical single-loop controllers are insufficient for handling multivariable dynamics and hard physical constraints. This book provides a comprehensive, practically oriented introduction to MPC. The first eight core chapters build progressively from mathematical foundations — convex optimisation, KKT conditions, state-space modelling and discretisation, controllability, observability — through the Linear Quadratic Regulator and Kalman filter, to the full formulation and implementation of constrained linear MPC with output feedback and integral action. Every chapter includes worked examples and complete MATLAB/YALMIP code so that readers can reproduce every result. Four supplementary chapters extend the framework into advanced territory: robust and tube MPC for systems subject to bounded disturbances; nonlinear MPC via direct transcription and sequential convexification (CasADi); and learning-based MPC covering Gaussian Process MPC, Koopman operator MPC, data-driven predictive control (DeePC), and neural network MPC with an autonomous racing case study. A Simulink implementation chapter demonstrates closed-loop MPC applied to a multi-zone building energy system. Two self-study appendices review the required linear algebra and probability theory. The book is suitable as a companion to a graduate course in control engineering or as a self-contained resource for independent study. Readers are expected to have a basic background in SISO control, state-space representations, linear algebra, and MATLAB. This book is also available as a free open-access PDF at https://doi.org/10.5281/zenodo.19310682