Heating System Case Study for Simulation-Based Control System Analysis

General nonlinear black-box models, hybrid first-principles combined with statistical nonlinear models, and physical component models with block diagrams can often provide detailed and accurate descriptions useful for simulating dynamical systems. The opaque nature of such models often makes them less suitable for traditional (linear) approaches to control system design and analysis. We explore a simulation-based approach to controller design for a heating process as an alternative. As models, we use a system of nonlinear differential equations presented in a recent paper and a linear (multiple-input multiple-output) model identified in the present study. Based on defined criteria, PI-controllers for a key quality variable are designed by optimizing simulations. Simulated and experimental results for the identified linear model are compared to a nonlinear grey-box model. Based on this comparison, we propose to include simulation-based control system analysis for developing more reliable criteria for simulation-based design.