Universal Controllers for Architecture Simulation

Abstract

For optimization studies of dynamical systems, it is common practice to model and tune local controllers for miscellaneous subsystems. For instance, a model of a chemical plant may contain a valve motor model, and a model of a PID controller may be included to control the motor. The associated controller tuning effort is ultimately wasted. The actual controller will be retuned anyway after finalization of the system design, or will be structurally different. For this reason, control algorithms are needed that just provide the functionality of the actual control algorithm that will be designed in a later phase of the system design. These temporary algorithms need to have low tuning requirements, and it must be possible for non-controlspecialist to generate them. On the other hand, they only need to function inside a simulation environment. Several mainstream control approaches are reviewed, and boundary layer sliding mode control is proposed as a suitable approach for this kind of task. This class of controllers can be used without any tuning effort, and is able to compete with tuned PID-controllers in terms of tracking performance. An end-user friendly implementation of a universal controller in the equation-based and object-oriented modelling language Modelica is presented. Several examples are shown to demonstrate the performance of the proposed approach.