We develop control systems and control and operating strategies for our customers using simulation-based methods. The basis for development are system models or simulation or measurement data. Systems are characterized and analyzed based on the models or data. Subsequently, suitable control engineering methods and algorithms are selected for controller design. The controllers, control systems, and operating strategies we develop are either made available to our customers as FMUs, or we implement them directly in simulation and development environments such as Simulink, Labview, Dymola, or Python. Among others, we have successfully implemented controllers, control systems, and operating strategies in the following applications:
Our controllers, operating strategies, and control software are also used on our own test benches. Furthermore, many of the system models created for our customers, as well as many models from our model library TIL (e.g. heat pumps or fuel cell systems), contain controllers and control systems we designed and implemented.
Our control development is based on system analysis using simulation or measurement data that characterizes the system in steady state and dynamic terms. Using this analysis, we work with our customers to select appropriate control methods (e.g. linear controllers, gain scheduling, decoupling controllers, feedforward controls, model-based controllers) for the application of interest. We use time and frequency-based methods in our analysis, such as relative gain arrays, FOTD substitution models, and Bode diagrams. For such analyses, we offer the add-on "Control Oriented Analysis" for our software package MoBA Automation.
Control development can take place continuously alongside product development. Different control methods and schemes can thus be tested and evaluated at an early stage.
For the design of model-based controllers (i.e. controllers that directly incorporate simulation models) simplified physical models or data-driven models may be used. Appropriate model type and algorithms are chosen based on the specific requirements of the application. For example, if an application requires a machine learning model for online computation of optimal steady-state operating points, a suitable algorithm can be Nonlinear-Model-Predictive-Control (NMPC) for the optimization of dynamic trajectories.