The aim of this course is to teach participants a broad motion control tuning framework for multivariable systems in an industrial setting. We will teach you a step-wise systematic procedure with user-friendly tools, enabling a systematic design within limited tuning time. At the same time, in the step-wise procedure, the complexity is only increased when necessitated by the control specifications or control problem, for instance, how critical the interaction is for the control problem at hand. This aspect is extremely important in an industrial setting, where both fast tuning is important as well as the resulting control performance.
The course builds on the basic course 'Motion control tuning' for a SISO situation, which is briefly summarized at the beginning of the course. The multivariable framework starts with an analysis framework using well-known frequency response functions, that participants can use to judge whether traditional SISO motion controller tuning techniques can be used blindly. If the interaction turns out to be problematic, then decoupling techniques can be used to reduce interaction. In a next step, we will use systematic techniques, including sequential loop closing, to continue using the well-developed motion control tuning framework using frequency response functions, while at the same time addressing interaction. If the interaction is severe, the full multivariable control problem is addressed in a model-based design (H2, Hinfinity, and mu-synthesis), where the key focus lies on design aspects for motion systems. A key emphasis will be on the associated modeling cost, which can be rather excessive. Finally, the feedback controller is complemented with an advanced feedforward design. These steps continuously alternate with its experimental implementation in a hands-on case of a mechanical (2 axes) servo.
After completion of the course, you are able to use a broad arsenal of tools for control design of industrial multivariable servosystems. Through efficient analysis tools, you can determine, based on inexpensive frequency response function measurements, if and to what extend the interaction between the axes needs to be addressed. You will be able to apply a stepwise approach to find adequate settings of a multivariable controller, to determine the achievable performance of the controlled system, and to understand what limits this performance.