A key aspect of adaptive buildings is their automated response to stationary and dynamic loads, both of internal or external origin and a priori unknown. This functionality has to be guaranteed over the entire lifetime. The adaptation requires a control design, in which the truss structure is manipulated using the available actuators based on estimated states and loads. Design and experimental validation of these methods is a primary goal of this project.
The project addresses the following research questions:
- Which mathematical models are suited for the control design of adaptive structures?
- What are adaptive control algorithms that can automatically adapt to changes of the system?
- How can we design local controllers, which together guarantee global stability?
- How can distributed control concepts replace current models that rely on a hierarchical structure, thus rendering a “master controller” obsolete. Functional groups are created dynamically, based on the current state.
- How can we validate the performance of the developed control algorithms in experiments?
From a control engineering point of view, adaptive structures differ in several aspects from mechatronic systems and systems of process engineering. Through their functional elements with integrated actuators and sensors, many inputs and outputs are present. Thus, the input – output behavior of the system is of high dimension, rendering the control design a problem of multiple-input, multiple-output (MIMO) control.
Classical control concept for MIMO-control assume a central processing unit. The controller is designed and implemented as a centralized controller with access to all sensor outputs and actuator inputs. However, this concept is not suitable for adaptive buildings due to their multitude of active elements distributed over the entire structure, with large distances in between leading to high communication delays if the connections are realized using a common bus topology, e.g. Ethernet. However, these elements are, of course, physically coupled. This presents a challenge, as the project B04 aims at developing distributed control concepts, which operate on local hardware and rely on local sensor measurements and actuators only. To enable the exchange of information on demand, these local controllers will be digitally interconnected. Consequently, optimizing the underlying network structure offline and online is part of our research in this project.
Conventional control concepts typically lack the ability to adapt to drastically changing structure geometries and parameters of a building. In this project, we want to develop reconfigurable control concepts, which adapt autonomously to changes of the underlying controlled system. By focusing especially on the reconfiguration in the case of actuator failures, this research significantly improves the reliability and usability of an entire adaptive building.
Prof. Dr.-Ing. habil. Dr. h.c. Oliver Sawodny, Institute for System Dynamics