For manipulating the structure of an adaptive building, knowledge of its current state is essential. Various quantities of interest are not directly measurable which is why sensor fusion and state estimation methods are required. As an addition to conventional sensors (accelerometers, gyroscopes, strain gauges), novel optical measurement techniques are developed. In combination with the estimation algorithms, they are to guarantee position measurement with high precision and reliability even under challenging environmental conditions for structures of large spatial extension. Two complementary approaches for realizing the optical distance measurement are investigated.
Aim of this project is to apply sensor fusion to all available sensor signals in order to obtain a state estimate of the system. Sensor fusion has to be carried out in real-time, as the state must be available for control (project B04) at all times.
Given the above, project B02 seeks to answer the following questions:
- Which approaches are suitable for realizing precise optical distance measurements of structures with large spatial extension?
- Can we avoid light scattering to a degree that allows for precise determination of position even under challenging environmental conditions?
How is sensor fusion for conventional sensors and the optical measurement system carried out in real-time?
The first and primarily investigated approach for the optical distance measurement is based on determining as precisely as possible the position of active light emitters attached to the building. The emitters are mounted for instance to the building’s façade or to elevator shafts. They are realized cost-effectively using computer-generated holograms that are projected onto a camera system. Advanced image processing techniques allow for partial elimination of disturbances and artefacts due to weather conditions.
In a complementary approach, it is investigated how light scattering can be avoided despite adverse weather conditions. Since the measurement is performed by scanning arbitrary points on the façade, the building does not need to be modified. The aim is to suppress the scattering due to unfavorable weather conditions with short coherent light obtaining an ideal distance measurement. Measurement points on structures with large spatial extension can be scanned sequentially. For this approach, preliminary studies will be carried out as part of the project.
Building deformation is determined by applying sensor fusion on the distance information obtained from both the optical measuring system and the conventional sensor technology. While the optical measuring system collects information on the outer faces of the structure, conventional sensor technology is usually installed inside the building. Compared to the high sampling rate of conventional sensors, the optical measuring system samples measuring points at lower frequencies. A filter that uses a mathematical model for behavior of the building is designed to dynamically adapt the location of measuring points in order to achieve an optimal signal-to-noise ratio. The aim is to make the system modular and reconfigurable using distributed approaches and to merge conventional sensors with novel measuring principles. In a first step, a global filter is designed. Then, feasible subsystems and interfaces are defined to implement the filters locally, thereby increasing robustness and performance.
The resulting estimated state variables of this project, are used to control the actuators in project B04 and passed on to project B05 for dynamic multiscale visualization.
Most adaptive structures are heterogeneous systems that also include non-mechanical parts. For such structures, port-Hamiltonian systems are a promising modeling technique. The energy-based approach allows for a coupling of subsystems from different domains via the power flow at the interfaces. This way, multi-domain models can be directly created which have the additional advantageous property of being structurally passive. This is helpful for control design. In cooperation with subproject A06, an open source framework for the modelling of truss structures and frames as port-Hamiltonian systems is being developed (https://github.com/awarsewa/ph_fem).
- Prof. Dr.-Ing. Cristina Tarín, Institute for System Dynamics
- Dr. Wolfgang Osten
- Dr.-Ing. Tobias Haist, Institute of Applied Optics