Actuator and sensor placement for structural testing and control

Abstract In this paper, the actuator and sensor locations of a structural test item are selected as a replacement of the disturbance inputs and the performance outputs of a real structure. The most straightforward approach is to place sensors in the areas of performance evaluation, and actuators in the areas of disturbance action. However, this solution is rarely possible due to technical and economical reasons. Therefore, the actuators and sensors need to be placed in preselected regions, and should duplicate as close as possible the disturbance action and the performance measurements. In this paper a placement problem with non-collocated actuators and disturbances, as well as non-collocated performance and sensor outputs is solved. The solution is determined by locating sensors (actuators) such that the Hankel singular value vector of a structure from actuator inputs to sensor outputs is closely correlated with the Hankel singular value vector of the structure from the disturbance inputs to performance outputs. It is shown that this approach improves additionally the cross-coupling between actuators and performance, and between disturbances and the sensors, thus improving overall closed loop performance. The method is illustrated with the determination of sensors of a truss structure, where two selected sensors replaced an original set of 36 sensors.