Adaptively Controlled Dynamical Behavior of Sensory Systems Based on Mechanoreceptors

This paper presents results on adaptive control strategies for a sensory system to identify (unknown) ground excitations which force the sensor and its seismic masses, so that acting forces can be measured and identified. The sensor system is modeled as a spring-mass-damper system within a rigid frame with two degrees-of-freedom. The seismic masses are under the load of internal control forces, which shall ensure stabilization of the mass point rest positions despite the continuing ground excitations. Using these (measured) resulting regulating forces, we are able to identify the excitation force. The control strategies are designed referring to the natural behavior of mechanoreceptors from biology. These are able to adapt their sensitivity to the environment, so that they filter the important information out of the flood of information. Mimicking this behavior, adaptive control strategies are used with time-varying controller gains. In this way, we are able to design controllers which are still sensitive while a constant stimuli affects. So new incoming information can be identified with a high quality. Further on, the sensor has to be universal and shall consume less energy as possible. Therefore, control strategies from literature are analyzed and modified, so that the most effective ones are used for the sensor system in this paper. Finally, the best working control strategies are tested for both their long-term behavior to an excitation which simulates different situations and for their response to different system parameters, chosen randomly.