Development of adaptive transducer based on biological sensory mechanism



Journal Title

Journal ISSN

Volume Title



An adaptive sensor concept and prototype has been developed based on a sensing element which is analogous to and inspired by the arrangement of outer hair cells and inner hair cells between the basilar membrane and tectorial membrane which form the organ of corti in mammalian cochlea. The bio-inspired design was supported by development of a bond graph model of the electromotility (active response) of outer hair cells. Outer hair cells perform like actuators and simulation results using this model are compared with physiological data found in the literature to verify its characteristic response. Insight gained from the model is used to develop a sensor structure analogous to the organ of corti and designed to measure acceleration. A piezoelectric bimorph was selected as the transducer basis, and a bond graph model of the bimorph in an accelerometer configuration was formulated to aid control design and simulation. There is no published data regarding the type of information transmitted among the inner hair cells, outer hair cells, and brain. Consequently, a controller intended to adjust the adaptation process similar to what might exist in the cochlear system has been developed for the sensor and based on a model referenced adaptive control algorithm. Simulations verify that the algorithm can successfully control and enhance performance of the sensor. Practicability of the design is evaluated by a series of experiments on a prototype. This study focused on using a controller structure that was programmed, implemented, and tested using programmable logic based on FPGA technology. The experiments evaluated how well the adaptive sensor could meet a specified performance requirement. Implementation issues that arise, such as the need for differentiators in the adaptive controller or internal propagation of vibration within the sensor structure, hinder the tuning ability. Nevertheless, the trends indicate that the algorithm can meet the desired performance if certain limitations can be overcome. Finally, recommendations have been made for expansion of the research in such fields as an alternative structure for tuning, sensor networking, and reference sensor configuration.