Adaptive filters for real-time system identification and control

An adaptive filter or a neural network concept has been used to develop an algorithm for real-time system identification and control. The algorithm permits not only identification of transfer functions of linear structural systems but also extraction of modal parameters such as natural frequencies, damping ratios, and mode shape coefficients in real time. Because the algorithm is capable of identifying a single mode at a time, bandpass filters are employed to make it work properly for systems with multiple modes. The algorithm has been implemented in a digital signal processor-based data acquisition and control system for the system identification and control experiments with the U.S. Naval Research Laboratory space truss. It was experimentally demonstrated that the algorithm was capable of identifying modal parameters of the truss accurately by using a truss tip-mounted accelerometer as a sensor and a piezoceramic active strut as a disturbance source. The algorithm was also successfully employed to conduct active vibration suppression and performance monitoring of the feedback controller by using the active strut and a collocated force transducer.