Synthesis of Optimized Adaptive Digital Filters for System Identification and Vibration Control.

Kurzfassung

In recent developments of controllers for active noise cancellation and active vibration suppression, digital real-time filters and different adaptation algorithms are use. The necessity of their optimized performance lead to algorithms which are strictly based on the application, but the digital filter type should also be considered in an optimizing investigation. A few adaptive filter types have been developed in the past, especially for acoustic and speech processing applications. In the field of adaptive structural systems, however, special advancements of filters are missing. This paper deals with the synthesis of adptive filter systems which should be optimally designed for the applications of system identification and active vibration control. It is extremely useful to take a priori knowledge into consideration when choosing a filter type. Therefore, the filter type should be optimal with respect to the structure to be controlled, but also to the algorithm, in oder to actually offer the possibility of good performance. Thus, special adaptive IIR, Lattice and Ladder filters are discussed, and a new adaptive filter type is presented, which is a combination of all of these filters. Their properties (stability, sensitivity, round-off-noise, canonical/non-canonical and transposed forms, suitability for adaptive algorithms, etc.) are studied in experimental investigations. Illustrated examples are given, also showing the synthesis procedures.