نبذة مختصرة : International audience ; The literature on active control, which has developed abundantly since the last 20 years, suffers from a lack of methodologies to compare the performance of different proposed algorithms. More precisely, the criteria introduced to evaluate them are often specific to their field of application, which makes them unsuitable for objective comparison with other algorithms. This work proposes a unified method which intensively compares algorithm performances. It relies on the definition of application- independent descriptors that reflect performances in the time domain both for transient and steady regimes. This evaluation method adapts to the user’s needs by letting him choose the signals on which the algorithms are tested and by allowing him to adjust the weight of descriptors in the final score. The method is used to compare two feed-forward adaptive control algorithms for vibration cancellation: the FxLMS and an adaptive algorithm using the Youla-Kucera parametrisation. Test signals are extracted from an experimental test bench, designed to mimic a museum stand, consisting of a cantilever beam driven at one end with a vibration shaker and controlled at the other end with an actuator. The results show through several examples how the choice of descriptors’ weight can be influenced by the user’s needs and can affect the final score assigned to the algorithms.
Processing Request
No Comments.