Noise source and transmission path identification via state-space system identification

Abstract

Various approaches have been taken to analyze noise/vibration source(s) and transmission path(s), such as structural analysis, direct measurement, signal processing, etc. In spite of their remarkable success in understanding noise source(s) and, to a lesser extent, transmission path(s), these approaches require further analysis to identify the transmission paths or contributions from each noise/vibration source. In the systems and control community, the so-called “system identification” has long been used to directly build a mathematical relation (or transfer function) between source(s) and response(s), from which the transmission path(s) can be readily inferred. From the various system-identification techniques, this paper proposes the use of the recently developed state-space system identification, due to its superior performance over existing identification techniques. The first step of identification is to define possible candidates for noise/vibration sources and response(s). The independence of the candidate sources is then tested using state-space system-identification in order to eliminate any redundant source(s) that may degrade the transfer function estimate between source(s) and response(s). Once the source independence is guaranteed, the state-space system-identification technique is again applied to identify the transmission paths. A scroll compressor is used as a testbed throughout this paper, while the applicability of the proposed approach is virtually unlimited.