A practical trial of stochastic system identification under existence of background noise based on equivalence of statistics and its application to response probability evalutation of acoustic system with music input

Abstract

A practical identification method of functional type for sound insulation systems is derived especially in the form matched to the prediction problem of output response probability distribution once after introducing some new evaluation criteria on output statistics. The probability distribution of output response with an arbitrary input sound under the contamination of background noise is principally able to predict theoretically by positively using the statistics of input and background noises based on the physical law of intensity linearity. More specifically, the system parameters of the objective sound insulation system are systematically identified by employing an equivalence property between the experimentally observed output probability distribution and the theoretically predicted one as a criterion of the parameter identification. As an application to the actual noise environment, the proposed method has been concretely applied to a double walls sound insulation system and experimentally confirmed under a music sound input.