A hierarchical regression analysis of extended type for response probability prediction of acoustic systems an introduction of regression parameters with the input dependency

Abstract

In general, it is necessary to establish a new theory taking account of the input dependency of system parameters when the stochastic input signal (different from the test signal used as an input for the system identification) is newly excited. In this paper, a prediction method for the output response probability of arbitrary stochastic acoustic systems is proposed based on the foundation of the previously reported extended rearession analysis method with less information loss. In this case, a reasonable method for estimating the system parameters contained in the extended regression analysis method is proposed by considering the above input dependency and the statistical property of stochastic input signal. Finally, the effectiveness of the proposed method has been experimentally confirmed by applying it to an actual room acoustic svstem.