Introduction of Information Criteria for Identification of Generalized Regression Functon Type System Model and Its Application to the Probabilistic Evaluation of Actual Environmental Systems(Experiment)

DOI Web Site オープンアクセス

この論文をさがす

説明

As one of evaluations for living environment, it is important to evaluate environmental acoustic systems. For this purpose, the prediction of response output fluctuation to an arbitrary random input must be considered. Then, since, corresponding to non-Gaussian random input, the response output fluctuates in non-Gausssian distribution form, not only mean and variance but also the whole output probability distribution form are necessary to evaluate the output fluctuation. However, the environmetal acoustic system is too complicated to construct the physical dynamics based on the mechanism from bottom up way viewpoint. In this paper, from top down way viewpoint, as a system model, a general regression function of output to input is introduced. By employing the additive property of intensity quantity, the system contaminated with the background noise is expressed in the sum of the above regression function and background noise. In order to identify this model as naturally as possible, the minimum entropy criterion and the Kullback's information criterion matched to non-Gaussian fluctuation are introduced. By noting that the output intensity is non-negative random variable, the response output probability density function can be assumed in a form of the statistical type Laguerre series expansion. Its expansion parameters can be estimated by use of identified model according to input data. Finally, the proposed method is experimentally confirmed by applying it to an actual sound insulation system.

収録刊行物

キーワード

詳細情報 詳細情報について

問題の指摘

ページトップへ