Abstract
One of the representative statistics actually used in the evaluation and regulation problems of the environmental noise and vibration is the so-called L_α(α=5, 10, 50, …) evaluationlevel, which is directly related to the probability distribution form. This fact makes it difficult to solve the estimation and prediction problems of the L_α level systematically in the actual engineering field of noise and viblation control. First, our main interest is devoted to solve the estimation problem in the form of non-sequential algorithm for L_αlevel. By introducing a generalized non-stationary model for the internal structure of random noise and vibration, a unified cumulative distribution function of the observed non-stationary random fluctuation is proposed in advance in the form of orthogonal and/or non-orthogonal expansion series type expression. The important evaluation index, L_α level, is directly calculated by use of the generalized expansion theorem derived from the above unified distribution function in the form of non-sequential algorithm. Next, we investigate a theoretical estimation procedure based on the stochastic approximation method in the form of sequential algorithm. Furthermore, we have experimentally confirmed the validity and effectiveness of the above two different estimation methods not only by means of digital simulation but also by experimentally observed road traffic noise data in Hiroshima City. All of the experimental results clearly show the good agreement with the present theory.
