Noise from time varying sources: Estimation of component variance

✍ Scribed by H. Davies

• Publisher: Elsevier Science
• Year: 1983
• Tongue: English
• Weight: 280 KB
• Volume: 16
• Category: Article
• ISSN: 0003-682X
• DOI: 10.1016/0003-682x(83)90036-1

✦ Synopsis

The methods available for combining noise level d&tributions are reviewed. The reverse problem of decomposing a net distribution into its constituent parts is discussed and information is presented which allows estimation of the variance of a constituent distribution, given the residual and net distributions.

INTRODUCTI ON

The prediction of net noise levels arising from existing and projected sources is a central feature of current noise planning practice. Such computations are standard if the sources are incoherent and if the levels arising from them are independent of time, but whilst the first of these conditions is usually met in practice, the second is not. In this case, energy computations involving the equivalent continuous sound level, Leg, are still straightforward, but if the time fluctuations as represented by, say, the various percentiles, L., or the noise pollution level, Lm,, are to be taken into account, the problem becomes that of combining probability distributions of levels.

Such a combination is properly achieved by a convolution-like process.

In particular, suppose that X(x) and Y(y) represent the probability density functions of the sound pressure levels x and y arising from two sources, and that the sources are incoherent so that particular values of x 147