Some precise evaluation theory of Leq using a wide sense digital filter and its application to traffic noise

In a usual modeling method for actual stochastic systems, it is very often to find some idealized model by paying attention to only the average form of stochastic phenomena. Namely, the deviations form this idealized model are artificially treated as some meaningless information. Especially in a case when considering an arbitrary non-linear system with non-Gaussian input signals, however, the available informations still remain in these deviations in order to estimate and predict precisely some stochastic characteristics, e.g., the probability density function (abbr., p.d.f.), the cumulative distribution function (abbr., c.d.f.) and the higher order non-linear correlation functions. In the usual modeling methods for actual systems, the deviations from the actual phenomena occur inevitably owing to the model constructions under the idealized conditions.