Dutilleux G., Ecotière D.

The study of long term acoustic performance of road pavements requires to access the evolution of the intrinsic parameters of these surfaces, like resistivity or tortuosity. In long range sound propagation aswell, it is well known that ground absorption is time-varying. The understanding of this variation is an important issue. In both application cases, non-destructive measurement methods are required. The present contribution focuses on the 2-parameter model by Delany and Bazley for outdoor surfaces and the 4-parameter one by Hamet-Berengier for porous road pavements. It is shown that for both models the absorption coefficient spectrum obtained by an ISO 13472-1 compliant measurement contains enough data for the automatic identification. For the 2-parameter model the identification is based on a suitable cost function. The complexity of the function is low enough for a resolution by exhaustive search. It turns out that, for a large set of soft ground samples, the cost function is convex. Therefore the resolution can be done quite efficiently by classical gradient descent methods. Regarding the 4-parameter model, although the cost function is no longer convex, the resolution can be performed using a global optimization tool, like an evolutionary algorithm.