Introduction. Impulse response testing is currently one of the most common methods of acoustic express diagnostics of the foundation slabs and other plate-like structures state, but is relatively uncommon in the Russian Federation. A modification of this technique, based on the analysis of normalized response parameters, is being actively developed by Russian specialists to assess the quality of soil-structure contact. The issue of developing of the new informative parameters and criteria for identifying of anomalies associated with contact violations is topical.

Aim. The purpose of the work is to inform specialists about the capabilities of the normalized acoustic response analysis method for localizing zones of soil-structure contact violation. It is proposed to apply the attribute "void index" and an auxiliary criterion for data analysis based on the nature of the distribution of attribute values.

Materials and methods. The features of the normalized acoustic response testing metodologia for soil-structure contact violations searching are described. The method for calculating the attribute "void index" for the data of the technique is shown. A criterion for the selection of anomalous areas is proposed, based on the data analysis of the nature of the distribution of attribute values.

Results. Verification of the "void index" attribute application for searching for zones of soil-structural contact violation is shown on the results of a survey of a tunnel liner physical model with a known position of the void. An example of the results of a survey of a slab at the base of an underground car park shows the use of a "void index" in conjunction with the normalized signal energy attribute. A method for statistical processing of the results is proposed, which supplements the traditionally used "three sigma" criterion.

Conclusions. A method of the "void index" attribute application to localize zones of soil-structure contact violation for the normalized acoustic response analysis method is shown. A method for statistical data processing is proposed, considering the nature of the distribution of response attributes.