A railroad track requires continuous monitoring of its technical condition. In the previously proposed work, an intelligent system for identifying track sections requiring unscheduled inspection, installed in one of the rolling stock cars of a railroad bed, has been developed. A noisy vibration signal enters the system, and the characteristics of the noise are computed, which are further used as informative attributes to identify the initial latent period of malfunction initiation. For this purpose, the noise characteristics are transmitted via radio channel to the control station, where a knowledge base reflecting the technical condition of each section is formed. In this paper, in order to increase the adequacy of the control results, we propose using ratios of the estimates of the high-order characteristics of the useful signal and noise as informative attributes. Due to the complexity of implementation of these characteristics, it is recommended to send the control result to the control station, where technical means are installed that allow to perform such calculations. Introduction of the proposed ratios in the knowledge base allows to reveal the dynamics of malfunction development. The use of the proposed technologies increases the degree of train traffic safety.