The paper covers creating the algorithms for calculating the probability of various types of defects in tunnels, the development of which can lead to accidents. Tunnels are an important and complex part of the transport and communication system, through which heavy traffic is carried out. Determining the probability of defects in the latent period of their initiation in individual sections of tunnels is an important problem. The formation of defects is accompanied by the appearance of noise that distorts the useful signals coming from sensors and measuring instruments installed to control the stability of the tunnel and the reliability of its structures. Traditionally measuring instruments register noisy signals, and the technical condition of the tunnels is assessed on the basis of the values of their characteristics. It is shown in the paper that the more reliable indicators of fixing the onset of dangerous changes in the latent period of initiation are the characteristics of the noise, which cannot be extracted from the noisy signal. It is noted that the probability with which the noise takes on admissible and critical values is an indicator of changes in the technical condition of tunnels. Algorithms have been developed for calculating the probabilities of the noise values getting in the given intervals. These probabilities are stored as reference sets for the initiation of tunnel defects. After the training has been carried out, the values of the probabilities with which the noise takes on the given values at different time instants are matched to the type of defect and one of the possible technical states: serviceable, operational, partially operational, inoperable; pre-emergency; emergency, etc. It is also shown that the differences in the probabilities with which the noise takes on the same values at different times are indicators of the dynamics of changes in the malfunction in the tunnels. A database ofinformative attributes of the intensity of the development of failures is also created in the paper. For this database, the indicators of the dynamics of the development of a defect are determined, such as insignificant, slow, significant, intensive.
