The use of artificial intelligence for dependability assurance

The Aim of the paper is to analyse the state of the art of artificial intelligence application in Russia as regards technological dependability, as well as to propose new promising areas of research and development. Methods. The methods of contextual information search, system analysis, and dependability theory are used. Results . A review of domestic publications in the area of interest was conducted and showed the applicability of various artificial intelligence methods, in particular machine learning, to improve the dependability of various technological items. Two main t asks are identified to be solved: identification of pre failures in order to prevent failures by conducting preventive maintenance or repair; rapid detection of failures that have already occurred and their localisation. Examples of existing similar solutions are provided. The possible ways to overcome the absence of initial learning data associated with rare failures, are analysed. For more accurate prediction of failures, it is proposed to collect and use not only the parameters that characterise an examined item itself, but also environmental parameters that can also affect the condition of the item. The paper shows the relevance of studies aimed at generalized and systematic results to serve as guidelines for preferred application of certain artificial intelligence methods. New promising areas of artificial intelligence application are indicated, i.e., identifying possible common causes in cases of multiple failures, which will help reduce recovery time, and analysing the root causes of failures in order to take measures to eliminate them or reduce their future impact. Conclusion . The conducted analysis and the propose recommendations will contribute to the cross‑industry exchange of experience, the expansion and deepening of work on the use of artificial intelligence for dependability assurance and make them more practical.