Availability Markov Models of Complex Equipment with Relearning Diagnostic Means after Detecting Latent Failures

Availability assessment is one of the core processes in reliable systems. Downtime can be reduced in many ways, from preventing errors through proper testing to reducing maintenance and repair time. However, this article considers a different approach - improving the reliability of error detection by providing the diagnostic system with a learning component that should increase the reliability of the system. While the learning phase may increase system downtime, it may be beneficial to have a more robust error detection system that begins to improve reliability at a certain point in time significantly. To investigate this hypothesis and analyze the dependencies, the diagnostic system was modeled as a multifragment Markov chain (MFMC) in R. The results of this work can be summarized as follows: 1) A general approach to the construction of a linear multifragment Markov chain in R was developed. 2) A general-purpose diagnostic system was built and represented as an MFMC. 3) Investigate how the learning state of the model, the intensity of error detection improvement, and other parameters affect the availability function, as well as what combinations and approaches are appropriate for specific business cases.