ANN- Assisted Reliability and Availability of a Warm Standby Repairable Computer Network System

This study investigated the reliability and availability of a computer lab network consisting of three computer labs connected through a server in a parallel configuration under a 2-out-of-3: G policy utilizing a random process, employing a supplementary variable technique, and an artificial neural network (ANN) approach. The proposed complex system may experience failure because of the failure of at least two computer labs, a server, or a catastrophic event occurring at any given time t. The failure rates of the units were constant and expected to follow an exponential distribution. The repair rates are assumed to be general; however, a completely failed system is coupled using the Gumbel-Hougaard family copula. Each type of failure and repair rate is regarded as a neural weight when analyzed using the ANN approach, which is managed using an exponential distribution. The system state probabilities, up and down state probabilities, and availability and reliability of the model are evaluated using a Markovian process and Laplace transforms. Finally, the reliability and availability of the model were analyzed using an ANN framework to estimate the analytical curves. The results obtained using MAPLE software (analytical computation) and MATLAB software (ANN training and simulation) confirm that the ANN provides reliable estimates of the reliability and availability measures that closely match the analytical outcomes.