Modeling and testing of immunity of computerized equipment to fast electrical transients

One of the major problems associated with testing computerized equipment for immunity stems from the fact that their susceptibility to electrical transients is time-variant. This paper develops an improved statistical model for the susceptibility of computerized equipment to electrical transients and illustrates a few affordable methods for estimating the mean and maximum malfunction probability in an operational cycle. Reinforcement learning algorithms are used to estimate the maximum susceptibility efficiently. These algorithms automatically restrict testing to the most susceptible time windows. This allows affordable testing for safety-related equipment and also supports effective redesign.