Machine Learning Applications for Reliability

Published literature reports a number of Artificial Neural Networks (ANN) applications to solve problems in the area of circuit design, modelling and fault identification. Motivated by these findings and building upon previous applications of device modelling through ANN, we identify in this article promising areas for reliability processes improvement using ANNs. Because the run faster than simulators, ANN models can provide an impact in the area of Worst-Case Analysis, where a large number of computationally intensive transient simulations may be required. Failure diagnosis is particularly promising in the area of flight circuits, both for ground testing and also for in-flight fault detection and circuit self-correction.