Compact artificial neural network approach for multiple fault location in digital circuits

The authors report a new approach for the diagnosis of multiple faults using artificial neural networks (ANNs). The approach utilises a compact and efficient set of data derived from a set of test patterns generated for a given circuit. It is demonstrated that the approach will lead to a higher multiple faults diagnosis performance, in addition to a significant reduction in the size of the ANN and the training data. Hence enabling failure diagnosis of larger and more complex systems.