Artificial neural network for testing analog circuits

Abstract

A method for testing analog circuits with the aid of neural networks is described. It takes advantage of the high parallel information processing capabilities of these networks. The testing problem is efficiently formulated in a neural-network context. A back-propagation neural network and a functional link net have been trained to synthesize the complicated mapping from the circuit measurements space to the circuit elements space. A testing example is presented to demonstrate the proposed method.< >