Diagnosis of telephony line card component failures using an artificial neural network

Abstract

This paper describes the development of a Modular Artificial Neural Network (MA") that is suitable for the diagnosis of failures in high-volume, printed circuit boards using both analog and digital components. A case study comparing the MA" to a n existing expert system used for diagnosing telephony line card component fatlures is presented. The structure and training strategy for the MA", as well as the current status of the project are discussed.