Research of Fault Detection System Combining with Fault Tree Analysis with Artificial Neural Network in Large Scale Die-Forging Press

Abstract

This paper research on a new fault detection method used in large scale die-forging press, based on the national major science and technology project (2009ZX04005-011). This method combines fault tree analysis (FTA) with artificial neural network (ANN), it need to pick up training samples of ANN from fault tree constructing by analyzing system first. An effective network model can be achieved through using BP algorithm to train the samples based on the tree-layer feed-forward network structure. This network model can output fault reasons according to fault symptoms.