Class-based neural network method for fault location of large-scale analogue circuits
Abstract
A new method for fault diagnosis of large-scale analogue circuits based on the class concept is developed in this paper. A large analogue circuit is decomposed into blocks/sub-circuits and the nodes between the blocks are classified into three classes. Only those sub-circuits related to the faulty class need to be treated. Node classification reduces the scope of search for faults, thus reduced after-test time. The proposed method is more suitable for real-time testing and can deal with both hard and soft faults. Tolerance effects are taken into account in the method. The class-based fault diagnosis principle and neural network based method are described in some details. Two non-trivial circuit examples are presented, showing that the proposed method is feasible.