FAILED PORTION POINTING SYSTEM OF ANALOG ELECTRONIC CIRCUIT APPLYING NEURAL NET

摘要

PURPOSE:To enable a failed portion to be pointed out by using a neural net, capturing a target analog electronic circuit as a black box, and then leading a degree of failure confidence of a possibly faulty portion as a result of measurement of a plurality of measurement items for warranting characteristics and functions of a target circuit. CONSTITUTION:Measurement results for each measurement item regarding and analog electronic circuit are allocated to an input layer unit of a neural net 1 as an input value and at the same time a degree of failure confidence at a possibly faulty portion is allocated to an output layer unit of the neural net 1 as a teach value, thus enabling the neural net 1 to perform learning. Therefor, an unknown actually measured value for an analog electronic circuit is input, an input value which was already used for learning is taken into the actually measured value in the form of interpolation, and at the same time it is recognized as an approximation of an already learned pattern and then the degree of failure confidence at a possible faulty portion to be output according to this approximation pattern, thus enabling a failed portion of the analog electronic circuit to be pointed out.