Abstract
A novel method of generating multi-frequency test stimuli for incipient faults is presented to improve the fault detection accuracy of analog circuits. This paper analyzes the primary cause of low incipient fault detection accuracy and indicates that the high aliasing between the incipient fault response and the normal response seriously impairs the fault recognition ability of fault classifiers. Therefore, this paper builds an aliasing measuring model (AMM) to generate the multi-frequency test stimuli set for incipient faults. The principle part of the AMM is the aliasing measuring algorithm (AMA), which uses the response aliasing as the pivotal index to evaluate the test frequency. The test frequencies with smaller response aliasing will be selected. The other part of the AMM contains the genetic algorithm and the greedy algorithm, which can advance the AMA to quickly generate the multi-frequency test stimuli set for the incipient fault and can ensure that the obtained test set covers the entire circuit and contains fewer test frequencies. The simulation experimental results validate two conclusions: the test frequencies obtained by the AMM remarkably increase the incipient fault detection accuracy and reduce the time for test frequency generation and the multi-frequency test stimuli set contains fewer elements. The hardware experimental results demonstrate that the proposed method is practicable and effective.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have