Abstract: With the improvement of power electronic technology,the characteristics of line fault arc become more complex,and the traditional fault arc identification method has been unable to cope with the new power system.Compared with the traditional fault arc identification scheme,the generation type and new detection technology of fault arc are studied,a digital fault arc recognition scheme based on AI+neural network is proposed,which is based on typical time-domain and frequency-domain characteristics such as harmonic factor,total harmonic distortion rate,current zero rest time,current change rate,current periodicity,etc.In this scheme,the regression algorithm vector machine is used to classify the feature map,and the convolution neural network is implemented through three steps of forward transmission operation,reverse transmission operation and iterative regression operation.Frequent iterative regression is used to obtain the best feature recognition map.This paper takes the fault arc identification of dimmer as an example to verify the fault identification scheme.The results show that this scheme can achieve more efficient,more accurate and more stable fault arc identification.

Key words: Fault arc, Digitization, AI+ neural network, Iterative regression, Feature recognition diagram