Research on fault identification method of primary and secondary fusion equipment based on data drive

In this paper, the data-driven and deep learning technology is integrated into the fault identification of distribution network, and the fault identification under multiple scenarios of distribution network is realized. Mainly for single-phase grounding, small resistance grounding, arc suppression coil and other faults. This paper constructed a structural model, analyzed the features of the model, selected the eigenvalues, and conducted deep learning training. The training results show that the combination of data-driven and deep learning can effectively identify typical faults. In complex phase states, the small resistance system can ensure a $\mathbf{9 9 \%}$ accuracy rate, and can complete correct line selection in various situations such as fault location of $\mathbf{2 0 \%}$, $\mathbf{5 0 \%}, \mathbf{6 0 \%}$, etc. At the same time, the fault training of a single case is less than 2 seconds, which can achieve rapid self growth.