Substation risk state detection method considering human-machine collaboration

In order to solve the problem of low intelligence of grid operation and maintenance risk identification system, this paper proposes a substation risk state detection method. Firstly, to address the problem of insufficient feature extraction capability, S-ELAN network is designed to obtain more substation risk state features. Secondly, adaptive spatial fusion is used to reduce the inconsistency of feature information. Then, the gradient information is utilized to adaptively adjust the weight information of different samples. The test results show that the overall test accuracy of the method in this paper reaches more than $80 \%$, which is significantly better than the existing detection methods, and verifies the effectiveness of the method.