In the distribution network accessed by photovoltaic

the existing leakage protection devices cannot distinguish the leakage current of photovoltaic equipment and the leakage current of biological electric shock fault

resulting in potential safety hazards in the system. In order to solve these problems

a leakage identification method for photovoltaic connected to distribution network based on optimized KELM is proposed in this paper. First

a 9-dimensional original fault feature set is constructed

and a 4-dimensional high-correlation feature subset is selected from the 9-dimensional feature set by NCA. Then

the optimized feature subset is used as input to establish an electric shock recognition model based on KELM

and the parameters in the KELM model are optimized by SSA. Finally

the proposed SSA-KELM method is applied to electric shock recognition

and is compared with the standard Kernel Extreme Learning Machine (KELM)

Support Vector Machine (SVM)

BP Neural Network (BPNN) The comparison results show that SSA-KELM has the highest recognition rate for the type of electric shock of photovoltaic access distribution network compared with other methods

with an average recognition accuracy of 97.98%

which provides a certain theoretical reference for the effective recognition of biological electric shock and photovoltaic leakage.