The evaluation of the severity of partial discharge is a multi-objective decision problem. Considering thenon-linear relationship among the different characteristic parameters

it is the core of this paper to extract the partialdischarge characteristic parameters and evaluate the severity. The experimental results show that the developmentprocess of partial discharge of solid insulation switchgear under typical gap defects has been studied experimentally

and the φ-u scatter plot and φ-n spectrum of different PD stages have been constructed. Extracting nine feature pa-rameters to characterize the severity of PD

and the physical meaning of each characteristic parameter and the partialdevelopment stage are analyzed. The results show that the partial characteristic parameter value interval is put downin different degree

and the neural network algorithm is used to construct of judgment system for severity of partial dis-charge in solid insulated switchgear

which avoids only considering the single factor leads to the irrationality of the judgment result. Finally

the correctness and validity of the method are verified by an example test.