Abstract: To improve the accuracy of partial discharge type diagnosis in of gas insulated switchgear(GIS), a method of partial discharge diagnosis in GIS based on EWT-FE and IHPO-SVM algorithm is proposed. To deeply explore the internal features of the partial discharge signal, the empirical wavelet transform(EWT) combined with the fuzzy entropy(FE) algorithm is used to decompose the signal and extract the effective feature quantity. To improve the adaptive capability and classification recognition accuracy of support vector machine(SVM) algorithm, the cosine decay calculation method and the exponential descent function are used to improve the hunter-prey algorithm, resulting in the improved hunter-prey optimizer(IHPO) algorithm, and the parameters of SVM are optimally selected by this method. Finally, an experimental model of GIS partial discharge is constructed, and a partial discharge identification model based on EWT-FE signal analysis combined with IHPO-SVM is established to verify the effectiveness of the proposed algorithm. The experimental results show that the diagnostic accuracy of the proposed algorithm for partial discharge type of GIS is more than 95%, which is better than the traditional diagnostic algorithm.