For the characteristics of the traditional statistical parameters, due to the number of discharge and discharge phase distribution and other factors, there will be invalid information, and reduce the partial discharge recognition rate. In order to reduce the influence of these factors, this paper introduces the texture feature of gray-level co-occurrence matrix to make pattern recognition of partial discharge. First, the (j-q-n) three-dimensional spectrum of partial discharge phase-discharge capacity-discharge numbers is constructed through the experiment, and the discharge distribution matrix is acquired, and then the corresponding gray level co-occurrence matrix is calculated, and the corresponding texture features are extracted, finally, cluster analysis is used to classify different discharge defects effectively. Experimental results show that texture feature based on discharge distribution of gray-level co-occurrence matrix can effectively identify the different types of partial discharge, which provides a new idea for partial discharge pattern recognition. This work is supported by Fundamental Research Funds for the Central Universities (No. 2012QNA10) and China Postdoctoral Science Foundation (No. 2013M541755).