In recent years, as an effective positioning method, electromagnetic wave time reversal (EMTR) technology has begun to be applied in the field of partial discharge. Compared with traditional UHF positioning methods, EMTR technology only requires a single sensor, which has huge advantages and good application prospects. However, the currently commonly used maximum field strength and minimum entropy criteria are difficult to accurately obtain the focusing moment and position of time reversal under complex structures. In response to the above problems, this paper proposes a Density-Based Spatial Clustering of Applications with Noise (DBSCAN) time inversion localization method for electromagnetic waves based on the different characteristics of waveforms at signal and non-signal sources, and the location of partial discharges in gas-insulated switchgear (GIS) is verified using CST Studio Suite simulation software. In order to verify the field feasibility of EMTR, we built a laboratory model to conduct partial discharge experiments. The experimental results show that the average positioning error of EMTR method is less than 20 cm, which can realize the accurate positioning of partial discharge sources in GIS. Compared with traditional methods, EMTR method reduces the number of sensors used and improves the anti-interference performance, which has certain advantages.