Partial discharge (PD) localization is critical to ensure the safe operation of power equipment. In this paper, an optimization scheme is developed for PD positioning of external insulation equipment. The optimization scheme is based on an eight‐element cross sensor array to receive the ultrasonic signal generated by the PD, and uses the highly accurate 2‐D multiple signal classification (2‐D MUSIC) algorithm to locate the signal. Simulated discharge signal is adopted to test the optimization scheme, and the results show that the positioning error is less than 0.61° when the signal‐to‐noise ratio (SNR) is above −5 dB. In addition, three discharge models are tested at various locations. The average azimuth errors are 0.9°, 1.9°, and 1.55° for the needle‐plate, cone‐plate, and ball‐plate discharge models, respectively, and the average pitch angle errors are 0.75°, 2.1°, and 1.4°, respectively. These show that the needle‐plate model has the best positioning effect. Simulations and experiments are also carried out on the double discharge source scenario, and the error is within 2.2°. A PD location visualization equipment based on the proposed optimization scheme is developed and applied on‐site. The equipment is capable of satisfying the requirements of operation and maintenance.