Surface discharge of power equipment is an important indicator of latent faults and early defects. The microphone array-based acoustic detection method can acquire the discharging acoustic signals emitted by the equipment in a non-contact manner

and then the discharge position can be localized by array signal processing algorithms. However

conventional algorithms suffer from low resolution and poor robustness. To overcome these shortages

an acoustic localization method for surface discharge of power equipment based on compressive beamforming (CSBF) is proposed in this paper. First

the localization problem is formulated as a CSBF model constrained by a variant function of the generalized Gaussian distribution. Then

the model is solved efficiently based on the accelerated proximal gradient (APG) framework

and the estimated results of source counts and positions are obtained. To demonstrate the effectiveness of the proposed method

a series of simulations with different source distances

environmental observation conditions

and computational parameters are conducted; furthermore

the copper wire and defective insulator discharging localization experiments are carried out in the laboratory environment. Results show that CSBF-APG can achieve a position estimated error of less than 0.1 meter and the running time of less than 2 seconds

which demonstrates its favorable resolution and robustness

and also exhibits adaptability to practical engineering environments with reflections and reverberation. Finally

in a converter transformer high-voltage partial discharge test

the proposed method is adopted to successfully locate the discharge source of an experimental grounding wire

improving the test efficiency and also verifying its practicality. This paper can provide valuable guidance and reference for acoustic detection and localization of surface discharges of power equipment.