THREE-DIMENSIONAL PARTIAL DISCHARGE LOCALIZATION IN POWER TRANSFORMERS USING HYBRID UHF AND ACOUSTIC SENSORS: A DYNAMIC SIMULATION STUDY

Abstract

Partial discharges (PDs) in power transformers serve as early warning signs of insulation
deterioration and, if left undetected, can lead to severe failures. This study introduces a new simulation-based method for locating PDs using a combination of Ultra-High Frequency (UHF) and Acoustic Emission (AE) sensors. A three-dimensional transformer model was created, with sensors
strategically positioned to capture both electromagnetic and mechanical PD signals. The PD signals
were modeled to account for distance-dependent attenuation and Gaussian noise. PD localization
was performed using a least-squares triangulation approach based on the sensors’ time-of-arrival
measurements. The actual PD source was located at (0.6, 0.4, 0.8) m, while the estimated location
was (0.602, 0.398, 0.802) m, resulting in a very small spatial localization error of 0.003 m.
Triangulation residuals were under 0.01 m for all sensors, confirming the numerical stability of the
estimation. Analysis of signal amplitudes showed a strong inverse relationship with sensor distance,
validating the attenuation model. Also, a 3D heatmap was generated to visualize PD probability,
effectively revealing the PD source and other potential high-risk regions. The main innovation of this
work is the combination of dual-sensor fusion, optimization, and visualization (hybrid triangulation
and 3D probability heatmap), which together improve localization accuracy compared to traditional single-sensor methods. This methodology provides highly accurate, and effective PD localization, offering a foundational step toward future experimental validation for predictive maintenance of transformers.