Work

Adaptive phase-shift control of LCC resonant converters with transformer-insulation micro-arcing risk awareness

Year: 2026
Type: article
Abstract: High-voltage LCC resonant converters are widely used in applications such as X-ray devices, but they face severe reliability challenges. Conventional phase-shift control often exacerbates internal voltage fluctuations across the high-voltage transformer windings. This can induce internal transformer-insulation micro-arcing and accelerate dielectric failure. To address this issue, this paper proposes an adaptive phase-shift control method with transformer-insulation micro-arcing risk awareness. First, this study analyzes the disturbance caused by the micro-arcing incubation process on the resonant tank. This analysis reveals the physical coupling mechanism between phase-shift angle variations and local electric field distortion. Subsequently, key electrical features are selected to construct a physics-informed neural network (PINN). These features include the complementary winding voltage difference, the third harmonic of the resonant current, and the current change rate. The model embeds circuit law residuals and physical boundaries. This overcomes the generalization limits of data-driven methods in small-sample high-voltage scenarios. Consequently, it achieves real-time, accurate, and highly interpretable quantification of micro-arcing risk. Finally, an adaptive phase-shift control law is designed. It utilizes the risk probability and its rate of change as feedforward compensation signals. This strategy dynamically and smoothly adjusts the phase-shift angle to optimize the electric field distribution. Therefore, it proactively suppresses the occurrence of micro-arcing. Compared to conventional phase-shift control, this method significantly narrows the voltage peak distribution at critical insulation points. It effectively reduces the probability of micro-arcing events whilst maintaining a stable output voltage.
Language: English
Open Access status: gold
Estimated APC paid: $1,800
OpenAlex ID: W7237217240
FWCI:
Cites: 34
Cited by:
Related to: 10