Research on Ultrasonic Detection Technology for Transformer Windings Based on BKA-ICCEMDAN Optimization Algorithm

BKA-ICCEMDAN joint algorithm is proposed to address the issue of ultrasonic signal noise interference caused by the significant rise in transformer oil temperature under high load conditions. This temperature increase affects signal quality, introducing noise disturbances.The proposed method first applies the Improved Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (ICCEMDAN) to decompose the ultrasonic signal into Intrinsic Mode Functions (IMFs). Then, the Black Kite Algorithm (BKA) is utilized to optimize the IMF selection by adjusting its parameters, such as population size and iteration count. After selecting the optimized IMF components, the denoised signal is reconstructed.Finally, comparative experiments validate that this optimization algorithm can effectively mitigate the adverse effects of oil temperature on ultrasonic signals, significantly enhancing the accuracy of ultrasonic winding detection.