Ultrasonic Localization of Transformer Patrol Robot Based on Wavelet Transform and Narrowband Beamforming

The large size and metal-enclosed casings of oil-immersed power transformers present significant challenges for patrol robots attempting to accurately locate their position within the transformer. Therefore, this paper proposes a three-dimensional spatial localization method for transformer patrol robots using a nine-element ultrasonic array. This method is based on wavelet decomposition and weighted filter beamforming (WD-WFB) algorithms. To address the issue of strong noise interference in the field, the ultrasonic localization signals are adaptively decomposed into wavelet coefficients at different frequencies and scales. An improved semi-soft thresholding function is applied to the decomposed wavelet coefficients to reduce noise and reconstruct the localization signals, resulting in localization signals with low distortion and a high signal-to-noise ratio(SNR). To overcome the limitations of traditional beamforming algorithms regarding interference resistance and signal resolution, this paper presents an improved WFB algorithm. By obtaining the energy distribution of the scanning area and determining the position of the maximum energy point, the spatial position of the transformer patrol robot can be determined. The test results show that the proposed improved semi-soft threshold function demonstrates superior denoising performance compared to traditional threshold functions. When compared to the soft threshold function, it achieves improvements of 15.32% in SNR and 15.57% in normalized correlation coefficient (NCC), along with a 48.91% reduction in root mean square error (RMSE). Compared with the hard threshold function, the improvement is even more significant: the SNR is improved by 60.55%, the NCC is improved by 24.90%, and the RMSE is reduced by 58.77%. The denoising effect was significantly improved compared to the traditional threshold function. In a 1200 mm × 1000 mm × 1000 mm transformer test box, the improved WFB algorithm in this paper was used to perform multiple localizations of the transformer patrol robot at different positions after denoising the field signals using the semi-soft threshold function. The maximum relative localization error was 3.47%, and the absolute error was within 2.6 cm, meeting engineering application requirements.