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

Area of Science:

• Robotics and Automation
• Signal Processing
• Electrical Engineering

Background:

• Power transformers present significant localization challenges for robots due to their size and metal enclosures.
• Accurate robot positioning is crucial for effective inspection and maintenance of critical infrastructure.

Purpose of the Study:

• To develop a robust 3D spatial localization method for transformer patrol robots.
• To address challenges posed by noise interference in ultrasonic signals within transformer environments.

Main Methods:

• Utilized a nine-element ultrasonic array for spatial localization.
• Employed wavelet decomposition for adaptive signal denoising.
• Applied an improved semi-soft thresholding function for signal reconstruction.
• Implemented an improved weighted filter beamforming (WFB) algorithm for position determination.

Main Results:

• The improved semi-soft threshold function significantly enhanced signal-to-noise ratio (SNR) and normalized correlation coefficient (NCC), while reducing root mean square error (RMSE) compared to traditional methods.
• The improved WFB algorithm achieved a maximum relative localization error of 3.47% and an absolute error within 2.6 cm in a transformer test box.
• Denoising performance showed substantial improvements, with SNR increasing by up to 60.55% and RMSE decreasing by up to 58.77%.

Conclusions:

• The proposed WD-WFB method offers superior denoising and localization accuracy for transformer patrol robots.
• The technique effectively overcomes noise interference, enabling reliable robot positioning in challenging industrial settings.
• The achieved accuracy meets practical engineering application requirements for transformer inspection.