Search for: All records

Abstract Ensuring the structural integrity of the overhead power line conductor is crucial for maintaining the safety and reliability of the electrical transmission system. Exposure to environmental hazards like moisture, dust, and Wind-Induced Vibrations (WIV) can lead to defects and corrosion in power line conductors, which are primary contributors to fatigue and shortened lifespan. Thus, this paper presents a vision-based health inspection of power line conductors for a maintenance robot. The method involves image filtering techniques such as Sobel, Scharr, and Gray-scale Variance Normalization (GVN). After filtering the image, row and column analysis is conducted to identify relevant patterns that distinguish healthy and unhealthy conductors, utilizing histograms for data representation. From the histogram data analysis, 10 features were chosen from observation. Subsequently, the collected image data is classified into either healthy or unhealthy categories through supervised machine learning models, including Random Forest (RF), Multi-Layer Perception (MLP), and Gradient Boosting (GB). The best combination of features is extracted to optimize each machine-learning models accordingly. Experimental results validated the effectiveness of our method, which has been specifically fitted for the Mobile Damping Robot (MDR), presenting its potential for enhancing power line maintenance.