AUTONOMOUS TRANSMISSION LINE ROUTE OPTIMIZATION WITH CNN-R2 -NET AND OPTIMIZER

Abstract

The proper positioning of transmission lines is essential for maximising energy distribution and reducing environmental impact. This study aims to enhance the identification of prospective locations for transmission line installation by utilising sophisticated classification methods, particularly integrating Convolutional Neural Networks (CNN) and R2-UNet architectures with hyperspectral and LiDAR data. Our methodology incorporates many remote sensing technologies to classify land cover types precisely, producing a binarised image that emphasises appropriate installation locations. The framework utilises a stringent CNN and R2-UNet-based classification method, succeeded by an optimal path selection approach for transmission line routing, guaranteeing environmental compliance. Our results indicate elevated categorisation accuracy across many land cover categories, attaining accuracy rates of 98%, 95%, and 97% for "Trees," "Road," and "Water," respectively. The analysis of misclassification rates provided suggestions for prospective model modifications. The visualisations illustrate optimal transmission line routes, confirming the model's dependability. This study offers a complete framework for transmission line planning that employs CNN and R2-UNet models with remote sensing data to deliver actionable insights, hence aiding sustainable infrastructure development. Subsequent research should concentrate on optimising classification algorithms and integrating real-time data to improve adaptability and accuracy in transmission line routing.