Research on Hexagonal Grid-Based Path Planning Method for Transmission Lines

Driven by the “Dual Carbon” strategy, the planning of transmission lines in complex terrains urgently requires overcoming the technical limitations of traditional square grid modeling methods, such as insufficient geometric compatibility and high path redundancy. Conventional approaches suffer from frequent path deviations and redundant tower placements due to mismatches between orthogonal adjacency relationships and natural terrain continuity. Additionally, distorted diagonal distance calculations lead to cost estimation errors, while Qlearning algorithms face state space explosion issues under 8neighborhood movement patterns. To address these challenges, this study proposes an intelligent transmission line planning method that integrates hexagonal grids with reinforcement learning. By constructing a dynamic mapping model between hexagonal grids and geographic space under an axial coordinate system, geometric compatibility between natural terrain morphology and honeycomb-like topology is achieved, eliminating diagonal distance distortion. A multi-level dynamic reward function driven by land-type costs is innovatively designed and integrated with an adaptive <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\varepsilon$</tex>-greedy exploration strategy to balance algorithmic exploration and exploitation efficiency. Validated in Guangzhou's Conghua District, the method integrates multi-source data including a digital elevation model, land-use grids, and water system protection zones to establish a hexagonal grid cost quantification system. Experimental results in the case study of Guangzhou's Conghua District demonstrate the hexagonal model's superiority: total path cost decreases from 11.665 to 7.598 (Q-learning results), route length shortens from 12.295 km to 11.108 km (Q-learning results), and bending coefficient improves from 1.231 to <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{1. 1 0 3}$</tex> (Q-learning results). The hexagonal grid Q-learning path achieves continuous and smooth trajectories through <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{1 2 0}$</tex>-degree diagonal movements, effectively avoiding high-cost areas such as steep slopes and water bodies. This path optimization significantly reduces redundant turns and the need for corner tower placements, thereby substantially lowering construction difficulty and cost. Compared to square grid paths which frequently exhibit multiple turns and redundant segments, particularly in obstacle-dense areas, the hexagonal model generates path geometries with superior smoothness and efficiency. This study confirms that hexagonal grid modeling overcomes terrain-matching defects of traditional methods, providing high-precision, low-cost planning solutions for transmission projects in mountainous and riverine environments, with substantial engineering application value.