A* Algorithm-Based 3D Path Planning Method for UAVs in Substations

Abstract

In the context of accelerating integration between artificial intelligence and industrial automation, intelligent operation and maintenance has become a key enabler for smart grid infrastructure. To address the issues of cumbersome traditional modeling processes, manual specification of task point cruising sequences, and insufficient adaptability of path planning in substation UAV inspections, this paper proposes a 3D path planning method for substation UAVs based on point cloud rasterization and an improved A* algorithm. Taking substation point cloud data as the core input, this method eliminates the need for traditional manual modeling. Through point cloud preprocessing and rasterization conversion, it automatically extracts feature information such as equipment contours and spatial positions, rapidly constructing a 3D grid environment model with safety distance constraints to achieve automated and precise scene representation. Based on the improved A* algorithm framework, an improved Traveling Salesman Problem (TSP) model is introduced to build an autonomous task point sequencing module. Combined with the spatial distance between task points and inspection priorities, a multi-objective optimization function is constructed to solve for the globally optimal cruising sequence, thereby minimizing overall inspection mileage. Simultaneously, the cost function is optimized by incorporating grid threat weights and UAV kinematic constraints to balance the safety, optimality, and search efficiency of individual path segments. Finally, cubic spline interpolation is applied to smooth and optimize the path, reducing operational energy consumption. Using point cloud data from a 220 kV substation as experimental samples, the proposed method was validated against the traditional DijkstrA* algorithm and manual planning methods. Results show that the rasterization modeling time is reduced by 72.5% compared to traditional modeling; TSP-based task point sequencing reduces overall inspection mileage by 10.2%, with a sequencing rationality of 98.6% and an obstacle avoidance success rate of 99.2%. The planning time is controlled within 0.4 seconds. This method can efficiently meet the requirements of multi-task substation inspections, providing technical support for the intelligent upgrade of power grid inspections.