针对复杂山区环境下无人机覆盖搜索任务中路径冗余与地形阻碍问题，该文提出一种基于混合策略的覆盖搜索方法。该方法融合传统固定式路径规划方法与基于自适应惯性权重改进的粒子群优化算法，当无人机遇到死点时，利用改进粒子群优化算法回溯路径。对比分析了队列回溯与栈回溯两种搜索策略及其对应的优化后的混合策略在不同地形条件下的适用性，结果表明：混合策略在复杂山区任务中展现出显著优势，其搜索路径长度较基于栈回溯的固定式搜索方式减少31.8%，且路径覆盖率稳定增长至100%；改进粒子群优化算法通过自适应权重调整，较传统粒子群优化算法和人工蜂群优化算法收敛速度更快。在不同地形特征环境中进行模拟的结果验证了混合策略在覆盖搜索路径优化与地形适应性上的有效性。该研究为无人机在应急救援等场景中的应用提供了可靠的路径规划方法。

Objective: In complex mountainous environments, unmanned aerial vehicle (UAV) coverage search tasks often encounter two core challenges: path redundancy and terrain obstructions. Although fixed-pattern search methods offer convenience and high efficiency in simple scenarios, they struggle to effectively avoid dead points and obstructures in complex terrains due to their rigid pre-planned trajectories. As a result, path repetition and reduced search efficiency become particularly prominent. To address the challenges of path redundancy and terrain obstructions in UAV coverage search tasks within complex mountainous environments, this study proposes a hybrid strategy that integrates traditional fixed-pattern search with an improved particle swarm optimization (PSO) algorithm. This strategy optimizes return path planning, minimizes path redundancy, and enhances adaptability in complex terrains. Methods: This research adopts a grid-based modeling approach to discretize complex terrains, constructing a simulation environment using real-world digital elevation model data from a specific area of Luding County, Sichuan Province, China. During data preprocessing, high-precision terrain data are converted into 3D surfaces via bi-linear interpolation, and threshold segmentation algorithms create binary representations of obstacle zones and passable areas. To address the challenge of dead points in fixed-pattern searches, this study introduces a hybrid backtracking mechanism that integrates queue-based and stack-based backtracking. When encountering dead points, an improved PSO algorithm with adaptive inertia weights is introduced to plan safe and efficient cross-regional paths. In the early iterations, the algorithm assigns larger inertia weights to enhance global exploration. Subsequently, these weights are reduced to refine local searches. In addition, path safety is ensured through various constraint functions, including mathematical models to avoid terrain blockages, maintain safe distances from obstacles, and ensure path continuity. Results: The experimental results indicate that the proposed hybrid strategy exhibits significant advantages in complex mountainous enviornments. This strategy, which combines queue-based backtracking and stack-based backtracking, reduces total path length by 0.66% and 21.1%, respectively. Path coverage gradually increases from initial levels to full coverage (100%), demonstrating robust performance across various terrain conditions. Notably, in highly complex environments, the improved PSO algorithm exhibits faster convergence speed and higher path-planning accuracy than the traditional PSO and the artificial bee colony algorithms. Comparative analysis reveals that stack-based backtracking performs better in complex terrains, whereas queue-based backtracking is more suitable for regions with greater local connectivity. Furthermore, this research is the first to demonstrate that the hybrid strategy can automatically adjust the number of backtrackings without prior information, ensuring flight safety while achieving optimal coverage. The overall optimization reaches 21.1%. Conclusions: This paper presents a hybrid-strategy-based UAV coverage search method for complex mountainous areas and validates its applicability and superiority across various terrain features through experiments. The findings reveal that the hybrid strategy maintains strong terrain adaptability while balancing efficiency and feasibility. In addition, the selection of backtracking methods directly influences the frequency of heuristic algorithm invocations and ultimately affects the quality of path planning. The successful application of the improved PSO algorithm demonstrates its potential for multi-objective optimization in complex environments, laying a foundation for further exploration of more intelligent and flexible UAV path planning technologies. This study holds significant implications for UAV applications in critical scenarios such as emergency rescue and disaster reconnaissance and provides new perspectives for autonomous UAV navigation.