To address the issues in the dung beetle optimization algorithm, such as falling into local optima and insufficient global search capability, which lead to suboptimal performance in 3D UAV path planning, a multi-strategy improved dung beetle optimization algorithm was designed. A 3D spatial model was constructed, and a comprehensive evaluation function was developed by considering factors such as path length, threat, altitude, and smoothness. First, a hybrid chaotic sequence was employed to enhance the initial population diversity. Then, during the dung beetle rolling stage, a “differential mutation” operator was introduced to improve the algorithm′s local search ability. This was combined with an improved sine algorithm to update individuals via a probability switching mechanism, further enhancing the global search capability. Finally, an improved spiral search strategy was incorporated during the breeding stage to strengthen the algorithm′s ability to escape local optima. Through optimization of six benchmark functions and analysis of particle motion trajectories in the search space, the results demonstrated that the improved algorithm performed better in terms of convergence speed, accuracy, and robustness. When applied to 3D UAV path planning, the optimal, average, and worst values of path length improved by 0.41%, 5.67%, and 18.03%, respectively, further validating the effectiveness of the improvement strategies and the superiority of this algorithm in practical engineering applications.