To address the challenges of complex terrain, diverse threats, and premature convergence in UAV path planning for mountain disaster rescue, a path planning method based on Adaptive Combined Differential Evolution (ACDE) is proposed. A mountainous environment model incorporating no-fly zones is established, and a comprehensive weighted cost function is formulated to transform the path planning problem into a constrained optimization problem. Three targeted improvements are introduced into the conventional differential evolution framework. First, a rapidly-exploring random tree initialization strategy is employed to enhance the spatial coverage of the initial population. Second, a lightweight Actor-Critic reinforcement learning framework is introduced to construct an adaptive parameter mechanism, by which the scaling factor and crossover probability are dynamically adjusted according to the population state. Third, a dynamic hybrid mutation mechanism based on population stratification is developed to balance global exploration and local exploitation. The effectiveness of the proposed algorithm is demonstrated by simulation results, and safe obstacle-avoidance paths with shorter length and lower cost are obtained under complex mountainous conditions.