To address the problems of low search efficiency, insufficient obstacle avoidance capability, and excessive path turning points in the traditional A* algorithm, as well as the strong convergence blindness and poor convergence stability of the traditional ACO algorithm in multi-target visiting-order optimization, an improved A*-IACO hierarchical fusion path planning method is proposed. First, two-dimensional mapping, obstacle inflation, and grid-based modeling are performed for the main working surface of the brake valve. Second, a safety potential field cost and a turning-angle penalty are introduced into the traditional A* algorithm. Combined with key anchor point extraction, visibility pruning, and cubic B-spline smoothing, a local path library is constructed. Finally, based on path cost and endpoint direction information, an IACO algorithm integrating inter-segment turning-angle heuristic information, non-uniform pheromone initialization, adaptive pheromone evaporation, and 2-opt optimization is designed. Simulation results show that, in scenarios with 10 and 16 target points, the proposed method reduces the path length by 12.13% and 17.42%, the cumulative intra-segment turning angle by 9.94% and 15.42%, the cumulative inter-segment turning angle by 20.78% and 16.87%, and the comprehensive cost by 13.12% and 16.74%, respectively, compared with the traditional ACO algorithm. The experimental results verify the effectiveness of the proposed method for path planning in brake valve surface cleaning.