Abstract: In response to the challenges of ultra-wideband indoor vehicle localization in complex underground parking environments, a multi-anchor deployment system based on the Decawave DWM1000 module was developed, together with a unified solution spanning static geometric layout and dynamic optimization. A three-dimensional localization model was first established, in which regular triangular, quadrilateral, pentagonal, and hexagonal anchor deployment strategies were investigated. Through geometric dilution of precision analysis and experimental validation, the regular hexagonal configuration was shown to reduce geometric dilution of precision by approximately 40.5% compared with the conventional equilateral triangular layout under identical conditions. To address non-line-of-sight propagation, severe multipath effects, and signal attenuation, an improved particle swarm optimization algorithm with linearly decreasing inertia weight was further proposed to dynamically optimize anchor radius and height, reducing the average geometric dilution of precision from 1.25 to 0.829, corresponding to an additional improvement of 33.9% over traditional layouts. Finally, in a realistic underground parking scenario covering 20 m × 20 m and incorporating obstacles such as walls and pillars, an improved extended Kalman filter was compared with the least squares method, demonstrating superior performance in terms of dynamic responsiveness and robustness against positioning errors. （请作者规范缩写使用方法, 除了类似于GDP, CPU之类的第一次出现需用全称, 太多缩写了, PSO中文里是粒子群算法, 为什么其它的还是缩写? ）