针对复杂城市环境下利用三维高斯波束跟踪进行电波传播预测时,由于波束数目多且传播场景复杂,碰撞检测耗时长,导致预测效率低的问题,提出了一种基于网格分区的有效区域选择加速算法,该方法先对传播场景进行三维网格分区,并根据有效射线集中区域选择有效场景区域,以减少无效射线的跟踪,同时通过准确定位射线所在分区,减少相交测试的数目。仿真结果表明,加速后模型预测误差小于8dB,满足工程实际精度需求,而预测效率可提高约85%,适用于大规模复杂场景下的无线网络的规划、设计和优化。 Aiming to solve the problem of low forecasting efficiency due to the large number of beams, complicated propagation scene and long collision detection time, aiming at the problem of radio propagation prediction using 3D Gaussian beam tracking in complicated urban environment, an effective region selection acceleration based on grid partitioning is proposed In this method, the three-dimensional grid of propagation scenes is firstly partitioned and the effective scene area is selected according to the effective ray concentration area to reduce the tracking of ineffective rays. At the same time, the number of intersections tests is reduced by accurately locating the partition where rays are located. The simulation results show that the prediction error of the model after acceleration is less than 8dB, which meets the actual project accuracy requirements, while the prediction efficiency can be increased by about 85%. It is suitable for the planning, design and optimization of wireless networks in large-scale and complex scenarios.