基于改进粒子群优化神经网络的高程异常拟合方法  

Elevation Anomaly Fitting Method Based on Improved Particle Swarm Optimization Neural Network

作　　者：张文君 师文杰 ZHANG Wenjun;SHI Wenjie(Qinghai Provincial Institute of Basic Surveying and Mapping,Xining 810000,China;Qinghai Nonferrous First Geological Exploration Institute,Xining 810000,China)

机构地区：[1]青海省基础测绘院,青海西宁810000 [2]青海省有色第一地质勘查院,青海西宁810000

出　　处：《地理空间信息》2025年第3期1-4,共4页Geospatial Information

摘　　要：针对BP神经网络高程异常拟合时存在的局部最优、收敛速度慢等问题,提出了基于改进粒子群优化神经网络的高程异常拟合方法。通过结合惯性权值因子的非线性调整策略与学习因子的线性调整策略,能有效提高粒子群算法的收敛速度,避免出现局部最优解情况。利用改进粒子群算法优化BP神经网络初始参数,以提高高程异常拟合精度。通过两组典型实验区验证了该方法的精度与有效性。In response to the problems of local optima and slow convergence speed of BP neural network in elevation anomaly fitting,we proposed an elevation anomaly fitting method based on improved particle swarm optimization neural network.By combining the nonlinear adjustment strategy of inertia weight factor and the linear adjustment strategy of learning factor,this method can effectively improve the convergence speed of particle swarm optimization algorithm and avoid local optima.We used the improved particle swarm optimization algorithm to optimize the initial parameters of BP neural network to improve the accuracy of elevation anomaly fitting,and verified the accuracy and effectiveness of this method by two sets of typical experimental areas.

关 键 词：改进粒子群 自适应惯性权值 线性学习因子 BP神经网络 高程异常拟合 

分 类 号：P208[天文地球—地图制图学与地理信息工程]