基于改进鲸鱼算法优化神经网络的GPS高程拟合方法  

GPS Height Fitting Method Based on Improved Whale Algorithm Optimized Neural Network

作　　者：钱建国[1] 徐志文 赵玉国 郭洁 王志强 赵金来 QIAN Jianguo;XU Zhiwen;ZHAO Yuguo;GUO Jie;WANG Zhiqiang;ZHAO Jinlai(School of Mapping and Geographical Science,Liaoning Technical University,88 Yulong Road,Fuxin 123000,China;Guoneng Baorixile Energy Co Ltd,751 Shenbao Road,Hulun Buir 021599,China;Jalainur Coal Industry Co Ltd,17 Yulin Street,Hulun Buir 021410,China)

机构地区：[1]辽宁工程技术大学测绘与地理科学学院,辽宁阜新123000 [2]国能宝日希勒能源有限公司,内蒙古自治区呼伦贝尔021599 [3]扎赉诺尔煤业有限责任公司,内蒙古自治区呼伦贝尔市021410

出　　处：《大地测量与地球动力学》2024年第2期122-127,共6页Journal of Geodesy and Geodynamics

摘　　要：采取混沌映射和自适应惯性权重结合的策略对标准鲸鱼算法进行改进,从而提高算法的全局寻优能力和收敛速度,并针对BP神经网络的劣势,利用改进鲸鱼算法对BP神经网络进行优化处理。在此基础上建立改进鲸鱼算法优化BP神经网络的GPS高程异常拟合预测模型,并通过两组不同地形特征工程中的GPS数据对模型进行验证。结果表明,利用改进鲸鱼算法优化的BP模型进行GPS高程拟合时可取得更高的精度和稳定性。By combining chaotic mapping and adaptive inertial weight,we improve the standard whale algorithm so as to improve the global optimization ability and convergence speed of the algorithm.Aiming at the disadvantage of BP neural network,we use the improved whale algorithm to optimize BP neural network.On this basis,we establish the improved whale algorithm to optimize the BP neural network GPS elevation anomaly fitting prediction model,and the model is verified by two groups of GPS data in different terrain feature engineering.The results show that the BP model optimized by the improved whale algorithm can achieve higher accuracy and stability in GPS height fitting.

关 键 词：改进鲸鱼算法 混沌映射 自适应惯性权重 高程拟合 BP神经网络 

分 类 号：P228[天文地球—大地测量学与测量工程]