摘要
为了构建围堰监测预警的人工智能模型,保障围堰安全高效运行,提出一种用于处理受约束非线性单调算子方程的混合三项共轭梯度算法。搜索方向由Hestenes-Stiefel(HS)和无记忆Broyden-Fletcher-Goldfarb-Shanno(BFGS)计算的方向组合而成。在没有任何附加条件的情况下,搜索方向拥有充分下降性和信赖域性。在温和的假设条件下,证明了该方法的全局收敛性。该算法可解决机器学习问题,实验结果也验证了所提方法的优越性。
In order for an artificial intelligence model for monitoring and early warning of cofferdam to be constructed,and the safe and efficient operation of the cofferdam to be ensured,a hybrid triple conjugate gradient algorithm was proposed in this paper for dealing with constrained nonlinear monotone operator equations.The search directions are combined from those computed by Hestenes-Stiefel(HS)and memoryless Broyden-Fletcher-Goldfarb-Shanno(BFGS).In the absence of any additional conditions,the search direction possesses sufficient descent and trust domain properties.Global convergence of the method is proved under mild assumptions.The algorithm solves the machine learning problem and the experimental results validate the superiority of the proposed method.
作者
陈登峰
杨恩祈
罗维宏
欧亿鹏
韦作仪
黄旻洁
袁功林
CHEN Dengfeng;YANG Enqi;LUO Weihong;OU Yipeng;WEI Zuoyi;HUANG Minjie;YUAN Gonglin(College of Traffic&Transportation,Chongqing Jiaotong University,Chongqing 400074,China;Baise Hub Navigation Investment Co.,Ltd.,Baise 53300o,China;Guangxi Bagui Engineering Supervision and Consulting Co.,Ltd.,Nanning 530021,China;School of Pre-undergraduate Education,Baise University,Baise 533000,China;School of Mathematics,Guangxi University,Nanning 530004,China;Center for Applied Mathematics of Guangxi(Guangxi University),Nanning 530004,China)
出处
《广西大学学报(自然科学版)》
北大核心
2026年第1期203-214,共12页
Journal of Guangxi University(Natural Science Edition)
基金
国家自然科学基金项目(11661009)
广西重点研发计划项目(AB24010011,AB25069188)
广西科技基地和人才专项项目(桂科AD22080047)
广西八桂学者团队项目(GXR-6BG2424008)。
关键词
全局收敛
水文预测
机器学习
global convergence
hydrological forecasting
machine learning
