Please wait a minute...
浙江大学学报(理学版)  2022, Vol. 49 Issue (4): 435-442    DOI: 10.3785/j.issn.1008-9497.2022.04.007
数学与计算机科学     
求解大规模矛盾方程组的最小二乘支持向量机算法
郑素佩(),闫佳(),宋学力,陈荧
长安大学 理学院,陕西 西安 710064
A least square support vector machine algorithm for solving huge contradictory equations
Supei ZHENG(),Jia YAN(),Xueli SONG,Ying CHEN
School of Science,Chang'an University,Xi'an 710064,China
 全文: PDF(2943 KB)   HTML( 7 )
摘要:

房价预测、共享单车出租数量预测、空气污染情况预测等常涉及矛盾方程组求解,对其数值求解方法研究具有重要的理论意义与应用价值。当矛盾方程组规模过大时,用传统的最小二乘法求解,不仅计算量大,而且由于误差积累使最终结果的准确性不高。鉴于此,采用机器学习中的最小二乘支持向量机(least squares support vector machine,LS-SVM)算法求解大规模矛盾方程组,并分别针对线性、非线性、单变量、多变量矛盾方程组进行了数值求解。数值结果表明,数据类型和数据量的变化对结果的影响不大,因此只要选取适当的参数就可建立合适的模型,得到高精度的预测结果。

关键词: 大规模矛盾方程组机器学习LS-SVM最小二乘法    
Abstract:

Contradictory equations often appear in the predictions of housing price, the number of shared bike rentals, air pollution and other problems. It is of important theoretical significance and practical application value to conduct research on the related numerical method. When the number of contradictory equations is huge, it is too difficult to use the traditional least square method to solve the problem due to the accumulation of errors. In view of this, this paper adopts the least square support vector machine (LS-SVM) algorithm, which is suitable for machine learning of big data process, to solve the huge contradictory equations, and applies the algorithm to problems with practical application background. Experimental results show that the linear, nonlinear, univariate and multivariable contradictory equations can be solved numerically, the change of data type and data volume does not affect the results much, and the appropriate model can be built to obtain high accuracy results as long as the appropriate parameters are selected.

Key words: huge contradictory equations    machine learning    LS-SVM    least square method
收稿日期: 2021-07-29 出版日期: 2022-07-13
CLC:  O 241.2  
基金资助: 国家自然科学基金资助项目(11971075);陕西省自然科学基金青年项目(2020JQ-338)
通讯作者: 闫佳     E-mail: zsp2008@chd.edu.cn;yj_yan_jia@163.com
作者简介: 郑素佩(1978—),ORCID: https://orcid.org/0000-0003-2502-6998,女,博士,副教授,主要从事微分方程数值解法研究,E-mail:zsp2008@chd.edu.cn.
服务  
把本文推荐给朋友
加入引用管理器
E-mail Alert
RSS
作者相关文章  
郑素佩
闫佳
宋学力
陈荧

引用本文:

郑素佩,闫佳,宋学力,陈荧. 求解大规模矛盾方程组的最小二乘支持向量机算法[J]. 浙江大学学报(理学版), 2022, 49(4): 435-442.

Supei ZHENG,Jia YAN,Xueli SONG,Ying CHEN. A least square support vector machine algorithm for solving huge contradictory equations. Journal of Zhejiang University (Science Edition), 2022, 49(4): 435-442.

链接本文:

https://www.zjujournals.com/sci/CN/10.3785/j.issn.1008-9497.2022.04.007        https://www.zjujournals.com/sci/CN/Y2022/V49/I4/435

图1  算例1数值结果实验1的核函数参数为γ=17,σ=2;实验2的核函数参数为γ=40,σ=2。
图2  算例2数值结果实验1和实验2的核函数参数均为γ =20,σ=2。
图3  算例3数值结果实验1的核函数参数为γ=10,σ=1.47×10-7;实验2的核函数参数为γ=10,σ=1×10-7
图4  算例4数值结果核函数参数为γ=150, σ=0.1。
图5  算例5数值结果核函数参数为γ =1 800,σ=30。
图6  算例6数值结果核函数参数为γ =1 500,σ=13。
图7  算例7数值结果核函数参数为γ =200, σ=25。
图8  算例8数值结果核函数参数为γ =190,σ=20。
1 JIANG B N. The Least-Squares Finite Element Method[M]. Berlin/Heidelberg:Springer-Verlag Press, 1998. doi:10.1007/978-3-662-03740-9_8
doi: 10.1007/978-3-662-03740-9_8
2 李玉良, 何宇轩, 楚盖, 等. 基于最小二乘的工程结构边界载荷反求误差消减方法[J]. 计算力学学报,2020, 37(3): 284-292. DOI:10.7511/jslx20190629001
LI Y L, HE Y X, CHU G, et al. An error reduction method based on the least squares method for inverse computation of boundary loads of engineering structures[J]. Chinese Journal of Computational Mechanics, 2020, 37(3): 284-292. DOI:10.7511/jslx20190629001
doi: 10.7511/jslx20190629001
3 TELLINGHUISEN J. Least squares methods for treating problems with uncertainty in x and y [J]. Analytical Chemistry, 2020, 92(16): 10863-10871. DOI:10.1021/acs.analchem.0c02178
doi: 10.1021/acs.analchem.0c02178
4 张燕平, 张铃. 机器学习理论与算法[M]. 北京: 科学出版社, 2012: 18-55.
ZHANG Y P, ZHANG L. Machine Learning Theory and Algorithm[M]. Beijing: Science Press, 2012: 18-55.
5 SHAWE-TAYLOR J, CRISTIANINI N. Kernel Methods for Pattern Analysis[M]. Cambridge: Cambridge University Press, 2005: 47-82.
6 LEMARÉCHAL C, BOYD S, VANDENBERGHE L. Convex optimization[J]. European Journal of Operational Research, 2006, 170(1): 326-327.
7 SUYKENS J A K, VANDEWALLE J. Least squares support vector machine classifiers[J]. Neural Processing Letters, 1999, 9(3): 293-300. DOI:10. 1023/A:1018628609742
doi: 10. 1023/A:1018628609742
8 XU F, LIU Z P, ZHENG H B, et al. On-line modeling of ship maneuvering motion based on LS-SVM[J]. Ship Mechanics, 2021, 25(6):752-759.
9 MIAO F, SHARMA A. Design and implementation of construction cost prediction model based on SVM and LSSVM in industries 4.0[J]. International Journal of Intelligent Computing and Cybernetics, 2021, 14(2): 145-157. DOI:10.1108/IJICC-10-2020-0142
doi: 10.1108/IJICC-10-2020-0142
10 赵庆志, 刘洋, 姚顽强. 利用最小二乘支持向量机的短临降雨预测模型构建[J]. 大地测量与地球动力学, 2021, 41(2): 152-156. DOI:10.14075/j.jgg. 2021.02.008
ZHAO Q Z, LIU Y, YAO W Q. Establishment of short-term rainfall forecast model by least square support vector machine[J]. Journal of Geodesy and Geodynamics, 2021, 41(2): 152-156. DOI:10. 14075/j.jgg.2021.02.008
doi: 10. 14075/j.jgg.2021.02.008
11 封建湖, 车刚明, 聂玉峰, 等. 数值分析原理[M]. 北京: 科学出版社, 2006.
FENG J H, CHE G M, NIE Y F, et al. Principles of Numerical Analysis[M]. Beijing: Science Press, 2006.
12 徐仲, 张凯院, 陆全, 等. 矩阵论简明教程[M]. 3版. 北京: 科学出版社, 2014.
XU Z, ZHANG K Y, LU Q, et al. Matrix Theory Concise Course[M]. 3rd ed. Beijing: Science Press, 2013.
13 王萼芳, 石生明. 高等代数 [M]. 4版. 北京: 高等教育出版社, 2013:137-140.
WANG E F, SHI S M. Advanced Algebra[M]. 4th ed. Beijing: Higher Education Press, 2013: 137-140.
14 王华忠, 俞金寿. 核函数方法及其在过程控制中的应用[J]. 石油化工自动化, 2005, 41(1): 25-30. DOI:10.3969/j.issn.1007-7324.2005.01.008
WANG H Z, YU J S. Kernel-based methods and their applications to process control[J]. Automation in Petro-Chemical Industry, 2005, 41(1): 25-30. DOI:10.3969/j.issn.1007-7324.2005.01.008
doi: 10.3969/j.issn.1007-7324.2005.01.008
15 RASMUSSEN C E. Gaussian Processes in Machine Learning[M]. Berlin: Springer, 2003: 63-71. DOI:10.1007/978-3-540-28650-9_4
doi: 10.1007/978-3-540-28650-9_4
16 荣海娜, 张葛祥, 金炜东. 系统辨识中支持向量机核函数及其参数的研究[J]. 系统仿真学报, 2006, 18(11): 3204-3208, 3226. doi:10.3969/j.issn.1004-731X.2006.11.050
RONG H N, ZHANG G X, JIN W D. Selection of kernel functions and parameters for support vector machines in system identification[J]. Journal of System Simulation, 2006, 18(11): 3204-3208, 3226. doi:10.3969/j.issn.1004-731X.2006.11.050
doi: 10.3969/j.issn.1004-731X.2006.11.050
[1] 康俊锋,符悦,方雷,李咪咪,谢玉静,周朝阳. 基于最小二乘法赋权的ARIMA-LSTM模型预测入境旅游人数——以上海市为例[J]. 浙江大学学报(理学版), 2023, 50(4): 508-520.
[2] 常晓洁,张华. 一种基于V-TGRU模型的资源调度算法[J]. 浙江大学学报(理学版), 2022, 49(4): 467-473.
[3] 郏丽丽,孙婷婷. 紫色球杆菌视紫红质光谱特性的机器学习研究[J]. 浙江大学学报(理学版), 2022, 49(3): 280-286.
[4] 李君轶, 任涛, 陆路正. 游客情感计算的文本大数据挖掘方法比较研究[J]. 浙江大学学报(理学版), 2020, 47(4): 507-520.
[5] 潘水洋, 刘俊玮, 王一鸣. 基于神经网络的股票收益率预测研究[J]. 浙江大学学报(理学版), 2019, 46(5): 550-555.
[6] 叶兴德 江金 生. 双调和方程边值问题的一种新的伪谱逼近[J]. 浙江大学学报(理学版), 1995, 22(1): 13-21.