基于对偶神经网络的动力方程精细积分法

摘要
图/表
参考文献(16)
相关文章 (15)

全文: PDF (849 KB) HTML (1 KB)
输出: BibTeX | EndNote (RIS)

摘要针对一般动力学方程，提出了一种用于求解具有任意非齐次项动力学方程的对偶神经网络精细积分算法。在时间域内，对动力学非齐次方程求解中涉及到的积分运算，选用一组神经网络同时逼近被积函数和原函数，然后通过牛顿莱布尼茨公式实现积分项的求解。该方法利用神经网络的函数拟合优势，具有对时间步长不敏感，不需要对矩阵求逆，不对非齐次项进行假设等优点。通过算例与精细积分法、威尔逊- 、广义精细积分法等方法进行比较，计算结果表明该方法精度较高，适用范围广。
关键词：精细积分法；对偶神经网络；动力方程；直接积分

	服务

	把本文推荐给朋友
	加入我的书架
	加入引用管理器
	E-mail Alert
	RSS
	作者相关文章
	杨永1
	李海滨1
	2

关键词 ：精细积分法, 对偶神经网络 , 动力方程, 直接积分

Abstract：For dynamics equation, the existing common solution method is the precise integration method, but the result of the equation given by the precise integration method contains complex integrals, which is difficult to solve by the traditional numerical integration method. In this manuscript, an integration algorithm is proposed based on neural networks. The algorithm builds two neural networks, one neural network is used to approximate the integral function and the other neural network is used to approximate the original function. While one neural network is trained, the other neural network is trained at the same time. The result of the integration can then be obtained by Newton Leibniz function. The algorithm of this manuscript takes advantage of the function fitting of the neural network, which is insensitive to the selected time step during the solution of the kinetic equations, does not require the inverse of the matrix, and does not require the assumption of inhomogeneous term. With computational examples, the algorithm of this manuscript is compared with a variety of existing commonly used methods. The calculation results show that the method has high accuracy and wide applicability.
Key words: high precision integration; dual neural networks; dynamic equation; direct integral

Key words： high precision integration dual neural networks dynamic equation direct integral

收稿日期: 2021-06-22 出版日期: 2022-08-28

引用本文:

杨永1，李海滨1,2. 基于对偶神经网络的动力方程精细积分法[J]. 振动与冲击, 2022, 41(16): 188-193.
YANG Yong1,LI Haibin1,2. A precise integration method for dynamic equations based on dual neural networks. JOURNAL OF VIBRATION AND SHOCK, 2022, 41(16): 188-193.

链接本文:

http://jvs.sjtu.edu.cn/CN/ 或 http://jvs.sjtu.edu.cn/CN/Y2022/V41/I16/188

[1] 钟万勰. 暂态历程的精细计算方法[J]. 计算结构力学及其应用, 1995, 12(1): 1-6.
Zhong wanxie. Precise computation for transient analysis[J]. Computational structural mechanics and applications, 1995, 12(1): 1-6.
[2] 钟万勰. 结构动力方程的精细时程积分法[J]. 大连理工大学学报, 1994, 34(2): 131-136.
Zhong wanxie. On precise time-integration method for structural dynamics[J]. Journal of Dalian University of Technology, 1994, 34(2): 131-136.
[3] Zhong W,Cheung Y,Li Y. The Precise Finite Strip Method[J]. Computers & Structures, 1998, 69(6): 773-783.
[4] 高强,谭述君,钟万勰. 精细积分方法研究综述[J]. 中国科学:技术科学, 2016, 46(12): 1207-1218.
Gao Qiang, Tan Shujun, Zhong Wanxie. A survey of the precise integration method[J]. Scientia Sinica Technologica, 2016, 46(12): 1207-1218.
[5] 顾元宪,陈飚松,张洪武. 结构动力方程的增维精细积分法[J]. 力学学报, 2000, 32(4): 447-456.
Gu Yuanxian, Chen Biaosong, Zhang Hongwu. Precise time integration with dimension expanding method[J]. Acta Mechanica Sinica, 2000, 32(4): 447-456.
[6] 高小科,邓子辰,黄永安. 基于三次样条插值的精细积分法[J]. 振动与冲击, 2007, 113(9): 75-77, 82, 171.
Gao Xiaoke, Deng Zichen, Huang Yongan. A high precise direct integration based on cubic spline interpolation[J]. Journal of Vibration and Shock, 2007, 113(9): 75-77, 82, 171.
[7] 汪梦甫,周锡元. 结构动力方程的高斯精细时程积分法[J]. 工程力学, 2004,21(4): 13-16.
Wang Mengfu, Zhou Xiyuan. Gauss precise time-integration of structural dynamic analysis[J]. Engineering mechanics, 2004, 21(4): 13-16.
[8] 刘伟伟,刘玲,胥祥. 高斯精细积分法在结构动力方程的应用[C]//北京力学会第二十五届学术年会会议论文集, 2019: 754-758.
[9] 王永,马骏,李靖翔,等. 非齐次动力学方程的一种精细积分单步方法[J]. 计算力学学报, 2020, 37(2): 212-217.
Wang Yong, Ma Jun, Li Jingxiang, et al. A precise integration single -step method for nonhomogeneous dynamic equations [J]. Chinese Journal of Computational Mechanics, 2020, 37(2): 212-217.
[10] 储德文,王元丰. 精细直接积分法的积分方法选择[J]. 工程力学, 2002, 19(6): 115-119.
Chu Dewen, Wang Yuanfeng. Integration formula selection for precise direct integration method[J]. Engineering mechanics, 2002, 19(6): 115-119.
[11] 富明慧,林敬华,刘祚秋. 结构动力分析的广义精细积分法[C]//第九届全国振动理论及应用学术会议论文集, 2007: 888-892.
[12] 富明慧,刘祚秋,林敬华. 一种广义精细积分法[J]. 力学学报, 2007,39(5): 672-677.
Fu Minghui, Liu Zuoqiu, Lin Jinghua. A generalized precise time step integration method[J]. Acta Mechanica Sinica, 2007, 39(5): 672-677.
[13] 王海波,何崇检,贾耀威. 线性动力分析的一种通用积分格式[J]. 振动与冲击, 2019, 38(10): 48-53.
Wang Haibo, He Chongjian, Jia Yaowei. General integration scheme for linear dynamic analysis [J]. Journal of Vibration and Shock, 2019, 38(10): 48-53.
[14] Li H,He Y,Nie X. Structural Reliability Calculation Method Based on the Dual Neural Network and Direct Integration Method[J]. Neural Computing and Applications, 2018, 29(7): 425-433.
[15] Du J,Li H. Direct Integration Method Based on Dual Neural Networks to Solve the Structural Reliability of Fuzzy Failure Criteria[J]. Proceedings of the Institution of Mechanical Engineers, Part C: Journal of Mechanical Engineering Science, 2019, 233(19): 7183-7196.
[16] Li H, Li Y, Li S. Dual Neural Network Method for Solving Multiple Definite Integrals.[J]. Neural Comput, 2018, 31(1): 208-232.