|本期目录/Table of Contents|

[1]王京涛,陆金桂.基于多策略改进神经网络的力学性能近似估算[J].南京工业大学学报(自然科学版),2018,40(06):63-69.[doi:10.3969/j.issn.1671-7627.2018.06.010]
 WANG Jingtao,LU Jingui.Approximate estimation of mechanical properties based on multi-strategy back propagation algorithm[J].Journal of NANJING TECH UNIVERSITY(NATURAL SCIENCE EDITION),2018,40(06):63-69.[doi:10.3969/j.issn.1671-7627.2018.06.010]
点击复制

基于多策略改进神经网络的力学性能近似估算()
分享到:

《南京工业大学学报(自然科学版)》[ISSN:1671-7627/CN:32-1670/N]

卷:
40
期数:
2018年06期
页码:
63-69
栏目:
出版日期:
2018-11-20

文章信息/Info

Title:
Approximate estimation of mechanical properties based on multi-strategy back propagation algorithm
文章编号:
1671-7627(2018)06-0063-07
作者:
王京涛陆金桂
南京工业大学 机械与动力工程学院,江苏 南京 211800
Author(s):
WANG JingtaoLU Jingui
College of Mechanical and Power Engineering,Nanjing Tech University,Nanjing 211800,China
关键词:
Q235 多策略BP算法 力学性能 近似估算
Keywords:
Q235 multi-strategy BP algorithm mechanical property approximate estimation
分类号:
TP183;TH136
DOI:
10.3969/j.issn.1671-7627.2018.06.010
文献标志码:
A
摘要:
介绍了应用于Q235焊接力学性能的神经网络近似分析方法,开展了焊接力学性能的样本数据对神经网络近似估算的实验研究,为提高估算的准确度,提出了一种新的基于融合多策略混合粒子群优化BP算法(MSBPA)应用于焊接性能数据的非线性映射处理,建立了估算模型。将估算模型应用于Q235焊接力学性能估算分析。为验证算法的有效性,分别运用了融合多策略混合粒子群优化BP算法、遗传算法优化BP算法和传统BP算法对焊接性能近似估算问题进行对比仿真分析。结果表明:基于融合多策略混合粒子群优化BP估算模型对Q235焊接力学性能有较好的非线性拟合能力,估算值与实验值间最大相对误差仅为3.5%,具有较优的估算准确性。
Abstract:
The approximate analysis method of neural network was applied for the mechanical properties of Q235 welding,and the experiments was carried out to estimate the sample data of welding on the mechanical properties of the neural network approximation.In order to improve the accuracy of estimation,a new multi-strategy hybrid particle swarm optimization back-propagation(BP)algorithm(MSBPA)for nonlinear mapping process of the welding performance data was proposed.The estimation model was applied to estimate the mechanical properties of Q235 welding.In order to verify the effectiveness of the algorithm,the fusion strategy,the hybrid particle swarm optimization BP algorithm,the genetic algorithm optimization BP algorithm and the traditional BP algorithm were used to analyze and compare the approximate estimation of the welding performance.Results showed that the multi-strategy fusion hybrid particle swarm optimization BP estimation model of Q235 welding mechanical properties had a good non-linear fitting capability.The maximum relative error between the estimated values and the experimental ones was only 3.5%,and the proposed method had better estimation accuracy.

参考文献/References:

[1] 陈果.用结构自适应神经网络预测航空发动机性能趋势[J].航空学报,2007,28(3):535.
[2] 吴良晨,王东坡,王康,等.Q420B焊接接头超高周疲劳性能[J].天津大学学报(自然科学版),2008,41(12):1480.
[3] CHEN G Y,ZHANG M J,ZHAO Z,et al.Measurements of laser-induced plasma temperature field in deep penetration laser welding[J].Optics & laser technology,2013,45(3):551.
[4] 董丹阳,王观军,马敏,等.车用双相钢激光焊接接头组织性能研究[J].中国激光,2012,39(9):70.
[5] 陈根余,陈飞,张屹,等.对接间隙对车用高强钢光纤激光焊接的影响[J].中国激光,2011,38(6):154.
[6] XU W,WESTERBAAN D,NAYAK S S,et al.Tensile and fatigue properties of fiber laser welded high strength low alloy and DP980 dual-phase steel joint[J].Materials and design,2013,439(11):373.
[7] FARABI N,CHEN D L,ZHOU Y.Micro-structure and mechanical properties of laser welded dissimilar DP600/DP980 dual-steel joint[J].Journal of alloys and compounds,2011,509(3):982.
[8] YAO X.Evolutionary artificial neural networks[J].International journal of neural systems,1993(3):203.
[9] POURMOUSAVI KANI S A,AREDHALI M M.Very short-term wind speed prediction:a new artificial neural network-Markov chain model[J].Energy conversion and management,2011,52(1):738.
  
[10] LI G,SHI J,ZHOU J Y,Bayesian adaptive combination of short-term wind speed forecasts from neural network models[J].Renew energy,2011,36(1):352.
[11] GUO Z H,ZHAO W G,LU H.Y,et al.Multi-step prediction for wind speed using modified EMD-based artificial neural network model[J].Renew energy,2012,37(1):241.
[12] 蒋丰骏,徐月同,邓小雷,等.手动托盘车焊接工艺的焊接变形仿真及预测[J].浙江大学学报(工学版),2013,43(9):1625.
[13] 魏微,高谦.改进的 BP 神经网络模型预测充填体强度[J].哈尔滨工业大学学报,2013,45(6):90.
[14] ZENG G M,QIN X S,HE L,et al.A neural network predictive control system for paper mill waste water treatment[J].Engineering applications of artificial intelligence,2003,16(2):121.
[15] 李农,李建芬,刘宇平.一类参数未知混沌系统的追踪控制与参数辨识[J].物理学报,2011(5):106.
[16] 吕艳萍,李绍滋,陈水利,等.自适应扩散混合变异机制的微粒群算法[J].软件学报,2007,18(11):2740.
[17] CHIANG C L.Improved genetic algorithm for power economic dispatch of units with valve-point effects and multiple fuels[J].IEEE transaction power system,2005,20(4):1690.
[18] GAING Z L.Particle swarm optimization to solving the economic dispatch considering the generator constraints[J].IEEE transaction power system,2003,18(3):1187.
[19] GUO Z H,ZHAO W G,LU H Y,et al.Multi-step prediction for wind speed using modified EMD-based artificial neural network model[J].Renew energy,2012,37(1):241.

备注/Memo

备注/Memo:
收稿日期:2017-09-01
基金项目:国家自然科学基金(50975133); 国家“十二五”科技支撑计划(2013BAF02B11)
作者简介:王京涛(1990—),男,山东聊城人,硕士,主要研究方向为智能算法; 陆金桂(联系人),教授,E-mail:lujg@njtech.edu.cn.
引用本文:王京涛,陆金桂.基于多策略改进神经网络的力学性能近似估算[J].南京工业大学学报(自然科学版),2018,40(6):63-69..
更新日期/Last Update: 2018-11-30