正在加载图片...
第13卷第5期 智能系统学报 Vol.13 No.5 2018年10月 CAAI Transactions on Intelligent Systems Oct.2018 D0:10.11992/tis.201704007 网络出版地址:http:/kns.cnki.net/kcms/detail/23.1538.TP.20180424.0923.003html 基于改进KH算法优化ELM的目标威胁估计 傅蔚阳,刘以安,薛松2 (1.江南大学物联网工程学院,江苏无锡214122;2.中国船舶重工集团公司第七研究院电子部,北京 100192) 摘要:为了提高目标威胁度估计的精确度,建立了反向学习磷虾群算法(OK田优化极限学习机的目标威胁估 计模型(OKH-ELM),提出基于此模型的算法。该模型使用反向学习策略优化磷虾群算法,并通过改进后的磷 虾群算法优化极限学习机初始输入权重和偏置,使优化后的极限学习机能够对威胁度测试样本集做更好的预 测。实验结果显示,OKH算法能够更好地优化极限学习机的权值与阈值,使建立的极限学习机目标威胁估计 模型具有更高的预测精度和更强的泛化能力,能够精准、有效地实现目标威胁估计。 关键词:目标威胁估计:磷虾群算法:极限学习机:反向学习:神经网络;权值;阈值:威胁估计模型 中图分类号:TP391.9文献标志码:A文章编号:1673-4785(2018)05-0693-07 中文引用格式:傅蔚阳,刘以安,薛松.基于改进KH算法优化ELM的目标威胁估计.智能系统学报,2018,13(5): 693-699 英文引用格式:FU Weiyang,LIU Yi'an,XUE Song.Target threat assessment using improved Krill Herd optimization and extreme learning machine[J.CAAI transactions on intelligent systems,2018,13(5):693-699. Target threat assessment using improved Krill Herd optimization and extreme learning machine FU Weiyang',LIU Yi'an',XUE Song? (1.School of Internet of Things Engineering,Jiangnan University,Wuxi214122,China,2.Electronic Department,The Seventh Re- search Institute of China Shipbuilding Industry Corporation,Beijing 100192,China) Abstract:To improve the accuracy of target threat estimation,the opposition-based learning Krill Herd optimization (OKH)and extreme learning machine(OKH-ELM)model is established,and the algorithm based on the model is presented.The OKH-ELM adopts opposition-based learning(OBL)to optimize KH,and then the improved KH and ex- treme learning machine are employed to simultaneously optimize the initial input weights and offsets of the hidden lay- er in ELM.A target threat database is adopted to test the performance of OKH-ELM in target threat prediction.The ex- perimental result shows that OKH Algorithm can better optimize the weights and thresholds of the hidden layer in ELM and improve the prediction precision and generalization ability of the target threat assessment model;therefore,it can accurately and effectively estimate target threat. Keywords:target threat assessment;Krill Herd algorithm;extreme learning machine;opposition-based learning:neural networks;weights:thresholds:threat estimation model 严格来讲,目标威胁估计是一个NP困难问的结果,但BP神经网络也有着明显缺点,比如训 题山。在进行威胁估计时,给出一个各种因素与 练时间长、易陷入局部极值、学习率选择敏感 威胁程度的函数关系困难很大。文献[2]使用 等。所以本文提出了改进的磷虾群算法(opposition- BP神经网络处理目标威胁估计问题获得了不错 based learning Krill Herd optimization,,OKH)优化极 收稿日期:2017-04-12.网络出版日期:2018-04-24 限学习机的目标威胁估计模型。磷虾群算法是 基金项目:江苏省自然科学基金项目(BK20160162) 通信作者:傅蔚阳.E-mail:18806186287@163.com. 2012年由Gandomi等提出的一种新的仿生优化DOI: 10.11992/tis.201704007 网络出版地址: http://kns.cnki.net/kcms/detail/23.1538.TP.20180424.0923.003.html 基于改进 KH 算法优化 ELM 的目标威胁估计 傅蔚阳1 ,刘以安1 ,薛松2 (1. 江南大学 物联网工程学院,江苏 无锡 214122; 2. 中国船舶重工集团公司第七研究院 电子部,北京 100192) 摘 要:为了提高目标威胁度估计的精确度,建立了反向学习磷虾群算法 (OKH) 优化极限学习机的目标威胁估 计模型 (OKH-ELM),提出基于此模型的算法。该模型使用反向学习策略优化磷虾群算法,并通过改进后的磷 虾群算法优化极限学习机初始输入权重和偏置,使优化后的极限学习机能够对威胁度测试样本集做更好的预 测。实验结果显示,OKH 算法能够更好地优化极限学习机的权值与阈值,使建立的极限学习机目标威胁估计 模型具有更高的预测精度和更强的泛化能力,能够精准、有效地实现目标威胁估计。 关键词:目标威胁估计;磷虾群算法;极限学习机;反向学习;神经网络;权值;阈值;威胁估计模型 中图分类号:TP391.9 文献标志码:A 文章编号:1673−4785(2018)05−0693−07 中文引用格式:傅蔚阳, 刘以安, 薛松. 基于改进 KH 算法优化 ELM 的目标威胁估计[J]. 智能系统学报, 2018, 13(5): 693–699. 英文引用格式:FU Weiyang, LIU Yi’an, XUE Song. Target threat assessment using improved Krill Herd optimization and extreme learning machine[J]. CAAI transactions on intelligent systems, 2018, 13(5): 693–699. Target threat assessment using improved Krill Herd optimization and extreme learning machine FU Weiyang1 ,LIU Yi’an1 ,XUE Song2 (1. School of Internet of Things Engineering, Jiangnan University, Wuxi 214122, China; 2. Electronic Department, The Seventh Re￾search Institute of China Shipbuilding Industry Corporation, Beijing 100192, China) Abstract: To improve the accuracy of target threat estimation, the opposition-based learning Krill Herd optimization (OKH) and extreme learning machine (OKH-ELM) model is established, and the algorithm based on the model is presented. The OKH-ELM adopts opposition-based learning (OBL) to optimize KH, and then the improved KH and ex￾treme learning machine are employed to simultaneously optimize the initial input weights and offsets of the hidden lay￾er in ELM. A target threat database is adopted to test the performance of OKH-ELM in target threat prediction. The ex￾perimental result shows that OKH Algorithm can better optimize the weights and thresholds of the hidden layer in ELM and improve the prediction precision and generalization ability of the target threat assessment model; therefore, it can accurately and effectively estimate target threat. Keywords: target threat assessment; Krill Herd algorithm; extreme learning machine; opposition-based learning; neural networks; weights; thresholds; threat estimation model 严格来讲,目标威胁估计是一个 NP 困难问 题 [1]。在进行威胁估计时,给出一个各种因素与 威胁程度的函数关系困难很大。文献[2]使用 BP 神经网络处理目标威胁估计问题获得了不错 的结果,但 BP 神经网络也有着明显缺点,比如训 练时间长、易陷入局部极值、学习率 η 选择敏感 等。所以本文提出了改进的磷虾群算法 (opposition￾based learning Krill Herd optimization, OKH) 优化极 限学习机的目标威胁估计模型。磷虾群算法是 2012 年由 Gandomi 等 [3]提出的一种新的仿生优化 收稿日期:2017−04−12. 网络出版日期:2018−04−24. 基金项目:江苏省自然科学基金项目 (BK20160162). 通信作者:傅蔚阳. E-mail:18806186287@163.com. 第 13 卷第 5 期 智 能 系 统 学 报 Vol.13 No.5 2018 年 10 月 CAAI Transactions on Intelligent Systems Oct. 2018
向下翻页>>
©2008-现在 cucdc.com 高等教育资讯网 版权所有