基于粒子群改进神经网络的舰艇磁场推算模型

上海交通大学学报（自然版） ›› 2011, Vol. 45 ›› Issue (06): 809-813.

基于粒子群改进神经网络的舰艇磁场推算模型

连丽婷，肖昌汉，杨明明，周国华

(海军工程大学电气与信息工程学院，武汉 430033)

收稿日期:2010-07-10 出版日期:2011-06-29 发布日期:2011-06-29

The Model of Ship’s Magnetic Field Extrapolation Based on Neural Network Improved by Particle Swarm Optimization

LIAN Li-Ting, XIAO Chang-Han, YANG Ming-Ming, ZHOU Guo-Hua

(School of Electrical and Information Engineering, Naval University of Engineering, Wuhan 430033, China)

Received:2010-07-10 Online:2011-06-29 Published:2011-06-29

摘要/Abstract

摘要： 针对目前线性建模解决舰艇内外磁场推算问题时存在的困难，从非线性优化的角度出发，建立了内外磁场之间的误差反向传播神经网络预报模型.为了改善网络的固有缺陷，利用粒子群算法优化网络的初始权值与阈值，使其能够逃离局部最优点，增强了网络的鲁棒性.该方法避免了利用线性化方法存在的诸多困难，可实现舰艇内外磁场推算.利用船模实验对网络预测的准确性进行了验证，结果表明其换算精度较线性方法有所提高，满足工程实际需求.

关键词: 舰艇；磁场；闭环消磁；粒子群算法；误差反向传播

Abstract: The magnetic anomaly created by ferromagnetic submarines may endanger their invisibility. Nowadays, a new technique called closedloop degaussing system can reduce the magnetic anomaly especially permanent one in realtime. To achieve it, a model which is able to predict offboard magnetic field from on board measurements was required. Many researchers settle the problem by a linear model. A back propagation neural network model was proposed to solve it. The model can escape local optimum thanks to optimizing the initial weight values and threshold values by particle swarm optimization algorithm. The method can avoid many problems from linear model and its high accuracy and good robustness was tested by a mockup experiment.

Key words: ship, magnetic field, closed loop degaussing, particle swarm optimizer, error back propagation

中图分类号:

TM153.1

连丽婷, 肖昌汉, 杨明明, 周国华. 基于粒子群改进神经网络的舰艇磁场推算模型[J]. 上海交通大学学报（自然版）, 2011, 45(06): 809-813.

LIAN Li-Ting, XIAO Chang-Han, YANG Ming-Ming, ZHOU Guo-Hua. The Model of Ship’s Magnetic Field Extrapolation Based on Neural Network Improved by Particle Swarm Optimization[J]. Journal of Shanghai Jiaotong University, 2011, 45(06): 809-813.

参考文献

［1］Oliver C, Jeanlouis C, Jeant B, et al. Recent improvements for solving inverse magneto static problem applied to thin shells［J］. IEEE Transactions and Magnetics, 2002, 38(2): 10051008.

［2］Carl S. Closedloop multisensor control system and method［P］. US Patent： 5189590, 19930223.

［3］Richard M, Robert A. Ship degaussing system and algorithm［P］. US Patent： 6965505 B1, 20051115.

［4］Vuillerunet Y. Closed loop degaussing applied to double hull submarine magnetization identification from near magnetic fields measurements［D］. Grenoble： Grenoble Electrical Engineering Laboratory, 2008: 2530.

［5］王鲸，刘胜道，苏广东．一种潜艇内外磁场换算算法的实验验证［J］．海军工程大学学报，2010, 22(1)：8386.

WANG Jing, LIU Shengdao, SU Guangdong. An experimental verification of a conversion of magnetic field inside and outside submarine［J］. Journal of Navy University of Engineering, 2010, 22(1): 8386.

［6］连丽婷，肖昌汉，刘胜道,等．基于粒子群算法的薄钢板两侧磁场推算中的位置优化［J］．上海交通大学学报， 2010， 44(7)： 975979.

LIAN Liting, XIAO Changhan, LIU Shengdao， et al. The measurement positions optimization in the extrapolation of steel’s magnetic field based on particle swarm optimization algorithm［J］. Journal of Shanghai Jiaotong University, 2010, 44(7): 975979.

［7］董长虹．Matlab神经网络与应用［M］．北京：国防工业出版社， 2007： 64110.

［8］Kennedy J, Eberhart R C. Particle swarm optimization ［C］// Proceedings of the IEEE International Conference on Neural Network. Perth, Australia: IEEE, 1995: 3943.

［9］WANG Ping, HUANG Zhenyi, ZHANG Mingya, et al. Mechanical property prediction of strip model based on PSOBP neural network［J］. Journal of Iron and Steel Research, 2007, 15(3): 8791.

［10］史东生，弟宇鸣，周春林．粒子群优化算法在神经网络识别γ能谱中的应用［J］．核技术，2007，30(7)：615619.

SHI Dongsheng, DI Yuming, ZHOU Chunlin. Application of particle swarm optimization to identify gamma spectrum with neural network［J］. Nuclear Techniques, 2007, 30(7): 615619.