遗传优化小波神经网络在组合导航系统中的应用  

Application of wavelet neural networks based on genetic algorithms in integrated navigation system

作　　者：刘建娟[1] 徐晓苏[1] 刘锡祥[1] 

机构地区：[1]东南大学仪器科学与工程系,南京210096

出　　处：《中国惯性技术学报》2006年第2期33-37,共5页Journal of Chinese Inertial Technology

基　　金：国家自然科学基金项目(60374046);总装备部国防预研基金项目(514090101035W0609)

摘　　要：实时性是组合导航系统的一个重要指标,而神经网络的优化学习问题是决定网络效率的关键技术。遗传优化小波神经网络不仅继承了小波分析良好的局部性及其神经网络的学习和推广能力,而且具有遗传算法全局寻优的特点,是多层前向神经网络学习的一种理想算法。将它应用于组合导航系统中并进行了仿真,结果表明,该算法能够根据实际情况自适应确定网络结构,实时性好,精度与常规方法相当。One of the important indexes in integrated navigation system is the real-time performance, and the key technology is the optimum learning method, which determines the efficiency of the Neural Networks. Wavelet neural networks based on the genetic algorithms（GAWNN） not only has the local property of the wavelet analysis and the generalization capability of the artificial neural networks, but also has the advantages of the fast global searching of the genetic algorithms. So the GAWNN can be treated as an ideal algorithm for the training of the multi-layer feeding forward neural networks. In this paper, it was used in integrated navigation system, and the simulation results show that this algorithm can adaptively determine the network structure, has good real-time performance, and has the similar precision to that of former systems.

关 键 词：组合导航 遗传算法 小波神经网络 

分 类 号：U666.1[交通运输工程—船舶及航道工程]