转换GPS高程的遗传神经网络方法  

Conversion of Height by Genetic Neural Network Method

作　　者：吴良才[1] 危志明[1] 

机构地区：[1]东华理工大学地球科学与测绘工程学院,江西抚州344000

出　　处：《测绘科学技术学报》2007年第4期244-246,共3页Journal of Geomatics Science and Technology

基　　金：地球动力学国家重点实验室开放基金项目资助(LED0506)

摘　　要：遗传算法是一类借鉴生物界自然选择和自然遗传机制的随机搜索算法,其主要特点是群体搜索策略和群体中个体之间的信息交换,搜索不依赖于梯度信息。把遗传算法和神经网络智能技术相结合,利用神经网络作为模型,以遗传算法作为权值进化算法进行GPS高程转换,提出了该算法的基本思想和算法实现过程。并通过实例进行计算,结果表明该算法用于GPS高程转换具有较好的精度,具有一定的实用价值。Genetic algorithm is a kind of evolution developing from the idea of biological options and natural genetic free searching algorithm. Its main characteristics are exchanges of information between group searching strategy and group-individual one, which cannot be relied on gradient information. With the combination of the genetic algorithm and neural network, the paper studies the conversion of GPS height based on genetic neural network and updated algorithm taking genetic algorithm as the weight. The paper also discusses the basic idea of the algorithm and its realization of algorithm process. In some cases, it turned out to be useful to adopt the conversion of GPS height in height precision and as a result.

关 键 词：GPS大地高 高程异常 遗传算法 神经网络 遗传神经网络 

分 类 号：P228[天文地球—大地测量学与测量工程]