基于BP神经网络的空间坐标视觉测量误差控制与补偿研究

来源期刊:中南大学学报(自然科学版)2011年第z1期

论文作者:李凯 袁峰 丁振良

文章页码:60 - 66

关键词:空间坐标;神经网络;视觉测量;误差补偿

Key words:spatial coordinates; neural network; vision measurement; error compensation

摘    要:基于在利用由2台经纬仪组成的空间三维坐标测量系统进行几何量测量的过程中,经纬仪的垂直角读数在特定位置会产生系统误差,并对精度带来不利影响,以空间三坐标测量系统作为研究对象,利用空间交汇原理建立了待测点坐标以及待测距离与经纬仪观测角的数学模型,分析了系统误差产生的原因。提出将系统误差作为非线性项,通过选取满足误差存在条件的数据训练网络令其逼近非线性函数,激励网络从而对系统误差进行有效控制和补偿。在仿真中根据精度指标不断优化BP神经网络的权值,获得了较好的神经网络参数配置。实验中补偿后测量结果与理论计算值较吻合,说明了这一方法的有效性。

Abstract: Based on the fact that during the geometric variables measurement course of spatial three-dimension coordinates by bino-theodolites, system errors of vertical observing angles’ reading occur at certain position by experiment data, the result of which leads to bad effects to precision, objected to spatial coordinates measurement system, the mathematic model of point coordinate and distance to be measured was established, and the cause of system error by means of space rendezvous and docking technology was analyzed. The idea was proposed of regarding system error as nonlinear component, training neural network’s (NN) by samples met the condition of system error existence to approach nonlinear function, and exciting NN to efficiently control and compensate system errors. In simulation, the weights of NN were optimized according to the precision performance index, and better parameters’ configurations were obtained. In experiment, consistence degree of compensated results and theoretical results validates the effectiveness of this method.

有色金属在线官网  |   会议  |   在线投稿  |   购买纸书  |   科技图书馆

中南大学出版社 技术支持 版权声明   电话:0731-88830515 88830516   传真:0731-88710482   Email:administrator@cnnmol.com

互联网出版许可证:(署)网出证(京)字第342号   京ICP备17050991号-6      京公网安备11010802042557号