基于AGA⁃GRNN的三维室内定位研究  

Research on 3D indoor positioning based on AGA⁃GRNN

作　　者：马翠红 徐天天 杨友良 MA Cuihong;XU Tiantian;YANG Youliang(North China University of Science and Technology,Tangshan 063210,China)

机构地区：[1]华北理工大学,河北唐山063210

出　　处：《现代电子技术》2020年第14期90-93,共4页Modern Electronics Technique

基　　金：国家自然科学基金(61171058)。

摘　　要：针对传统的测距定位模型容易受到外部因素的干扰,故而降低了定位精度的问题,提出自适应遗传算法广义传播神经网络模型(AGA⁃GRNN)。引入自适应遗传算法(AGA)用于优化广义回归神经网络(GRNN)的参数,通过AGA⁃GRNN构建无线信号强度(RSSI)与目标位置之间的关系进行定位,利用对应的映射关系判断目标位置。仿真结果表明,该算法在15 m×15 m×5 m范围内的平均定位误差为26.2 cm。与GRNN及BP相比,计算精度分别提高了45.4%和53.6%。同时将AGA优化GRNN三维定位模型与GA优化GRNN三维定位模型的优化时间进行了比较,结果表明,AGA⁃GRNN的平均定位时间减少了0.5 s,有效地提高了三维定位的精度和效率。In allusion to the traditional ranging and positioning model is easy to be disturbed by the external factors,which reduce the positional accuracy,an adaptive genetic algorithm generalized propagation neural network model(AGA⁃GRNN)is proposed.The adaptive genetic algorithm(AGA)is introduced to optimize the parameters of generalized propagation neural network(GRNN),and the relationship between received signal strength indicator(RSSI)and target location is established with AGA⁃GRNN to conduct the positioning.The target position is judged by means of the corresponding mapping relation.The simulation results show that the algorithm′s average positioning error in the range of 15 m×15 m×5 m is 26.2 cm.In comparison with GRNN and BP,the calculation accuracy of this algorithm is increased by 45.4%and 53.6%,respectively.The lengths of optimizing time of the GRNN 3D positioning models optimized with AGA and GA are compared.The results show that the average positioning time of AGA⁃GRNN is reduced by 0.5 s,which effectively improves the accuracy and efficiency of 3D positioning.

关 键 词：三维定位模型 室内定位 AGA⁃GRNN 射频识别 目标位置 仿真实验 

分 类 号：TN976-34[电子电信—信号与信息处理] TP391.9[电子电信—信息与通信工程]