面向室内定位的DHOHF-Elman神经网络算法  

DHOHF-Elman Neural Network Algorithm for Indoor Localization

作　　者：毛永毅[1] 阴颖 Mao Yongyi;Yin Ying(Xi’an University of Posts&Telecommunications,Xi’an,Shaanxi 710061,China)

机构地区：[1]西安邮电大学

出　　处：《信号处理》2019年第8期1358-1365,共8页Journal of Signal Processing

基　　金：陕西省重点研发计划项目(2018KW-004)

摘　　要：针对传统Elman神经网络算法在室内存在定位精度低的问题,提出了一种基于UWB(Ultra Wideband)改进的DHOHF-Elman(Elman neural network with Double Hidden layers and Output-Hidden Feedback,DHOHF-Elman)神经网络算法。该算法改进了神经网络拓扑结构增加了第二隐含层和第二承接层,达到了双隐含层反馈的效果,通过信道模型模拟大量实验数据,对构造的神经网络模型进行了训练与测试,表明了改进后的神经网络算法较传统神经网络算法有更高的定位精度和较好的收敛性,最后通过仿真结果分析验证了改进算法在有无高斯白噪声环境中的优良性和有效性。Traditional Elman neural network algorithm has relatively low positioning accuracy under indoor environment,which is a problem for indoor positioning system.To enhance the accuracy,an improved DHOHF-Elman(Elman neural network with Double Hidden layers and Output-Hidden Feedback,DHOHF-Elman)algorithm based on UWB(Ultra Wideband)is proposed.Through improving the nerve net topology structure by adding the second hidden layer and the second receiving layer,the algorithm improves the performance of the neural network,result in achieving the effect of double hidden layer feedback.Then,large amount of data simulated though channel model is used to train and test the constructed neural network model,indicating that the improved neural network algorithm has higher positioning accuracy and better convergence speed than traditional neural network algorithm.Finally,the simulation results verify the superiority and effectiveness of the new algorithm in different environments under the condition of consisting with and without WGN(White Gaussian Noise)respectively.

关 键 词：室内定位 ELMAN神经网络 超宽带定位技术 信道模型 

分 类 号：TN927.3[电子电信—通信与信息系统]