查询结果:   刘晓晨,张静.基于改进BP神经网络的室内无线定位方法[J].计算机应用与软件,2016,33(6):114 - 117.
中文标题
基于改进BP神经网络的室内无线定位方法
发表栏目
网络与通信
摘要点击数
765
英文标题
INDOOR WIRELESS POSITIONING BASED ON IMPROVED BP NEURAL NETWORK
作 者
刘晓晨 张静 Liu Xiaochen Zhang Jing
作者单位
上海师范大学信息与机电工程学院 上海 200234     
英文单位
School of Information,Mechenical and Electrical Engineering,Shanghai Normal University,Shanghai 200234,China     
关键词
室内定位 BP神经网络 思维进化算法 接收信号强度指示
Keywords
Indoor positioning BP neural network Mind evolutionary computation (MEC) Received signal strength indicator
基金项目
国家自然科学基金项目(61101209);上海市自然科学基金项目(11ZR1426600);上海师范大学一般科研项目(DYL201406);上海师范大学重点学科基金项目(DZL126)
作者资料
刘晓晨,硕士生,主研领域:无线室内定位,神经网络。张静,副教授。 。
文章摘要
针对在室内无线定位中采用加权质心定位法时精度较低且难以克服信号不稳定的问题,提出改进的BP神经网络方法。以接收信号强度(RSSI)为输入、二维平面坐标为输出建立网络结构,网络的初始权值和阈值用思维进化算法优化,并用边长3 m的正方形区域内的196个样本数据训练。实验结果表明,在27个预测点上可达到定位精度0.1 m。相比于BP网络以及BP网络和遗传算法的结合算法,该定位方法训练收敛时间短,定位结果稳定。
Abstract
For indoor wireless positioning, we used an improved BP neural network to overcome the low accuracy and signal instability when a weighted centroid positioning method being adopted. We established the BP network structure by using the received signal strength indication (RSSI) as input and the two-dimensional position as output. The mind evolutionary computation was used to optimise its initial weights and thresholds. The network was trained by 196 sample data within a square area of 3 m side length. Experimental results showed that it was able to achieve the positioning accuracy by 0.1 m at 27 predictive test points. Compared with a standard BP neural network as well as with a combination of BP network and genetic algorithm, this positioning method had the performance of short training and convergence time, the positioning result was stable as well.
下载PDF全文