CAJ | 학술논문

考虑到基于Zigbee传感器网络信号强度指示的指纹库定位算法只适用于静态条件下的定位，无法有效去除受噪声污染严重的有害数据，定位精度不尽人意，根据噪声的分布特点，引入粒子滤波技术，优化了该算法，有效去除了受噪声污染的有害数据，实现了目标的动态精确定位跟踪。传统粒子算法在应用过程中对重要性分布函数做了高斯分布假设，使得定位精度受到影响，将多层神经网络引入滤波算法进行补偿优化。实验结果验证了改进算法的优越性，其在静态条件下平均定位误差小于1．14m，动态跟踪定位精度也有很大幅度的提高，具有实际应用价值。
고필도기우Zigbee전감기망락신호강도지시적지문고정위산법지괄용우정태조건하적정위，무법유효거제수조성오염엄중적유해수거，정위정도불진인의，근거조성적분포특점，인입입자려파기술，우화료해산법，유효거제료수조성오염적유해수거，실현료목표적동태정학정위근종。전통입자산법재응용과정중대중요성분포함수주료고사분포가설，사득정위정도수도영향，장다층신경망락인입려파산법진행보상우화。실험결과험증료개진산법적우월성，기재정태조건하평균정위오차소우1．14m，동태근종정위정도야유흔대폭도적제고，구유실제응용개치。
Considering fingerprint database positioning algorithm based on signal strength indicator by Zigbee sensor network was only suitable for positioning under static conditions and more-polluted data could not be removed effectively causing unsatisfied positioning accuracy,particle filter was used to optimize the fingerprint database positioning algorithm based on the distribution of the noise characteristics,and harmful data polluted by noise was effectively removed,at the same time,dynamic precise loca-tion tracking of obj ectives was achieved.While the important distribution function of filtering was assumed to be Gaussian distri-bution in application,multilayer neural network was introduced in filter algorithm to compensate.Experimental result proved, the mean positioning error of improved algorithm was less than 1. 14m,and dynamic positioning accuracy of tracking process had also been greatly improved.