多传感器数据融合的概率权方法  

Weight of probability method for multi-sensor data fusion

作　　者：万树平[1] 

机构地区：[1]江西财经大学信息管理学院,江西南昌330013

出　　处：《传感器与微系统》2009年第8期12-13,17,共3页Transducer and Microsystem Technologies

基　　金：国家自然科学基金资助项目(10626029);江西省自然科学基金资助项目(0611082);江西省教育厅科技计划资助项目(GJJ08350)

摘　　要：针对多传感器对某一特性指标进行多次测量实验的数据融合问题,提出了一种基于概率基本理论的融合方法。该方法把各传感器的测量数据作为一随机变量,利用测量值的样本数据,确定测量值随机变量的取值范围,估计出样本均值和样本标准差,从而得到各传感器数据融合的概率权,并给出数据融合公式。应用实例验证了该方法的有效性和稳健性。Aimed at data fusion of multi-sensor experiment of many times on some characteristic index, a new fusion method is proposed based on the basic theory of probability. The measured data of every sensor is taken by the method as a random variant. According to the samples of measured data, the range of the random variant for the measured value is determined, and the sample mean and standard variance are estimated. The weight of probability is derived and the formula of data fusion is presented. The applied example proves the effectiveness and stableness of the method.

关 键 词：多传感器 数据融合 特征指标 概率权 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置]