摘要
传统传感器测量数据处理方法只采用算术平均值的数字滤波法,虽然这种方法具有一定的抗干扰性,但从统计理论和实际应用情况分析来看,这种方法处理的数据不是测量结果的最好表示,尤其对于多传感器测量情况甚至更糟糕.针对这种情况提出了通过2次利用偏度分析建立动态检测门限判别并剔除粗差,再进行数据融合的分批估计方法.数据分析结果表明,这样处理后的数据测量误差和方差均更小,测量结果更接近测量真值.
The method of digital filter by arithmetic mean is usually used to preprocess the datum ameasured by sensors. Although such a method is good for preventing interference, it is not best way to show the result from the theory of statistics and the actual application, especially under the condition of multi-sensor. Thus, a new method to preprocess the datum measured by multi-sensor is put forward, by which the data that contains gross error is judged and deleted by using a dynamic threshold and the double skewness analysis, and then data fusion is finished by means of division estimation. The data analysis shows that measure error and mean square error of data got by the method are both lesser than that of the traditional one, and the measured results approach the true data more compared with the that of the method of digital filter.
出处
《海军工程大学学报》
CAS
北大核心
2005年第4期80-82,96,共4页
Journal of Naval University of Engineering
关键词
预处理
多传感器
偏度
数据融合
preprocessing
multi-sensor
skewness
data fusion
