贝叶斯统计在测量数据拟合处理中的应用  

The Application of Bayesian Statistic in Data Fit Processing

作　　者：管兴胤[1] 李真富[1] 宋朝晖[1] 

机构地区：[1]西北核技术研究所,西安710024

出　　处：《核电子学与探测技术》2010年第4期503-505,523,共4页Nuclear Electronics & Detection Technology

摘　　要：分析了在测量数据处理过程中广泛使用的最小二乘拟合的合理性及其缺点,并给出了贝叶斯统计应用在测量数据拟合处理中的原理和一般方法,即贝叶斯拟合。分析表明,贝叶斯拟合方法避免了最小二乘拟合对测量数据的限定性假设,比最小二乘拟合具有更宽的适用范围,其拟合结果具有科学、直观的特点,可以很好地替代最小二乘拟合应用在数据处理中。The rationality and disadvantage of least squares fitting that is usually used in data processing is analyzed , and the theory and commonly method that Bayesian statistic is applied in data processing is shown in detail. As it is proved in analysis, Bayesian approach avoid the limitative hypothesis that least squares fitting has in data processing, and the result has traits that it is more scientific and more easily understood, may replace the least squares fitting to apply in data processing.

关 键 词：最小二乘拟合 贝叶斯统计 马尔可夫链蒙特卡罗方法 

分 类 号：TB115[理学—数学]