摘要
本文利用较实用的时变自回归（ＡＲ）模型，来拟合动态测试（尤其是长过程测试）中的二阶非平衡数据，并且不采用当前广泛使用的递推算法，而探讨应用移动Ｍａｒｐｌｅ算法来拟合该时变模型。经各种类型数据的仿真试验及某些实例的计算结果表明，该方法具有与常用递推算法同样的时变跟踪能力，而无明显的滞后现象，且拟合精确度较高、稳定性好，只是计算速度略低。可见对于非平稳数据不便应用成批算法的观点，需作重新认识。
In this paper, a more practical time varying AR model is applied to fit the second order nonstationary data in dynamic measurement. And this model is also flited by using moving Marple's algorithm instead of the widely applied recursive algorithm. Simulation test of assorted data and calculation of some examples shows: compared with the recursive algorithm, moving Marple's algorithm not only has the same tracking ability of time varying without lag, but also has a higher calculation speed.Therefore, what generally believed that nonstationary data cannot be managed with batch processing,but only with recursive algorithm should be reconsidered.
出处
《计量学报》
CSCD
1994年第2期92-98,共7页
Acta Metrologica Sinica
关键词
AR模型
Marple算法
几何量
测试
Second order nonstationarity
Time varying AR model
Recursive algorithm
Moving Marple's algorithm
