相关系数平稳过程方法

工程实际中遇到的随机过程,其均值和方差多数都随时间而变化,因此,无法用传统的平稳随机过程理论描述。文中通过对大量随机过程实测数据的分析研究,发现它们的相关系数函数基本上不随时间平移而改变,即相关系数函数是平稳的。在此基础上建立一种新的相关系数平稳随机过程理论和方法,并给出相关系数平稳正态过程的极大似然估计。从而在性能测试中,采用本文方法只要测定一个样本函数就能够对产品性能的均值和方差进行估计,与传统的需要测定一组样本函数的成组试验法相比,可以节省大量试验。

METHOD FOR CORRELATION COEFFICIENT STATIONARY PROCESS

The means and variances of random processes in the engineering vary with time in the majority, therefore they can not be solved by the traditional stationary random process theory. In this paper,a large number of test data from random processes have been analyzed and researched, and their functions of correlation coefficient are found not to change basically when the time intervals are identical. A new theory of correlation coefficient stationary process is established, and a method of maximum likelihood estimate for the correlation coefficient stationary normal process is presented on this foundation. This method has been used in the performance tests of aircraft and satellite successfully. It shows that a great number of specimens and time have been saved.