正态分布序列变点识别问题探讨

Discussion on Variable Point Identification of Normal Distribution Sequence

文章对含有变点的正态分布随机变量序列进行了研究。构造了正态分布多变点模型,运用ASAMC(退火随机近似蒙特卡洛)方法得到了变点的估计,通过随机模拟进行验证,模拟结果表明对于变点个数及其位置的识别精准。实证检验了乌鲁木齐市1973—2017年共45年间平均温度序列的温度变点。

This paper studies the sequence of normal distribution random variables with variable points. Firstly, the paper constructs a multi-point model of normal distribution, and then uses ASAMC (annealing random approximation Monte Carlo) method to obtain the estimation of variation points. Finally, a random simulation is made for verification. The result shows that the identification of the number and location of variation points is accurate. The paper also empirically tests the temperature variation points of average temperature series in Urumqi city during the period of 45 years from 1973 to 2017.