关键词: GM（1,1）模型, 马尔可夫模型, 优化, 呼伦贝尔市, 年降水量, 预测

Abstract: By using annual precipitation data of Hulunbuir grassland from 2002 to 2016, taking the method of Weighted slip optimization, unbiasing grey model improvement and residual sequence correction to modify the GM（1,1） model, the Grey-Markov combination model of sliding unbiased residuals correction is constructed. The results show that after comparing the actual precipitation of Hulunbuir in 2017 and 2018, finding the combined model is ideal in the degree of prediction fitting and accuracy, it can be used to reasonably predict the annual precipitation of Hulunbuir. In 2019—2021, the annual precipitation of Hulunbuir will increase. The most likely in 2019 will be in accurate assessment and overestimation state, with a probability of 41% and 38% respectively. However, in 2020 and 2021 is the overestimated state, that is, the actual annual precipitation in these two years will be lower than the predicted result of the prediction model with the same probability of 42%. On the whole, the annual precipitation in Hulunbuir will be abundant in the future, which will play a positive role in promoting the ecological development of Hulunbuir grassland.

Key words: GM（1,1） model, Markov model, optimize, Hulunbuir city, annual precipitation, forecast