基于马尔科夫的小麦干热风年型预测

Prediction of Dry-hot Wind Annual Pattern for Wheat Based on Markov

摘要: 基于马尔科夫原理“无后效性”特点，利用河南省滑县2001—2021年每年5月13日—6月10日的气象数据，建立马尔科夫小麦干热风年型预测模型。通过回代检验对滑县2004—2021年干热风进行预测结果检验，选取相同的数据与BP神经网络模型进行结果对比。结果表明，马尔科夫模型预测概率77.78%，精度较高，并且在相同的数据基础上比BP神经网络预测模型的表现更好，因此马尔科夫模型可以更好地对小麦干热风进行预警，可以起到防灾抗灾的效果，对提高小麦产量具有重要意义。

Abstract: Based on "no aftereffect" characteristic of Markov principle, a Markov model for predicting annual pattern of wheat dry hot wind was established using meteorological data from May 13 to June 10, 2001 to 2021 in Hua County, Henan Province．Through backpropagation testing, prediction results of dry-hot wind in Hua County from 2004 to 2021 were tested, and the same data was selected for comparison with BP neural network model．Results showed that the Markov model has a prediction probability of 77.78%, it had high accuracy, and performed better than the BP neural network prediction model on the same data basis．Therefore, the Markov model could better warn against wheat dry and hot winds, and could play a role in disaster prevention and resistance, which was of great significance for improving wheat yield．