数据挖掘技术在精细化温度预报中的应用

摘要

简要介绍了精细化天气预报和气象数据挖掘应用的现状,在对BP神经网络预测方法详细分析的基础上,研究了基于时间序列数据挖掘实现精细化温度预报的方法。该方法基于时序分析技术,建立起适合于BP神经网络的输入样本模型,通过反复学习从温度时序中建立预测模型,将其用于未来24 h的精细化温度预报。同时,对BP神经网络算法和步骤做了简要介绍,针对原有的BP算法存在的不足,做了一些改进。最后,通过对预测挖掘系统的设计和在Matlab6.5仿真平台上的试验,建立了温度预报模型,以兰州市观测站数据为时间序列研究对象,对精细化温度预报进行了仿真实现。对基于时序的数据挖掘理论的应用和开发精细化温度预报方法做了有益的探索。%The paper introduces the domestic and international current situation about the development of refined weather forecast and data mining application.On the basis of detailed analysis of BP neural network forecasting method,this paper researches a data mining method based on time series analysis technology which can be used on refined temperature forecast.This method can build an input sample pattern which is suit for the BP neural networks of data mining and finally establish a predictive model by studying temperature time series again and again,which used for the next 24 hours refined temperature forecast.At the same time,a brief introduction of the algorithm and steps of the BP neural network is given out in the paper,and some further improvement is made aiming at the deficiency of the original BP algorithm.Finally,through design data mining system and test on the Matlab6.5 simulation platform,the temperature forecast model was established.This study had done some helpful exploration on application of data mining theory based on time series analysis technology and developed method of refined weather forecast.