太阳逐时总辐射混沌优化神经网络预测模型研究

根据影响太阳逐时总辐射的气象、地理等方面因素的分析以及对太阳逐时总辐射历史数据的相关性分析,确定了建立太阳逐时总辐射的神经网络预测模型输入因素项。根据全年最大可照时数统一了太阳逐时总辐射各天的历史数据,并对宝山气象站的太阳逐时总辐射建立了混沌优化神经网络预测模型(CONN),编制了计算机程序。模型输出反映了太阳逐时总辐射的变化规律,预测结果也足够准确。

STUDY OF CHAOS OPTIMIZATION NEURAL NETWORKS FOR THE FORECAST OF HOURLY TOTAL SOLAR IRRADIATION

According to the knowledge of the influences on the hourly total solar irradiation from meteorological, geograph-ical, and the correlation analyses of the historical data sequence of such irradiation, the input terms to the artificial neural network were determined for the forecast of hourly total solar irradiation. The modeling of a chaos optimization neural network (CONN) was carried out to forecast the hourly total solar irradiation of Baosan Meteorological Station in Shanghai with a supposed base of unified sunshine duration, and the computer program was established. The CONN model has given forecasts that reflect the actual patterns of the curves of hourly total solar irradiation. The forecasted results are accurate enough.