基于神经网络的太阳辐照度预测方法研究

Prediction method of solar irradiance based on neural network

以天数、时间、天气情况、温度作为输入数据,太阳辐照度作为输出数据,提出采用人工神经网络建立太阳辐照度预测模型,并通过这一模型进行太阳辐照度预测。通过在青海大学内进行的实验建立适用于西宁地区的太阳辐照度神经网络模型。并根据这一模型对2012年5月12日太阳辐照度情况进行了预测。结果表明,对于预测天气和预测温度与实际情况接近时,平均误差为10%左右。

A neural network was used to build a prediction model of the solar irradiance with the number of days,time,weather conditions,temperature as input data,solar irradiance as output data,and solar irradiance was predicted with the model.According to the model,an experiment conducted in Qinghai University in Xining to build the model and predict solar irradiance at May 12,2012.The results showed that the average error is about 10% when the predicted situation was close to the actual situation.