光伏板温度预测与仿真

Photovoltaic Board Temperature Prediction and Simulation

针对PV/T系统工作时气象因子的随机性和波动性造成PV/T组件温度预测精度降低的问题,充分考虑气象因子对预测精度的影响,建立了改进的BP神经网络预测模型。采用神经网络与遗传算法相结合,在分析组件温度与气象因素以及相邻时间序列温度相关性的基础上,提出了基于气象因子与组件历史温度相结合,提取出对PV/T组件温度影响较大的因子作为模型输入的同时优化神经网络结构的研究方法。以实验室光伏光热综合利用系统及其历史数据为研究对象,研究了各气象因子、组件历史温度等因子对PV/T组件温度的影响关系,对比分析了在晴天及多云天气下预测模型的预测效果,得到了较为明显的优化效果。研究结果表明,加入组件历史温度同时提取对组件温度影响最大的因子作为预测模型输入,同时优化神经网络结构能够有效地提高PV/T组件温度预测的精度。

Regarding to the problem that randomness and volatility of meteorological factors in PV/T system work- ing time cause low temperature prediction accuracy of PV/T components, and fully considering the impact of meteoro- logical factors on prediction precision, an improved BP neural network prediction model was established. By combing the neural network and genetic algorithm, and on the basis of analyzing the components temperature, meteorological factors and the correlation of adjacent time series, the paper proposes a research method which takes the factors that have greater influence on the component temperature as the input model and, at the same time, optimizes the neural network structure. Taking the photovohaic photo thermal utilizing system of the laboratory and its historical data as the research object, the paper studied the impact of all meteorological factors, historical temperature of components on PV/T components, compared and analyzed the predicting effect of predicting model in sunny and cloudy weather conditions. Then a rather distinct optimizing effect was obtained. The results show that adding the historical temperature of the components, the method can effectively improve the accuracy of PV/T component temperature prediction.