NWP卡尔曼滤波光伏功率的预测模型

Photovoltaic power prediction model based on NWP Kalman filter

针对超短期光伏功率预测问题,提出一种基于NWP的卡尔曼滤波光伏功率预测模型。将太阳总辐照度视为系统状态变量,光伏发电功率视为系统观测量,利用实测数据拟合确定预测模型的观测方程,引入激励噪声作为控制方程输入量,确定系统差分控制方程,在递推迭代过程中,应用测量功率数据实时修正和限制系统剩余白噪声,实现光伏功率的超短期预测。结果表明,模型的预测误差除个别畸点幅值较大外,其余点误差都在6%以内,天气晴朗时在2%以内,呈现出较好的准确性。

This paper proposes a photovoltaic power prediction model based on NWP Kalman filter in response to ultra short-term photovoltaic power prediction.The study consists of regarding the total solar irradiance as the state variable of the system and taking the photovoltaic power as the system observation measurement and determining the observation equation of the prediction model by fitting the measured data,and then introducing the excitation noise as the input of the control equation to determine the differential control equation of the system;and in the process of recursive iteration,performing real time correction and limitation of system residual white noise using measured power data to realize the ultra short-term prediction of photovoltaic power.The results show that the model demonstrates a better accuracy despite the prediction error within 6%except for the larger amplitude of some abnormal points,and the other points with the error of less than 6%,and less than 2%with clear weather.