基于EMD-ANFIS模型的短期光伏出力模拟研究

Short-term output prediction of photovoltaic power based on EMD-ANFIS model

为降低光伏发电并网对电网运行造成的不稳定影响,亟需进一步提高光伏出力预测的精度。本文以ANFIS模型为基础,结合EMD的数据预处理器技术,提出了改进的EMD-ANFIS模型。以龙羊峡光伏电站为研究对象,采用光伏电站实测小时出力数据,建立短期光伏出力模拟的ANFIS模型和EMD-ANFIS模型。实例计算结果表明:(1)ANFIS模型和EMD-ANFIS模型模拟结果的确定性系数分别为0.93和0.96,模拟结果均为甲等;(2)相比ANFIS模型,EMD-ANFIS模型适应能力更强,精度更高,能够更好的拟合不同气象条件下的光伏出力曲线,推荐为光伏出力预测的应用模型。研究成果可提高光伏出力的预测精度,有助于确保大规模光电并网运行的安全性与可靠性,提高绿色能源利用效率,促进光伏产业的发展。

For maintaining the safety and stability of network operation caused by photovoltaic power generation paralleling in the Grid,it is necessary to improve the output prediction precision of photovoltaic power.In this paper,based on data preprocessing technology,the Adaptive Networks based Fuzzy Inference System（ANFIS）model is improved by the Empirical Mode Decomposition（EMD）.Taking aphotovoltaic power station,Longyangxia,for example,short-term photovoltaic power prediction models,ANFIS model and EMD-ANFIS model are established based on measured hour output data in the past.It is demonstrated that：（1）the deterministic coefficients of ANFIS model and EMD-ANFIS model are 0.93 and 0.96,respectively,both of which are first-class models;（2） The EMD-ANFIS model is better than the ANFIS model,both in adaptability and perdition accuracy,which can fit the photovoltaic power output curve better under different weather conditions.For greatly improved the perdition accuracy,the EMD-ANFIS model is recommended on output prediction of photovoltaic power.The results are of significance in maintaining the safety and stability of network operation by photovoltaic power generation paralleling in the Grid,improving the efficiency of green energy and the developing of photovoltaic power industry.