频域分解和深度学习算法在短期负荷及光伏功率预测中的应用

Applications of Frequency Domain Decomposition and Deep Learning Algorithms in Short-term Load and Photovoltaic Power Forecasting

由于光伏发电和负荷的时变特性,光伏发电功率的消纳存在动态的过程,提升精细化的光伏及负荷预测技术对配电网的实时调度运行至关重要。该文在负荷和光伏发电精确预测的基础上,对光伏消纳能力进行分析。首先提出一种基于频域分解的短期负荷预测方法;其次,应用基于孤立森林和长短期记忆神经网络,预测短期光伏发电功率;然后,分析负荷特性及光伏发电特性;最后,结合消纳指标对安徽省某地级市国庆期间进行消纳能力分析。实验结果表明,所提出的短期负荷及光伏预测方法可达到理想的预测效果,消纳分析结果为该地区配电网火电机组调配提供参考意义。

Due to the time-varying characteristics of photovoltaic power generation and load, the photovoltaic power consumption is a dynamic process. The improvement and refinement of the photovoltaic and the load forecasting technology is critical to the distribution networks real-time dispatching. Based on the accurate prediction of load and photovoltaic power generation, this paper analyzed the photovoltaic consumption capacity. Firstly, a method of the short-term load forecasting was proposed, based on frequency domain decomposition. Secondly, the short-term photovoltaic power generation forecasting was applied via the isolation forest and the long short-term memory neural network algorithms. Then the load and photovoltaic power generation characteristics were analyzed. Finally, indicators of distributed PV capacity were proposed and calculated. The consumption capacity of a city in Anhui Province during the National Day was analyzed. Experimental results verify that the short-term load and photovoltaic forecasting methods can achieve the desired forecasting effect. The results of the consumption analysis provide a reference for the deployment of thermal power units in the distribution network.