考虑季节变化对负荷特征影响的电力系统短期负荷预测

Short-Term Load Forecasting of Power System Considering Seasonal Variation Impact on Load Characteristic

随着微电网用电设备的复杂化,产生的环保、经济效益问题日益突出,电力负荷的短期预测对于区域的精细化调度至关重要。当前的负荷预测方法缺乏对不同区域季节性变化因素的表征,导致预测精度较低。提出了一种基于改进的自适应噪声完备集合经验模态分解一霜冰优化算法——双向循环神经网络(ICEEMDAN-RIME-BiGRU)考虑季节差异的短期负荷预测方法。首先,采用ICEEMDAN方法对四季的电力负荷进行分解;其次,结合RIME算法的软霜搜索策略、硬霜穿刺机制和正向贪婪选择机制,分别学习不同季节下电力负荷的分量特征,实现对BiGRU模型的参数寻优,并将特征分量输入网络模型,所得结果相加得到时间序列预测值;最后,以某地区微电网的负荷数据为例进行算例分析。结果显示,所提出的方法相较于其他3种典型相关预测方法,对于区域季节性差异对负荷的影响具有显著的表征能力,可以提升负荷预测精度。

With the increasing complexity of microgrid electrical equipment,environmental and economic issues are becoming more prominent.Short-term load forecasting is crucial for the refined regional dispatching.Current load forecasting methods often lack characterization of seasonal variation factors across different regions,resulting in lower prediction accuracy.Therefore,this paper proposes a short-term load forecasting method based on ICEEMDAN-RIME-BiGRU that considers seasonal differences.Firstly,the ICEEMDAN method is used to decompose the electrical load into seasonal components.Secondly,integrating the soft frost search strategy,hard frost puncture mechanism,and forward greedy selection mechanism of the RIME algorithm,the method learns the characteristics of load components in different seasons to optimize the parameters of the BiGRU model.These feature components are then input into the network model,and the results are aggregated to obtain the time series forecast values.Finally,a case study using load data from a microgrid in a specific region is conducted.The results demonstrate that compared to three other typical correlation forecasting methods,the proposed method effectively characterizes the impact of seasonal differences on load,thereby improving load forecasting accuracy.