基于联合时序场景和改进TCN的高比例新能源电网负荷预测

Short-term Load Forecasting for Power System with High Proportion New Energy Based on Joint Sequential Scenario and Improved TCN

为充分挖掘新型电力系统建设过程中高比例新能源并网对负荷预测的影响,以风光负荷数据为研究对象,提出一种基于联合时序场景和改进型时间卷积网络的短期负荷预测方法。首先,基于3σ准则对风光负荷历史数据进行分析,剔除异常数据,然后应用联合时序场景刻画负荷需求与风光出力的相关性,分类出不同负荷预测场景。接着,利用随机森林算法进行负荷预测特征量提取,构建随机森林时间卷积网络(RF-TCN)预测模型,并采用Bootstrap算法对预测结果进行修正。最后,以甘肃省2022年数据为例进行仿真,并设置4种对比算例。仿真结果证明了所提方法的有效性,以期在新型电力系统建设过程中发挥积极作用。

In order to fully explore the influence of high proportion of new energy in the power grid on load forecasting,wind power solar power load data is used as the research object to forecast the load changes through a load forecasting method combining modified time convolution network(TCN)and joint sequential scenario.Firstly,the historical data is analyzed based on the 3σcriterion to eliminate the abnormal data,then the joint sequential scenario is applied to depict the correlation between load demand and power output of new energy,and classify different load prediction scenarios.Afterwards,load prediction feature extraction is performed based on random forest(RF)algorithm to construct RF-TCN network prediction model,and the prediction results are corrected by Bootstrap algorithm.Finally,the data of a region of Gansu province is taken for example simulation and four comparative examples are set.The simulation results prove the effectiveness of the proposed method and expect to play a positive role in the construction of the new power system.