基于小样本数据差分扩容的微电网负荷预测方法

Method of Load Forecasting in Microgrid Based on Differential Expansion of Small Sample Data

面向新能源微电网负荷预测,针对中长期负荷预测中样本数据不足导致预测精度不高的问题,提出了一种新的数据扩容方法,将原始数据样本两两差分作为新数据样本,通过设置阈值进行数据筛选,在扩充数据样本的同时保证数据的准确性,分析了基于差分运算的数据扩容方法对于负荷数据不确定性的消除效果。考虑气象、人口等影响因素,利用多元回归分析拟合各种关联因素对负荷的影响,将原始数据负荷预测模型作为对照,通过比较预测结果与实际负荷之间的误差,验证了数据扩容方法的实用性和准确性。

For load forecasting in renewable energy microgrid,aiming at the problem that the prediction accuracy is not high because of the lack of sample data in mid-and long-term load forecasting.This paper proposes a new data expansion method.The original data sample is used as a new data sample,and the data is filtered by setting a threshold.The data sample is expanded while ensuring the accuracy of the data,and the data based on the difference operation is analyzed.The capacity expansion method can eliminate the uncertainty of load data.Considering the influencing factors such as meteorology and population,using multiple regression analysis to fit the impact of various related factors on the load,the original data load forecasting model was used as a control,and practicality and accuracy of the expansion method in this paper was verified by comparing the error between the predicted result and the actual load.