售电侧开放市场环境下基于多分位鲁棒极限学习机的短期负荷预测技术

该文基于极限学习机算法设计一种用于短期负荷预测的多分位鲁棒极限学习机模型,该模型能解决传统预测模型抗干扰能力差的缺陷,可以在面临不确定性因素干扰的情况下准确预测负荷。对传统模型和多分位鲁棒极限学习机模型的鲁棒性和多分位性进行验证,对比结果表明,多分位鲁棒极限学习机模型的鲁棒性更好,在不同分位下的预测精度更高。

In this paper,a multi-quantile robust extreme learning machine model for short-term load forecasting is designed based on the extreme learning machine algorithm.This model can solve the defect of poor anti-interference ability of the traditional forecasting model.The load can be predicted accurately in the face of uncertain factors.The robustness and multi-quantile of the traditional model and the multi-quantile robust limit learning machine model are verified.The comparison results show that the multi-quantile robust limit learning machine model has better robustness and higher prediction accuracy under different quantiles.