摘要
在新型电力系统稳步发展与“双碳”目标持续构筑背景下,“源荷互动新模式”成为电力系统能源转型的必然趋势,以多样性、灵活性为主要特征的电力负荷作为电力系统的重要组成部分,其预测场景分析及预测模型研究对新型电力系统的运行、维护和规划至关重要。为深入研究人工智能背景下负荷预测领域取得的进展与突破,以新型电力系统负荷预测为着眼点,总结归纳当今负荷预测必要性与实用性,分类介绍五个典型负荷预测场景,针对基于人工智能技术的智能负荷预测算法模型进行系统化分析,结合机器学习中的神经网络、深度学习算法、集成学习算法对比单一预测模型及组合预测模型的特点,详细阐述各类模型在负荷预测领域的应用现状,以期为“双碳”目标下新型电力系统源荷互动的新模式构建提供合理化参考。
With the steady development of the new-type power system and the continuous construction of the"dual carbon"targets,the"new interactive mode of source and load"has emerged as an inevitable trend in the energy transformation of the power system.The power load,characterized by diversity and flexibility,is an important component of the power system.Research on its prediction scenario analysis and prediction model is crucial for the operation,maintenance,and planning of the new power system.In order to conduct in-depth research on the progress and breakthroughs made in the field of load forecasting under the background of artificial intelligence,we focus on load forecasting in the new-type power system.We summarize the necessity and practicality of current load forecasting and introduced five typical load forecasting scenarios in a classified manner.In addition,we conduct a systematic analysis on the load forecasting model based on artificial intelligence,and compare the characteristics of single forecasting models and combination forecasting models with artificial neural network,deep learning algorithms and ensemble learning algorithms in machine learning.Furthermore,we elaborate on the current application status of various models in the field of load forecasting,aiming to provide a reasonable reference for the construction of a new mode of source load interaction in the new-type power system within the framework of the"double carbon"targets.
作者
杨佳泽
王灿
王增平
YANG Jiaze;WANG Can;WANG Zengping(School of Electrical and Electronic Engineering,North China Electric Power University,Beijing 102206,China;State Grid Beijing Electric Power Company,Beijing 100301,China)
出处
《华北电力大学学报(自然科学版)》
北大核心
2025年第3期54-67,共14页
Journal of North China Electric Power University:Natural Science Edition
基金
国家电网公司总部科技项目(5100-202199529A-0-5-ZN)。
关键词
新型电力系统
人工智能
神经网络
负荷预测
机器学习
深度学习
new-type power system
artificial intelligence
neural networks
load forecasting
machine learning
deep learning
