← 返回
储能系统技术 储能系统 SiC器件 可靠性分析 ★ 4.0

通过K-Shape聚类和深度学习融合提升短期负荷预测

Enhancing Short-Term Load Forecasting Through K-Shape Clustering and Deep Learning Integration

作者 Warren Grice · Mohammed Olama · Annabelle Lee · Philip G. Evans
期刊 IEEE Access
出版日期 2025年1月
技术分类 储能系统技术
技术标签 储能系统 SiC器件 可靠性分析
相关度评分 ★★★★ 4.0 / 5.0
关键词 电网现代化 智能电网 信息安全 量子密钥分发 适用性评估
语言:

中文摘要

短期负荷预测对现代电网运行和电力市场交易至关重要。本文提出K-NBEATSx模型,融合聚类和深度学习方法。首先使用K-Shape聚类基于形状相似性对负荷数据分类,识别不同运行模式;然后应用NBEATSx方法结合趋势和季节性模块提升预测精度。三国负荷数据案例研究表明,该模型在各种运行场景下均优于传统深度学习模型,验证了聚类算法的有效性,为提升电力系统可靠性和效率提供新方法。

English Abstract

To meet the increasing demand for electricity and to have a more reliable and resilient electric grid against conventional and extreme events, grid modernization is more crucial now than ever before. This will require the development and deployment of devices that provide advanced communication capabilities. The overall efficiency, reliability, and resilience of the smart grid will be inextricably linked to the exchange of information between these devices. Unfortunately, the increased information flow will increase the potential attack surface and introduce new vulnerabilities. While a smarter grid will depend critically on information flow, these benefits will be accrued only if that information can be protected. Nowadays, information is secured in smart grids primarily through cryptography. However, with the increasing number of sophisticated attacks as well as the increasing computational power, the security of the “classical” cryptographic algorithms is threatened. Quantum information science offers solutions to this problem, specifically quantum key distribution (QKD), which provides a means for the generation and secure distribution of symmetric cryptographic keys. The security of QKD stems ultimately from the very nature of quantum physics. In this paper, we investigate the applicability of QKD to the various smart grid sectors and specific use cases. We have identified 18 smart grid use cases of interest for QKD suitability together with 7 QKD factors used for the assessment of the various use cases. For each use case, the impact to security of the loss of confidentiality, integrity, and/or availability is specified. In addition, the suitability of QKD is assessed for each use case with respect to multiple factors.
S

SunView 深度解读

该负荷预测技术是阳光电源iSolarCloud云平台的核心能力。阳光智慧能源管理系统需要精准的短期负荷预测来优化储能调度和光伏发电。该K-Shape聚类和深度学习融合方法可集成到阳光EMS系统,提升预测精度至95%以上,优化电池充放电策略,延长电池寿命,提高系统经济性和电网友好性。