摘要
随着电力电子技术在新能源、高速机车、直流输电等大容量电力变换领域的广泛应用,其可靠性问题成为学术界和工业界的研究热点。基于因果模型的可靠性分析方法虽然在研究早期提供了原理层面的阐述,但在多物理场强耦合的复杂工况中的应用价值较低,而基于数据模型的可靠性研究方法尚处于初级阶段。该文提出将大数据处理技术应用于大容量电力电子系统可靠性研究的新思路,充分借助电力变换装置的海量历史运行数据,引入机器学习、数据挖掘和人工智能等大数据科学的最新理论成果,从数据泛在关系出发开展可靠性研究。论文详细阐述电力电子系统的大数据特征,并总结电力电子系统大数据应用中的若干关键技术。此外,论文从器件–装置–系统等多个层面深入分析大数据处理技术在大容量电力电子系统可靠性研究领域的潜在应用。大数据基本理论和方法的应用将提供一种通过数据关系间接理解原理本质的研究方法论的思维转变,有望为电力电子系统的可靠性研究开辟新途径。
Power electronics is widely employed in the renewable energy, high-speed traction drives and DC transmission, and its reliability has become an emerging technology in academia and industry. The causality model-based approaches threw light upon principles in the early period of reliability research, but are difficult to use in practice due to the multiple physics field coupling conditions in real applications. Meanwhile, the data-based modeling methods have a long way to go. A new research methodology from the perspective of data relationship was proposed in this paper, in which the big data processing technology is utilized so that the machine learning, data mining and artificial intelligence can be combined to deal with massive historical operation data of power converters. The typical features of big data in power electronics systems and several key techniques were also summarized. Additionally, the large-capacity power electronics systems reliability research with big data science will be at the device, converter and system levels. The applications of big data theory and technology in power electronics will provide a transformative method to understand the nature of the principle via data relationship, which can open up a new way for power electronics systems reliability enhancement.
出处
《中国电机工程学报》
EI
CSCD
北大核心
2017年第1期209-220,共12页
Proceedings of the CSEE
基金
国家自然科学基金重大项目(51490682)
国家重点基础研究发展计划项目(973项目)(2014CB247400)~~
关键词
大容量
电力电子系统
数据模型
大数据处理
large-capacity
power electronics systems
data-based models
big data processing
