A review on performance of artificial intelligence and conventional method in mitigating PV grid-tied related power quality events

详细信息    查看全文

• 作者：Ken Weng Kow^a ; ; Yee Wan Wong^a ; ^{kecx4kkg@exmail.nottingham.edu.my" class="auth_mail" title="E-mail the corresponding author} ; Rajparthiban Kumar Rajkumar^b ; Rajprasad Kumar Rajkumar^a
• 关键词：PV Grid Tied System ; Power quality ; Mitigation ; Artificial intelligence ; High penetration
• 刊名：Renewable and Sustainable Energy Reviews
• 出版年：2016
• 出版时间：April 2016
• 年：2016
• 卷：56
• 期：Complete
• 页码：334-346
• 全文大小：369 K

文摘

Integration of renewable energy resources into power networks is the trend in power distribution system. It is to reduce burden of centralized power plant and global emissions, increase usage of renewable energy, and diverse energy supply market. However, solar photovoltaic which is a type of renewable energy resource, is found to generate peak capacity for a short duration only. Next, its output is intermittent and randomness. In addition, it changes behavior of power distribution system from unidirectional to bidirectional. As a result, it causes different types of power quality events to the power networks. Therefore, these power quality events are urged to be mitigated to further explore the potential of solar photovoltaic system. This paper aims to investigate negative impacts of photovoltaic (PV) grid-tied system to the power networks, and study on performance of artificial intelligence (AI) and conventional methods in mitigating power quality event. According to the surveys, power system monitoring, inverter, dynamic voltage regulator, static synchronous compensator, unified power quality conditioner and energy storage system are able to compensate power quality events which are caused by PV grid-tied system. From the studies, AI methods usually outperform conventional methods in terms of response time and controllability. They also show talent in multi-mode operation, which is to switch to different operation modes according to the environment. However, they require memory to achieve abovementioned tasks. It is believed that unsupervised learning AI is the future trend as it can adapt to the environment without the need of collecting large amount of data before the AI is implemented.