基于多层感知机的智能软开关就地控制方法

Local Control Method of Soft Open Point Based on Multi-layer Perceptron

分布式电源在配电网中的高渗透率接入,直接加剧了系统功率波动和电压越限的风险,传统的调节手段难以满足控制要求。为了有效应对分布式电源出力的间歇性和不确定性,提出了一种基于多层感知机的智能软开关SOP(soft open point)就地控制方法。以降低网络损耗和减小电压偏差为目标,通过历史数据训练建立多层感知机模型,利用本地量测信息确定"电压-无功"曲线的参数,从而制定智能软开关的运行策略,实现配电网功率调节的快速响应和精细化控制。与集中式控制方法相比,基于"电压-无功"曲线的就地控制方法的通信和数据处理负担小,响应速度快,且具有较好的鲁棒性。最后,采用PG&E 69节点测试算例验证了所提方法的有效性。

With the high-penetration integration of distributed generations(DGs)into distribution network,the risks of system power fluctuations and voltage violations increase sharply. As a result,the conventional regulation means cannot meet the control requirements. To effectively cope with the intermittent and uncertain characteristics of output from DGs,a local control method of soft open point(SOP)based on multi-layer perceptron(MLP)is proposed in this paper.To reduce network loss and voltage deviation,an MLP model is trained with the historical data,and the parameters of Q-V curve are determined by using the local measurement information. In this way,the operation strategy of SOP can be formulated,thereby realizing the fast response and fine power regulation of distribution network. Compared with the centralized control methods,the local control method based on the Q-V curve has lower burden of communication and data processing,fast response,as well as better robustness. Finally,a modified PG & E 69-node system is taken as a test example,and results verify the effectiveness of the proposed method.