面向变电站综合控制的分布式光伏发电并网策略研究  

Research on distributed photovoltaic power generation grid connection strategy for substation integrated control

作　　者：李宝强[1,2] 卢阳 LI Baoqiang;LU Yang(Guilin University of Electronic Technology,Guilin 541000,China;Guangxi Longyuan Renewables Co.,Ltd.,Nanning 530000,China)

机构地区：[1]桂林电子科技大学,广西桂林541000 [2]广西龙源新能源有限公司,广西南宁530000

出　　处：《现代电子技术》2024年第22期113-118,共6页Modern Electronics Technique

基　　金：广西龙源新能源有限公司智慧化企业建设项目(BJTY-ZC-2023-0251)。

摘　　要：传统比例积分控制策略的控制效果不佳、精度较低,因此以分布式光伏发电为基础的变电站在光伏强度突变、环境温度波动等现象的影响下,会出现系统并网电流谐波畸变率高、系统并网成功率低、负荷预测以及调度准确率差等缺陷。针对上述问题,设计一种自适应滑模控制策略。通过引入滑模观测器代替传统的比例积分控制,提升控制器对外界扰动的抵抗能力,从而提高系统的抗扰性。同时采用长短期记忆神经网络算法辅助滑模观测器进行并网控制,形成自适应滑模控制策略,对发电侧和用户侧的数据进行采集与预测。以分布式光伏变电站的数字仿真模型为基础,与比例积分、传统滑模控制进行同工况对比实验,结果表明,系统的并网成功率可达95.62%,负荷预测与调度准确率均提升至95%以上。The traditional proportional integral control strategy has poor control effectiveness and low control accuracy.Therefore,substations based on distributed photovoltaic power generation may suffer from high harmonic distortion rate of system grid connected current,low success rate of system grid connection,and poor accuracy of load forecasting and scheduling under the influence of sudden changes in photovoltaic intensity and environmental temperature fluctuations.On this basis,an adaptive sliding mode control strategy is designed.By introducing a sliding mode observer instead of traditional proportional integral control,the controller's resistance to external disturbances is improved,thereby enhancing the system's anti-interference ability.A long short-term neural network algorithm is used to assist the sliding mode observer in grid connection control,forming an adaptive sliding mode control strategy to collect and predict data on the power generation and user sides.Based on the digital simulation model of distributed photovoltaic substations,the comparative experiments with proportional integral and traditional sliding mode control under the same operating conditions are conducted,and the results show that the success rate of the system grid connection can reach 95.62%,and the accuracy of load prediction and scheduling can be improved to over 95%.

关 键 词：分布式光伏系统 变电站 滑模观测器 长短期记忆神经网络 并网成功率 负荷预测 

分 类 号：TN322.8-34[电子电信—物理电子学] TM615[电气工程—电力系统及自动化]