基于动态参数等效的风电场并网系统振荡稳定性评估的数据驱动方法  

A Data-driven Method for Oscillation Stability Assessment of Grid-connected Wind Farm Based on Dynamic Parameters Equivalent

作　　者：陈剑 杜文娟 王海风 CHEN Jian;DU Wenjuan;WANG Haifeng(State Key Laboratory of Alternate Power Systems with New Energy Resources(North China Electric Power University),Changping District,Beijing 102206,China;School of Electric Engineering,Sichuan University,Chengdu 610065,Sichuan Province.China)

机构地区：[1]新能源电力系统国家重点实验室(华北电力大学),北京市昌平区102206 [2]四川大学电气工程学院,四川省成都市610065

出　　处：《中国电机工程学报》2022年第19期6958-6972,共15页Proceedings of the CSEE

基　　金：国家自然科学基金项目(52077144)。

摘　　要：开环模式谐振理论为风机接入电力系统引起的振荡稳定性提供了清晰的机理分析基础。然而,如何利用实测数据来评价风电场并网后电力系统的振荡稳定性仍然是一个悬而未决的问题。针对多风电场并网电力系统的振荡稳定性问题,该文提出一种基于等效系统和数据驱动的振荡稳定性评估方法。该方法能够在全系统参数不确定情况下,实现对大规模风电场系统的动态等效,利用联合分布适配算法,在等效系统中离线建立数据驱动的振荡稳定评估模型,并泛化到实际系统中,实现在线评估。通过典型仿真系统算例与实际大规模风电场算例,结合多方面的对比测试,验证所提方法的有效性和适应性,证明建立的数据驱动模型具有优良的评估性能。所提方法通过深度学习,将评估模型的在线计算负担转移到离线训练中,对实现基于量测的大型风电场振荡风险评估具有重要意义。Open-loop modal resonance theory provides a clear mechanism analysis for the oscillation stability caused by wind turbine generators(WTGs) connected to power system.However, how to use measured data to evaluate the oscillation stability in a real power system with large-scale wind farm connected still remains an open issue. For the oscillation stability in a multi-wind-farm connected power system, this paper proposed an equivalent-system-based and data-driven method for oscillation stability assessment(OSA). This method could achieve dynamic equivalence of large-scale wind farm systems in the case of some uncertain parameters of the full system, use the joint distributed adaptation algorithm to establish a data-driven oscillation stability evaluation model offline in the equivalent system, and generalize it to online evaluation in the real system. Through a typical simulation system example and a real large-scale wind farm example, and combined with various comparative tests, the efficacy and adaptability of the proposed approach for OSA were demonstrated, proving the excellent evaluation performance of the established data-driven model. The method proposed in this paper can transfer the burden of online calculation of the evaluation model to offline training through deep learning,which is of great significance for realizing the measurementbased OSA of large wind farms.

关 键 词：数据驱动 振荡稳定性 等效系统 开环模式谐振 

分 类 号：TM744[电气工程—电力系统及自动化]