文章摘要
李常刚,李华瑞,刘玉田,等.计及低频减载动作的最大暂态频率偏移快速估计[J].电力系统自动化. DOI: 10.7500/AEPS20180630003.
LI Changgang,LI Huarui,LIU Yutian, et al.Fast Estimation of Maximum Transient Frequency Deviation Considering Under-frequency Load Shedding[J].Automation of Electric Power Systems. DOI: 10.7500/AEPS20180630003.
计及低频减载动作的最大暂态频率偏移快速估计
Fast Estimation of Maximum Transient Frequency Deviation Considering Under-frequency Load Shedding
DOI:10.7500/AEPS20180630003
关键词: 电力系统  频率偏移  低频减载  支持向量机  集成学习  特征降维
KeyWords: Power systems  Frequency deviation  Under-Frequency Load Shedding (UFLS)  Support vector machine (SVM)  Ensemble learning  Feature dimension reduction
上网日期:2019-04-11
基金项目:国家重点研发计划
作者单位E-mail
李常刚 电网智能化调度与控制教育部重点实验室山东大学 lichgang@sdu.edu.cn 
李华瑞 电网智能化调度与控制教育部重点实验室山东大学 lihuarui_sdu@163.com 
刘玉田 电网智能化调度与控制教育部重点实验室山东大学 liuyt@sdu.edu.cn 
吴海伟 国网江苏省电力有限公司 wuhaiwei@js.sgcc.com.cn 
徐春雷 国网江苏省电力有限公司 xuchunlei@js.sgcc.com.cn 
摘要:
      随大容量远距离直流输电和大规模可再生能源接入,受端电网频率安全风险增大。针对大容量直流闭锁等可能触发低频减载的严重扰动,提出基于机器学习的电力系统最大暂态频率偏移快速估计方法。将问题分解为低频减载响应判断和最大频率偏移估计两个子问题,通过子模型交替求解估计最大暂态频率偏移;基于支持向量回归方法构建最大频率偏移估计子模型,以支持向量机为个体学习器构建基于Bagging集成学习的低频减载响应判断子模型;以运行方式和扰动信息为输入,采用ReliefF算法和主成分分析法对运行方式特征进行选择和提取,降低模型复杂度。以某多直流馈入受端系统为例构建最大暂态频率偏移估计模型,验证所提方法准确性和适应性。
Abstract:
      With integration of large capacity High Voltage Direct Current (HVDC) transmission and large-scale renewable generation, risk of receiving-end power systems frequency security is increasing. For such severe disturbances as large-scale HVDC blocking which may trigger Under-Frequency Load Shedding (UFLS), this paper presents a method for estimating maximum transient frequency deviation based on machine learning. The problem is decomposed into two sub-problems to estimate execution of UFLS and maximum frequency deviation, respectively. The maximum transient frequency deviation is estimated by solving the sub-models alternately. The support vector regression method is used to establish the maximum frequency deviation estimation sub-model, the Bagging ensemble learning method based on support vector machine is used to establish the UFLS response judgement sub-model. With operating condition and disturbance information as inputs, ReliefF method is introduced to select important features, and principal component analysis is used to extract features into lower dimension to reduce model complexity. With a receiving-end power system in China with multi HVDC links as example, transient frequency deviation estimation models are built to verify accuracy and adaptability of the proposed method.
查看全文   查看/发表评论  下载PDF阅读器