摘要
为了构建准确的风电功率时间序列模型,提高风电功率的建模精度,本文提出一种基于状态数决策模型的马尔科夫链-蒙特卡洛(MCMC)法。首先,对原始功率序列进行滤波处理,利用Metropolis-Hastings算法抽样生成风电功率状态序列,提高风电建模的计算效率和精度,进而根据生成的功率状态序列,利用前一时刻的功率值叠加波动量及噪声,提高生成风电功率序列的相关性;其次,根据两种评价指标构建状态数决策模型,确定最优风电功率,避免人工选取状态数难以获取最优生成功率的缺陷;最后,以宁夏某风电场为例,对比分析生成风电功率的不同特性及不同抽样方法,该方法生成的风电功率序列在各评价指标上均优于现有的方法,能更好地复现历史功率的数据特征。
In order to establish an accurate time series model of wind power and raise the accuracy of the model,an Markov chain Monte Carlo(MCMC)method based on the decision model of state number is proposed.First,the original power sequence is filtered,and the power state sequence is generated by Metropolis-Hastings sampling to improve the calculation efficiency and accuracy of wind power modeling.Secondly,according to the generated state sequence of wind power,the power value of the previous time is used to superimpose fluctuation quantity and noise,which improves the correlation of the generated wind power sequence.Finally,the decision model of state number is constructed according to the two evaluation indexes to determine the optimal wind power,avoiding the defect that it is difficult to obtain the optimal wind power sequence by manually selecting state number.Finally,taking a wind farm in Ningxia as an example,the different characteristics of wind power generation and different sampling methods are compared and analyzed.The wind power sequence generated by this method is superior to the existing methods in each evaluation index,and can better reproduce the data characteristics of historical power sequence.
作者
李娇
杨伟
LI Jiao;YANG Wei(School of Automation,Nanjing University of Science and Technology,Nanjing 210094)
出处
《电气技术》
2022年第1期70-77,共8页
Electrical Engineering
基金
国家电网公司科技项目资助(JSDL-XLFW-SQ-2016-10-092)。
