基于扩大区间和改进矩阵的风电日内出力极端场景模拟
Simulation of Extreme Scenarios of Wind Power Intraday Output based on Expanded Sampling Intervals and Improved Sampling Matrices
罗倩 1侯佳辰 2刘亚茹 3喻冉 4梅亚东 5王现勋 3谭永杰
作者信息
- 1. 长江水利委员会长江中游水文水资源勘测局,湖北省 武汉市 430012
- 2. 长江水利委员会长江中游水文水资源勘测局,湖北省 武汉市 430012;长江大学资源与环境学院,湖北省 武汉市 430100
- 3. 长江大学资源与环境学院,湖北省 武汉市 430100
- 4. 中国葛洲坝集团股份有限公司,湖北省 武汉市 430033
- 5. 武汉大学水资源工程与调度全国重点实验室,湖北省 武汉市 430072
- 折叠
摘要
为模拟大波动、强随机的风电日内出力极端场景,提出一种基于扩大抽样区间和改进抽样矩阵的风电日内出力极端场景模拟方法.通过指定分位点扩大拉丁超立方抽样区间,通过量化风电波动程度和控制抽样矩阵最大波动幅度的方式改进抽样概率矩阵,从而进行逆运算抽样得到体现风电时序特征的场景集合;进而使用由单时段极端值、日均极端值和连续多时段极端值组成的指标组合筛选出风电日内出力的极端场景.以我国西北地区某风电场为对象进行了算例分析,结果表明:所提方法模拟的风电场景在仿真程度以及丰富度方面,优于蒙特卡罗模拟结果,且在风电出力极端场景模拟方面具有较好的模拟效果,为极端场景模拟提供了技术参考.
Abstract
To simulate the extreme intra-day wind power output scenarios characterized by large fluctuations and strong randomness,we propose a method based on expanding the sampling range and improving the sampling matrix.By specifying quantiles,the Latin hypercube sampling range is expanded,and the sampling probability matrix is improved by quantifying the degree of wind power fluctuation and controlling the maximum fluctuation amplitude of the sampling matrix.Subsequently,a set of scenarios reflecting the temporal characteristics of wind power is obtained through inverse sampling.Extreme intra-day wind power output scenarios are selected using a combination of indicators consisting of single-period extreme values,daily average extreme values,and continuous multi-period extreme values.A case study of a wind farm in the northwest region of China shows that the wind power scenarios simulated using our method outperform those simulated by the Monte Carlo method in terms of both simulation accuracy and scenario richness.Moreover,our method exhibits good performance in simulating extreme wind power output scenarios,providing a technical reference for extreme scenario simulation.
关键词
风电/场景模拟/极端场景/蒙特卡罗抽样/拉丁超立方抽样Key words
wind power/scenario generation/extreme scenarios/Monte Carlo sampling/Latin hypercubes sampling引用本文复制引用
出版年
2024
NETL
NSTL
万方数据