| 引用本文: | 徐沈智,艾小猛,邹佳芯,张舒捷,李湃,黄越辉,文劲宇.优选状态数的MCMC算法在风电功率序列生成中的应用[J].电力自动化设备,2019,39(5): |
| XU Shenzhi,AI Xiaomeng,ZOU Jiaxin,ZHANG Shujie,LI Pai,HUANG Yuehui,WEN Jinyu.Application of optimizing state number Markov chain Monte Carlo algorithm in wind power generation[J].Electric Power Automation Equipment,2019,39(5): |
|
| 摘要: |
| 传统马尔科夫链蒙特卡洛(MCMC)法的状态数选择常依赖于人工经验，应用于风电功率序列建模时难以较好地同时模拟原始风电功率序列的概率分布特性和自相关特性。针对该问题，提出一种优选状态数的MCMC(OSN-MC)算法。首先给出MCMC方法状态数的选取范围，其次在该范围内以生成序列与原始序列的自相关函数的误差平方和最小为原则确定优选状态数，然后利用各状态对应功率范围内的累积分布函数抽样生成随机风电功率，提高优选状态数下生成风电功率序列对于原始序列分布特性的模拟精度。应用OSN-MC法和MCMC法对中国、美国和欧洲的12个风电场生成风电功率序列，并与原始实测序列进行特性比较，结果表明：OSN-MC法生成的风电功率序列对原始序列的分布特性和自相关特性的模拟效果均优于MCMC法所生成的风电功率序列。 |
| 关键词: 风电功率 马尔科夫链 蒙特卡洛法 优选状态数 序列生成 |
| DOI：10.16081/j.issn.1006-6047.2019.05.009 |
| 分类号:TM614 |
| 基金项目:国家电网有限公司科技项目(考虑新能源发电不确定性的随机优化调度关键技术研究与示范) |
|
| Application of optimizing state number Markov chain Monte Carlo algorithm in wind power generation |
|
XU Shenzhi1, AI Xiaomeng1, ZOU Jiaxin2, ZHANG Shujie2, LI Pai3, HUANG Yuehui3, WEN Jinyu1
|
|
1.Hubei Electric Power Security and High Efficiency Key Laboratory, State Key Laboratory of Advanced Electromagnetic Engineering and Technology, School of Electrical and Electronic Engineering, Huazhong University of Science and Technology, Wuhan 430074, China;2.Key Laboratory of Photovoltaic Power Generation and Grid Integration, State Grid Qinghai Electric Power Research Institure, Xining 810008, China;3.State Key Laboratory of Operation and Control of Renewable Energy & Storage Systems, China Electric Power Research Institute, Beijing 100192, China
|
| Abstract: |
| The state number selection of traditional MCMC(Markov Chain Monte Carlo) method often depends on personal experience, it is difficult to simulate the probability distribution and autocorrelation characteristics of original wind power series at the same time when the method is applied in wind power series modelling, for which, an OSN-MC(Optimizing State Number Markov Chain Monte Carlo) algorithm is proposed. Firstly, the state number selection range of MCMC method is given. Secondly, the optimal state number is determined based on the principle of minimizing the square sum of autocorrelation function error of generated series and original series. Then the cumulative distribution function in the corresponding power range of each state is adopted to generate random wind power, which improves the simulation accuracy of generated wind power series to the probability distribution of original series. The OSN-MC and MCMC methods are applied to generate wind power series for 12 wind farms in China, the United States and Europe, and the comparison with the original measured series shows that the simulation effect of wind power series generated by OSN-MC algorithm on the distribution and autocorrelation characteristics of original series is better than that of MCMC algorithm. |
| Key words: wind power Markov chain Monte Carlo method optimizing state number series generation |
