考虑风-光-荷不确定性的数据驱动型机组组合模型

doi:10.3969/j.issn.2097-0706.2022.01.003

摘要/Abstract

摘要：

传统的概率分布模型在不确定性建模方面主要依赖于模型的选择,但假定的单一模型往往无法准确刻画随机量的复杂变化规律。除此之外,参数概率模型对随机量之间时间、空间相关性的描述不够,而基于连接函数等相关性建模方法在描述多个随机量的相关性方面又过于复杂,给模型实际应用增加了难度。通过采用一种数据驱动的不确定性建模方法,提出了一种数据驱动下的两阶段机组组合模型,基于非参数狄利克雷过程高斯混合模型（DPGMM）和变分贝叶斯推断（VBI）方法来描述风电、光伏和负荷的不确定性,考虑了多风电场和各节点负荷间的相关性,并采用传统数学优化方法对所构建的机组组合模型进行求解。最后,在IEEE-30节点算例系统上进行了仿真验证,结果表明,采用DPGMM模型能够较好地拟合随机量的概率分布并描述随机量之间的相关性。

关键词: 风电, 光伏, 负荷, 不确定性, 随机量, 数据驱动, 机组组合, 狄利克雷过程, 高斯混合模型

Abstract:

The traditional probability distribution model mainly relies on the choice of model in uncertainty modelling, but the assumed single model often cannot accurately describe the complex variations of random quantities. In addition, the parametric probability model does not adequately describe the temporal and spatial correlations between random variables, while the correlation modelling methods like Copula functions are too complicated in describing the correlations of multiple random variables, which adds difficulty to the practical applications of the model. Taking a data-driven uncertainty modelling method, a data-driven two-stage unit commitment model is proposed. The non-parametric Dirichlet process Gaussian mixture model （DPGMM） and the variational Bayesian inference （VBI） method are used to describe the uncertainty of wind power, photovoltaics and load. Taking the correlations between multiple wind farms and the load of each node into consideration, the traditional mathematical optimization method is applied to solve the established unit commitment model. Finally, the simulation verification is carried out on the IEEE-30 node test system. The results show that the DPGMM model can fit the probability distribution of random quantities and describe the correlation between them.

Key words: wind power, PV, load, uncertainty, random variable, data-driven, unit commitment, DP, GMM

中图分类号:

TK01⁺9:TM732

石立宝, 翟放. 考虑风-光-荷不确定性的数据驱动型机组组合模型[J]. 综合智慧能源, 2022, 44(1): 18-25.

SHI Libao, ZHAI Fang. Data-driven unit commitment model incorporating the uncertainty of wind-PV-load[J]. Integrated Intelligent Energy, 2022, 44(1): 18-25.

图/表 12

图1

图2

图3

图4

表1

表2

图5

图6

图7

图8

表3

表4

参考文献 17

[1]	TUOHY A, MEIBOM P, DENNY E, et al. Unit commitment for systems with significant wind penetration[J]. IEEE Transactions on Power Systems, 2009, 24(2):592-601. doi: 10.1109/TPWRS.2009.2016470
[2]	WANG J, SHAHIDEHPOUR M, LI Z. Security-constrained unit commitment with volatile wind power generation[J]. IEEE Transactions on Power Systems, 2008, 23(3):1319-1327. doi: 10.1109/TPWRS.2008.926719
[3]	POZO D, CONTRERAS J. A chance-constrained unit commitment with an n-K security criterion and significant wind generation[J]. IEEE Transactions on Power systems, 2013, 28(3):2842-2851. doi: 10.1109/TPWRS.2012.2227841
[4]	胡杰, 唐静, 谢仕义, 等. 基于大数据技术的电厂设备状态评估和预警应用研究[J]. 华电技术, 2020, 42(2):1-6.
	HU Jie, TANG Jing, XIE Shiyi, et al. State assessment and early warning application for power plant equipment based on big data technology[J]. Huadian Technology, 2020, 42(2):1-6.
[5]	王凤, 周铁梁. 基于大数据的大型装备全生命周期管理研究[J]. 华电技术, 2020, 42(2):12-16.
	WANG Feng, ZHOU Tieliang. Explore the value of data for lifecycle management of large equipment[J]. Huadian Technology, 2020, 42(2):12-16.
[6]	张徐亮, 万里冰, 钱伟中, 等. 基于区块链的电力大数据安全保障体系[J]. 华电技术, 2020, 42(8):68-74.
	ZHANG Xuliang, WAN Libing, QIAN Weizhong, et al. Security assurance system for electric power big data based on blockchain technology[J]. Huadian Technology, 2020, 42(8):68-74.
[7]	NING C, YOU F. Data-driven adaptive robust unit commitment under wind power uncertainty: A bayesian nonparametric approach[J]. IEEE Transactions on Power Systems, 2019, 34(3):2409-2418. doi: 10.1109/TPWRS.59
[8]	鲁卓欣, 徐潇源, 严正, 等. 不确定性环境下数据驱动的电力系统优化调度方法综述[J]. 电力系统自动化, 2020, 44(21):172-183.
	LU Zhuoxin, XU Xiaoyuan, YAN Zheng, et al. Overview on data-driven optimal scheduling methods of power system in uncertain environment[J]. Automation of Electric Power Systems, 2020, 44(21):172-183.
[9]	赵鑫, 郑文禹, 侯智华, 等. 基于粒子群优化算法的多能互补系统经济调度研究[J]. 华电技术, 2021, 43(4):14-20.
	ZHAO Xin, ZHENG Wenyu, HOU Zhihua, et al. Research on economic dispatch of multi-energy complementary system based on Particle Swarm Optimization[J]. Huadian Technology, 2021, 43(4):14-20.
[10]	ZHENG X, CHEN H. Data-driven distributionally robust unit commitment with Wasserstein metric: Tractable formulation and efficient solution method[J]. IEEE Transactions on Power Systems, 2020, 35(6):4940-4943. doi: 10.1109/TPWRS.59
[11]	ZHANG C, BÜTEPAGE J, KJELLSTRÖM H, et al. Advances in variational inference[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2018, 41(8):2008-2026. doi: 10.1109/TPAMI.34
[12]	SUN W, ZAMANI M, HESAMZADEH M R, et al. Data-driven probabilistic optimal power flow with nonparametric Bayesian modeling and inference[J]. IEEE Transactions on Smart Grid, 2019, 11(2):1077-1090. doi: 10.1109/TSG.5165411
[13]	DRAXL C, CLIFTON A, HODGE B M, et al. The wind integration national dataset (WIND) toolkit[J]. Applied Energy, 2015, 151:355-366. doi: 10.1016/j.apenergy.2015.03.121
[14]	SENGUPTA M, XIE Y, LOPEZ A, et al. The national solar radiation data base (NSRDB)[J]. Renewable and Sustainable Energy Reviews, 2018, 89:51-60. doi: 10.1016/j.rser.2018.03.003
[15]	NYISO. New York independent system operator: Load data[EB/OL]. [2021-07-20]. https://www.nyiso.com/load-data.
[16]	ULLAH Z, WANG S, RADOSAVLJEVIĆ J, et al. A solution to the optimal power flow problem considering WT and PV generation[J]. EEE Access, 2019, 7:46763-46772.
[17]	BISWAS P P, SUGANTHAN P, AMARATUNGA G A. Optimal power flow solutions incorporating stochastic wind and solar power[J]. Energy Conversion and Management, 2017, 148:1194-1207. doi: 10.1016/j.enconman.2017.06.071

概率模型	2006-06-10	2006-06-11
Gamma	58.06	214.01
Weibull	61.48	215.12
Lognorm	58.29	198.14
GMM-1	58.44	198.14
GMM-2	72.75	144.04
GMM-3	108.20	247.10
GMM-4	107.48	231.95
DPGMM	115.53	251.63

概率模型	2006-06-10	2006-06-11
Gamma	134.78	81.80
Weibull	129.60	81.08
Lognorm	135.00	98.45
GMM-1	134.81	234.95
GMM-2	111.03	122.70
GMM-3	30.13	35.44
GMM-4	20.80	27.91
DPGMM	18.57	20.09

机组编号	启停计划(01:00—24:00)
G1	1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
G2	1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
G3	1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
G4	0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
G5	0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0
G6	0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

穿透率/%	负荷波动率/%
穿透率/%	10	15	20
10	1.198 7×10⁴	1.283 9×10⁴	1.356 4×10⁴
20	1.029 3×10⁴	1.112 0×10⁴	1.193 7×10⁴
30	9.042 0×10³	9.850 1×10³	1.056 0×10⁴
40	8.348 0×10³	9.117 2×10³	9.743 0×10³