梯度下降法在水库调度函数优化中的应用

doi:10.16180/j.cnki.issn1007-7820.2020.01.013

摘要/Abstract

摘要：

调度函数是制定水库中长期运行调度的常用方法,然而现行的通过回归分析法提取出的调度函数拟合精度不佳。针对这一问题,文中建立了以调度期内水库发电量最大为目标函数的优化调度模型,通过线性回归提取各时段水库调度函数系数作为初始值,再采用梯度下降法优化各时段调度函数系数,并在拟合精度差的时段引入流量分级策略,最终获得优化后的水库调度函数。实例计算结果表明,文中采用的方法不进可以提供精度较高的调度函数系数,还保证了优化调度的发电效益,证明了该方法的有效性和实用性。

关键词: 水库调度, 发电调度, 水库调度函数, 线性回归, 梯度下降法, 流量分级策略

Abstract:

The dispatching function is a common method for formulating long-term operation scheduling of reservoirs. However, the method used to extract the dispatching function by binary linear regression usually gets poor fitting accuracy. Aiming at this problem, the optimal scheduling model with the maximum generating capacity of the reservoir during the dispatching period was established. The reservoir scheduling function coefficients of each period were extracted as the initial values by linear regression. Then the gradient descent method was used to optimize the scheduling function coefficients of each period. The inflow classification strategy was introduced in the period with poor fitting accuracy, and the optimized reservoir scheduling function was finally obtained. The calculation results showed that the effective and practical method used in the paper provided higher precision scheduling function coefficients and ensured the optimal power generation efficiency of the dispatching.

Key words: reservoir operation, power generation scheduling, reservoir operation function, linear regression, gradient descent method, inflow classification strategy

中图分类号:

王炜,胡一鸣,石强. 梯度下降法在水库调度函数优化中的应用[J]. 电子科技, 2020, 33(1): 68-72.

WANG Wei,HU Yiming,SHI Qiang. The Application of Gradient Descent Method in Optimizing the Dispatching Function of Reservoirs[J]. Electronic Science and Technology, 2020, 33(1): 68-72.

图/表 6

表1

表2

图1

表3

表4

图2

参考文献 16

[1]	赵志鹏, 廖胜利, 程春田 , 等. 梯级水电站群中长期优化调度的离散梯度逐步优化算法[J]. 水利学报, 2018,49(10):1243-1253.
	Zhao Zhipeng, Liao Shengli, Cheng Chuntian , et al. Discrete gradient progressive optimality algorithm for mid-long-term optimal operation of multi-reservoir system[J]. Journal of Hydraulic Engineering, 2018,49(10):1243-1253.
[2]	徐斌, 卢庆文, 钟平安 , 等. 梯级水电站群联合优化调度函数的逐步回归模型[J]. 水力发电, 2017,43(8):99-103.
	Xu Bing, Lu Qingwen, Zhong Ping’an , et al. Stepwise regression model for extracting joint optimal operation rules of cascade hydropower stations[J]. Water Power, 2017,43(8):99-103.
[3]	张雪映, 吴小婷, Daniel Eliote Mbanze, 等. 基于SEOA算法的抽水蓄能梯级水库调度研究[J]. 电子科技, 2019,32(7):1-5.
	Zhang Xueying, Wu Xiaoting, Daniel Eiliote Mbanze , et al. Research on scheduling of pumped storage cascade reservoir based on SEOA algorithm[J]. Electronic Science and Technology, 2019,32(7):1-5.
[4]	李文武, 张雪映, Daniel Eliote Mbanze, 等. 基于SARSA算法的水库长期随机优化调度研究[J]. 水电能源科学, 2018,36(9):72-75.
	Li Wenwu, Zhang Xueying, Daniel Eliote Mbanze , et al. Reserch on long-term stochastic optimal operation of reservoir based on SARSA algorithm[J]. Water Resources and Power, 2018,36(9):72-75.
[5]	纪昌明, 喻杉, 周婷 , 等. 蚁群算法在水电站调度函数优化中的应用[J]. 电力系统自动化, 2011,35(20):103-107.
	Ji Changming, Yu Shan, Zhou Ting , et al. Application of ant colony algorithm for hydropower dispatching function optimization[J]. Automation of Electric Power Systems, 2011,35(20):103-107.
[6]	李承军, 陈毕胜, 张高峰 . 水电站双线性调度规则研究[J]. 水力发电学报, 2005(01):11-15,46.
	Li Chengjun, Chen Bisheng, Zhang Gaofeng . Study on bilinear dispatching rule of hydropower station[J]. Journal of Hydroelectric Engineering, 2005(1):11-15,46.
[7]	刘志刚, 胡斌奇, 伍永刚 , 等. 基于云模型的水库调度函数拟合方法研究[J]. 水电能源科学, 2017,35(3):53-56,23.
	Liu Zhigang, Hu Binqi, Wu Yonggang , et al. Research on reservoir operation function fitting method based on cloud model[J]. Water Resources and Power, 2017,35(3):53-56,23.
[8]	孙秋戎, 陈森林, 董建凡 , 等. 改进的RAGA在水库调度函数优化中的应用[J]. 中国农村水利水电, 2013(09):5-7,11.
	Sun Qiurong, Chen Senlin, Dong Jianfan , et al. The application of improved RAGA in optimizing the dispatching function of reservoirs[J]. China Rural Water and Hydropower, 2013(9):5-7,11.
[9]	郭玉雪, 方国华, 闻昕 , 等. 水电站分期发电调度规则提取方法研究[J]. 水力发电学报, 2018(12):1-13.
	Guo Yuxue, Fang Guohua, Wen Xin , et al. Study on deriving the staged dispatching rules of hydropower station[J]. Journal of Hydroelectric Engineering, 2018(12):1-13.
[10]	马锴, 于洋庆, 朱善迎 , 等. 基于梯度下降和一致性的电网分布式经济调度算法[J]. 中国科学:信息科学, 2018,48(10):1364-1380.
	Ma Kai, Yu Yangqing, Zhu Shanying , et al. Distributed algorithm for economic dispatch based on gradient descent and consensus in power grid[J]. Scientia Sinica:Informationis, 2018,48(10):1364-1380.
[11]	郭跃东, 宋旭东 . 梯度下降法的分析和改进[J]. 科技展望, 2016, 26(15): 115, 117.
	Guo Yuedong, Song Xudong . Analysis and improvement of gradient descent method[J]. Science and Technology 2016, 26(15): 115, 117.
[12]	刘方, 张粒子 . 流域梯级水电优化调度模型与方法研究综述[J]. 华北电力大学学报(自然科学版), 2017,44(5):81-90.
	Liu Fang, Zhang Lizi . Review on optimization scheduling model and method of cascaded hydropower stations[J]. Journal of North China Electric Power University(Natural Science Edition), 2017,44(5):81-90.
[13]	张睿, 张利升, 覃晖 , 等. 梯级水电站多目标兴利调度建模及求解[J]. 水电能源科学, 2016,34(6):39-42.
	Zhang Rui, Zhang Lisheng, Qin Hui , et al. Modeling and solution of multi-objective operation of cascade reservoirs[J]. Water Resources and Power, 2016,34(6):39-42.
[14]	周华艳, 周建中, 何中政 , 等. 基于烟花量粒子群算法的水库群联合优化调度[J]. 水电能源科学, 2018,36(10):84-87,38.
	Zhou Huayan, Zhou Jianzhong, He Zhongzheng , et al. Joint optimal operation of reservoir group based on hybrid fireworks quantum particle swarm algorithm[J]. Water Resources and Power, 2018,36(10):84-87,38.
[15]	陈洋波 . 水库群隐性随机优化调度研究[J]. 葛洲坝水电工程学院学报, 1994,16(2):45-52.
	Chen Yangbo . Research on the implicit stochastic optimization of multireservoir systems[J]. Journal of Gezhouba Institute of Hydro-Electric Engineering, 1994,16(2):45-52.
[16]	梁志明, 万飚, 陈森林 , 等. 基于流量分级策略的水电站分期调度规则提取方法[J]. 水力发电学报, 2016,35(1):8-18.
	Liang Zhiming, Wan Biao, Chen Senlin , et al. Derivation of staged operation rules for hydropower stations using inflow classification strategy[J]. Journal of Hydroelectric Engineering, 2016,35(1):8-18.

月份	常数项b	可用水量系数a	回归系数R²
5	-156.137	24.157	0.807 3
6	-214.208	25.541	0.859 5
7	-368.856	28.083	0.933 9
8	-288.737	23.613	0.820 1
9	-167.003	20.545	0.678 7
10	-154.713	21.818	0.585 8
11	-29.144	11.048	0.435 1
12	45.185	4.039	0.053 4
1	13.627	5.892	0.060 8
2	62.037	4.472	0.030 5
3	3.926	11.336	0.252 3
4	-95.989	20.998	0.707 5

月份	调度线类型	常数项b	可用水系数a	回归系数R²
12	上调度线	3.137	0.472	0.467 1
12	下调度线	-10.745	1.052	0.654 8
1	上调度线	-0.279	0.702	0.802 3
1	下调度线	1.311	0.494	0.466 6
2	上调度线	0.964	0.679	0.463 3
2	下调度线	-3.513	0.863	0.425 6
3	上调度线	3.448	0.417	0.515 1
3	下调度线	-6.446	0.669	0.953 6

月份	优化后常数项b	优化后可用水量系数a	回归系数R²
5	-176.181	24.729	0.846 4
6	-214.242	25.783	0.877 7
7	-368.887	28.525	0.964 9
8	-308.879	24.098	0.855 2
9	-187.086	21.347	0.733 1
10	-174.806	21.543	0.624 9
11	-50.243	12.072	0.537 1
4	-108.098	21.529	0.743 7

月份	调度线类型	常数项b	可用水系数a	回归系数R²
12	上调度线	2.795	0.499	0.522 2
12	下调度线	-10.947	1.064	0.669 9
1	上调度线	-0.289	0.702	0.803 4
1	下调度线	0.996	0.518	0.513 8
2	上调度线	0.559	0.713	0.511 4
2	下调度线	-4.416	0.866	0.516 1
3	上调度线	3.102	0.458	0.558 7
3	下调度线	-6.647	0.678	0.978 6