Page 39 - 《中国电力》2026年第4期
P. 39

第 59 卷 第 4 期                                                                          Vol. 59, No. 4
               2026 年 4 月                            ELECTRIC POWER                                   Apr. 2026

              引用格式:崔一宸, 王鹤, 王莉丽, 等. 基于改进生成对抗网络场景生成的跨流域水风光互补中长期优化调度方法[J]. 中国电力, 2026, 59(4):
              35−46.
              Citation: CUI Yichen, WANG He, WANG Lili, et al. Medium- and long-term optimal scheduling method for inter-basin hydro-wind-photovoltaic
              complementary systems based on scenario generation with improved generative adversarial network[J]. Electric Power, 2026, 59(4): 35−46.


                         基于改进生成对抗网络场景生成的跨流域


                                 水风光互补中长期优化调度方法


                                   崔一宸 ,王鹤 ,王莉丽 ,银涛 ,黄山松                                 2
                                                      1
                                            1
                                                                            2
                                                                  2
                     (1. 东北电力大学 电气工程学院,吉林 吉林 132012;2. 国网四川省电力公司,四川 成都 610041)
              Medium- and long-term optimal scheduling method for inter-basin hydro-wind-
                    photovoltaic complementary systems based on scenario generation with

                                     improved generative adversarial network
                                         1
                                                                 2
                                                     1
                                                                           2
                              CUI Yichen , WANG He , WANG Lili , YIN Tao , HUANG Shansong     2
                              (1. School of Electrical Engineering, Northeast Electric Power University, Jilin 132012, China;
                                     2. State Grid Sichuan Electric Power Company, Chengdu 610041, China)

              Abstract:  With  the  continuous  increase  of  wind-solar  energy base in the Southwest region as an example. The results
              penetration,  how  to  address  the  volatility  and  uncertainty  of  show that, compared with conventional methods, the proposed
              their power output and formulate reasonable medium- and long-  method improves the consumption level of new energy and the
              term  scheduling  strategies  has  become  the  core  challenge  for  power  supply  reliability  of  the  system  while  considering  the
              the  efficient  operation  of  inter-basin  hydro-wind-solar  hybrid  utilization  of  water  resources  in  basins;  the  comprehensive
              systems.  To  this  end,  this  study  firstly  proposes  a  wind-solar  power curtailment rate and power shortage risk are reduced by
              joint  scenario  generation  method  considering  spatio-temporal  1.53  percentage  points  and  24.5%  respectively,  achieving
              correlation.  By  improving  the  network  structure  of  the  collaborative  optimization  and  effective  balance  among
              Wasserstein  generative  adversarial  network  (WGAN),  the  multiple objectives.
              model  can  hierarchically  extract  the  spatio-temporal  coupling  This work is supported by Science and Technology Project of
              characteristics  of  wind-solar  power  output,  and  generate  a  SGCC (Research and Application of Wide-Area Complement-
              representative  joint  scenario  set.  On  this  basis,  with  the  ary  Optimal  Scheduling  Technology  for  Wind-Solar-Hydro
              objectives of maximizing the expected energy consumption of  Clean  Energy  Bases  with  a  High  Proportion  of  Hydropower,
              the  complementary  system,  minimizing  the  curtailment  and  No.4000-202426080A-1-1-ZN).
              power  shortage,  and  minimizing  the  loss  of  spilled  water,  an  Keywords:  scenario  generation;  uncertainty;  generative
              adaptive  weight  determination  method  based  on  feedback  adversarial network; hydro-wind-solar complementary system;
              regulation is proposed to dynamically adjust the priority of each  medium- and long-term optimal scheduling; dynamic weight
              objective  in  different  time  periods,  and  a  medium-  and  long-
              term  multi-objective  stochastic  optimization  model  for  inter-  摘 要:随着风光渗透率的持续增长,如何应对其出力
              basin  hydro-wind-solar  hybrid  systems  is  constructed.  A  波动性与不确定性并制定合理的中长期调度策略已成为
              simulation analysis is carried out by taking an inter-basin clean  跨 流 域 水 风 光 互 补 系 统 高 效 运 行 面 临 的 核 心 挑 战 。 为
                                                                此,首先提出一种考虑时空相关性的风光联合场景生成
              收稿日期:2025−10−14; 修回日期:2026−02−02。
                                                                方法,通过改进       Wasserstein  生成对抗网络的网络结构,实
              基金项目:国家电网有限公司科技项目(含高比例水电的                         现模型对风光出力时空耦合特征的分层提取,生成具有
              风光水清洁能源基地广域互补优化调度技术研究及应用                          代表性的联合场景集;在此基础上,以互补系统消纳电
              4000-202426080A-1-1-ZN)。                          量期望最大化、弃电与缺电量最小化及弃水损失最小化

                                                                                                           35
   34   35   36   37   38   39   40   41   42   43   44