• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文OA文献 >Scenario generation and reduction for long-term and short-term power system generation planning under uncertainties
【2h】

Scenario generation and reduction for long-term and short-term power system generation planning under uncertainties

机译：不确定情况下长期和短期电力系统发电计划的方案生成和减少

页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

This dissertation focuses on computational issues of applying two-stage stochastic programming for long-term and short-term generation planning problems from the perspective of scenario generation and reduction. It follows a three-paper format, in which each paper discusses approaches to generating probabilistic scenarios and then reducing the substantial computational burden caused by a huge number of scenarios for different applications in power systems.The first paper investigates a long-term generation expansion planning model with uncertain annual load and natural gas price. A two-stage stochastic program is formulated to minimize the total expected expansion cost, generation cost and penalties on unserved energy while satisfying aggregated operational constraints. A statistical property matching technique is applied to simulate plausible future realizations of annual load and natural gas price over the whole planning horizon. To mitigate the computational complexity of a widely used classic scenario reduction method in this context, we firstly cluster scenarios according to the wait-and-see solution for each scenario and then apply the fast forward selection (FFS) method.The second paper prepares a basis for load scenario generation for the day-ahead reliability unit commitment problem. For the purpose of creating practical load scenarios, epi-splines, based on approximation theory, are employed to approximate the relationship between load and weather forecasts. The epi-spline based short-term load model starts by classifying similar days according to daily forecast temperature as well as monthly and daily load patterns. Parameters of the epi-spline based short-term load model are then estimated by minimizing the fitted errors. The method is tested using day-ahead weather forecast and hourly load data obtained from an Independent System Operator in the U.S. By considering the non-weather dependent load pattern in the short-term load model, the model not only provides accurate load predictions and smaller prediction variances in the validated days, but also preserves similar intraday serial correlations among hourly forecast loads to those from actual load.The last paper in this dissertation proposes a solution-sensitivity based heuristic scenario reduction method, called forward selection in recourse clusters (FSRC), for a two-stage stochastic day-ahead reliability unit commitment model. FSRC alleviates the computational burden of solving the stochastic program by selecting scenarios based on their cost and reliability impacts. In addition, the variant of pre-categorizing scenarios improves the computational efficiency of FSRC by simplifying the clustering procedure. In a case study down-sampled from an Independent System Operator in the U.S., FSRC is shown to provide reliable commitment strategies and preserve solution quality even when the reduction is substantial.
机译：本文从情景的产生和减少的角度，针对长期和短期发电计划问题，应用两阶段随机规划的计算问题。它遵循三篇论文的格式，其中每篇论文都讨论了生成概率情景的方法，然后减少了针对电力系统中不同应用的大量情景所导致的大量计算负担。第一篇论文研究了长期发电扩展计划年负荷和天然气价格不确定的模型。制定了一个两阶段的随机计划，以在满足汇总操作约束的同时，将总的预期扩展成本，发电成本和对无用能源的罚款降至最低。应用统计属性匹配技术来模拟整个计划范围内可能出现的年负荷和天然气价格的未来实现。为了减轻这种情况下广泛使用的经典场景缩减方法的计算复杂性，我们首先根据每种场景的观望解决方案对场景进行聚类，然后应用快速前向选择（FFS）方法。提前可靠性单元承诺问题的负荷方案生成的基础。为了创建实际的载荷场景，基于近似理论的落落样条被用于近似载荷与天气预报之间的关系。基于Epi-spline的短期负荷模型通过根据每日预报温度以及每月和每日负荷模式对相似日期进行分类开始。然后，通过使拟合误差最小化，估计基于Epi-spline的短期负荷模型的参数。该方法使用日间天气预报和从美国独立系统运营商获得的每小时负荷数据进行了测试。通过考虑短期负荷模型中与天气无关的负荷模式，该模型不仅可以提供准确的负荷预测，而且体积更小验证日内的预测方差，但也保持了小时预测负荷与实际负荷之间的日内序列相关性。本文的最后一篇论文提出了一种基于解决方案敏感度的启发式情景约简方法，称为资源簇中的前向选择（FSRC） ，用于两阶段随机日前可靠性单位承诺模型。 FSRC通过根据方案的成本和可靠性影响来选择方案，从而减轻了求解随机程序的计算负担。此外，预分类方案的变体通过简化聚类过程提高了FSRC的计算效率。在从美国独立系统运营商处向下采样的案例研究中，FSRC被证明可提供可靠的承诺策略，即使减少的幅度很大，也可以保持解决方案的质量。

著录项

  • 作者

    Feng Yonghan;

  • 作者单位
  • 年度 2014
  • 总页数
  • 原文格式 PDF
  • 正文语种 en
  • 中图分类

相似文献

  • 外文文献
  • 中文文献
  • 专利

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号