High-Fidelity, High-Performance Multi-Stage Transmission Planning with Spatio-Temporal Uncertainty Models

利用时空不确定性模型进行高保真、高性能多级传输规划

• 批准号: 1709094
• 负责人: Pascal Van Hentenryck
• 金额: $ 43.12万
• 依托单位: Regents of the University of Michigan - Ann Arbor
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2017
• 资助国家: 美国
• 起止时间: 2017-07-01 至 2019-04-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1709094&HistoricalAwards=false
• 关键词: Fidelity; Performance; Multi; Stage; Transmission

The transmission system of the United States, mostly built in the 1960s and 1970s, provides the backbone infrastructure to deliver electricity with great reliability to consumers all over the country. However, the drive towards renewable energy, power electronics, and changes in fuel costs are fundamentally altering the energy landscape. Significant cost and environmental benefits would result from a holistic study of transmission planning at the national scale, based on a high-fidelity modeling of the grid and accurate forecasting models for renewable energy. The goal of this project is to realize a step change in the fidelity, scalability, and performance of transmission planning systems. The research will have scientific as well as broad societal and educational impact. The project will also pioneer methodologies for solving important problems in energy optimization and management, and generate both short-term and long-term technical impacts. Moreover, the research results will be disseminated through education initiatives. Several education plans, including promoting K-12 education and participation of female and underrepresented minority groups in science and engineering, will be undertaken through involvement in various initiatives at the University of Michigan. This research proposes a new generation of transmission planning systems to amplify the benefits of renewable energy, while addressing the challenges created by increasing stochasticity and power electronics. The project relies on high-fidelity models of the grid and novel, hierarchical predictive models for renewable energy and substation loads that capture complex spatio-temporal correlations that are critical in obtaining realistic characterizations of uncertainties. The project brings together four PI and Co-PIs with their expertise in power system optimization, uncertainty quantification, algorithm design, and large-scale distributed computing. It proposes multi-stage stochastic programs over various risk and robustness measures for transmission planning and adopts a prioritization methodology to express planner preferences as the uncertainties are being revealed. It is expected to achieve high computational performance through the use of convex relaxations, large neighborhood search, and parallel implementations of decomposition algorithms. The proposed algorithms will be evaluated on real test cases offered by the largest transmission operator in Europe, ranging from 2,000 to 20,000 buses, as well as synthetic versions from these test cases that are adapted to the realities of the United States.

美国的输电系统大多建于20世纪60年代和70年代，为向全国各地的消费者输送可靠性极高的电力提供了骨干基础设施。然而，转向可再生能源、电力电子和燃料成本的变化正在从根本上改变能源格局。以电网的高保真建模和可再生能源的准确预测模型为基础，在国家范围内对输电规划进行全面研究将产生显著的成本和环境效益。该项目的目标是实现输电规划系统在保真度、可扩展性和性能方面的阶段性变化。这项研究将产生科学和广泛的社会和教育影响。该项目还将开创解决能源优化和管理中重要问题的方法，并产生短期和长期的技术影响。此外，研究成果将通过教育举措传播。将通过参与密歇根大学的各种倡议，实施若干教育计划，包括促进K-12教育以及女性和代表不足的少数群体参与科学和工程。这项研究提出了新一代输电规划系统，以扩大可再生能源的好处，同时应对随机性和电力电子设备增加带来的挑战。该项目依赖于电网的高保真模型，以及可再生能源和变电站负荷的新颖分层预测模型，这些模型捕捉复杂的时空关联，这些关联对于获得不确定因素的现实特征至关重要。该项目汇集了四个PI和Co-PI，他们在电力系统优化、不确定性量化、算法设计和大规模分布式计算方面具有专业知识。该方法针对输电规划中的各种风险和稳健性指标，提出了多阶段随机规划，并在不确定性暴露时，采用了优先排序方法来表达规划者的偏好。通过使用凸松弛、大邻域搜索和分解算法的并行实现，它有望获得高的计算性能。提议的算法将在欧洲最大的输电运营商提供的真实测试用例以及这些测试用例的合成版本上进行评估，测试用例的范围从2,000到20,000辆公交车，这些测试用例适应美国的现实。

项目成果

Distributionally robust facility location problem under decision-dependent stochastic demand

• DOI: 10.1016/j.ejor.2020.11.002
• 发表时间: 2021-03-07
• 期刊: EUROPEAN JOURNAL OF OPERATIONAL RESEARCH
• 影响因子: 6.4
• 作者: Basciftci, Beste;Ahmed, Shabbir;Shen, Siqian
• 通讯作者: Shen, Siqian

Multistage distributionally robust mixed-integer programming with decision-dependent moment-based ambiguity sets

• DOI: 10.1007/s10107-020-01580-4
• 发表时间: 2020-02
• 期刊: Mathematical Programming
• 影响因子: 2.7
• 作者: Xian Yu;Siqian Shen