| 王祥祯,李亚鹏,程春田,李刚,崔长跃.基于逆优化的邻接对手水电站参数反推模型及求解方法[J].水利学报,2025,56(6):759-770 |
| 基于逆优化的邻接对手水电站参数反推模型及求解方法 |
| An inverse optimization-based parameter estimation model and solution method for neighboring rival’s hydropower station |
| 投稿时间：2024-09-29 |
| DOI：10.13243/j.cnki.slxb.20240628 |
| 中文关键词: 电力市场 梯级水电 逆优化 双层优化 微粒子群算法 |
| 英文关键词: electricity market cascade hydropower inverse model bilevel optimization micro particle swarm optimization |
| 基金项目:辽宁省自然科学基金博士启动项目（2023-BSBA-066）；国家自然科学基金重点项目（52039002）；中央高校基本科研业务费项目（DUT24BS046） |
| 作者 | 单位 | E-mail | | 王祥祯 | 大连理工大学 水电与水信息研究所, 辽宁 大连 116024
大连市清洁能源高效利用与电力市场研究中心, 辽宁 大连 116024 | | | 李亚鹏 | 大连理工大学 水电与水信息研究所, 辽宁 大连 116024
大连市清洁能源高效利用与电力市场研究中心, 辽宁 大连 116024 | yplidut@foxmail.com | | 程春田 | 大连理工大学 水电与水信息研究所, 辽宁 大连 116024
大连市清洁能源高效利用与电力市场研究中心, 辽宁 大连 116024 | | | 李刚 | 大连理工大学 水电与水信息研究所, 辽宁 大连 116024
大连市清洁能源高效利用与电力市场研究中心, 辽宁 大连 116024 | | | 崔长跃 | 大连理工大学 水电与水信息研究所, 辽宁 大连 116024
大连市清洁能源高效利用与电力市场研究中心, 辽宁 大连 116024 | |
|
| 摘要点击次数: 39 |
| 全文下载次数: 23 |
| 中文摘要: |
| 梯级水电具有高度耦合的水力电力联系,上游电站的下泄流量直接影响下游电站发电和竞价计划的可行性,尤其当上下游归属不同利益主体时,上游电站从自利角度出发制定发电决策,会给下游电站带来高度不确定性。针对该问题,本文研究了一类由两个利益主体管理的三座电站的局部水电梯级,基于可观测的上游电站下泄流量、发电量等历史序列数据,提出了一个反推其运行参数及历史序列数据的逆优化模型,并根据工程经验确定待定参数的可行域。该模型为双层优化模型,其上层模型负责确定目标电站的主要运行参数,下层模型利用上层模型给定的参数模拟目标电站的自利调度过程,天然径流在流域内的分配比例也融入到了待推断参数中。为求解这一具有高度非凸及多变量特征的数学模型,本文提出了基于改进微粒子群算法的变维搜索算法。算例表明,所提反推模型和求解算法可准确推断上游邻接电站的运行参数和历史序列数据,推断误差在2%以内,所提求解算法比原始算法有明显提升,且始终保持了较强的搜索能力。仿真实验和统计分析也论证了本文所提模型及求解算法的可靠性。 |
| 英文摘要: |
| Cascade hydropower stations have a highly coupled hydro connection, where the discharge from upstream stations directly affects the feasibility of power generation and bidding plans for downstream stations, especially when the upstream and downstream stations belong to different entities. Upstream stations make generation decisions based on their own interests, which brings high uncertainty to downstream stations. To address this issue, this paper studies a local hydropower cascade consisting of three stations. Based on the observable historical behavior data of the upstream stations, an inverse optimization model is proposed to estimate the operational parameters and historical data sequences, with the feasible domain of the undecided parameters determined by engineering experience. This model is a bilevel optimization model, where the upper-level model determines the main operational parameters of the objective station, and the lower-level model uses the parameters provided by the upper-level model to simulate the self-interested scheduling process of the objective station. The allocation ratio of natural runoff within the basin is also integrated into the parameters to be inferred. To solve the model with highly non-convex and multi-variable mathematical characteristics, a variable-dimensional search method based on particle swarm optimization is proposed. Numerical examples demonstrate that the proposed inverse estimating model and variable-dimensional search method can accurately estimate the operational parameters and historical series data of upstream neighboring stations, with an estimation error within 2%. The proposed solving method consistently maintains strong search capabilities during the search process, showing significant improvement over the original algorithm. Simulation experiments and statistical analysis further validate the reliability of the proposed model and solution algorithm. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|