【作者】 卢有麟；

【导师】 周建中；

【作者基本信息】 华中科技大学 ， 水利水电工程， 2012， 博士

【摘要】 流域梯级大规模水电站群是一个开放的复杂巨系统,其联合优化运行不是孤立的,而是处在一定的自然环境和社会经济环境之中,受水文过程、用水需求、发电控制等因素影响,是一类高维、时变的大规模复杂决策优化问题,非线性、不确定性、多目标、多属性、多层次、多阶段是其根本特征,一直是水利科学与系统科学交叉发展的前沿问题之一。流域梯级电站群不仅担负着流域生活、生产、生态供水等重要任务,而且在防洪减灾、充分利用水能资源发电、流域水量调度以及缓解局部地区生态环境退化中发挥了关键作用,联合调度的潜力与效益巨大。然而,随着梯级水电站群规模的扩大、梯级拓扑结构的日趋复杂化以及调度目标的多元化,梯级电站之间的水力、电力联系日趋复杂,不同调度需求、调度目标之间存在的相互制约与竞争关系日益凸显,传统优化调度的理论与方法已难以适应有效发挥大规模梯级水电站群综合效益的工程需求,因此,亟需研究并发展新的优化理论与方法。本文围绕长江流域水能资源快速开发背景下大规模梯级水电站群的联合优化调度决策问题,以梯级枢纽综合效益的充分发挥为目标,通过引入复杂系统建模理论、群智能优化理论以及Pareto多目标优化与决策技术,对流域梯级大规模水电站群的多目标优化调度与多属性决策问题进行了深入的研究,取得了一些具有理论意义及工程应用价值的研究成果。通过系统集成和实地部署,部分研究成果在三峡梯调中心获得了工程应用。主要研究工作及创新成果如下：(1)围绕流域梯级大规模水电站群联合优化调度的工程需求,建立了长江中上游大规模控制性水电站群联合优化调度模型、金沙江下游梯级与三峡梯级联合蓄水优化调度模型以及区域电网水火电联合优化调度模型,并针对已有理论和模型应用于工程实际时模型求解方法的制约,首先研究了一种改进差分进化优化方法,实现了多重变量耦合和复杂约束优化模型求解技术的突破。该方法在差分进化寻优的理论框架中引入了混沌算子,实现了参数的自适应调整以及二次局部寻优,以此提高了算法的优化能力,有效避免了“早熟收敛”现象的发生。以长江中上游梯级水电能源系统为研究对象进行了实例研究,调度结果以及对比分析表明,该方法无论是在求解效率、求解精度还是处理多重复杂约束的能力等方面与现有方法相比,性能均有显著改善,为梯级水电站群联合优化调度问题的求解提供了一条新途径。(2)考虑到流域梯级电站群联合优化调度需要综合考虑多个调度目标,从本质上而言是一类复杂的多目标约束优化问题,研究工作通过引入Pareto优化的相关概念,研究并论述了多目标联合调度建模的理论基础,并针对多重约束优化模型的特点,提出了适用于多目标调度问题的智能优化方法。通过对改进差分进化算子的多目标拓展,研究了一种循环更新策略实现多目标非劣外部种群的规模控制,设计了一种基于Pareto向量对比的多目标混沌搜索策略,实现多目标优化问题的高效求解。为测试所提出方法的优化性能,采用一组模态各异的测试函数对其进行了性能测试与对比分析研究,数值仿真结果验证了所提出方法具有优异的多目标并行优化性能。(3)综合考虑金沙江下游梯级以及三峡梯级多目标运用的工程需求,通过分析梯级水电站群防洪、发电以及生态调度模式下各调度目标间的竞争与冲突关系,建立了流域梯级水电站群联合多目标防洪、发电以及生态调度模型,并运用改进多目标差分进化算法对模型进行了高效求解,快速得到了关于不同目标的非劣调度方案集,在此基础上,对各调度目标间的相互影响规律进行了解析,为调度方案最终的决策会商优选提供了技术支撑与数据支持。(4)为快速、科学、合理的评价各非劣调度方案,以确定多目标均衡调度的最优实施方案,研究工作将联合调度方案决策会商过程中的主、客观偏好信息纳入决策模型,同时结合Vague集理论对模糊信息的描述能力,研究了一种新的多属性决策方法。该方法通过构建属性重要性差异度矩阵实现了主观偏好的集结,提出一种改进熵权公式,从而准确计算出决策指标的客观权重；此外,通过计算各候选方案相对正、负理想方案综合贴近程度的Vague值,实现各候选方案的优劣排序。以所提出方法为决策评价的数学工具,对流域梯级水电站群的多目标联合防洪、发电以及生态非劣调度方案的评价优选问题进行了研究,实现了不同决策偏好、不同决策情景下各候选方案快速、科学的评价优选。更多还原

【Abstract】 As a complex nonlinear large scale system, the optimal operation of large scale cascade hydroelectric stations is a high dimensional constrained optimal multi-objective problem with complex characteristics such as strong-coupling of the decision variables and nonlinear objective functions. This problem is affected by large amount of complex hydrological and electrical factors, thus it is an interdisciplinary research field in hydro science and complex system engineering, which is also attracting more and more attention of the researchers. Nowadays, cascade hydroelectric stations not only undertake the water supply responsibility for the area, but also play an important in power generation, ecological improvement of the river basin area and flood control. Hence, there are great benefits if the joint operation can be implemented by those large scale cascade hydroelectric stations. However, along with the rapidly exploitation of hydropower resources, the topological structure as well as the hydraulically and electrical relationship among cascade hydroelectric stations become more and more complex, moreover, the conflicts among multiple operational objective are more obliviously. Traditional optimal operation theories and techniques can not fully exert the comprehensive benefit of the large-scale cascade hydroelectric stations, and new joint operation and decision-making theories and methods are necessitated in present. In this thesis, aiming at the joint optimal operation problems of large-scale cascade hydroelectric stations, we adopt complex system modeling, multi-objective optimization and decision-making theory as well as intelligent evolutionary optimal method into our research, and a series of research results in joint optimal operation and decision-making of large-scale cascade hydroelectric stations are obtained, The main content of the innovation and contribution in this thesis are listed as follows:(1) Considering the complex characteristics of the joint optimal operation of large-scale cascade hydroelectric stations, we establish optimal models of large-scale cascade hydropower stations of upstream cascaded reservoirs of Yangtze River and Impounding dispatch for the lower cascade reservoirs as well as optimal dispatch of hydro-thermal system. Meanwhile, in order to overcome the limitation of existing theories and models when applied to solve practical constrained operational problems, a modified differential evolution method (MDE) which is designed for solving the optimal operation of cascade hydropower stations with coupled decision variables while avoiding " curse of dimensionality " efficiently is proposed. The proposed method adopts differential evolution as the basic framework for solving the problem, and dynamic parameter adjust strategy as well as local search operator based on chaotic sequences are introduced into the modified method to improve the optimal capacity. The obtained results show that MDE can solve those problem with fast convergence rate and high precision while handing the complex constraints effectively, thus we provide a new efficient way to solve optimal operation problems of large-scale cascade hydroelectric stations.(2) By analyzing the complex features of the optimization problem with multiple objectives, this thesis introduce the concept of Pareto optimal to discusses the theoretical basis of the multi-objective joint operation modeling, and a modified multi-objective differential evolution (MMODE) is presented to solve multi-objective problems (MOPs) with complex constraints. In MMODE, we adopt archive technology to store Pareto optimal solutions during the searching process, and the size of archive set is maintained by using a iterative strategy which is based on proposed "μ+1" selection operation. Meanwhile, a chaotic multi-objective local search operator based on Tent chaotic sequence is designed to improve the efficiency of the proposed method. MMODE is tested by a series of widely used benchmark MOPs, and the comparing case study is implemented by analyzing the results obtained by MMODE and the results obtained by some other multi-objective optimal algorithms, The results verify the high efficiency and accuracy of the proposed method when dealing with MOPs with complex characteristics.(3) In view of the conflict and competitive relationships among multiple operational requirements with Jinsha river downstream cascade project and Three Gorges cascade project, based on correlation analysis of multiple operational objectives, we establish a multi-objective flood control model, a multi-objective power generation model, and a multi-objective ecological dispatch model to consider multiple dispatch requirements of the cascade hydropower stations at the same time under different schedule modes. Afterwards, the proposed MMODE method is applied to solve these multi-objective models. The results show that MMODE can deal with above multi-objective operation model effectively, and a set of non-dominated dispatch schemes can be obtained rapidly, and based on which we analyze the mutual influence among those operational objectives, thus providing alternatives dispatch schemes for operational decision makers.(4) In order to determine the multi-objective comprehensively optimal operation scheme quickly, scientifically and reasonably, this thesis combined subjective and objective preference during the decision-making process comprehensively and a novel decision-making method based Vague set is proposed. The standard attribute importance matrix is established to quantify their difference to achieve subjective preference, and a modified entropy calculating equation is proposed to calculate entropy weight precisely. Meanwhile, we describe the closeness degree between the alternative schemes and the ideal schemes by using Vague value. By applying the proposed method to solve the multi-objective power generation, flood control and ecological dispatch decision-making problem and analyzing the decision result, we obtain the optimal scheme under different operational situations and preferences rapidly and correctly.更多还原