《Water resources management》上刊登的“多水库调度规则：基于粒子群的多蜂群优化算法”

作者：Leila Ostadrahimi, Miguel A. Mariño, Abbas Afshar

刊物：《Water resources management》,2012年,26卷,第2期, 407-427页

关键词：水库调度；多库群；PSO；模拟优化；调度规则

摘要：水库调度规则能够帮助和指导调度人员进行水库泄水和蓄水的操作，使整个系统的目标尽可能地达到最优。在多水库联合系统中，可能存在着许多可行的调度策略。系统工程和优化技术能够帮助我们在众多可行的调度策略中识别出最优的调度策略。本文针对多水库系统提出了若干条调度规则，这些调度规则的确定均借助基于多蜂群的粒子群优化方法，并且结合了较为熟知的HEC-ResPRM模型，并沿袭的是调参-模拟-优化的技术模式。为了提高标准粒子群算法的效率，本文嵌入了“多蜂群”的的算法机制。调度规则的相关参数是采用调参-模拟-优化的技术方法，在该方法中，文章使用了一种全尺度的模拟模型对通过粒子群算法得到的每个实验组的参数进行模拟。文章还使用哥伦比亚河流域上现存的Mica、Libby和Grand Coulee三个水库构成的水库系统来检验MSPSO方法在开发水库调度规则时的有效性。我们比较了基于规则的水库调度结果和由HEC-ResPRM得到的结果。结果表明，使用本文提出的方法对水库系统进行实时调度得到的结果要强于常见的隐随机优化方法得到的结果。

Multi-reservoir Operation Rules: Multi-swarm PSO-based Optimization Approach

Authors： Leila Ostadrahimi, Miguel A. Mariño, Abbas Afshar

Journal：Water resources management, 2012, 26(2): 407-427.

Key Words：Reservoir operation;Multi swarm;PSO;Simulation-optimization;Operating rule

Abstract： Reservoir operation rules are intended to help an operator so that water releases and storage capacities are in the best interests of the system objectives. In multi-reservoir systems, a large number of feasible operation policies may exist. System engineering and optimization techniques can assist in identifying the most desirable of those feasible operation policies. This paper presents and tests a set of operation rules for a multi-reservoir system, employing a multi-swarm version of particle swarm optimization (MSPSO) in connection with the well-known HEC-ResPRM simulation model in a parameterization–simulation–optimization (parameterization SO) approach. To improve the performance of the standard particle swarm optimization algorithm, this paper incorporates a new strategic mechanism called multi-swarm into the algorithm. Parameter of the rule are estimated by employing a parameterization–simulation–optimization approach, in which a full-scale simulation model evaluates the objective function value for each trial set of parameter values proposed with an efficient version of the particle swarm optimization algorithm. The usefulness of the MSPSO in developing reservoir operation policies is examined by using the existing three-reservoir system of Mica, Libby, and Grand Coulee as part of the Columbia River Basin development. Results of the rule-based reservoir operation are compared with those of HEC-ResPRM. It is shown that the real-time operation of the three reservoir system with the proposed approach may significantly outperform the common implicit stochastic optimization approach.

原文链接：http://link.springer.com/article/10.1007/s11269-011-9924-9