【作者】 李翔硕；

【导师】 王淳；

【作者基本信息】 南昌大学 ， 电力系统及其自动化， 2012， 硕士

【摘要】 输电网规划是根据电力系统的负荷及电源发展情况对输电系统的主要网架做出的发展规划,是电力系统规划的重要组成部分。输电网规划的决策变量维数较高,约束条件复杂,求解相对困难。目前,已有众多人工智能算法应用于求解输电网规划问题,但到目前为止,仍未得到一种公认的理想的输电网规划求解方法。本文尝试将模拟自然界物种迁移规律而构成的生物地理学优化算法(Biogeography-based Optimization, BBO)应用于输电网规划,采用18节点系统和19节点系统进行测试,深入分析了各种操作机制作用下的算法性能。1)对比了不同迁移模型对BBO算法性能的影响,得到了最适合输电网规划的余弦迁移模型。2)测试了不同初始参数设置对算法性能的影响,由于BBO算法采用了根据不同栖息地物种数量选择不同操作强度的生物激励机制,使得算法内部搜索机制独特,对初始参数要求宽松。3)借鉴模拟退火算法(Simulated Annealing, SA)与其他智能算法融合的方法,将SA引入BBO形成混合算法,提高了算法收敛速度。4)提出以待选线路为决策变量的一维编码形式,将多阶段输电网的动态规划转化为静态规划,解决了决策变量维数随着规划阶段数的增加而几何增长的问题,同时,也有利于BBO算法中各操作机制的实现。更多还原

【Abstract】 Transmission network planning, which plans transmission system according to power system load forecast and power planning, plays a critical role in power system planning. It is relatively difficult to solve Transmission network planning because of its high decision variable dimension and complex constraint condition.At present, there are many artificial intelligence algorithms which are being used to solve transmission network planning. However, we have not got a established ideal one. Biogeography-based optimization (BBO), which simulates the natural migration phenomenon, and ecology evolutionary algorithm of food chain (EEAFC), which simulates energy transmission phenomenon, are applied to solve transmission network planning problem. Tests are performed on18-bus system and19-bus system so as to comprehensively analyze abilities of algorithms under various mechanisms.1) Find that cosine migration model works best for transmission network planning through comparing the optimization abilities of different migration models.2) Tests optimization abilities of BBO based on different initial parameters show that it depends on initial parameters little because of its different biological incentive mechanism.3) Taking example by simulated annealing(SA) and other intelligence algorithms fusion, SA is proposed to improve BBO’s convergence speed.4) A real number encoding method, which adopts the lines as the decision variables and the serial number of the planning stage as the search domain, is proposed and multisage optimal planning of the transmission network is transformed into static optimization problem. By using the proposed method, it solves the problem that decision variables’dimension increases as the stage increases and it is suitable to BBO algorithm implement.更多还原