![a8a6f4e0647dbbd48af640cf5a1c1f87.png](https://i-blog.csdnimg.cn/blog_migrate/d29740dbca22c9c68c0605fe1191e565.png)
![483933a10367e7baaa8a8bbcc058f920.png](https://i-blog.csdnimg.cn/blog_migrate/f61044715aacab0b7349b61b817a52e3.png)
![145fa267e6faf2c145d829583daf7282.png](https://i-blog.csdnimg.cn/blog_migrate/7a976b66627d20cb3c6d1cd7840737ca.jpeg)
文章信息
1
论文标题:
风电场电缆布线问题的启发式算法
Heuristic Algorithms for the Wind Farm Cable Routing Problem
作者信息:
Davide Cazzaro1,2,Martina Fischetti1,Matteo Fischetti3
1 Vattenfall BA Wind, Jupitervej 6, 6000 Kolding, Denmark
2 DTU, Akademivej 358, 2800 Kongens Lyngby, Denmark
3 DEI, Università di Padova, via Gradenigo 6/A, 35100 Padova, Italy
原文链接:
http://www.enerarxiv.org/page/thesis.html?id=1998
论文简介
2
文章亮点
本文中描述,实现和测试了针对风电场电缆布线问题的五种不同的元启发式算法:模拟退火算法,禁忌搜索算法,可变邻域搜索算法,蚂蚁算法和遗传算法。
描述了一种名为Sweep的构造试探法,该方法通常会在很短的计算时间内找到初始的高质量解决方案。
摘要
风电场电缆布线问题在海上风电场设计中起着关键作用。给定风力发电场中涡轮机和变电站的位置,以及将涡轮机产生的电力传输到变电站所需的一组电缆,任务是定义一个电缆连接树,以使总电缆成本最小化。在本文中,我们描述,实现和测试了针对该问题的五种不同的元启发式算法:模拟退火算法,禁忌搜索算法,可变邻域搜索算法,蚂蚁算法和遗传算法。我们还描述了一种名为Sweep的构造试探法,该方法通常会在很短的计算时间内找到初始的高质量解决方案。我们在两个数据集上比较了启发式方法的效果:一个包含文献中的实例,并用作训练集来调整我们的代码,而第二个是一组非常大的新的现实实例集(我们公开提供),用作测试集。最后对启发式算法提出了一些实用的建议:根据我们的实验,可变邻域搜索获得了最佳的整体性能,而禁忌搜索算法紧随其后。
ABSTRACT
The Wind Farm Cable Routing problem plays a key role in offshore wind farm design. Given the positions of turbines and substations in a wind farm and a set of electrical cables needed to transfer the electrical power produced by the turbines to the substation, the task is to define a cable-connection tree that minimizes the overall cable cost. In the present paper, we describe, implement and test five different metaheuristic schemes for this problem: Simulated Annealing, Tabu Search, Variable Neighborhood Search, Ants Algorithm, and Genetic Algorithm. We also describe a construction heuristic, called Sweep, that typically finds an initial high-quality solution in a very short computing time. We compare the performance of our heuristics on two datasets: one contains instances from the literature and is used as a training set to tune our codes, while the second is a very large new set of realistic instances (that we make publicly available) used as a test set. Some practical recommendations on the proposed heuristics are finally provided: according to our experiments, Variable Neighborhood Search obtains the best overall performance, while Tabu Search is our second best heuristic.
Keywords
Wind farm optimization
Cable routing problem
Metaheuristics
Computational analysis
文章附图
3
![110f82671725b99893d523061bcab6b0.png](https://i-blog.csdnimg.cn/blog_migrate/04aae6df2eb9feb0c7889001bcbc2b13.png)
Figure 1: Example of a solution for a WFCRP instance with n = 80 turbines. The substation is node 0, all other nodes are turbines. Cable capacities are normalized: cable type 1 can support up to 7 turbines, while cable type 2 is more expensive but can support up to 12 turbines. At most C = 7 cables can be connected to the substation, and the built cables are pairwise noncrossing
![436ff0505e8442d9ed161f6f4d05debc.png](https://i-blog.csdnimg.cn/blog_migrate/fe1ac457801b7198bdc34bfe3e1b102d.png)
Figure 2: An example of the turbine partition of the Sweep heuristic: starting from turbine 2, in anti-clockwise direction, with 7 turbines per group. It forms five groups of turbines (with different colors in the figure): four groups have 7 turbines each, namely {2, 3, 4, 5, 25, 29, 30}, {19, 20, 23, 24, 26, 27, 28}, {14, 15, 16, 17, 18, 21, 22}, and {7,8, 9, 10, 11, 12, 13}, while the last one is composed of just two turbines {1, 6}.
![816f730e13391b7857323110c89e87ce.png](https://i-blog.csdnimg.cn/blog_migrate/31f87cad5917f89c9986abc2f0dc7fe7.png)
Figure 3: Gap obtained by Sweep (multistart) with respect to the best known solutions [4] on the training instances(the lower the better). The heuristic, in less than one second of computing time, reaches quite good solution costs, all within 8% from the best known ones.
![bbf53d393140b24a0b3ee85239118124.png](https://i-blog.csdnimg.cn/blog_migrate/dd69617155d60bbe9f74f83dd3719b19.png)
Figure 6: Performance profile [36] on the training set instances for our six heuristics: Simulated Annealing, Tabu Search, Variable Neighborhood Search, Sweep, Ants Algorithm, and Genetic Algorithm.
![3fb1df202c9c41694b850ea01c31487c.png](https://i-blog.csdnimg.cn/blog_migrate/d524e5f95a86188235727e93e4a87999.png)
Figure 7: Performance prole on the training instances which compares the MILP-based matheuristic technique results in [4] (1 hour time limit) with the six implemented heuristics with 10-minute time limit. All heuristics are close to the (near) optimal solutions computed in 1 hour by the matheuristic. The only exception is the Ants Algorithm, whose curve lies further on the horizontal axis and therefore does not appear in the graph. This is also the case for the following plots, in which the Ants Algorithm is present but is not visible.
小编按
4
与陆地风力发电相比,海上风力发电具有突出的特点,海上风电由于其资源丰富、风速稳定、开发利益相关方较少、不与其他发展项目争地、可以大规模开发等优势。但是因为海上环境恶劣,情况复杂,所以在开发海上风电时应当在设计阶段就提出可靠安全的方案,同时降低运营成本,使海上风电有更好的发展条件。集电系统拓扑优化,其物理意义是希望找到海缆长度最优的集电系统拓扑连接方 式,认为运时成本也能达到最优。
关于我们
本期小编:杨宏伟
EnerarXiv是一个开放获取(open access)的能源领域预印本平台,涵盖了包括能源科学与技术以及与能源相关的经济学、社会科学及政策和法律的多学科跨科领域的前沿研究。致力于为能源领域的学者及爱好者们打造优质的分享和交流的社区。EnerarXiv提供预印本上传及分享的平台,并科普最新的领域动态、前沿知识、和相关资讯,以激发更多的讨论与思考、促进交流与碰撞,为能源领域的科研及产业增添动力。
在浏览器中输入下方链接,进入EnerarXiv:
www.enerarxiv.org
点击阅读原文查看EnerarXiv原文