• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文期刊>International Journal of Swarm Intelligence and Evolutionary Computation >Multi-objective Particle Swarm Optimization with Gradient Descent Search
【24h】

Multi-objective Particle Swarm Optimization with Gradient Descent Search

机译：梯度下降搜索的多目标粒子群算法

获取原文
           
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Routing problems are classical combinatorial optimization tasks that find much applicability in numerous industrial and real-world scenarios. One challenging variant of the routing problem is the Fuel Distribution Problem (FDP) that a transportation company must face in its everyday operations. The main activity of a transportation fuel company is restocking all its stores, i.e. petrol stations, along a geographical map, with the goal to minimizing its overall costs. In this research work we present a hybrid heuristic based on the metaphor of the immune system for solving the FDP, which basically asks to find a set of routes as shorter as possible for a fixed number of company’s vehicles in order to satisfy the several received demands of customers. In particular, the presented immunological algorithm takes inspiration by the clonal selection principle, whose key features are cloning, hyper- mutation, and aging operators. Such algorithm is also characterized, in having a (i) deterministic approach based on the Depth First Search (DFS) algorithm - used in the scheme of assigning a vertex to a vehicle - and (ii) a local search operator, based on the exploration of the neighborhood. The algorithm has been tested on one real data instance, with 82 vertices, and 25 others artificial different instances, taken from DIMACS graph coloring benchmark. The experimental results presented in this work, not only prove the robustness and efficiency of the developed algorithm, but show also the goodness of the local search, and the approach based on the DFS algorithm. Both methodologies help the algorithm to better explore the complex search space.
机译：路由问题是经典的组合优化任务，可在许多工业和实际场景中找到很多适用性。路由问题的一个具有挑战性的变体是运输公司在日常运营中必须面对的燃料分配问题（FDP）。运输燃料公司的主要活动是沿地理地图补充其所有商店，即加油站，以使总成本最小化。在这项研究工作中，我们提出了一种基于免疫系统隐喻的混合启发式方法来解决FDP问题，该方法基本上是要求为固定数量的公司车辆找到尽可能短的一组路线，以满足一些已收到的需求客户。特别地，所提出的免疫学算法从克隆选择原则中得到启发，克隆选择原则的主要特征是克隆，超突变和衰老算子。这种算法的特征还在于，它具有（i）基于深度优先搜索（DFS）算法的确定性方法-用于将顶点分配给车辆的方案中-（ii）基于探索的本地搜索运算符的附近。该算法已在一个真实数据实例上进行了测试，该实例具有82个顶点，另外25个人工不同实例来自DIMACS图形着色基准。这项工作给出的实验结果，不仅证明了所开发算法的鲁棒性和有效性，而且还证明了局部搜索的优点以及基于DFS算法的方法。两种方法都有助于算法更好地探索复杂的搜索空间。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号