Optimisation par essaims particulaires pour la logistique urbaine

Abstract : In this thesis, we are interested in the management of goods flows in urban areas, also called last mile logistics, and associated with various current issues: economic, environmental, and societal. Four main stakeholders are involved by these challenges: shippers, customers, carriers and local authorities, each with different priorities (improving service quality, minimizing the travelling distance, reducing greenhouse gas emissions, etc.). Faced with these challenges in the city, one possible action lever is to optimize the routes for the pickup and/or delivery of goods. Three types of urban flows are considered: from or to the city, and intra-urban. For outgoing and incoming flows into the city, the goods are first grouped in a warehouse located on the suburban area of the city. If there are several warehouses, the associated planning problem is the Location Routing Problem (LRP). We are studying one of its variants called the Capacitated Location Routing Problem (CLRP). In this problem, by respecting the capacity constraint on vehicles and depots, the location of depots and route planning are considered at the same time. The objective is to minimize the total cost, which consists of the cost of opening depots, the cost of using vehicles, and the cost of the travelling distance. For all flows, we are also looking to solve a Pickup and Delivery Problem (PDP), in which a fleet of vehicles simultaneously carries out pickup and delivery operations. We focus on two of its variants: the selective variant where not all requests are satisfied, in a context of paired demands and time windows on sites (Selective Pickup and Delivery Problem with Time Windows and Paired Demands, or SPDPTWPD). The second studied variant is the extension of the first one by adding the possibility of carrying out transport in several stages by introducing operations for the exchange of goods between vehicles at transfer sites (Selective Pickup and Delivery with Transfers or SPDPT). The considered objectives for these two variants of PDP are to maximize profit and to minimize distance. Each studied problem is formally described, mathematically modelled as a linear program and then solved by exact, heuristic and/or metaheuristic methods. In particular, we have developed algorithms based on a metaheuristic called Particle Swarm Optimization, which we have hybridized with local search operators. The approaches are validated on instances of different sizes from the literature and/or on instances that we have generated. The results are critically analyzed to highlight the advantages and drawbacks of each method.
Document type :
Complete list of metadatas

Cited literature [100 references]  Display  Hide  Download

Contributor : Abes Star <>
Submitted on : Tuesday, October 22, 2019 - 5:54:09 PM
Last modification on : Wednesday, October 23, 2019 - 1:25:05 AM
Long-term archiving on: Thursday, January 23, 2020 - 8:33:58 PM


Version validated by the jury (STAR)


  • HAL Id : tel-02327313, version 1


Zhihao Peng. Optimisation par essaims particulaires pour la logistique urbaine. Automatique. Université Bourgogne Franche-Comté, 2019. Français. ⟨NNT : 2019UBFCA010⟩. ⟨tel-02327313⟩



Record views


Files downloads