Download presentation
Presentation is loading. Please wait.
Published byLorraine Marshall Modified over 9 years ago
1
PSO algorithms for Generalized Multi-depot VRP with pickup & delivery requests Pandhapon Sombuntham 108026
2
Background Proposed Approaches Experiment Summary Q&A Contents
3
Transportation of Material Right Time Route Planning Right Place Right Quantity ?
4
Background Location A Location B Depot otb-games.com To ATo B VRP (Vehicle Routing Problem ) Shipment Done Shipment Done To Depot VRPSPD (Vehicle Routing Problem with simultaneous pickup & delivery ) Shipment Done To B No direct shipment between Locations (except depot) No direct shipment between Locations (except depot)
5
Background Depot otb-games.com Location A Location B To B PDP (Pickup & Delivery Problem) Finished Shipment Empty truck Vehicle Station Direct shipment 3 Roles of Locations Pickup Location Delivery Location Vehicle Station One role for a location Direct shipment 3 Roles of Locations Pickup Location Delivery Location Vehicle Station One role for a location To B clipartof.com Delivery Location Pickup Location
6
Background Location A Location B 1 1 To A 2 2 To B Location C 7 7 To A Allow … Direct shipment … Can pickup more than 1 item from a location to deliver to more than one destinations. … Location can play multiple roles Allow … Direct shipment … Can pickup more than 1 item from a location to deliver to more than one destinations. … Location can play multiple roles 6 6 To D 3 3 To B 4 4 To C 5 5 8 8 To B GVRP – MDPDP Generalized Vehicle Routing problem for multi-depot with pickup and delivery requests Location D Any location can play multiple roles
7
Deliver Background Depot (Supply Node) Customer ( Demand Node) VRP (Vehicle Routing Problem ) PDP (Pickup & Delivery Problem) 1 1 2 2 3 3 4 4 GVRP – MDPDP Generalized Vehicle Routing problem for multi-depot with pickup and delivery requests Pickup Deliver Vehicle Station Pickup Location Deliver Location Location with Vehicle Location w/o Vehicle Allow Multiple roles for each location Many pickups at a location
8
Limousine service at Airport in Big city Example mitchellslimousines.net Airport To C clipartof.com To B clipartof.com To A clipartof.com A B C Airport Station Many pickups at airport Airport is both pickup and delivery location
9
Pooling Vehicle SME Location A Location B 1 1 To A 2 2 To B Location C 7 7 To A 6 6 To D 3 3 To B 4 4 To C 5 5 8 8 To B GVRP – MDPDP Generalized Vehicle Routing problem for multi-depot with pickup and delivery requests Location D SME Sharing Fleet of vehicles Among alliances Any location can play multiple roles + Many Pickups at locations
10
Daily operation 100s items Consider ? VRP Vehicle capacity Heterogeneous vehicle On-Time delivery Maximum Route time Direct shipments Many pickups items at any locations Multiple-role locations PDP GVRP-MDPDR Experience Poor Utilization
11
Proposed Approach Based on Particle Swarm Optimization (PSO) framework for solving the vehicle routing problems,i.e. CVRP VRPSPD,and VRPTW (Ai & Kachitvichyanukul, 2009a,2009b,2009c) PSO with multiple social learning terms of Pongchairerks & Kachitvichyanukul [10],[11] PSO rigasturists.lv istockphoto.com http://talkfeeleez.files.wordpress.com/ Initialize particles with random position and zero velocity Evaluate fitness value Update pbest and gbest Meet stopping criterion? Update velocity and position Start End YES NO
12
Thesis Framework GVRP-MDPDR GLNPSO Application Encoding (Solution representation) Encoding (Solution representation) Decoding SD1 SD2 SD3 Preliminary TestEffect on Time Appropriate Swarm size & steps Comparison algorithms
13
Test Instances PDPTW (Li & Lim,2001) Special cases of GVRP-MDPDR Newly Generated Instances 20-100 locations with 30-100 items involved Half-random-half-clustered Randomly distributed Clustered
14
Test on PDPTW Case Best known solution Best of 5 Replications Average NVDistanceNVDistanceNVDistance lc10110828.9410828.9410.00828.94 lc10210828.9410828.9410.00828.94 lc10391035.3591063.639.40977.47 lc1049860.019863.369.00884.14 lc10510828.9410828.9410.00828.94 lc10610828.9410828.9410.20890.53 lc10710828.9410828.9410.00828.94 lc10810826.4410826.4410.20839.41 lc10991000.6010827.8210.00828.04 lc2013591.563 3.00591.56 lc2023591.563 3.00591.56 lc2033585.563591.173.00591.17 lc2043590.603 3.00616.26
15
Application on Real case Vehicle capacity Heterogeneous vehicle On-Time delivery Maximum Route time Direct shipments Many pickups items at any locations Multiple-role locations 128 items Consider Vehicle ID27 Items18 No. of Visit12 Route Time534 Sequence Location Visit Pickup ItemsDeliver Items 165 26262,63 35555,57,58 45455 5596162 66157,61,63 75352,53 846 53 95046 104552 116258 1265
16
Application with Multi-Objective PSO (MOPSO) GVRP-MDPDR GLNPSO Application Encoding (Solution representation) Encoding (Solution representation) Decoding MOPSO w 1 x Cost 1 x NV + w 2 x Cost 2 x Total distance = Total Cost
17
Multi-Objective PSO Two objective functions Number of Vehicle used Total distances MOPSO Trade-off solutions Pareto front optimality (Nguyen et al., 2010)
18
Multi-Objective PSO Erc1 ParameterValue No. of Particle 50 Iteration500 No. of Neighbors 5 Cp1.3 Cg1.3 Cl1 Cn2 Inertia weight 0.9 to 0.4 Provide Alternatives for the decision maker to analyze the tradeoff.
19
More Generalized case of VRP Add practical consideration Extend PSO Framework Experiments Application with Real-word Decision Supports tools Summary Vehicle capacity Heterogeneous vehicle On-Time delivery Maximum Route time Direct shipments Many pickups items at any locations Multiple-role locations
20
Further study Develop encoding and decoding Procedure Randomness & Logical methods Analyze more about properties of the problem More Practical consideration MOPSO More objectives to considered Adaptive PSO Recommendation
21
Q&A
22
Best Wishes For your attentions
24
Output Vehicle ID27 Items18 No. of Visit12 Route Time534 Sequence Location Visit Waiting TimePickup ItemsDeliver Items 1650 2627762,63 3557855,57,58 4543655 55906162 661057,61,63 753052,53 8460 53 950246 1045352 1162058 12650
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.