Вы находитесь на странице: 1из 22

DSS for the planning of urban freight

transportation based on heuristic and


metaheuristic algorithms.
DSS para la planeacin de transporte de mercanca a nivel urbano
basado en algoritmos heursticos y metaheursticos

DSS for the planning of urban freight transportation based on heuristic and
metaheuristic algorithms

Presentation Outline

Brief introduction
Problem description
Mathematical model formulation
Solution approaches given in literature
Heuristic and Metaheuristic Algorithms implemented in the
DSS
Software Development Methodology Applied
Development and Implementation

DSS for the planning of urban freight transportation based on heuristic and
metaheuristic algorithms

Brief Introduction
Optimal planning of the logistic network has been a broad area of study and most
interesting topic among researches in the field
This study was based on the preliminary results obtained from the ongoing project
entitled Desarrollo de Sistemas Inteligentes de Gestin de Transporte para soporte
a la toma de decisiones operativas urbanas, which implies the construction of
software that directs the optimal routing and allocation of trucks in the city in their
daily planning program.
A programming methodology is used, which has been proven to facilitate the
integration of many decision procedures or algorithms in the decision making tool,
thus making it more flexible to implement in real life situations
Algorithms selected for each functionality of the program are presented
The integration of these algorithms in the DSS tool are proven to generate good
solutions

DSS for the planning of urban freight transportation based on heuristic and
metaheuristic algorithms

Problem Description
The logistic operation of a company at a daily basis implies a series of activities:
Orders management

Fleet administration allocation of vehicles


Distribution management
Capacity planning
Route design
Control and follow-up

DSS for the planning of urban freight transportation based on heuristic and
metaheuristic algorithms

Shortest Path
Problem

Network Flow
Problem

Multiobjecti
ve
Transportati
on Network
Design

Transportation
Problem

Spanning Tree
Problem

Optimal
Network
Design
Problem

Assignment
Problem

Transshipmen
t Problem

Vehicle
Routing
Problem

Fig 1. Multiobjective Transportation network and its basic instances (source:


Current & Marsh, 1993)

DSS for the planning of urban freight transportation based on heuristic and
metaheuristic algorithms

Problem Description Designing optimal routes


One of the main subjects of attention in the scientific community has been with
respect to the vehicle routing problem (VRP), first defined by Dantzig and Ramser in
1959
The vehicle-routing problem (VRP) consists on assigning sequences to the vehicles
that must visit each one of the customers in the service area in order to minimize
time of the total route or transportation costs.
The VRP is described as a directed graph G(A,V), where V= {0,1,n} represents a
set of nodes and A is the set of arcs that connect each one of the nodes.
In real life applications, this problem has been approached under multiple variants:
VRPTW

HFVRP
MDVRP
CVRP
VRPPD
Fig 2. VRP as the TSP problem

DSS for the planning of urban freight transportation based on heuristic and
metaheuristic algorithms

Problem Description allocation of vehicles and capacity planning


The daily operation also implies capacity planning and allocation of vehicles, even
more if the planning involves multiple depots were merchandize is dispatched
A mathematical allocation problem is usually formulated, where capacities are also
considered. This classical allocation problem is considered NP-complete.
When integrating the classical allocation problem with VRP, its complexity
increases.
Most authors are implementing the location-routing problem (LRP), due to the fact
that the facilities location or allocation of clients in several clusters can be
dynamically changed and thus the routes considered previously.
The location-routing problem (LRP) implies both the allocation and routing of
vehicles, allowing the program to determine the capacity and/or size of the facilities
and vehicles.
Given a set of potential depots with opening costs, a fleet of identical vehicles and
a set of customers with known demands, the classical LRP consists in opening a
subset of depots, assigning customers to them and determining vehicle routes, to
minimize a total cost including the cost of open depots, the fixed costs of vehicles
used, and the total cost of the routes [Prodhon and Prins, 2014].

DSS for the planning of urban freight transportation based on heuristic and
metaheuristic algorithms

Mathematical Model Formulation


Variables.
Conjuntos.
: Types of products
: Deposits.

: available vehicles

: customers
Objective function.
: Periods of time

Parmetros.

Wi= weight per unit of product


Voli= volume per unit of product
CapPk : Capacity of vehicle (in weight)
CapVk : Capacity of vehicle (in volume)
Capij : Capacity of product in deposit
Demilt: Demand of product by customer at time .
CTk : Cost of transportation ($/veh*km) of vehicle per km travelled
CAkj : Setup cost ($/veh) of vehicle in deposit
Distjl : Distance (in km) travelled between deposit j and client l.
NVl : Service level for client .

DSS for the planning of urban freight transportation based on heuristic and
metaheuristic algorithms

Mathematical Model Formulation


Restrictions:

DSS for the planning of urban freight transportation based on heuristic and
metaheuristic algorithms

Solution approaches given in literature VRP and its variants


Exact algorithms

Heuristic algorithms

DSS for the planning of urban freight transportation based on heuristic and
metaheuristic algorithms

Solution approaches given in literature VRP and its variants


Metaheuristic

Hybrid

DSS for the planning of urban freight transportation based on heuristic and
metaheuristic algorithms

Solution approaches given in literature LRP


Special attention given
to:
Multiobjective problems
Breakthrough:
Hybrid Algorithms!

Combining 2
metaheuristics
For example:
GRASP with Path
Relinking
Combining a
metaheuristic with an
exact method

DSS for the planning of urban freight transportation based on heuristic and
metaheuristic algorithms

Solution approaches given in literature LRP

Table 1. Source: PRODHON & PRINS, A survey on recent research on location-routing problems, EJOR, 238
(2014), 1-17.

DSS for the planning of urban freight transportation based on heuristic and
metaheuristic algorithms

Heuristic and Metaheuristic Algorithms implemented in the DSS

Route construction and optimization VRPTW


Ant Colony Optimization (ACO)

Allocation of customers and vehicles


EPSO Evolutionary Particle Swarm Optimization

Replication
Mutation
Reproduction
Fitness
Selection

LRP

Simulates Ants
behavior when looking
for foot sources
Likewise the algorithm
builds routes to visit
costumers

DSS for the planning of urban freight transportation based on heuristic and
metaheuristic algorithms

Heuristic and Metaheuristic Algorithms implemented


Generating new feasible solutions for
multiple objectives Cost and Time
objectives
Genetic Algorithm

Genetic Algorithm GA is an evolutionary


approach for a large type of combinatorial
optimization problems that uses the
concepts of mutation and selection from
Darwinian theory of evolution.

Proposed GA pseudo-code
Create an initial population of individuals
While termination condition not satisfied do
Repeat
If crossover condition satisfied then
Select parents
Perform crossover
End if
If mutation condition satisfied then
Perform mutation
End if
Evaluate fitness of offspring
Until sufficient offspring created
Select new population
End while

DSS for the planning of urban freight transportation based on heuristic and
metaheuristic algorithms

Software Development Methodology Applied

Six
Sigma
ITIL
CMMI
Others

Approach
QA

Approach
to
Software
and
Hardware
Industry
Agile
developm
ent
approach

HARD SNOW
DEVELOPMENT

Ericcson
Objectory
UML SysML
RUP -Open UP
Others

Agile
Manifesto
XP
SCRUM
ICONIX
Lean
Others

DSS for the planning of urban freight transportation based on heuristic and
metaheuristic algorithms

Development and Implementation


Domain Model

DSS for the planning of urban freight transportation based on heuristic and
metaheuristic algorithms

Development and Implementation


Use Case Model

Sequence Diagram

Robustness Model

DSS for the planning of urban freight transportation based on heuristic and
metaheuristic algorithms

Development and Implementation


Data Model

DSS for the planning of urban freight transportation based on heuristic and
metaheuristic algorithms

Development and Implementation


Architecture

View
Controll
er
Model

ExtJS FrameWork
HTML
AJAX

PHP
Laravel Framework

MySQL
DAO

DSS for the planning of urban freight transportation based on heuristic and
metaheuristic algorithms

ACKNOWLEDGEMENTS

RESEARCH GROUP:
Carlos Javier Uribe
Ruben Snchez
Nilson Herazo
Lauren Castro
Fairuz Ospino
Henry Maury
Heyder Paez
Diana G. Ramrez

This research was possible under the financial


support of the Atlantic Department Government
and the Colombian General Royalties System,
through the project LOGPORT.

DSS for the planning of urban freight transportation based on heuristic and
metaheuristic algorithms

THANK YOU!
Gracias!
Diana G. Ramrez-Ros, MSc.
Industrial Engineer
PhD student in Transportation Engineering starting
Fall 2015
Research Assistant LOGPORT-CUC
Adjunct Assistant Professor
dramirez7@cuc.edu.co
diana.gineth.ramirez@gmail.com
Skype: diana.g.ramirez

Any questions?

Heyder Paez Logreira


Electronic Engineer
MSc. Student in Systems Engineering
and CS
Research Assistant LOGPORT-CUC
Assistant Professor
hpaez@cuc.edu.co
heyderpaez@gmail.com
Skype: heyderpaez

Вам также может понравиться