Download as ppt, pdf, or txt
Download as ppt, pdf, or txt
You are on page 1of 60

Research Presentation

Hybrid Multiobjective
Evolutionary Algorithm Solving
Vehicle Routing Problem
Lecturer: Prof. Martin Henz
CS 6211 Special Topics in Computer Science II

Chew Yoong Han


Agenda
 Introduction to MOEA
 Vehicle routing problem with time windows
 HMOEA: algorithm design and benchmark
 Real world TTVRP optimization
 Summary
Introduction
 Evolutionary algorithms
 Global search optimization technique
 Based on natural selection and reproduction
EAs flow
Selection
Parents

Recombination

Population
Mutation

Offspring
Replacement
EA Application In:
 Bio- computing
 Evolvable hardware
 Game playing
 Job-shop scheduling
 Management sciences
 Non-linear filtering
MOEA
 Multiobjective optimization problem (MOP)
 A problem to formulate a design in which there are
several criteria or design objectives
 MOEA
 Multiobjective evolutionary algorithm
 EA that can solve MOP
MOEA
 Minimize [f1(x), f2(x), …, fk(x)]
 Subject to the m inequality constraints
gi ( x )  0 i  1, 2,..., m
 And subject to the p equality constraints
hi ( x)  0 i  1, 2,..., p
 k is the number of objectives
Conventional MO handling
 Weighting method
 Constraints method
 Penalty method
 Disadvantages
 Priori
 Trial and error (weights)
Pareto Concept
 Two objectives
minimization.
 Point A and point B
are dominating
points.
 Point C is
dominated.
 Curve from A to B
is Pareto front.
MOEA characteristics
 Random vs. Deterministic
 Population vs. Single best solution
 Creating new solutions through crossover
 Creating new solution through mutation
 Selecting solutions via “Survival of the fittest”
MOEA algorithms
 MOEA in literature
 VEGA, David Schaffer (1985)
 MOGA, Fonseca and Fleming (1993)
 NPGA, Horn et al. (1994)
 NSGA, Srinivas and Deb (1994)
 PAES, Knowled and Corne (2000)
 SPEA, Zitler (2001)
Research

 HMOEA to solve VRPTW

 New model TTVRP + optimization


Vehicle Routing Problem
 Determining the best routes or schedules for
pickup/delivery of passengers or goods in a
distribution system
 Objectives: Minimize time/distance/cost
 Relevant parameters: size of the fleet, capacity,
number of drivers, customers’ information and
timing constraints.
VRPTW
 A set of vehicles with limited capacity
 Route from a central depot to a set of
geographically dispersed customers with
known demands
 Predefined time windows
 Capacity constraint
VRPTW (cont2)
 Depot, customer
 Capacity
 Time windows
 Objectives:
 Vehicle (Route)
 Routing cost
Vehicle Routing Solution
5

6
7 4
3
10

Depot
8
1

9
12
11
R1 R2
VRPTW: Methods
 Graph theory
 Minimal spanning tree + shortest path
 Saving method
 Branch and bound
 Simulated annealing
 Tabu search
 Ant colony
VRPTW: Survey
 Solomon (1987)
 Golden and Assad (1988)
 Desroiser et al. (1995)
 Laporte et al. (2000)
 Kilby et al. (2000)
 Toth and Vigo (2002)
Why MOEA in VRPTW ?
 Optimize for a set of solutions
 Minimum dependence to priori
 Global search ability
 Single objective vs. Multiobjective
 Better solution !
Concerns in MOEA Design
 Representation of chromosome
 Creative or innovative evolutionary operators
 Performance metrics to compare
 Feasibility handling policy
 Improve existing elitism reservation
HMOEA components
 Variable length representation
 Route exchange cross over
 Multi mode mutation
 Pareto ranking
 Tournament selection
 Local exploitation
HMOEA flow
Start

Pre-processing

Build initial
population

Cost function Stopping


End
evaluations criteria met?

Selection Local search


exploitation

Route-exchange
Crossover
Elitis m

Multimode
mutation
HMOEA Representation
Chromosome A Chromosome encodes
a complete routing
0 0 0 solution

2 1 6

5 3 10

7 4 0 A route (vehicle)
contains sequence of
0 8 customers

0
Route-Exchange Xover
R1

R1

R2
R2

Selected routes
On certain criteria
Multimode mutation
PM - Mutation rate

PE - Elastic rate No mutation

PS - Squeeze rate Partial_swap

Merge_Routes Split_Route
Local exploitation
 Intra_Route
 Generate two different numbers based on the size of routing sequence of
both vehicles.
 Pick two routes randomly and swaps two nodes from each route.
 Lambda interchange
 Scan through nodes in route A and moves the node into route B.
 Repeat until a pre-defined number of nodes are shifted or the the last
node of route A is met.
 Shortest path first
 Rearrange the order of nodes in a particular route.
 Based on its distance from the previous node.
 Feasibility must be ensured
Pareto Ranking
 Assigns the same smallest rank (highest fitness)
for all non-dominated individuals
 The individuals are inversely ranked according
to how many individuals in the population
dominating them, -
 Smaller number of vehicles but equal cost of routing
 Smaller routing cost but equal of number of vehicles
 Smaller routing cost and smaller number of vehicles
Tournament Selection
 Tournament selection.
 A tournament size of 2.
 All individuals in the population are randomly
grouped into pairs.
 The individual with a lower rank (higher
fitness) will be selected for reproduction.
Discussion (1)
 Better population distribution
Discussion (2)
 Comparison to best known
  Rochat Chiang Potvin Potvin and Taillard Chiang Homberger Tan Tan  
Problem and and Russell, et al., Bengio, et al., and and et al., et al., HMOEA
Class Taillard, (1996) (1996) (1996) (1996) Russell, Gehring, 2001d 2001e
(1995) (1997) (1999)

C1 10.00 10.00 10.00 10.00 10.00 10.00 10.00 10.00 10.00 10.00
828.45 909.80 861.00 838.00 828.45 828.38 828.38 851.96 841.96 827.00

C2 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.20 3.00 3.00
590.32 684.10 602.50 589.90 590.30 591.42 589.86 620.12 611.2 590.00

R1 12.58 12.50 12.60 12.60 12.25 12.17 11.92 13.20 12.91 12.92
1197.42 1308.82 1294.70 1296.83 1216.70 1204.19 1228.06 1220.0 1205.0 1187.0

R2 3.09 2.91 3.10 3.00 3.00 2.73 2.73 4.40 5.00 3.51
954.36 1166.42 1185.90 1117.70 995.38 986.32 969.95 985.69 929.6 951

RC1 12.38 12.38 12.60 12.10 11.88 11.88 11.63 13.30 12.60 12.74
1369.48 1473.90 1465.00 1446.20 1367.51 1397.44 1392.57 1366.62 1392.3 1355

RC2 3.62 3.38 3.40 3.40 3.38 3.25 3.25 5.20 5.80 4.25
1139.79 1401.50 1476.10 1360.60 1165.62 1229.54 1144.43 1108.50 1080.10 1067.00

All 427 422 427 422 416 411 406 470 471 441
57120 65201 64679 62572 57993 58502 57876 57903 56931 56262
Discussion (3)

 Routing cost for different


fitness functions
Discussion (4)

 Routing cost for local


exploitation
 C201 instance
Routing result

 Routing solution
 For RC2-07
What’s Next

 New transportation model


 Optimization ……
 Analysis and discussion
TTVRP
 Transportation problem for moving empty or
laden container for a logistic company in a city
with local port
 A routing plan consists a sets of selected jobs,
scheduled to internal trucks
 Constraints: time windows, trailers (type)
 Objectives: number of trucks, cost
TTVRP model
1
2

5 6

12

4
8

11
10
9
Comparison to existing models
 VSP (2000)
 TTRP (2002)
 SDVRP (1998)
 PACVRP (1981)
 HVRP (1999)
Job models (1)
Free Trade Customer Depot
Zone in Warehouse
Port
Loaded Trip (first leg) Empty Trip (second leg)

Free Trade Customer Depot


Zone in Warehouse
Port
Loaded Trip (second leg) Empty Trip (first leg)
Job models (2)
Free Trade Zone Depot
in Port

Empty Trip

Depot
Tasks Type (1)
 Importation job

1. Port to warehouse with trailer type 20


2. Port to warehouse with trailer type 40
3. Warehouse to depot with trailer type 20
4. Warehouse to depot with trailer type 40
Tasks Type (2)
 Exportation job

5. Depot to warehouse with trailer type 20


6. Depot to warehouse with trailer type 40
7. Warehouse to port with trailer type 20
8. Warehouse to port with trailer type 40
Tasks Type (3)
 Empty containers movement

9. Port to depot with trailer type20


10. Depot to port/other depot with trailer type 20
11. Port to depot with trailer type 40
12. Depot to port/other depot with trailer type 40
Trailer issue (Export)
 Export job order
 From initial location, a truck has to pick up a
correct trailer at trailer exchange point,
followed by a container at the depot
 Then, it proceeds to warehouse and it leaves the
trailer there.
 After that, a truck (any) will be assigned to
move that trailer to port. The truck then has to
leave the port with its trailer.
Trailer issue (Import)
 Import job order
 From initial location, a truck has to pick up a
correct trailer at TEP before it proceeds to port.
 Then it moves trailer to warehouse and leaves it
there.
 Finally a truck (any) will move this trailer from
warehouse to depot. Again the truck has to
leave depot with the trailer.
When to visit TEPs ?
 A truck has to pick up a correct trailer from
nearest TEP when
 It serves task type 1,2,5,6, 9 to 12 and
 It does not have a trailer with it OR the trailer it has
is not the correct type.
 A truck has to leave its trailer at TEP
 It serves task type 3,4,7,8 but
 It has a trailer with it.
Trailer number
 Trailer number changes due to pick-up and
return activities (segmentation of resource)
 Number of trailer available for pick-up =
number of trailer of previous time slot
+ trailers returned in previous time slot
- trailers picked-up in current time slot
HMOEA in use
 Test case generated: cover different scenario as
much as possible
 Altogether 28 test cases
 Perform computation on PCs
 Using HMOEA to solve
 Computational results
Discussion (1)
 Convergence test Cost improvement for test case 100_3_4
4650

 Average cost in Average routing cost

population
4600

Average routing cost


 Reduced over 4550

generations 4500

 Overall
4450

improvement
4400 10

13

16

19

22

25

28

31

34

40

46

49
1

37

43
Iteration
Discussion (2)
Average truck number for 100_3_4
50

45
Average truck number  Convergence test
40 for 2nd objective
35
 Average truck
Truck number

30

25
number
20  Reduced over
15

10
generations
1

10

13

16

19

22

25

28

31

34

37

40

43

46

49

Generation
Discussion (3)
Truck vs. Cost : Pareto-100-1-2
5100
First
Int 1
5000
Int 2
Final
4900
 Pareto dominance
4800
diagram
4700
Cost of routing

 Pseudo Pareto
4600
front
4500
 Improvement
4400
observed
4300

4200

4100
0 5 10 15 20 25 30 35 40
Number of truck
Discussion (4)
 Comparison of Comparison of Outsourced Jobs
trailers effect 30
1 or 2
25 2 or 3
3 or 4

Minimum outsourced jobs


20

15

10

0
100 112 120 132
Test cases category
Conclusion to TTVRP
 TTVRP maneuvers routing of containers within
specified time and resource allocated.
 Interesting effect of trailer: cost, truck number
required and utilization.
 Performance shows dependence on distribution
pattern of trailer resource.
Summary
Summary
 VRPTW solved using HMOEA
 Involves the optimization of routes for multiple
vehicles so as to meet all given constraints and to
minimize the objectives of travel distance and number
of vehicles
 Lower routing cost
 Better population distribution
 Good convergence trace
Summary (2)
 TTVRP inherits common properties of a
industry model VCPS
 Trailer resource in VRP
 Solved using HMOEA
 Useful information derived
Future Work
 Improve algorithm by distributed
computing technology: to reduce
computation time
Distributed MOEA (1)
 Why distributed ?
 Low cost but high performance in computation
Distribute MOEA (2)
 No free lunch
 Distributed algorithm MOEA design
 Research on partitioning algorithm
 Reduce communication overhead
 Achieve similar or better performance
 Involve heterogeneous system properties
References
 Evolutionary optimization
Ruhul Sarker, Masoud Mohammadian, Xin Yao, Boston : Kluwer academic
publishers, c2002.

 Introduction to management science


Bernard W. Taylor III, new jersey : prentice hall, 1999.
Q&A

You might also like