Scheduling by Applying Tabu Search: A Textile Case

1
Scheduling by Applying Tabu SearchA Textile Case

• Prepared by
• Ali Orhan AYDIN

2
Scope of the Presentation

• To present a study
• which
• explains job scheduling example
• in a textile system
• by applying Tabu Search

3
Introduction

• Systems are set of components which are related
  by some form of interaction, and which act
  together to achieve some objectives.
• One of the most important man-made systems is
  production system.
• Major aim of the most of the manufacturing and
  service systems is to make profit.
• Therefore, they produce goods and services by
  using scarce resources.
• To achieve this purpose efficiently use of
  resources becomes key succession factor.
• Scheduling promises in helping production systems
  to pursue this goal.

4
Introduction

• Scheduling is a series of activities aim of which
  is to assign jobs and/or resources to men and/or
  machines in production systems.
• By performing these activities, it is targeted to
  minimize production time and costs, by providing
  information to a production system on what to
  make, when to make, with which staff, and on
  which machine.

5
Introduction

• Many tools and approaches are proposed to achieve
  a good schedule.
• Pioneer tool for scheduling and planning is Gantt
  Chart.

6
Introduction

• Most basic method for scheduling purposes is to
  use dispatching rules.
• SIRO Service in Random Order
• ERD Earliest Release Date First
• EDD Earliest Due Date First
• MS Minimum Slack First
• WSPT Weighted Shortest Processing Time First
• LPT Longest Processing Time First
• SST Shortest Setup Time First
• CP Critical Path
• LNS Largest Number of Successors
• SQNO Shortest Queue at the Next Operation

7
Introduction

• There are also some composite dispatching rules.
• ATC Apparent Tardiness Cost is a rule that
  combines WSPT and MS.
• ATCS Apparent Tardiness Cost with Setups is a
  rule that combines WSPT, MS and SST.
• All of these dispatching rules prioritize all the
  jobs that are waiting for processing on a
  machine.

8
Introduction

• To find optimum schedule there are exact
  optimization methods.
• If the scheduling problem is inherently easy
  linear programs can be used to solve them.
• Integer programming
• Branch-and-bound methods
• Cutting plane methods
• Some hybrid methods

9
Introduction

• On the other hand, scheduling problems are
  usually Non-Deterministic Polynomial-Time Hard
  (NP-Hard).
• In some cases, solving NP-Hard problems may take
  enormous time. Usually, in real life that amount
  of computer time is not available. Therefore,
  finding optimal solution is nearly impossible.
• Many methods proposed to find a good solution to
  scheduling problems of production systems in a
  relatively short time.
• Those solutions are acceptable and feasible
  solutions that presumably is not far from optimal.

10
Introduction

• Many methods proposed to find a good solution to
  scheduling problems of production systems in a
  relatively short time.
• In such cases, beam search, local search and
  global search heuristics can be applied.

11
Introduction

• Local Search Heuristics
• Hill-Climbing
• Min-Conflicts
• Min-Conflicts-Random-Walk
• Steepest-Descent-Random-Walk
• GSAT
• WalkSat
• Simulated Annealing
• Tabu-Search.

12
Introduction

• Global Search Heuristics
• Evolutionary algorithms
• Genetic algorithms
• Memetic algorithms
• Population based algorithms.
• Particle swarm algorithms
• Ant colony algorithms
• Bees algorithms

13
Introduction

• All of these approaches have some advantages and
  disadvantages.
• It is a fact that global search heuristics
  requires more time to achieve acceptable
  solution.
• On the other hand, global and local search
  heuristics give nearly the same results.
• Therefore, in this paper one of the local search
  heuristic tabu search is applied in scheduling
  problem of a textile system.

14
Literature Review Greedies

• Among many heuristic algorithms local search
  algorithms are reviewed in this section.
• In local search, an initial configuration
  (valuation of variables) is generated and the
  algorithm moves from the current configuration to
  a neighborhood configurations until a solution
  (decision problems) or a good solution
  (optimization problems) has been found or the
  resources available are exhausted.
• Usually, local search algorithms are called as
  greedy algorithms since, they try to reach
  global maximum or minimum by searching the space
  locally.
• Local search algorithms move from solution to
  solution in the space of candidate solutions (the
  search space) until a solution deemed optimal is
  found or a time bound is elapsed.

15
Literature Review Greedies

• General Psuedo Code of local-search algorithms

16
Literature Review Greedies

• Hill-climbing is probably the most known
  algorithm of local search.
• Algorithm of hill-climbing

17
Literature Review Greedies

• Psuedo Code of Min-Conflicts

18
Literature Review Greedies

• Min-Conflicts-Random-Walk algorithm

19
Literature Review Greedies

• Steepest-Descent-Random-Walk algorithm

20
Literature Review Greedies

• Tabu Search algorithm

21
Literature Review Greedies

• Tabu search is applied in many resource
  allocation problems like
• cell formation problem
• designing manufacturing cells
• optimization of Process Plans
• Parallel Flowshop Scheduling
• project scheduling
• flow-shop scheduling
• job shop scheduling

22
Literature Review Greedies

• There are also hybrid algorithms which are
  developed by combining tabu search and other
  heuristics to find better solutions to the
  problems.
• Tabu Search and Simulated Annealing is combined
  by Zolfaghari and Liang
• Tabu Search and Genetic Algorithm is hybridized
  by Ombuki and Ventresca
• Tabu and Scatter Search is combined by Blazewicz,
  Glover, and Kasprzak

23
Literature Review Greedies

• Moreover, like it is aimed in this paper, tabu
  search is used in scheduling jobs in textile
  manufacturing systems and proved to be efficient.
  
• In their paper, Tucci and Rinaldi 31 describes
  a typical fabric production system and applies
  tabu search.

24
Applying Tabu Search A Textile Case

• Aim of this paper is to apply tabu search in a
  textile manufacturing industry.
• Specifically, fabric production is taken under
  consideration. In such systems, to produce fabric
  first raw strings are dyed. Afterwards, they are
  tied to conics and sent to fabric department to
  be weaved. Final stage is quality control and
  while controlling fabric, they are rolled on
  cylinders.

25
Applying Tabu Search A Textile Case

• This paper focuses on scheduling jobs in weaving
  process while trying to minimize maximum lateness
  on single machine.

26
Applying Tabu Search A Textile Case

• Bill-of-Material of such fabric product

27
Applying Tabu Search A Textile Case

• As an example, here 20 jobs are described.

28
Applying Tabu Search A Textile Case

• Length of ordered fabrics and total processing
  time of each job.

29
Applying Tabu Search A Textile Case

• Due dates of jobs.

30
Applying Tabu Search A Textile Case

• In the frame of the given information, initial
  solution for the problem is obtained by applying
  dispatching rule Earliest Due Date.

31
Applying Tabu Search A Textile Case

• Afterwards, Tabu Search is applied. 20 iterations
  are performed by using OpenTS open source code in
  JAVA environment.

32
Conclusion

• Tabu Search is an effective algorithm to achieve
  job scheduling objectives like maximum
  lateness/tardiness minimization.
• It requires less processing time than global
  search heuristics.
• As the job size increase, time requirement also
  increases to generate a schedule.
• Therefore, Tabu Search can be suggested to mass
  production systems since, they deal with many
  jobs.

33
Conclusion

• The study just tried to simply explain the Tabu
  Search concept.
• The problem provided is oversimplified.
• Actual textiles systems deal with much more
  higher amount of jobs.
• Because, it is aimed to show how Tabu Search can
  be used in minimize maximum tardiness problem.
• In this manner, a good feasible solution but not
  optimal is found.

34

• THANK YOU
• FOR
• LISTENING