15
© 2007, Global Institute of
Flexible Systems Management
Introduction
The rapid advancement of technology is resulting in
shortened life cycles and putting great pressure on
organizations to achieve shorter and shorter manufacturing
lead times. Hence, one of the important focuses of
manufacturing enterprises is to find ways and means of
reducing the manufacturing lead times to be able to respond
quickly for changing market demands. Towards this end,
scheduling has been always been considered as one of the
most important issues for lead time reduction in a mixed
model flexible system. Due to this complexity of scheduling
problems an efficient and effective advanced scheduling
approach is in great demand. Considering a mixed model
flow shop scheduling problem, where ‘n’ independent jobs
required to be processed on ‘m’ different machines, as a case
representing a flexible system. Flow shop scheduling has
been proved to be NP-hard in strong sense [Garey et. al,
1979]. Till now, mathematical programming, constructive
heuristics and meta-heuristics have been proposed for a
generic flow shop scheduling [Reeves, 1995; Ogbu, 1990;
Wang et. al, 2003, Chan et. al 2005], where it is usually
assumed that the buffer size between every two successive
machines is infinite. However, in real practice, the
A Genetic Algorithm Based Scheduling for a Flexible System
S. Wadhwa
Department of Mechanical Engineering
Indian Institute of Technology Delhi, New Delhi, India
J. Madaan
Department of Mechanical Engineering
Indian Institute of Technology Delhi, New Delhi, India
R. Raina
Department of Mechanical Engineering
Indian Institute of Technology Delhi, New Delhi, India
Abstract
Based on the rapid technical development in the last decade, the automation in many areas of production and distribution has
developed significantly. This leads to complex situations where decisions have to be taken within a short time and among
several alternatives – often without the human intervention. This paper considers flexible system as a mixed model flexible flow
line, where ‘n’ independent jobs are required to be processed on ‘m’ different machines, where all the jobs have the same
processing order on the machines. Here the objective is to find the ordering of the jobs on the machines that minimizes the
make-span. Above objective can be achieved through evolutionary based heuristic of Genetic Algorithm (GA) on the flow line
scheduling problem. The advantage of the GA approach here is the ease with which it can handle constraints and objectives;,
making it easy to adapt the GA scheduler to the particular requirements of a very wide range of possible line scheduling
problem. The proposed architecture has been developed to depict the application of genetic algorithms to optimize production
schedules in a flexible flow line system representing a flexible system. Results show that the implementation of the genetic algorithm
is very effective as compared to standard sequencing rules like shortest processing time, total processing time, etc. and at the
same time easy to use. Finally this paper intends to discuss some of these interesting results with a focus on application of lead-
time reduction in an interesting flexible system like RES (Reverse Enterprise System).
Keywords: flexible system, flexible flow line, genetic algorithm, reverse enterprise system, simulator
petrochemical processing industries, supply chains system
with returns (reverse logistics) and electronic chip
manufacturing system can be considered as example of
system where buffer is non-existent. In such case, a job is
called blocked if its operation on a certain machine has
finished but the machine after that is still busy. In these
situations GA approach is the easiest because it can handle
variety of constraints and objectives of flexible systems; all
such cases can be handled as weighted components of the
fitness function, making it easy to adapt the GA scheduler
to the particular requirements of a very wide range of
possible overall objectives of generic flow shop scheduling.
A number of knowledge-based system (KBS) techniques
for reconfiguring flexible systems have been well proposed
in the literature. Wadhwa & Brown (1989) had proposed
potential of basic Petri net concepts for graphic
representation, specification purposes, and control of flow
in a flexible system. Since these KBS technique rely heavily
on past experience. The required information sometimes may
not be available or difficult to be obtained. Hence genetic
algorithm (GA) is well a suited tool to tackle such complex
reconfiguration problems, because GA uses information-
randomized (genes) exchange to exploit and tackle complex
giftjourn@l
Global Journal of
Flexible Systems Management
2007, Vol. 8, No. 3, pp 15-24
Model