1
INTRODUCTION
Flexible job shop scheduling problem (FJSP) is an
extension of classical job shop scheduling problem (JSP)
and exists in many real-life situations, e.g. manufacturing
and remanufacturing. Remanufacturing is the process of
disassembly and recovery at the module level and,
eventually, at the component level [1]. It requires the repair
or replacement of worn out or obsolete components and
modules. Parts subject to degradation affecting the
performance or the expected life of the whole are replaced.
Remanufacturing is a form of a product recovery process
that differs from other recovery processes in its
completeness: a remanufactured machine should match the
same customer expectation as new machines [2]. Guide Jr
[3] considered the production planning when inputs have
different and uncertain quality levels and discussed
different decision variables in remanufacturing
engineering. Junior [4] reviewed the literatures on
production planning and control in remanufacturing.
Seventy-six papers were examined and classified.
However, there are few literatures on reprocessing
scheduling in remanufacturing. Uncertainty in timing of
returns is one of seven major complicating characteristics
in remanufacturing [5]. Li [6] proposed a dynamic
programming approach to derive the optimal solution in the
case with large returned products and different arriving
time. Li [7] proposed a simulation optimization model with
a prioritized stochastic batch arrival mechanism to plan and
control the remanufacturing process. The uncertainties in
processing timing and quantity are factors cannot be
controlled by remanufacturers. It means that new returned
product(s) or job(s) may need to be inserted into the
ongoing existing scheduling solution. It is therefore
important to handle this uncertainty in remanufacturing
This work is supported by National Nature Science Foundation under
Grant
61603169, 61503170.
scheduling. In our research, the uncertainty of the returns is
model as new job insertion. After new job insertion,
rescheduling is executed based on the insertion time and the
available time of machines and existing jobs. The authors
of this study have studied on the remanufacturing
rescheduling for new job insertion with makespan and
maximum machine workload objectives [8, 9]. In real-life
shop floor, if the processing machine of one operation is
changed, the preparation work will have to be repeated on
new processing machine. The components or semi-finished
products will have to be moved from the current machine to
the new one. In this way, the time cost and labour cost will
increase. Hence, the stability of rescheduling solution is an
important metrics in real shop floor. This is the motivation
of this study and the instability of rescheduling solution
should be minimized.
As many discrete manufacturing systems, remanufacturing
processes can be modeled as FJSP problem. Due to the
complexity and multiple constraints of FJSP, the high
computational time in optimization becomes the major
hurdle for the real-time scheduling strategy. Among
different approaches, meta-heuristics are attractive for the
reasonable computation time and high quality solution. Jaya
algorithms is a simple and new meta-heuristic, which is
proposed by R. Venkata Rao [10]. Comparing to the most
existing meta-heuristic algorithms, the advantage and
competitiveness of Jay algorithm is that Jaya does not
require any algorithm-specific parameters and require only
common controlling parameters like population size and
number of generation for the launching. Jaya is
comparatively simpler to apply. In literature [10], Jaya has
been compared with several mete-heuristics, e.g. Genetic
Algorithm (GA), Differential Evolution (DE), Particle
Swarm Optimization (PSO), Artificial Bee Colony (ABC)
and TLBO. The experiments and discussions verify the
Discrete Jaya algorithm for solving flexible job shop rescheduling problem
Jing
Guo
1
, Kaizhou
Gao
2
, Chao Wang
1
, Hongyan Sang
2
, Junqing Li
2
, Peiyong Duan
2
1. College of
electrical and information engineering, Yangzhou Polytechnic Institute, Yangzhou, P.R. China 225000
E-mail: cnguojing@qq.com
2. School of computer, Liaocheng University, Liaocheng, P.R. China 252000
E-mail: gaokaizh@aliyun.com
Abstract: Rescheduling is a necessary procedure when new jobs come and are inserted into existing schedule in
remanufacturing. The stability is an important metrics to evaluate the quality of rescheduling solution. This paper
proposed a novel discrete Jaya algorithm to solve flexible job shop rescheduling problem with five objectives, including
Makespan, total flow time, maximum machine workload, total machine workload, and stability metrics. The purpose is
to analyze the influence or conflict between rescheduling stability and other objectives. Experiments are executed on a
large scale remanufacturing case. When the stability metric is optimized, the convergence history of other objectives are
recorded. The convergence curves of stability metric are counted when the four objectives are optimized individually.
The results and discussions show that the stability metrics is conflict or conflict to a certain extent with other objectives.
Key Words: Jaya algorithm, flexible job shop scheduling, rescheduling, remanufacturing, instability
6010
978-1-5090-4657-7/17/$31.00
c
2017 IEEE