Advances in Intelligent Systems Research, volume 154

6th International Conference on Measurement, Instrumentation and Automation (ICMIA 2017)

Migrating birds optimization for hybrid flow shop scheduling problem

with makespan
Caile Ren1, a, Chaoyong Zhang1,b , Yanbin Zhao2,c and Leilei Meng11, d
1
School of Mechanical Science & Engineering, Huazhong University of Science and Technology
Luoyu Road No.1037, Wuhan, 430074, China
2
School of Mechanical and Electronic Engineering, WuHan University of Technology
Luoshi Road N0.122, Wuhan, 430070, China
a
[email protected], [email protected], c [email protected], [email protected]

Keywords: Hybrid flowshop; Makespan; Metaheuristics; Migrating birds optimization

Abstract. This paper use migrating birds optimization algorithm to minimize makespan on hybrid
flow shop problem firstly in the literature. According to the characteristics of hybrid flow shop
problem, this paper proposed a job sequence adjustment model based on probability. Then combined
the adjustment model with migrating birds optimization algorithm to solve standard benchmark
problems and proved its validity.

Introduction
Hybrid flow shop scheduling problem(HFSP for short), first proposed by Salvador in 1973[1]. For
HFSP, there is at least one stage having more than one processor. It is due to the existence of HFSP
characteristics of parallel machine, HFSP can be used for machinery, logistics, steel, textile and other
industries scheduling problem. So the study of HFSP has important significance for the actual
production.
Since HFSP arises, many approaches have been proposed in the literature. Ruizab[2] divided these
approaches into three categories: exact algorithms, heuristics, and metaheuristics. exact algorithms,
such as branch and bound (B&B)[3], solve problems mathematically; Heuristics algorithms refer to
some scheduling rules, Panwalkar et al.[4] summarized 113 different rules. Metaheuristics, such as
Genetic Algorithm (GA)[5], Artificial Bee Colony (ABC)[6], Ant Colony Optimization (ACO)[7]
and so on.
Migrating birds optimization (MBO) algorithm is a new metaheuristic algorithm which is
proposed by Duman et al. [8] in 2012 for solving quadratic assignment problems. MBO is inspired
from migrating birds’ V flight formation, which can reduce energy consumption during the flight.

Problem Statement
The description of HFSP is as follows: there are n jobs J={1,2,…,N} will be processed through k
stages S={1,2,…,K}, each stage has parallel machines. At the same time, following constrains must
be satisfied: 1)all jobs go through all stages in the same order; 2)at least one stage has more than one
parallel machine; 3)each job only can be processed on one parallel machine at each stage; 4)each
machine only can process one job at any time.

Migrating Birds Optimization Algorithm for HFSP

Encoding. The encoding of HFSP is divided into two categories. The first class is the matrix
encoding, that is, a matrix represents an individual, the rows and the columns of the matrix represent
stages and jobs respectively.The second category is permutation-based encoding, that is, each solution
is represented by a sequence of all jobs’ index. For HFSP, Matrix encoding is cumbersome to deal
with and does not make use of the characteristics of HFSP. Therefore, this paper uses
permutation-based encoding.

Copyright © 2017, the Authors. Published by Atlantis Press.

This is an open access article under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/). 689
 Advances in Intelligent Systems Research, volume 154

Decoding. Decoding methods are divided into two categories. The first class is replacement decoding,
that is, all stages use the same job sequence. The second type is permutation decoding, of which only
the first stage assign jobs according to the given job sequence, the following stage assign jobs
according to the completion time of jobs at last stage, that is to say, jobs with earlier finish time at
stage s have the priority to be processed at stage s+1. This paper uses permutation decoding.
However, if you follow rules of permutation decoding strictly, you may miss a better solution. Fig.
1 is a gantt chart for a HFSP case, which has 4 jobs and 2 stages, each stage has two parallel machines
respectively, different colors represent different jobs, the number on rectangles represents job index.
As we can see from (a), the completion time of job 4 is 10, which is smaller than the completion time
of the job 3 that is 11 at the first stage. According to the principles of permutation decoding , we
should first assign job 4 and then job 3 at the second stage, under this circumstance, the makespan is
27. However, if we first assign job 3 on machine 3 and then assign job 4 on machine 4, the makespan
will become 25 which is smaller than 27.

(a) (b)
Fig. 1 The gantt chart for one HFSP case
It can be seen from the above that follow the rules of permutation decoding strictly can get a good
solution to a large extent, but we may miss a better solution. Therefore, on the basis of permutation
decoding, this article proposed an job sequence adjustment model(JSAM for short) based on
probability. As shown in Eq.1, where P is the probability, C is a coefficient, e is the natural constant.
Firstly, sort job sequence in an ascending order according to jobs’ completion time, and then Δ can be
calculated as the difference between two adjacent jobs’ completion time.
(1)
Permutation decoding with JSAM described as follows:

690
 Advances in Intelligent Systems Research, volume 154

Fig. 2 The Pseudo code for JSAM

Population Initialization. Taking the diversity of the population into account is very important in
population initialization. Therefore, the population are generated using a random strategy and require
the newly generated individual to be different from any individual in the current population.
Neighborhood Structure. In the MBO algorithm, the neighborhood of all followers is composed of
two parts, that is, the neighborhood solutions generated by itself and the unused neighborhood
solutions of the previous individual. The neighborhood structure used in this paper is based on the
optimal insertion operation(OIO for short) and the optimal switching operation(OSO for short). For
OIO, we remove job at position p1 and reinsert it into position p2; For OSO, we exchange two jobs at
position p1 and position p2.
The Process of MBO Algorithm. Define symbols as follows: EG is the evolution generation of the
population; Psize is the size of population; k is the number of neighborhood solutions of the leader; x
is the number of neighborhood solutions passed to the next bird; G represents tour rounds; 错误!未
找到引用源。 represents neighborhood solution set passed to the next bird.
Step 1: initial all parameters, initial population;
Step 2: perform OIO and OSO on the leader to generate k neighborhood solutions, select the best
neighborhood solution to replace the leader and keep the unused x neighborhood solutions to
.
Step 3: perform OIO and OSO on left queue followers to generate k-x neighborhood solutions, select
the best neighborhood solution among k-x and to replace the corresponding follower. Clear
and keep the unused x neighborhood solutions in k-x to ;
Step 4: a similar operation like step 3 is conducted to followers in right queue;
Step 5: if step2 to steps 4 have conducted G times, replace the leader with the best follower in left or
right queue;
Step 6: conduct step 2 to step 5 until the population have developed EG times.

Computational results
The MBO algorithm was coded by C++ on an Intel Core i5 3210M PC with 4GB of memory.
According to Duman, The parameters of MBO were set as follows: EG=200,
Psize=25,G=10,k=3,x=1.
Using JMBO to solve 10 large scale benchmark problems. Result is shown in Table 1. We can see
that JMBO gets better solutions than HVNS, HDBAC, PSO, AIS [9], thus we can come to the
conclusion than JMBO is effective to solve makespan minimisation problem in HFSP.

Table 1 Comparison results for ten large scale problems

makespan %Deviation
JMBO HVNS HDBAC PSO AIS IMBO HVNS HDBAC PSO AIS
j30c5e1 463 464 467 471 479 0.00 0.22 0.86 1.73 3.46
j30c5e2 616 616 616 616 619 0.00 0.00 0.00 0.00 0.49
j30c5e3 594 595 595 602 614 0.00 0.17 0.17 1.35 3.37
j30c5e4 565 566 568 575 582 0.00 0.18 0.53 1.77 3.01
j30c5e5 600 601 603 605 610 0.00 0.17 0.50 0.83 1.67
j30c5e6 602 603 605 605 620 0.00 0.17 0.50 0.50 2.99
j30c5e7 626 626 626 629 635 0.00 0.00 0.00 0.48 1.44
j30c5e8 674 674 675 678 686 0.00 0.00 0.15 0.59 1.78
j30c5e9 642 643 645 651 662 0.00 0.16 0.47 1.40 3.12
j30c5e10 573 577 580 594 604 0.00 0.70 1.22 3.66 5.41

691
 Advances in Intelligent Systems Research, volume 154

Acknowledgements
The authors would like to thank Jasques Carlier and Emmanuel Neron for their benchmark problems.
This research is supported by Cooperation and Exchange of the National Natural Science Foundation
of China(no.51561125002), the National Natural Science Foundation of China (nos. 51275190,
51575211).

References
[1] A solution to a special class of flow shop scheduling problems. Symposium on the Theory of
Scheduling and its Applications (North Carolina State Univ., Raleigh, N. C., 1972), pp. 83–91.
Lecture Notes in Econom. and Math. Systems, Vol. 86, Springer, Berlin, 1973.
[2] Ruizab R. The hybrid flow shop scheduling problem[J]. European Journal of Operational
Research, 2010, 205(1):1-18.
[3] Carlier, Neron J, Emmanuel. An Exact Method for Solving the Multi-Processor Flow-Shop[J].
RAIRO - Operations Research, 2000, 34(1):1-25.
[4] Panwalkar S S, Iskander W. A Survey of Scheduling Rules[J]. Operations Research, 1977,
25(1):45-61.
[5] Xiao W, Hao P, Zhang S, et al. Hybrid flow shop scheduling using genetic algorithms[C]//
Intelligent Control and Automation, 2000. Proceedings of the, World Congress on. IEEE,
2000:537-541 vol.1.
[6] Cui Z, Gu X. An improved discrete artificial bee colony algorithm to minimize the makespan on
hybrid flow shop problems[J]. Neurocomputing, 2015, 148(148):248-259.
[7] Alaykýran K, Engin O, Döyen A. Using ant colony optimization to solve hybrid flow shop
scheduling problems[J]. International Journal of Advanced Manufacturing Technology, 2007,
35(5-6):541-550.
[8] Duman E, Uysal M, Alkaya A F. Migrating Birds Optimization: A new metaheuristic approach
and its performance on quadratic assignment problem[J]. Information Sciences, 2012, 217(24):65-77.
[9] Li J Q, Pan Q K, Wang F T. A hybrid variable neighborhood search for solving the hybrid flow
shop scheduling problem[J]. Applied Soft Computing Journal, 2014, 24(24):63-77.

692