Evolutionary Intelligence

https://doi.org/10.1007/s12065-022-00767-2

RESEARCH PAPER

Fire and manoeuvrer optimizer for flow shop scheduling problems

Jabrane Belabid1

Received: 12 April 2022 / Revised: 7 July 2022 / Accepted: 22 July 2022

© The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2022

Abstract
The purpose of this paper is to introduce a new rich source of ideas and techniques that could be used to build new algorithms
capable to solve numerous encountered optimisation problems in different fields of science and engineering. The art of war is
one of the most rich disciplines in terms of already experimented strategies and tactics that can inspire researchers to design
new powerful and efficient metaheuristics. The framework of the proposed method are inspired by the main war phases and
contains seven components: initialization, intelligence, conception, suppression, advance, assault and exploitation. The basic
fire and manoeuvre tactic is adopted in the suppression and advance phases. The proposed fire and manoeuvre algorithm
(FMA) is a hybridization of a greedy algorithm with a multi-neighbouring search procedure. The developed algorithm has
been employed to minimize makespan of the classical flow shop scheduling problem. A mathematical model is presented
to describe the studied optimisation problem. Comparative experiments on Taillard’s data set confirmed that the (FMA)
results are more accurate than already published data. A comparison between the FMA and other popular nature-inspired
algorithms has been conducted. It was revealed that the proposed metaheuristic outperforms the classical genetic algorithm,
the migrating birds optimisation and the whale optimisation algorithm.

Keywords Hunting algorithm · Flow shop · Mixed-integer linear programming · Nature inspired algorithm

1 Introduction Nonetheless, these exact methods still limited and can’t solve
medium or large sized flow shop problems.
The flow shop problem has attracted huge attention as an Thus, the pioneering work of Johnson [3] presented a
important issue in the field of scheduling during the last method to solve a two and three machines flow shop produc-
decades. This attention is due to its wide range of engineer- tion system. The second category of algorithms is divided
ing applications in manufacturing environment. Many varia- into two subcategories, namely, heuristics and metaheuris-
tions on the classical flow shop problem have been proposed tics. However, these algorithms are used when we do not
to approach the different encountered scheduling problems look for the best solution but in fact an easily reachable fea-
under various constraints and criteria. The most fundamental sible one. Heuristics are methods that enable to discover
model of flow shop problem with the makespan minimiza- by trial or error a not guaranteed optimal solution of an
tion objective Fm ∕prmu∕Cmax has been studied extensively. optimization problem. In other words, heuristics are prac-
The commonly used approaches to solve the flow shop prob- tical methods inspired from empirical procedures used to
lem are classified into two categories, namely, exact and approach similar complex problems that can provide an
approximation methods. The first category that includes acceptable solution in a reasonable running time. Therefore,
linear programming [1] and branch-and-bound [2] method metaheuristics are the further development of heuristic algo-
showed good results for small-sized flow shop problems. rithms. In general, the metaheuristics outperform classical
heuristics. Lately, one can remarks easily that all algorithms
based on a combination of randomization and local search
are called metaheuristics by researchers. Since the probably
* Jabrane Belabid first heuristic [4] implemented by Alan Turing in 1945 to
[email protected] crack the Enigma code, heuristics have undergone crucial
1
Laboratory Mathematics and Applications, University
developments over time [5–12].
Hassan II of Casablanca, FST, PO Box 146, Mohammedia,
Morocco

13
Vol.:(0123456789)

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

 Evolutionary Intelligence

Recently, researchers have taken advantage of the pro- enhanced discrete fruit fly optimization algorithm (HED-
gress made in metaheuristics to better approach the flow FOA) is proposed by Shao et al. [33] to approach the dis-
shop problems [13, 14]. Framinan et al. [15] published a tributed blocking flow-shop scheduling problem. They found
review in which the heuristics employed to minimize makes- that the proposed HEDFOA provides the highest-quality
pan in a flow shop problem are classified into fixed func- solutions than some state-of-the-art methods.
tional heuristics, floating functional heuristics, and synthetic Recently, Li et al. [34] proposed a discrete artificial bee
functional heuristics. In 2004 Ying and Liao [16] were the colony (DABC) algorithm to solve a distributed hetero-
first to use the ant colony system to minimize the makespan geneous no-wait flow shop scheduling problem. Numeri-
in a flow shop production system. The authors found that cal results evince that the DABC outperform all existing
the ant behaviour inspired metaheuristic is more powerful algorithms. Yankai et al. [35] investigated the hybrid flow
in solving flow shop problem and present more effective shop scheduling problem with the objective to minimize
results than genetic algorithm, simulated annealing, and the makespan and the energy consumption by designing an
neighbourhood search. In many monographs dealing with improved multi-objective whale optimization algorithm to
the flow shop scheduling problems, researchers designed visualize the Pareto optimal set. Results demonstrate that
many techniques inspired by theoretical immunology to the IMOWOA provides better performance compared with
better approach the studied problems in a relatively reason- SPEA2 and NSGA-II. In a related work, Ding et al. [36]
able computational time (see for instance [17–19]). There- implemented a new hybrid particle swarm optimization
fore, various nature-inspired algorithms were developed to algorithm (HPSO) with integrated tabu search to minimize
solve flow shop sequencing problems such as genetic algo- makespan and energy cost in the flexible flow shop schedul-
rithms [20, 21], migrating birds optimization [22], cuckoo ing systems. Extensive computational results show that the
search [23], bat algorithm [24], ant lion optimization [25], HPSO is superior to the NSGA-II algorithm.
grasshopper optimization algorithm [26] and butterfly Besides, the secrets of natural phenomena and the
algorithm [27]. behaviors of living beings continue to inspire researchers
Marichelvam et al. [28] used a discrete fire fly algorithm to develop more efficient algorithms capable to better solve
to minimize the total tardiness in a hybrid flow shop system various optimization problems. In this investigation, we aim
with sequence dependent setup times. They found that the to enrich the metaheuristic portfolio with a new algorithm
proposed bio-inspired metaheuristic algorithm is successful inspired by the art of war. The steps of the proposed algo-
to approach the studied flow shop problem. Deb et al. [29] rithm are inspired by the phases of war. Indeed, the coor-
showed that the rhinoceros’ natural behaviour inspired dination between different army components is essential to
metaheuristic presents better performance in solving permu- carry out a winning strategy. This concept is adopted in this
tation flow shop problems than the particle swarm optimiza- paper to design a metaheuristic resulting from the better
tion. In the work of Zhu et al. [30] a quantum description coordination between several algorithms. It is important to
of the solution is adopted to propose a quantum-inspired note that Ayyarao et al. [37] designed a new optimization
cuckoo co-search algorithm (QCCS) in order to solve the no- algorithm based on ancient war strategy where the positions
wait flow shop scheduling problem. Results showed that the of soldiers during wartimes are updated according to the
quantum approach (QCCS) outperforms some well-perform- adopted strategy. In fact, the classical attack and defence
ing algorithms such as GA-VNS, HGA, TS-PSO, TMIIG. strategies have been implemented.
In a recent work, Wang et al. [31] used a multi-objective In addition, strategy may be defined as the logical
whale swarm algorithm ( MOWSA) to minimize makespan approach that leads to the goals in accordance with the avail-
and energy consumption in a distributed flow shop produc- able resources. Individuals, families, cities, states and organ-
tion system. The authors demonstrate that the proposed ( izations have their logical approach to achieve their goals.
MOWSA) provides better results than other well-known On the level of state, a country has its own national strategy
algorithms like NSGA-II, SPEA2, PAES and MDEA. which covers the diplomatic, economic, social and military
Marichelvam et al. [32] proposed an improved swarm opti- aspects. As far as this last pattern is concerned, the historical
mization algorithm (IPSO) to minimize the weighted sum roots of military strategy date back to the origins of human
of makespan and total flow time of a hybrid flow shop pro- warfare and the development of large-scale empires. The
duction environment under the influence of human factors, need of more resources have led ancient empires to recon-
namely, skills, age, learning and forgetting factors. Exten- sider their military engagements in a much more scientific
sive computational experiments prove the superiority of the approach.
proposed IPSO over the SPT dispatching rule, NEH, genetic Greek etymology defines the strategy as the fact of lead-
algorithm, hybrid water flow like algorithm, tabu search, ing the army. In this regard, Plato stressed on the fact that
variable neighbourhood search based on a priori approach, war and therefore strategy, must be subject to politics [38].
and water flow like algorithm. In another work, a hybrid In other words, the military art is a field that belongs to the

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Evolutionary Intelligence

government. Before the Greeks, Sun Tzu who lived during the makespan criterion. It can be shown as n∕m∕F∕Cmax
the period of the Warring States period (6th century B.C.) according to Conway et al. [48] notation or F∕prmu∕Cmax
was the first known as theorist of strategy. He wrote a book as denoted by Graham et al. [49] where n and m are the
on strategy named the Art of War [39]. This latter still being numbers of jobs and machines, respectively, F stands for the
studied by all strategists. It’s an up-to-date book which dealt flow shop, prmu refers the permutation constraint and Cmax
with the non-military means in warfare. 1794 marked the which characterizes the maximum completion time of all
appearance of the word strategy in the European countries. jobs designates the criterion to minimize. The assumptions
During the 18 and the 19th centuries, several military gener- of this scheduling problem are as follows:
als focused their attention to the study of war.
Antoine Jomini (1779–1869), defined strategy as the art • Each machine can handle only one job at a time and each
of making war upon the map [40].The Prussian general Carl job can be processed only by one machine at a time.
von Clausewitz (1780–1831) analysed the dual aspect of • The preemption and interruption are not permitted.
strategy that matches the means with the ends [41]. Whereas • All jobs are available at time zero.
Sir Basil Henry Liddell Hart (1895–1970) introduced the • All machines are permanently available.
concept of grand strategy [42]. New technologies together • Each machine processes the jobs in the same order.
with the Nuclear weapons have led to the emergence of • The operations set-up times are sequence independent
new strategies. The use of sea and air powers requested and considered as part of the processing times.
new approaches that integrated those fields. In this regard,
theorists such as Giulio Douhet (1869–1930) [43] and Billy The following notation will be used to build the mathemati-
Mitchell (1879–1936) [44] insisted on the fact that air cal model:
power alone could win wars whereas Alfred Thayer Mahan
(1840–1914) [45] introduced the importance of sea power • m is the number of machines.
in modern warfare. Nevertheless, the military strategy still • n is the number of jobs.
brings up new concepts. Therefore, military staff are now • J is a set of n jobs J = {J1 , J2 , ..., Jn }.
setting the rules of engagement in the digital battlefields in • M is a set of m machines M = {M1 , M2 , ..., Mm }.
order to face cyberwars (see for instance [46, 47]). • pi,k is the processing time of job Jk on machine Mi.
The coming parts of this paper are organized as follows: • Ci,j is the completion time of job j on machine Mi.
Sect. 2 provides the problem description and solution pres- • Cmax is the makespan.
entation with a mixed integer linear programming model.
Section 3 shows the new developed metaheuristic. Section 4 Since the objective is to minimize the completion time of all
presents the analysis of computational experiments. Finally, jobs, the following formulas are employed to calculate Cmax:
we draw some concluding remarks and present some sugges-
tions for the further works.
C1,1 = p1,1 (1)

The completion time of the job j on the first machine is:

2 Description of the problem C1,j = C1,j−1 + p1,j j = 2, ..., n (2)

In the highly competitive manufacturing environment, the The completion time of the first job on the other machines is:
produced jobs are divided into a group of small tasks oper- Ci,1 = Ci−1,1 + pi,1 i = 2, ..., m (3)
ated in the same production route. Defined as the flow shop
production system, this technique highly impacts the cost The end date of a job j on the other machines is defined by:
and speed of manufacturing of a product and influences the
profits of companies that they adopt it. Each operation may
Ci,j = max{Ci,j−1 , Ci−1,j } + pi,j (4)
require suitable raw materials, tools, equipment, experts
and workers, and is characterized by the processing time i = 2, ...m;j = 2, ..., n
needed to be executed. Therefore, the jobs pass through the
same route composed of work stations, or machine groups. The makespan of a given sequence in the search space can
In other words, the flow shop is an industrial system in be expressed as:
which a set of n jobs is processed in a set of m machines Cmax = max {Cm,j } (5)
M = {M1 , M2 , ...., Mm } going through the same route. It 1≤j≤n

should be noted that many variations of this classical flow It is worth mentioning that the job-permutation-based rep-
shop structure exist in the real-word production systems. resentation is adopted in this investigation. Therefore, each
The most fundamental model is the standard flow shop with

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

 Evolutionary Intelligence

permutation is represented by an array containing a string Constraint set (7) ensures that each position is occupied
of numbers which refer to jobs. by only one job. Constraint set (8) specifies that each job
In the recent past years, there has been a renewed interest must be assigned to a position. Constraint set (9) estimates
in the mathematical modelling of scheduling problems and the completion time of the first job on the first machine.
especially those of flow shop. The mixed integer linear pro- Constraint set (10) controls that the processing of a job on
gramming (MILP) (see [50–53]) is one of the most successful the first machine begins after the processing of its prede-
of these mathematical approaches. The MILP model for the cessor on the same machine. Similarly, constraint set (11)
permutation flow shop problem (PFSP) with the objective of controls that the processing of a job at each machine can
makespan minimization as shown in constraint 6 consists of only begin after its processing on the previous machine.
the following variables: Constraint set(12) enforces that each job can start only
when the processing of its predecessor job on the same
⎧ 1, if job j is assigned to the machine is completed. Expressions (13) and (14) define
⎪
xj,k = ⎨ position k of the sequence the domain of decision variables. Figure 1 illustrates an
⎪ 0, otherwise example of scheduling a set of five jobs on a production
⎩
line composed of there machines. The processing times are
i, j = 1, ..., n given in Table 1. The sequences 𝜋 1 = {J3 , J1 , J2 , J4 , J5 , } and
Ci,k = Completion time of job at position 𝜋 2 = {J4 , J5 , J2 , J3 , J1 , } are tested. It is clearly seen how the
k on machine i order of jobs can influence the completion time of all jobs.
One can remarks that the jobs are continuously processed on
k = 1, .., n; i = 1, .., m
the first machine. However, some waste of time is generated
Then, the model continues as follows: by the waiting time caused by scheduling jobs on the second
Objective function: and third machines. In conclusion, the job order in the first
machine is a determining factor that should guide the search
Minimize Cmax = Cm,n (6) for the optimal solution.
Subject to:
∑
n
3 Hypothesis and limitations
xj,k = 1, k = 1, ..., n (7)
j=1
Traditional mathematical techniques frequently encounter
difficulties in facing the escalating complexity of real world
∑
n
optimization problems. Therefore, researchers try to imi-
xj,k = 1, j = 1, ..., n (8)
k=1
tate nature to introduce new methods more efficient. Thus,
different nature inspired algorithms have been employed
∑
n lately to approach solutions in diverse complex problems
C1,1 ≥ xj,1 × p1,1 ; j = 1, ..., n (9) of science or engineering. Among these biological inspired
j=1 processes one can cite evolutionary algorithms [54], artifi-
cial immune systems [55], ant colony optimization [56], bee
∑
n colony [57, 58], particle swarm optimization [59], dwarf
C1,k ≥ C1,k−1 + xj,k × p1,j ; k = 2, ..., n (10) mongoose optimization [60], arithmetic optimization algo-
j=1 rithm [61], Aquila optimizer [62], reptile search [63], Ebola
optimization search [64] etc. In this investigation we try to
∑
n
introduce a new type of metaheuristics inspired by the art
Ci,k ≥ Ci−1,k + xj,k × pi,j , i = 2, ..., m; k = 1, ..., n (11) of war. Therefore, the annals of military strategy are rich in
j=1
successful techniques and methods that deserve to be used in
other fields of science and engineering such as optimization.
∑
n
Ci,k ≥ Ci,k−1 + xj,k × pi,j , i = 1, ..., m k = 1, ..., n (12)
j=1
Table 1  The processing time of Job J1 J2 J3 J4 J5
5 jobs on 3 machines
Ci,k ≥ 0, i = 1, ..., m; k = 1, ..., n (13) M1 9 8 7 3 5
M2 6 4 5 6 4
xj,k ∈ {0, 1}, j, k = 1, ..., n (14) M3 6 6 8 5 5

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Evolutionary Intelligence

Fig. 1  Gantt chart for a flow Cmax = 44

shop problem composed of five
jobs and three machines M1 J3 J1 J2 J4 J5

M2 J3 J1 J2 J4 J5

M3 J3 J1 J2 J4 J5

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 time
1
(a) : π = {J3 , J1 , J2 , J4 , J5 , }
Cmax = 40

M1 J4 J5 J2 J3 J1

M2 J4 J5 J2 J3 J1

M3 J4 J5 J2 J3 J1

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 time
2
(b) : π = {J4 , J5 , J2 , J3 , J1 , }

The designed metaheuristic, categorized as hybrid social- In the pseudo-code, the stages of intelligence and con-
based algorithm according to Alatas and Bingol classifica- ception are translated by the construction of good solutions
tion [65, 66], is derived from the tactics of classical armed obtained through a random search procedure. The employed
conflicts in which different strategic concepts are imple- tactic called fire and movement uses suppressive fire to desta-
mented in the battlefield (search space). Indeed, the algo- bilize the enemy and weaken the cohesion of their organiza-
rithm consists of seven phases, namely, initialization, intelli- tion. It consists of dividing the unit in two major element,
gence, conception, suppression, advance, assault and finally the supporting element and the moving one. The supporting
exploitation. The initialization phase consists of generating element of the military unit fires from behind to distract the
an initial solution of the problem at random or using a heu- enemy, while the moving element advances rapidly while
ristic like GRASP or NEH. The phase called intelligence hiding by terrain obstacles as illustrated in Fig. 2. Next, the
is widely documented in the works of Sun Tzu [39] and moving element will set up a base of fire in order to allow the
Clausewitz [41]. It is manifested by the collection of data supporting element to move forward. The two unit elements
about terrain, enemy and adversary’s system through the will exchange roles until they become close sufficiently to the
implementation of technical and human assets available to enemy position. After that the two unit elements launch an
the various components of the armed forces, namely, the assault to destroy the enemy or reach the desired objective. In
land, maritime and air forces. This phase enables a clear other words, the FM tactic could be performed in three stages,
vision about the terrain and the enemy. Hence, the study of namely, suppression, advance and assault.
enemy’s points of strength and weakness aims to prepare the The position of moving element in the FM tactic is cal-
friendly manoeuvrer by the choice of suitable means and tac- culated using the following equation:
tics and the analysis of the different feasible scenarios. This � | � |
latter point is known in the annals of war as the conception X(t + 1) = X (t) − M |X (t) − X(t)| (15)
| |
phase. The fourth phase, called execution, is characterized
by the coordination between the different components of the where t stands for the tth iteration, X designates the position
vector of the moving element, X is the enemy position and
′
armed forces to destroy sensitive and neuralgic points, con-
centrations and major enemy’s resistances. This step is pre- M is a linear function that decreases from 1 to 0 through iter-
sented in our algorithm by using the fire and manoeuvrer or ations. The equation (16) allows to update the parameter M.
fire and movement (FM) tactic [67]. This FM tactic is based ( )
1
on the combination between three military actions, namely, M =1−i (16)
maxit
suppression, advance and assault. Finally, the exploitation
phase consists of reducing the residual enemy’s element in where maxit is the maximum number of iterations.
the battlefield.

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

 Evolutionary Intelligence

Fig. 2  Fire and movement tactic

2 1 5 7 4 8 6 3 N1 or using the greedy randomized adaptive search (GRASP),

described in Algorithm 2, for small and medium instance
2 6 1 5 7 4 8 3 N2 sizes. The detected enemy position is treated with Algo-
2 6 5 7 4 8 1 3 rithm 5 to define more accurately the enemy concentration.
2 6 8 4 7 5 1 3 N3 The step 3 allows to define more precisely the objective to
deal with in the enemy zone. In order to imitate the first
2 5 7 4 8 1 6 3 N4 stage of the FM tactic mechanism, the step 4 of the FMA
is based on the multi-neighbouring local search (Algo-
3 8 2 4 5 6 1 7 12 10 11 9 rithm 6) and Eq. 15 to update the moving element position.
N5 The advance and assault FM mechanisms are performed by
3 8 12 10 11 2 4 5 6 1 7 9 applying Algorithm 4 on the current solution, respectively.
The last step of the designed procedure allows to discover
Fig. 3  Neighbourhood structures unexplored search spaces by shaking the best solution and
re-performing the FM tactic. This step is equivalent to the
Algorithm 1 detailed the main steps of the fire and search for another enemy element in the enemy zone. In
maneuver algorithm (FMA). It is worth noting that the fact, this last phase, known as exploitation in land combat,
designed metaheuristic relies on a multi-neighbouring concerns the pursuit of residual resistances in the enemy’s
search procedure which is classified as a variable neigh- depth to eliminate them. Figure 4 shows the framework of
bourhood structured search (VNS) according to Akyol and the FMA. The largest ranked value (LRV) [69] representa-
Alatas [68]. The employed VNS is based on various neigh- tion is employed to convert the continuous position vector
bourhood structures as shown in Fig. 3. The initialization to a sequence job permutation.
and intelligence steps of the proposed algorithm consist of It is well known that in combinatorial optimization
generating a set of solutions at random for big instances problems, researchers have designed many procedures

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Evolutionary Intelligence

Fig. 4  The framework of fire Generate an initial solution S0 at

and manoeuvrer algorithm random or using GRASP procedure

Update Sbest = S0

Apply the greedy algorithm on Sbest

to generate a set of search agents

Update Sbest and

for all search agent

Apply the multi-neigbouring lo-

cal search to improve solution

-Apply the LRV rule to convert the job

permutation to vector position,
-Update the vector position using Eq.
15

yes Apply the LRV rule to convert the

vector position to the job permutation

If solution
improvement?

no

Update Sbest and apply the local search to improve it

If termination yes
Shake Sbest Best solution Sbest
no test?

that build a solution incrementally. Among these algo- proposed and explained in detail in the work of Feo and
rithms one can find the greedy randomized adaptive search Resende [70] to solve optimization problem. Prabhakaran
procedure. This technique is a general iterative procedure et al. [71] applied the GRASP to approach the permutation

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

 Evolutionary Intelligence

flow shop scheduling problems. The computational results

demonstrated that the GRASP outperforms the Nawaz, Algorithm 3 Greedy Randomized Constructive Solution procedure
Enscore and Ham (NEH) heuristic [72]. Algorithm 2 out- Input: Data set and the greediness parameter α
lines the main steps of the GRASP procedure based on two Solution ← empty partial job sequences π ;
Initialize candidate set S;
phases, namely, the construction stage and local search. Compute the incremental cost h(x), ∀x ∈ S;
The construction phase aims to built iteratively and greed- while S = ∅ do
Calculate: smin ← min{h(x)\x ∈ S};
ily a partial solution element by element as described in Calculate: smax ← max{h(x)\x ∈ S};
Algorithm 3. As explained in the pioneering work of Feo RCL ← {x ∈ S\h(x) ≤ smin + α(smax − smin )};
Select randomly an element e from the RCL;
and Resende [70], the selection of the next element obey to Allocate e to the position giving the best partial job sequence π;
a greedy function characterized by a greediness parameter Update candidate set S;
end while
𝛼 that varies between 0 and 1. The ordering of the elements Output: π
according to the greedy function leads to the construction
of a restricted candidate list (RCL) which contains the
best elements as described in [73–76]. The probabilistic
behaviour of the GRASP is manifested by choosing at ran- Algorithm 4 LocalSearch(π)

dom from the RCL the element to be added at the current for i = 1 to l do
Choose a job at random and put it in a random position of the sequence
partial solution as in [77, 78]. Algorithm 3 describes the π  = Nrandom (π);
pseudocode of the construction stage. if F (π) ≤ F (π  ) then
π ← π ;
It should be noted that the greedy search method used in end if
this investigation consists of two phases, namely, destruc- end for
Output: π
tion and construction as presented in Algorithm 5. In the
destruction phase, a number z of jobs are removed from
the sequence and put into a list Φ . The number z is chosen
adequately depending on the instance size. In the construc- Algorithm 5 Greedy algorithm
tion phase, the eliminated elements are reinserted in order
Input: Generate a job sequence π
to improve the current solution. π (best) ← π
Step 1: (Destruction phase)
π ← π
Algorithm 1 Fire and manoeuvrer algorithm
for j = 1 to z do
Input: processing time pj1 , ..., pjm ; % z is the destruction parameter
Step 1: (Initialization) Extract randomly a job πj from the sequence π  and add it to the Φ subset
Generate an initial solution S0 at random or using a heuristic (Algorithm 2) end for
and update Sbest = S0 ; Step 2: (Construction phase)
Step 2: (Intelligence) for j = 1 to z do
Apply the greedy Algorithm 5 on Sbest to generate a set of search agents Extract the πj job from Φ subset.
S = {S1 , S2 , ..., Sp } and evaluate each element of S ; Test the job on all positions in the current sequence π  and choose the best
Step 3: (Conception) position giving the best completion time
Update Sbest ; end for
for all search agent do Update π (best)
while solution improvement do Output: π (best)
Step 4:(Suppression) Apply the multi-neigbouring local search (Algo-
rithm 6) to improve position;
Step 5: (Advance) Update the position using Eq. 15 ;
end while
end for
Step 6:(Assault) Apply the local search (Algorithm 4) to improve Sbest ; Algorithm 6 Multi Neighbouring Search Procedure(π)
Step 7: (Exploitation): (Termination test): If the stopping condition is
for i = 1 to M do
not satisfied shake solution S= Shake(Sbest ) using Algorithm 7 and return to
Generate π  = Ni (π);
Step 2;
if F (π) ≤ F (π  ) then
Output: Sbest .
π ← π ;
end if
end for
Output: π

Algorithm 2 GRASP procedure

Input: Data set and the greediness parameter α
while Stopping criterion is not met do
π ← Greedy Randomized Constructive Solution(Solution,α) (Algorithm
3);
π ←LocalSearch(π) (Algorithm 4);
end while
Output: π (best)

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Evolutionary Intelligence

verify the effectiveness and the independence of our numeri-

Algorithm 7 Shake procedure
cal codes. Thus, a very good concordance is noticed between
Input: Solution π
the data obtained by the mathematical model and the results
for i = 1 to Z do given by the metaheurstics. In order to assess the effectiveness
Choose at random k ∈ {1, 2, 3, 4, 5};
π  ← Nk (π); of the new metaheuristic to solve the classical flow shop prob-
end for

lems, the Relative Percentage Increase (RPI) is introduced as
Output: π
a response variable to evaluate the efficiency of the proposed
technique. The RPI is defined by:
Nevertheless, it should be noted that the proposed FMA
1 ∑ fi − fmin
10

can outperform other metaheuristics in solving classical flow RPIalgorithm = × 100 (17)
10 i fmin
shop problems but its efficiency in tackling other optimiza-
tion problems remains to be verified as stated by Wolpert The approach followed to calculate this parameter consists
and Macready in their No Free Lunch (NFL) theorem [79] in carrying out a series of 10 runs for each instance in order
which stipulates that none of these optimization algorithms to evaluate the value of f which is the average of the results
has proven to be the best to solve all types of problems. It obtained during the experimental calculations. In order to
is worth mentioning, that the FMA parameters should be guarantee a fair comparison, the stopping criterion for all
chosen carefully when dealing with very large instances. the metaheuristics tested is the computation time. Thus, the
Indeed, the good choice of the destruction parameter z may CPU time is considered as the main factor characterizing
influence the convergence speed of the designed procedure. the convergence. Hence, the used time limit is defined by
𝜏 = 𝜌 × n × m with n × m represents the size of the consid-
ered instance and 𝜌 is a calibration parameter used to fit the
4 Computational results and analysis calculation time to hardware capabilities and make the code
convergence possible in a reasonable duration as shown in
The aim of this section is to compare the new metaheuristic Table 2.
against some popular nature-inspired algorithms. The designed Table 3 shows the makespan values given by the FMA
hunting method is implemented using the Matlab R2018a lan- and those obtained by the GA and the novel particle swarm
guage in order to conduct the computational experiments on a optimization (NPSO) algorithm [81] which is based on ten
computer Intel (R) Core (TM) i7 CPU, 2.6 GHz, 8 GB RAM. types of mutation operators and three type of crossover
The LINGO 17 software is used to implement the mathemati- operators. The best values obtained by testing the differ-
cal model giving the exact solution. For the comparisons we ent combinations of crossover and mutation operators are
use the well-known standard benchmark set of Taillard [80]. compared by the data provided by the FMA. Table 3 illus-
The contains 120 instances of different sizes with a number of trates the average performance of ten runs on each problem
jobs between 20 and 500 and a number of machine that varies instance. The novel particle swarm optimisation method
between 5 and 20. For each size set, the benchmark provides seems to be efficient in dealing with small and medium
10 instances which are considered very difficult to solve. The instances. It is clear noticed that the FMA outperforms the
instances of a given size of n jobs scheduled on m machines
are marked as n × m combination. Hence, the size of search Table 3  Comparison of FMA results against published data of Lian
space can be calculated by n!. The Taillard’s benchmark uses et al. [81] for the PFSP
a uniform distribution in the range [1, 99] to generate the pro-
Problem Size Lower bound GA [81] NPSO [81] FMA
cessing time.
It should be noted that the implemented heuristics are com- TA001 20 × 5 1278 1303.8 1279.9 1278.0
pared against the mathematical model for small instance to TA011 20 × 10 1582 1690 1605.8 1582.0
TA021 20 × 20 2297 2434.1 2334.9 2297.2
TA031 50 × 5 2724 2766.1 2725 2724.0
Table 2  Calibration parameter 𝜌
TA041 50 × 10 2991 3265.2 3086.9 2991.2
n ∈ {10, ..., 90} n ∈ {100, ..., 500} TA061 100 × 5 5493 5549.3 5493.2 5493.0

m=5 m = 10 m = 20 m=5 m = 10 m = 20 Runs are performed by a computer Intel (R) Core (TM) i7 CPU, 2.6
𝜌 = 0.03 𝜌 = 0.05 𝜌 = 0.06 𝜌 = 0.1 𝜌 = 0.2 𝜌 = 0.3 GHz, 8 GB RAM
Best values are in bold

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

 Evolutionary Intelligence

Table 4  Comparison of FMA results against published data of Zobo- FMA provides better performances than NEH and NEGAVNS.
las et al. [82] on 20-job Taillard instances However, this comparison is not sufficient due to the rela-
Problem Size Lower bound NEH [82] NEGAVNS [82] FMA tively small sizes of the compared instances.
The convergence speed of implemented methods towards
TA001 20 × 5 1278 1286 1278.0 1278.0
their best solutions is of great importance insofar as it allows
TA002 20 × 5 1359 1365 1359.0 1359.0
first to verify the convergence of the used methods and then
TA003 20 × 5 1081 1159 1081.0 1081.0
to evaluate the speed of numerical codes.
TA004 20 × 5 1293 1325 1293.0 1293.0
Figures 5 show the variation of the objective function
TA005 20 × 5 1235 1305 1235.0 1235.0
versus the CPU time for different instance sizes. The fire
TA006 20 × 5 1195 1228 1195.0 1195.0
and maneuver algorithm convergence has been tested
TA007 20 × 5 1239 1278 1239.0 1239.0
against the three well-known nature-inspired algorithms,
TA008 20 × 5 1206 1223 1206.0 1206.0
namely, genetic algorithm (GA) [20], migrating birds opti-
TA009 20 × 5 1230 1291 1230.0 1230.0
mization (MBO) [83] and the whale optimization algo-
TA010 20 × 5 1108 1151 1108.0 1108.0
rithm (WOA) [84]. All curves begin from the same starting
TA011 20 × 10 1582 1680 1582.0 1582.0
search point. The tested algorithms converge to different
TA012 20 × 10 1659 1729 1659.3 1659.0
final values after a specific running time. Therefore, the
TA013 20 × 10 1496 1557 1496.0 1496.0
FMA provides better performances than other metaheuris-
TA014 20 × 10 1378 1439 1378.7 1378.0
tics. It is worth mentioning that the designed FM algo-
TA023 20 × 20 2326 2411 2326.4 2326.0
rithm converge more rapidly than the other metaheuris-
TA026 20 × 20 2226 2349 2226.7 2226.3
tics. This observation is verified by Fig. 6 which shows
TA027 20 × 20 2273 2362 2274.5 2273.0
the variation of the objective function as function of the
Runs are performed by a computer Intel (R) Core (TM) i7 CPU, 2.6 CPU time for relatively large instances. Moreover, it is
GHz, 8 GB RAM clear observed that the GA converges faster than the MBO
Best values are in bold when dealing with the large size instances. This behavior
is strictly linked to the parameters of the MBO algorithm
and particularly the size of the initial population and the
GA and the NPSO for all tested instances. However, this number of shared solutions. In addition, the WOA gives
comparison remains insufficient in the absence of the results the less quality solutions for all tested instances.
obtained for large instances. It should be mentioned that a Figure 7 illustrates the statistical results represented by
good choice of the destruction parameter z may affect the the ANOVA diagrams obtained for the instances 100 × 10
average time required by the FMA to reach the best solution. and 200 × 10. The analysis has been based on median value
Therefore, this parameter should be between n2 and n3 for small of RPI by performing 10 replications for each instance.
and medium instances. One can easily conclude that fire and maneuver algorithm
Besides, Zobolas et al. [82] used a Pentium IV at 2.4 presents better solutions than the other nature-inspired
GHz to conducted their computational experiments on Tail- metaheuristics. In fact, the median values of RPI given
lard’s data set. The authors designed a hybrid metaheuris- by the FMA are less than 2% for both tested instances.
tic referred as NEGAVNS which combines genetic algorithm However, the results provided by the WOA are the less
(GA) and a variable neighbourhood search (VNS) to mini- quality ones. In fact, the RPI given by the WOA is greater
mize the makespan criterion of classical flow shop produc- than 10% for instance 100 × 10 and bigger than 10% for
tion systems. Table 4 presents the average results of ten instance 200 × 10.
runs obtained by the NEH heuristic, NEGAVNS and FMA Moreover, Table 5 represents the relative percentage
for different problem’s size. One can easily remarks that the increase over the best known upper bound of the Taillard’s

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Evolutionary Intelligence

Fig. 5  Evolution of makespan 1450

FMA
2100
FMA
vs CPU time for FMA, MBO, MBO MBO
GA and WOA: a 20 × 5, b GA
WOA
2000 GA
WOA
1400
20 × 10, c 20 × 20, d 50 × 5, e
50 × 20, f 100 × 10, g 100 × 20 1900

and h 200 × 10 1350 1800

1700
1300

1600

1250 1500
0 10 20 30 40 50 60 0 20 40 60 80 100

(a) 20 × 5 (b) 20 × 10
2800 3100
FMA FMA
MBO 3050 MBO
GA GA
2700 WOA WOA
3000

2950
2600

2900

2500
2850

2800
2400
2750

2300 2700
0 50 100 150 200 250 300 0 100 200 300 400 500

(c) 20 × 20 (d) 50 × 5
5200 7000
FMA FMA
MBO MBO
5000 GA 6800 GA
WOA WOA
4800 6600

4600 6400

4400 6200

4200 6000

4000 5800

3800 5600
0 50 100 150 200 0 500 1000 1500

(e) 50 × 20 (f ) 100 × 10
4
10
8000 1.22
FMA FMA
7800 MBO MBO
GA 1.2 GA
WOA WOA
7600
1.18
7400
1.16
7200
1.14
7000
1.12
6800

6600 1.1

6400 1.08
0 500 1000 1500 0 500 1000 1500 2000 2500

(g) 100 × 20 (h) 200 × 10

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

 Evolutionary Intelligence

4
10
1.4 11
FMA
MBO
10
GA
1.35 WOA
9

8
1.3
7

1.25 6

5
1.2 4

3
1.15
0 500 1000 1500 2000 2500 2

(a) 200 × 20 1
104 FMA MBO GA WOA
3.05
FMA
3 MBO (a) 100 × 10
GA
WOA
2.95
8
2.9
7
2.85

2.8 6

2.75
5
2.7

4
2.65
0 500 1000 1500 2000 2500

(b) 500 × 20
3

2
Fig. 6  Convergence curves of FMA, MBO, GA and WOA for Tail-
lard’s large instances: a 200 × 20, b 500 × 20
1

instances, wherein the best result is marked in bold. We can FMA MBO GA WOA
find that the performance of the algorithms is highly influ-
enced by the instance size. In fact, the RPI values obtained (b) 200 × 10
by the FMA is less than 0.2% when the number of machines
equals 5. This RPI value generally tends to increase when Fig. 7  Means plot of FMA, MBO, GA and WOA for instances a
the number of machines or the number of jobs increase. It is 100 × 10 and b 200 × 10
clear obvious that the new proposed hunting procedure out-
performs the classical nature-inspired algorithms for most
tested instances (83.33%). It also should be noted that the 5 Conclusion
MBO is more powerful in solving PFSP than the GA or the
WOA. In this paper, a new war-inspired algorithm which combines
GRASP procedure, greedy algorithm and local search with a
multi-neighbouring strategy has been designed to approach

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Evolutionary Intelligence

Table 5  Relative percentage increase over the best known upper References
bound and 𝜏 elapsed CPU times for the genetic algorithm, migrating
bird optimization, whale optimization algorithm and FMA 1. Alan A, Pritsker B, Waiters LJ, Wolfe PM (1969) Multiproject
Jobs × machines RPIGA RPIMBO RPIWOA RPIFMA scheduling with limited resources: a zero-one programming
approach. Manag Sci 16(1):93–108. https://d​ oi.o​ rg/1​ 0.1​ 287/m​ nsc.​
20 × 5 1.49 1.02 0.99 0.06 16.1.​93
20 × 10 2.91 2.19 8.09 1.99 2. Ignall E, Schrage L (1965) Application of the branch and bound
technique to some flow-shop scheduling problems. Oper Res
20 × 20 2.45 1.59 5.30 1.93 13(3):400–412. https://​doi.​org/​10.​1287/​opre.​13.3.​400
50 × 5 1.11 0.37 3.14 0.18 3. Selmer MJ (1954) Optimal two-and three-stage production sched-
50 × 10 3.83 2.76 12.09 1.98 ules with setup times included. Naval Res Log Quart 1(1):61–68.
50 × 20 4.40 2.86 13.82 2.60 https://​doi.​org/​10.​1002/​nav.​38000​10110
4. Xin-She Y (2012) Artificial intelligence, evolutionary computing
100 × 5 0.02 0.01 1.96 0.00 and metaheuristics: in the footsteps of Alan Turing, volume 427.
100 × 10 1.51 0.75 9.74 1.20 Springer
100 × 20 4.57 2.32 13.90 2.05 5. Fouad B, Rajib Kumar B (2020) Nature inspired methods for
200 × 10 1.07 1.07 6.30 0.67 metaheuristics optimization: algorithms and applications in sci-
ence and engineering: vol 16. Springer. https://d​ oi.​org/​10.​1007/​
200 × 20 3.40 2.68 13.09 1.99 978-3-​030-​26458-1
500 × 20 2.31 11.16 2.61 1.94 6. Samaher A-J, Ayad A (2022) A novel optimization algorithm
(lion-ayad) to find optimal dna protein synthesis. Egypt Inform
Runs are performed by a computer Intel (R) Core (TM) i7 CPU, 2.6 J 23(2):271–290. ISSN 1110-8665. https://​doi.​org/​10.​1016/j.​eij.​
GHz, 8 GB RAM 2022.​01.​004
Best values are in bold 7. Samaher A-J, Ayad A, Ehab A-, Aseel A, Mustafa M (2021) Intel-
ligent forecaster of concentrations (pm2. 5, pm10, no2, co, o3, so2)
caused air pollution (ifcsap). Neural Comput Appl 33(21):14199–
the flow shop scheduling problems. The implemented 14229. https://​doi.​org/​10.​1007/​s00521-​021-​06067-7
algorithms have been tested using the Taillard’s data set. 8. Al-Janabi S, Alkaim AF, Adel Z (2020) An innovative syn-
A mixed integer linear programming model (MILP) has thesis of deep learning techniques (dcapsnet and dcom) for
generation electrical renewable energy from wind energy.
been proposed to describe the studied flow shop problem. Soft Comput 24(14):10943–10962. https://​d oi.​o rg/​1 0.​1 007/​
The performance of the new proposed metaheuristic was s00500-​020-​04905-9
compared to the outputs of MILP model, NEGAVNS , NEH, 9. Al-Janabi S, Mohammad M, Al-Sultan A (2020) A new method
NPSO, genetic algorithm, migrating birds optimisation and for prediction of air pollution based on intelligent computa-
tion. Soft Comput 24(1):661–680. https:// ​ d oi. ​ o rg/ ​ 1 0. ​ 1 007/​
whale optimisation algorithm. The results show that the pro- s00500-​019-​04495-1
posed fire and manoeuvre algorithm is much more efficient 10. Samaher A-J, Ahmed P, Hayder F, Kenan K, Ibrahim AS (2014)
than NEGAVNS , NEH, NPSO, classical GA, MBO and WOA Empirical rapid and accurate prediction model for data mining
in solving flow shop scheduling problems. tasks in cloud computing environments. In: 2014 International
Congress on Technology, Communication and Knowledge
The idea behind these war-inspired algorithms is to com- (ICTCK), pages 1–8. https://​d oi.​o rg/​1 0.​1 109/​I CTCK.​2 014.​
bine efficiently various techniques and procedures in order to 70334​95
produce new ones more efficient as in military tactic where 11. Muhammed AM, Samaher A-J (2020). A novel software to
different components, techniques and behaviours are jointly improve healthcare base on predictive analytics and mobile
services for cloud data centers. In: Yousef Farhaoui, editor,
combined and implemented to conduct successful opera- Big Data and Networks Technologies, pages 320–339, Cham, .
tions. Further research stems from the possibility of applying Springer International Publishing. https://​doi.​org/​10.​1007/​978-
this new algorithm to solve other scheduling or engineering 3-​030-​23672-4_​23
problems such as multi-objective production scheduling. 12. Samaher A-J, Mahdi Abed S, Maha M (2019)Pragmatic text
mining method to find the topics of citation network. In: Inter-
national Conference on Big Data and Networks Technologies,
pages 190–205. Springer. https://​doi.​org/​10.​1007/​978-3-​030-​
Data availability The data used the support the findings of this study 23672-4_​15
are available from the corresponding author upon request. 13. Jabrane B, Said A, Karam A (2020) Solving permutation flow
shop scheduling problem with sequence-independent setup time.
Declarations J Appl Math. https://​doi.​org/​10.​1155/​2020/​71324​69
14. Allali K, Aqil S, Belabid J (2022) Distributed no-wait flow shop
problem with sequence dependent setup time: Optimization of
Conflict of interest The authors declare that there are no conflicts of
interest regarding the publication of this paper.

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

 Evolutionary Intelligence

makespan and maximum tardiness. Simulat Model Pract Theory scheduling problem using a multi-objective whale swarm algo-
116:102455. https://​doi.​org/​10.​1016/j.​simpat.​2021.​102455 rithm. Swarm Evolut Comput 57:100716. https://​doi.​org/​10.​
15. Framinan JM, Gupta JND, Leisten R (2004) A review and clas- 1016/j.​swevo.​2020.​100716
sification of heuristics for permutation flow-shop scheduling with 32. Marichelvam MK, Geetha M, Ömür T (2020) An improved par-
makespan objective. J Oper Res Soc 55(12):1243–1255. https://​ ticle swarm optimization algorithm to solve hybrid flowshop
doi.​org/​10.​1057/​palgr​ave.​jors.​26017​84 scheduling problems with the effect of human factors—a case
16. Ying K-C, Liao C-J (2004) An ant colony system for permutation study. Comput Oper Rese 114:104812. https://​doi.​org/​10.​1016/j.​
flow-shop sequencing. Comput Oper Res 31(5):791–801. https://​ cor.​2019.​104812
doi.​org/​10.​1016/​S0305-​0548(03)​00038-8 33. Shao Z, Pi D, Shao W (2020) Hybrid enhanced discrete fruit fly
17. Alisantoso D, Khoo LP, Jiang PY (2003) An immune algorithm optimization algorithm for scheduling blocking flow-shop in dis-
approach to the scheduling of a flexible pcb flow shop. Int J tributed environment. Expert Syst Appl 145:113147. https://​doi.​
Adv Manuf Technol 22(11):819–827. https://​doi.​org/​10.​1007/​ org/​10.​1016/j.​eswa.​2019.​113147
s00170-​002-​1498-5 34. Li H, Li X, Gao L (2021) A discrete artificial bee colony algo-
18. Komaki GM, Teymourian E, Kayvanfar V (2016) Minimising rithm for the distributed heterogeneous no-wait flowshop schedul-
makespan in the two-stage assembly hybrid flow shop sched- ing problem. Appl Soft Comput 100:106946. https://​doi.​org/​10.​
uling problem using artificial immune systems. Int J Prod Res 1016/j.​asoc.​2020.​106946
54(4):963–983. https://​doi.​org/​10.​1080/​00207​543.​2015.​10358​15 35. Yankai W, Shilong W, Dong L, Chunfeng S, Bo Y (2021) An
19. Chakravorty A, Laha D (2017) A heuristically directed immune improved multi-objective whale optimization algorithm for the
algorithm to minimize makespan and total flow time in permu- hybrid flow shop scheduling problem considering device dynamic
tation flow shops. Int J Adv Manuf Technol 93(9):3759–3776. reconfiguration processes. Expert Syst Appl 174:114793. https://​
https://​doi.​org/​10.​1007/​s00170-​017-​0679-1 doi.​org/​10.​1016/j.​eswa.​2021.​114793
20. Reeves CR (1995) A genetic algorithm for flowshop sequenc- 36. Ding J, Schulz S, Shen L, Buscher U, Lü Z (2021) Energy aware
ing. Comput Oper Res 22(1):5–13. https://​doi.​org/​10.​1016/​0305-​ scheduling in flexible flow shops with hybrid particle swarm
0548(93)​E0014-K optimization. Comput Oper Res 125:105088. https://​doi.​org/​10.​
21. Wen-Jie X, He L-J, Zhu G-Y (2021) Many-objective flow shop 1016/j.​cor.​2020.​105088
scheduling optimisation with genetic algorithm based on fuzzy 37. Tummala SLV, Ayyarao, NSSR, Rajvikram ME, Nishanth P, MR,
sets. Int J Prod Res 59(3):702–726. https://​doi.o​ rg/1​ 0.​1080/​00207​ Gaurav S, Baseem K, and Bilal A (2022) War strategy optimiza-
543.​2019.​17054​18 tion algorithm: a new effective metaheuristic algorithm for global
22. Pan Q-K, Dong Y (2014) An improved migrating birds optimisa- optimization. IEEE Access 10:25073–25105. https://​doi.​org/​10.​
tion for a hybrid flowshop scheduling with total flowtime mini- 1109/​ACCESS.​2022.​31534​93
misation. Inform Sci 277:643–655. https://​doi.​org/​10.​1016/j.​ins.​ 38. Ostwald M (1996) Peace and war in plato and aristotle. Scripta
2014.​02.​152 Classica Israelica 15:102–118
23. Marichelvam MK (2012) An improved hybrid cuckoo search 39. Tzu Sun (1971) The art of war, vol 361. Oxford University Press
(ihcs) metaheuristics algorithm for permutation flow shop sched- 40. Jomini A-He (2007) George Henry Mendell, and William Price
uling problems. Int J Bio Inspired Comput 4(4):200–205. https://​ Craighill. The art of war. ourier Corporation
doi.​org/​10.​1504/​IJBIC.​2012.​048061 41. Von Clausewitz C (2008) On war. Princeton University Press
24. Tosun Ö, Marichelvam MK (2016) Hybrid bat algorithm for flow 42. Hart BHL (1991) Strategy. A Meridian book. Meridian
shop scheduling problems. Int J Math Oper Res 9(1):125–138. 43. Douhet G, Harahan JP, Kohn RH, Ferrari D (2009) The command
https://​doi.​org/​10.​1504/​IJMOR.​2016.​077560 of the air. Fire ant books, University of Alabama Press
25. Dian SW, Dana MU (2020) The hybrid ant lion optimization flow 44. Jones JR (2004) William Billy Mitchell’s Air Power. University
shop scheduling problem for minimizing completion time. J Phys Press of the Pacific
Conf Ser 1569: 022097. IOP Publishing. https://​doi.​org/​10.​1088/​ 45. Mahan AT(1918). The Influence of Sea Power Upon History,
1742-​6596/​1569/2/​022097 1660–1783. American century series, S-10. Little, Brown
26. Utama DM, Baroto T, Widodo DS (2020) Energy-efficient flow 46. Valeriano B, Jensen B, Maness RC (2018) Cyber strategy: the
shop scheduling using hybrid grasshopper algorithm optimiza- evolving character of power and coercion. Oxford University Press
tion. Jurnal Ilmiah Teknik Industri 19(1):30–38. https://​doi.​org/​ 47. Van Wie Davis E (2021) Shadow Warfare: Cyberwar Policy in
10.​23917/​jiti.​v19i1.​10079 the United States, Russia and China. Security and Professional
27. Qi X, Yuan Z, Han X, Liu S (2020) A discrete butterfly-inspired Intelligence Education Series. Rowman & Littlefield Publishers
optimization algorithm for solving permutation flow-shop sched- 48. Conway RW, Maxwell WL, Miller LW (2003) Theory of schedul-
uling problems. Neural Netw World 30(4):211. https://​doi.​org/​10.​ ing. Dover Books on Computer Science Series, Dover
14311/​nnw.​2020.​30.​015 49. Ronald LG, Eugene LL, Jan KL, Rinnooy Kan AHG (1979).
28. Marichelvam MK, Azhagurajan A, Geetha M (2018) Minimisation Optimization and approximation in deterministic sequencing and
of total tardiness in hybrid flowshop scheduling problems with scheduling: a survey. In: Annals of discrete mathematics, vol-
sequence dependent setup times using a discrete firefly algorithm. ume 5, pages 287–326. Elsevier, https://​doi.​org/​10.​1016/​S0167-​
Int J Oper Res 32(1):114–126. https://d​ oi.o​ rg/1​ 0.1​ 504/I​ JOR.2​ 018.​ 5060(08)​70356-X
091204 50. Stafford EF (1988) On the development of a mixed-integer linear
29. Deb S, Tian Z, Fong S, Tang R, Wong R, Dey Nilanjan (2018) programming model for the flowshop sequencing problem. J Oper
Solving permutation flow-shop scheduling problem by rhinoceros Res Soc 39(12):1163–1174. https://d​ oi.o​ rg/1​ 0.1​ 057/j​ ors.1​ 988.1​ 93
search algorithm. Soft Comput 22(18):6025–6034. https://d​ oi.o​ rg/​ 51. Šeda M (2007) Mathematical models of flow shop and job shop
10.​1007/​s00500-​018-​3075-3 scheduling problems. Int J Appl Math Comput Sci 4(4):241–246.
30. Zhu H, Qi X, Chen F, He X, Chen L, Zhang Z (2019) Quan- https://​doi.​org/​10.​5281/​zenodo.​10823​07
tum-inspired cuckoo co-search algorithm for no-wait flow shop 52. Débora PR, Ernesto GB (2012) Mixed-integer programming
scheduling. Appl Intell 49(2):791–803. https://​doi.​org/​10.​1007/​ models for flowshop scheduling problems minimizing the total
s10489-​018-​1285-0 earliness and tardiness. In: Just-in-Time systems, pages 91–105.
31. Guangchen W, Liang G, Xinyu L, Peigen L, Fatih Tasgetiren Springer, 2012. https://​doi.​org/​10.​1007/​978-1-​4614-​1123-9
M (2020) Energy-efficient distributed permutation flow shop

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Evolutionary Intelligence

53. Meng L, Zhang C, Ren Y, Zhang B, Lv C (2020) Mixed-integer 71. Prabhaharan G, Shahul Hamid Khan B, Rakesh L (2006)
linear programming and constraint programming formulations for Implementation of grasp in flow shop scheduling. Int J Adv
solving distributed flexible job shop scheduling problem. Comput Manuf Technol 30(11):1126–1131. https://​d oi.​o rg/​1 0.​1 007/​
Ind Eng 142:106347. https://​doi.​org/​10.​1016/j.​cie.​2020.​106347 s00170-​005-​0134-6
54. Genlin J (2004) Survey on genetic algorithm. Comput Appl Softw 72. Nawaz M, Enscore Jr EE, Ham I (1983) A heuristic algorithm
2:69–73 for the m-machine, n-job flow-shop sequencing problem. Omega
55. Guilherme CS, Eduardo EOC, Walmir MC (2020) An artificial 11(1):91–95. https://​doi.​org/​10.​1016/​0305-​0483(83)​90088-9
immune systems approach to case-based reasoning applied to fault 73. Mauricio GCR, Celso CR (2010) Greedy randomized adaptive
detection and diagnosis. Expert Syst Appl 140:112906. https://d​ oi.​ search procedures: Advances, hybridizations, and applications.
org/​10.​1016/j.​eswa.​2019.​112906 In: Handbook of metaheuristics, pages 283–319. Springer. https://​
56. Marco D, Luca MG (1997) Ant colony system: a cooperative doi.​org/​10.​1007/​978-1-​4419-​1665-5_​10
learning approach to the traveling salesman problem. IEEE Trans 74. Pardalos Panos M , Tianbing Q, Mauricio GCR (1998) A greedy
Evolut Comput 1(1):53–66. https://​doi.​org/​10.​1109/​4235.​585892 randomized adaptive search procedure for the feedback vertex
57. Emrah H (2020) Artificial bee colony: theory, literature review, set problem. J Combin Optim 2(4):399–412. https://​doi.​org/​10.​
and application in image segmentation. In: Recent Advances 1023/A:​10097​36921​890
on Memetic Algorithms and its Applications in Image Pro- 75. Gevezes Theodoros, Pitsoulis Leonidas (2015) A greedy ran-
cessing, pages 47–67. Springer, 2020. https://​doi.​org/​10.​1007/​ domized adaptive search procedure with path relinking for the
978-​981-​15-​1362-6_3 shortest superstring problem. J Combin Optim 29(4):859–883.
58. Kumar A, Kumar D, Jarial SK (2017) A review on artificial https://​doi.​org/​10.​1007/​s10878-​013-​9622-z
bee colony algorithms and their applications to data clustering. 76. Mauricio GC, Resende C, Ribeiro C (2019) Greedy randomized
Cybern Inform Technol 17(3):3–28. https://​doi.​org/​10.​1515/​ adaptive search procedures: Advances and extensions. In: Hand-
cait-​2017-​0027 book of metaheuristics, pages 169–220. Springer, 2019. https://​
59. Murat A, Kemal P (2020) Binary particle swarm optimization doi.​org/​10.​1007/​978-3-​319-​91086-4_6
(bpso) based channel selection in the eeg signals and its applica- 77. Claudio Arroyo José Elias, de Souza Pereira Ana Amélia (2011)
tion to speller systems. J Artifi Intell Syst 2(1):27–37. https://​doi.​ A grasp heuristic for the multi-objective permutation flowshop
org/​10.​33969/​AIS.​2020.​21003 scheduling problem. Int J Adv Manuf Technol 55(5–8):741–753.
60. Jeffrey OA, Absalom EE, Laith A (2022) Dwarf mongoose optimi- https://​doi.​org/​10.​1007/​s00170-​010-​3100-x
zation algorithm. Computer Methods Appl Mech Eng 391:114570 78. Shahul Hamid Khan B, Prabhaharan G, Asokan P (2007) A grasp
. ISSN 0045-7825. https://​doi.​org/​10.​1016/j.​cma.​2022.​114570 algorithm for m-machine flowshop scheduling problem with bic-
61. Laith A, Ali D, Seyedali M, Mohamed AE, Amir H (2021a) Gan- riteria of makespan and maximum tardiness. Int J Comput Math
domi. the arithmetic optimization algorithm. Comput Methods 84(12):1731–1741. https://​doi.​org/​10.​1080/​00207​16070​13313​76
Appl Mecha Eng 376:113609. ISSN 0045-7825. https://​doi.​org/​ 79. Wolpert DH, Macready WG (1997) No free lunch theorems for
10.​1016/j.​cma.​2020.​113609 optimization. IEEE Trans Evolut Comput 1(1):67–82. https://​doi.​
62. Laith A, Dalia Y, Mohamed AE, Ahmed AE, Mohammed AAA- org/​10.​1109/​4235.​585893
G, Amir HG (2021b) Aquila optimizer: a novel meta-heuristic 80. Taillard E (1993) Benchmarks for basic scheduling problems. Eur
optimization algorithm. Comput Ind Eng 157:107250. ISSN 0360- J Oper Res 64(2):278–285. https://d​ oi.o​ rg/1​ 0.1​ 016/0​ 377-2​ 217(93)​
8352. https://​doi.​org/​10.​1016/j.​cie.​2021.​107250 90182-M
63. Laith A, Mohamed AE, Putra S, Zong WG, Amir H (2022) Gan- 81. Lian Z, Xingsheng G, Jiao B (2008) A novel particle swarm opti-
domi. reptile search algorithm (rsa): a nature-inspired meta-heu- mization algorithm for permutation flow-shop scheduling to mini-
ristic optimizer. Expert Syst Appl 191:116158. ISSN 0957-4174. mize makespan. Chaos Solit Fract 35(5):851–861. https://d​ oi.o​ rg/​
https://​doi.​org/​10.​1016/j.​eswa.​2021.​116158 10.​1016/j.​chaos.​2006.​05.​082
64. Oyelade ON, Ezugwu AE-S, Mohamed TIA, Abualigah L (2022) 82. Zobolas GI, Tarantilis CD, Ioannou G (2009) Minimizing makes-
Ebola optimization search algorithm: a new nature-inspired pan in permutation flow shop scheduling problems using a hybrid
metaheuristic optimization algorithm. IEEE Access 10(16150– metaheuristic algorithm. Comput Oper Res 36(4):1249–1267.
16177):2022. https://​doi.​org/​10.​1109/​ACCESS.​2022.​31478​21 https://​doi.​org/​10.​1016/j.​cor.​2008.​01.​007 (ISSN 0305-0548)
65. Alatas B, Bingol H (2019) A physics based novel approach for 83. Duman E, Uysal M, Alkaya AF (2012) Migrating birds optimiza-
travelling tournament problem: optics inspired optimization. tion: a new metaheuristic approach and its performance on quad-
Inform Technol Control 48(3):373–388. https://​doi.​org/​10.​5755/​ ratic assignment problem. Inform Sci 217:65–77. https://​doi.​org/​
j01.​itc.​48.3.​20627 10.​1016/j.​ins.​2012.​06.​032
66. Bilal A, Harun B (2020) Comparative assessment of light-based 84. Mirjalili S, Lewis A (2016) The whale optimization algorithm.
intelligent search and optimization algorithms. Light Eng 28(6). Adv Eng Softw 95:51–67. https://​doi.​org/​10.​1016/j.​adven​gsoft.​
https://​doi.​org/​10.​33383/​2019-​029 2016.​01.​008
67. McAndrew WJ (1987) Fire or movement? Canadian tactical doc-
trine, sicily-1943. Military Affairs 51(3):140–145. https://d​ oi.o​ rg/​ Publisher's Note Springer Nature remains neutral with regard to
10.​2307/​19875​17 jurisdictional claims in published maps and institutional affiliations.
68. Akyol S, Alatas B (2017) Plant intelligence based metaheuristic
optimization algorithms. Artif Intell Rev 47(4):417–462. https://​ Springer Nature or its licensor holds exclusive rights to this article under
doi.​org/​10.​1007/​s10462-​016-​9486-6 a publishing agreement with the author(s) or other rightsholder(s);
69. Wang H, Wang W, Sun H, Cui Z, Rahnamayan S, Zeng S (2017) A author self-archiving of the accepted manuscript version of this article
new cuckoo search algorithm with hybrid strategies for flow shop is solely governed by the terms of such publishing agreement and
scheduling problems. Soft Comput 21(15):4297–4307. https://d​ oi.​ applicable law.
org/​10.​1007/​s00500-​016-​2062-9
70. Feo Thomas A, Resende Mauricio GC (1995) Greedy randomized
adaptive search procedures. J Global Optim 6(2):109–133. https://​
doi.​org/​10.​1007/​BF010​96763

13

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:

1. use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
2. use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
3. falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
4. use bots or other automated methods to access the content or redirect messages
5. override any security feature or exclusionary protocol; or
6. share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at

[email protected]