Received January 7, 2022, accepted February 11, 2022, date of publication February 15, 2022, date of current version

February 24, 2022.

Digital Object Identifier 10.1109/ACCESS.2022.3151641

Tasmanian Devil Optimization: A New

Bio-Inspired Optimization Algorithm
for Solving Optimization Algorithm
MOHAMMAD DEHGHANI 1, ŠTĚPÁN HUBÁLOVSKÝ2 , AND PAVEL TROJOVSKÝ 1
1 Department of Mathematics, Faculty of Science, University of Hradec Králové, 50003 Hradec Králové, Czech Republic
2 Department of Applied Cybernetics, Faculty of Science, University of Hradec Králové, 50003 Hradec Králové, Czech Republic

Corresponding author: Pavel Trojovský ([email protected])

This work was supported by the Excellence Project of Faculty of Science, University of Hradec Králové, under Grant 2210/2022-2023.

ABSTRACT In this paper, a new bio-inspired metaheuristic algorithm called Tasmanian Devil
Optimization (TDO) is designed that mimics Tasmanian devil behavior in nature. The fundamental
inspiration used in TDO is simulation of the feeding behavior of the Tasmanian devil, who has two
strategies: attacking live prey or feeding on carrions of dead animals. The proposed TDO is described, then
its mathematical modeling is presented. TDO performance in optimization is tested on a set of twenty-
three standard objective functions. Unimodal benchmark functions have analyzed the TDO exploitation
capability, while high-dimensional multimodal and fixed-exploitation multimodal benchmark functions have
challenged the TDO exploration capability. The optimization results indicate the high ability of the proposed
TDO in exploration and exploitation and create a proper balance between these two indicators to effectively
solve optimization problems. Eight well-known metaheuristic algorithms are employed to analyze the quality
of the obtained results from TDO. The simulation results show that the proposed TDO, with its strong
performance, has a higher capability than the eight competitor algorithms and is much more competitive.
For further analysis, TDO is tested in optimizing four engineering design problems. Implementation results
show that TDO has an effective performance in solving real-world applications.

INDEX TERMS Bio-inspired, exploitation, exploration, feeding, optimization, optimization algorithm,

Tasmanian devil.

I. INTRODUCTION searcher agents and using random operators. Metaheuristic

Optimization is the process of determining the best solution algorithms have become more popular than classical methods
among several candidate solutions for a problem with respect because of their advantages such as easy comprehension, easy
to its constraints. Advances in science and technology have implementation, good performance, ability to avoid local
led to the emergence of new and complex optimization optimization, and application to optimization problems in
problems as well as more details in existing optimization various sciences [3].
problems [1]. Many of these problems have features such as Two important features in the search and solution finding
unknown search space, discrete search space, non-derivative process in metaheuristic algorithms are exploration and
objective functions, high dimensions, and non-convexity. exploitation. Exploration is the concept of global search
This has led to the failure of traditional methods and math- in the problem-solving space to analyze different areas of
ematical analysis to effectively solve real-world optimization the search space and not get caught in the optimal local
problems [2]. Hence, researchers and scientists tend to areas, while exploitation is the concept of locally search in
introduce new optimization solving methods called meta- the neighborhood of the obtained solution to find a better
heuristic algorithms that can solve optimization problems solution [4]. Metaheuristic algorithms to have acceptable
without the need for derivative information, based on random performance in solving optimization problems, must have
a good balance between exploration and exploitation [5].
The associate editor coordinating the review of this manuscript and Metaheuristic algorithms are able to identify the optimal
approving it for publication was Massimo Cafaro . area based on the exploration phase and then converge

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
VOLUME 10, 2022 19599
 M. Dehghani et al.: Tasmanian Devil Optimization: New Bio-Inspired Optimization Algorithm

towards the global optimal solution based on the exploitation Algorithm (TSA) [17]. Some other swarm-based algorithms
phase. are Red Fox Optimization Algorithm (RFOA) [18], Raccoon
Due to the random nature of the search process in Optimization Algorithm (ROA) [19], Crow Search Algorithm
metaheuristic algorithms, the solution they provide may not (CSA) [20], Teaching-Learning Based Optimization (TLBO)
be the same as the global optimal. For this reason, the [21], and Grasshopper Optimization Algorithm (GOA) [22].
solution obtained from metaheuristic algorithms is called Modeling of various laws in physics and physical phe-
quasi-optimal solution [6]. The performance of metaheuris- nomena has been considered in the introduction of physics-
tic algorithms is different due to their randomness and based algorithms. Simulated Annealing (SA) algorithm is one
different nature in the search process. Thus, metaheuris- of the most prominent physical algorithms that is inspired
tic algorithms may offer different solutions in solving in metal melting operations by the process of melting
a given optimization problem [7]. This has led to the and cooling materials [23]. The simulation of Newton’s
design of numerous metaheuristic algorithms by researchers laws of motion with the use of physical forces has been
to provide better quasi-optimal solutions to optimization effective in the development of optimizers. Gravitational
problems. Search Algorithm (GSA) [24] using gravitational force,
Metaheuristic algorithms are stochastic optimization Spring Search Algorithm (SSA) [25] using elastic force,
problem-solving methods that have been inspired by various and Momentum Search Algorithm (MSA) using momentum
natural phenomena, the laws of physics, biological sciences, have been designed. Physical phenomena are the source
the rules of the game, and other evolutionary phenomena. of inspiration in the design of metaheuristics, such as
These algorithms first generate candidate solutions according Water Cycle Algorithm (WCA) [26] according to the water
to the constraints of the problem. Then, they improve cycle phenomenon, Small-World Optimization Algorithm
these candidate solutions based on the algorithm update (SWOA) [27] according to the mechanism of small-world
steps in an iterative-based process. The main difference phenomenon, and Black Hole (BH) [28] according to observ-
between metaheuristic algorithms is in the same process of able fact of black hole phenomena. Some other physics-
improving candidate solutions during algorithm iterations. based algorithms are Nuclear Reaction Optimization (NRO)
Metaheuristic algorithms in a general category based [29], Multi-Verse Optimizer (MVO) [30], Artificial Chemical
on main inspiration can be divided into four groups: Reaction Optimization Algorithm (ACROA) [31], Optics
evolutionary-based, swarm-based, physics-based, and game- Inspired Optimization (OIO) [32], Equilibrium Optimizer
based algorithms. (EO) [33], Atom Search Optimization (ASO) [34], and
Applying the biological sciences alongside the theory of Electromagnetic Field Optimization (EFO) [35].
natural selection and Darwin’s theory of evolution have Simulation of rules and behavior of players in different
inspired the design of evolutionary-based algorithms. Genetic games has led to the design of game-based algorithms.
Algorithm (GA) [8] and Differential Evolution (DE) [9] can Football Game Based Optimization (FGBO) [36] and Vol-
be considered as the most famous evolutionary algorithms. leyball Premier Ligue (VPL) [37] algorithms are game-
In the design of GA and DE, the random operators of based metaheuristics developed based on the simulation of
selection, crossover, and mutation play a key role in updating club competitions during a sports season. The behavior
the algorithm population. The mechanism of the human of players in collecting points and winning based on the
immune system in the face of disease has been a fundamental throwing mechanism is modeled on the design of Ring Toss
inspiration in the development of the Artificial Immune Game Based Optimizer (RTGBO) [38] and Darts Game
System (AIS) algorithm [10]. Optimizer (DGO [39].
The natural behaviors of various species of animals, The major research question in all studies of metaheuristic
birds, aquatic animals, and other living things have paved algorithms is whether, given the various algorithms that
the way for the development of swarm-based algorithms. have been developed, there is still a need to introduce
Particle Swarm Optimization (PSO) [11] and Ant Colony new algorithms. The No Free Lunch (NFL) theorem [40]
Optimization (ACO) [12] are among the most familiar and answers this question that the strong performance of an
widely used swarm-based algorithms. PSO has employed algorithm in solving a set of optimization problems provides
the natural behavior of swarm movement of birds or no guarantee of optimal performance in other problems.
fish. ACO has modeled the natural behavior of ants in Therefore, the superiority of a particular algorithm in solving
identifying the shortest path. The animals’ strategy in hunting all optimization problems is hypothesis rejected. The NFL
their prey in nature represents an optimization process. theorem provides a research path for scientists to design
Simulations of these natural behaviors have been employed new metaheuristic algorithms to solve optimization problems
in the design of metaheuristics such as Whale Optimiza- more effectively. The NFL theorem motivated the authors of
tion Algorithm (WOA [13], Marine Predators Algorithm this paper to come up with a new metaheuristic algorithm to
(MPA) [14], and Grey Wolf Optimization (GWO) [15]. effectively solve optimization problems.
Search behaviors of animals with access to food sources What is evident from all studies of literature review and
have led to the introduction of metaheuristics such as its obtained best knowledge is that Tasmanian devil behavior
Artificial Bee Colony (ABC) [16] and Tunicate Swarm simulation has not been employed in the design of any

19600 VOLUME 10, 2022

M. Dehghani et al.: Tasmanian Devil Optimization: New Bio-Inspired Optimization Algorithm

metaheuristic algorithm. However, the natural behavior of the

Tasmanian devil during feeding represents an optimization
process in achieving the main purpose of this animal, i.e.,
food source. This research gap prompted the authors to design
a new optimizer by simulating the Tasmanian devil feeding
strategy, which is discussed in the next section.
This paper introduces a new optimization algorithm called
Tasmanian Devil Optimization (TDO) that can be applied to
solve various science optimization problems. The scientific
contribution of this research can be expressed as follows:
1. The novelty of this paper is in the design of the new TDO
optimizer based on the simulation of the Tasmanian devil’s
natural behavior.
2. The fundamental inspiration of TDO is the Tasmanian
devil feeding mechanism in two strategies of live prey
hunting and carnivore eating.
3. The various stages of TDO are described and mathemati-
cally modeled.
4. TDO is evaluated by solving twenty-three benchmark
FIGURE 1. Tasmanian devil (take from Wikimedia Commons - Tasmanian
functions including unimodal, high-dimensional multi- Devil (33295981294)).
modal, and fixed-dimensional multimodal types.
5. TDO is used to solve four engineering design problems to
evaluate its performance in real-world problems.
6. To analyze the capability of the proposed algorithm, the
optimization results obtained from TDO are compared The optimization process is how to achieve the optimal
with eight well-known algorithms. solution for an optimization issue. The analogy of this process
In the following, the paper is organized in such a in the life and behaviors of the Tasmanian devil is like
way that in Section 2, the proposed TDO algorithm is access to food. In fact, just as in the optimization process,
introduced and modeled. Simulation studies and results are the goal is to find the optimal solution, in the Tasmanian
presented in Section 3. The capability of TDO in optimizing devil’s nutritional process, the goal is to find the food source.
engineering design problems is analyzed in Section IV. Two important principles in the optimization process are
Finally, in Section 5, conclusions and several research exploration in the comprehensive search the problem-solving
suggestions are presented. space and exploitation in approaching the optimal solution.
The Tasmanian devil’s search behavior in finding food
II. TASMANIAN DEVIL OPTIMIZATION sources in different spaces, in fact, indicates the exploration
In this section, the proposed metaheuristic Tasmanian Devil index in the optimization process in order to identify the
Optimization (TDO) is introduced and its mathematical optimal area of search space. On the other hand, the chasing
modeling is presented. process between the Tasmanian devil and the prey that occurs
in a limited area is similar to the exploitation index in the local
A. INSPIRATION AND BEHAVIOR OF TAMANIAN DEVIL search with the aim of converging to the optimal solution.
The Tasmanian devil is a carnivorous and marsupial wild This means that mathematical modeling of Tasmanian devil
animal belonging to the family Dasyuridae that lives in the strategies to reach the food source is prone to designing
island state of Tasmania. A photo of the Tasmanian devil an optimizer to achieve optimal solutions to optimization
is shown in Figure 1. Tasmanian devils are opportunistic problems.
animals, and although they are able to hunt prey, they feed
on carrion if present [41]. Tasmanian devil has two strategies
for feeding. In the first strategy, if Tasmanian devil finds a 1) INITIALIZATION
carrion, it feeds on it. In the second strategy, it hunts and feeds The proposed TDO is a population-based stochastic algo-
on prey by attacking it. rithm whose searcher agents are Tasmanian devils. The initial
The modeling of this Tasmanian devil feeding mechanism population of these agents is generated randomly based
is used in the TDO design. on the constraints of the problem. Population members of
TDO, who are searchers of problem-solving space, suggest
B. MATHEMATICAL MODELLING candidate values for problem variables based on their position
In this subsection, the process and how to simulate the natural in the search space. So mathematically, each member of a
behavior of Tasmanian demons during feeding is described to population is a vector with the number of elements equal to
design an optimizer. the number of problem variables. As a result, the set of TDO

VOLUME 10, 2022 19601

 M. Dehghani et al.: Tasmanian Devil Optimization: New Bio-Inspired Optimization Algorithm

members can be modeled using a matrix in (1). The concepts expressed in the Tasmanian devil strategy of
    eating carcasses are mathematically modeled using (3) to (5).
X1 x1,1 · · · x1,j · · · x1,m
In the TDO design, for each Tasmanian devil, the position
 ..   .. .. .. . .. 
 .   . . . .. .  of other population members in the search space is assumed
,
   
X = X i
 =  xi,1 · · · xi,j · · · xi,m  to be carrion locations. Random selection of one of these
 .   . . . . situations is simulated in (3) so that the k 0 th population
   
 ..   .. . .. .. .. 
..

member is selected as the target carrion for the i0 th Tasmanian
XN N ×m xN ,1 · · · xN ,j · · · xN ,m N ×m devil. Therefore, k must be chosen randomly from 1 to N
(1) while the opposite is i.
were X is the population of Tasmanian devils, Xi is the ith Ci = Xk , i = 1, 2, . . . , N , k ∈ {1, 2, . . . , N |k 6= i}, (3)
candidate solution while xi,j is its candidate value for the jth
where Ci is the selected carrion by ith Tasmanian devil.
variable, N is the number of searching Tasmanian devils, and
Based on the selected carrion, a new position is calculated
m is the number of variables of given problems.
for the Tasmanian devil in the search space. In the Tasmanian
The objective function of problem can be computed by
devil motion simulation in this strategy, if the objective
placing each of the candidate solutions in the values of the
function value of the carrion is better, the Tasmanian devil
variables of the objective function. As a result, the values
moves toward that carrion, otherwise it moves away from that
obtained for the objective function are modeled using a vector
carrion. This Tasmanian devil movement strategy is simulated
in (2).
    in (4). In the last step of the first strategy, after calculating the
F1 F(X1 ) new position for Tasmanian devil, this position is accepted
 ..   ..  if the value of the objective function is better in this new
 .   . 
position otherwise, Tasmanian devil remains in its previous
,
   
F =  Fi  =  F(Xi )  (2)
  
 ..   ..  position. This update step is modeled in (5).
 .   .  (
new,S1 xi,j + r · (ci,j − I · xi,j ), FCi < Fi ;
FN N ×1 F(XN ) N ×1 xi,j = (4)
xi,j + r · (xi,j − ci,j ), otherwise,
where F is the vector of values of the objective function (
and Fi is the value of the objective function obtained by the Xinew,S1 , Finew,S1 < Fi ;
Xi = (5)
ith candidate solution. The analysis of the values obtained Xi , otherwise,
for the objective function shows the quality of the candidate
solutions. The candidate solution that leads to the calculation Here, Xinew,S1 is the new status of the ith Tasmanian devil
new,S1
of the best value for the objective function is considered based on the first strategy, xi,j is its value for the jth
new,S1
the best member of the population. The best member of the variable, Fi is its objective function value, FCi is its
population is updated based on new values in each iteration. objective function value of selected carrion, r is a random
The population updating process in TDO is modeled on number in interval [0, 1], and I is a random number which
two Tasmanian devil feeding strategies. It is possible for can be 1 or 2.
any Tasmanian devil to eat carrion or feed on prey hunting.
In TDO, it is assumed that the probability of choosing any of 3) STRATEGY 2: FEEDING BY EATING PREY
these strategies is equal to 50%. According to this concept, (EXPLOITATION PHASE)
in each iteration of the TDO, each Tasmanian devil is updated The Tasmanian Devil’s second feeding strategy is to hunt
based on only one of these two strategies. and eat prey. Tasmanian devil behavior during the attack has
two stages. In the first stage, by scanning the area, it selects
2) STRATEGY 1: FEEDING BY EATING CARRION the prey and attacks it. Then, in the second stage, after
(EXPLORATION PHASE) approaching the prey, it chases it to stop it and start eating.
Sometimes the Tasmanian devil prefers to feed on carrion The modeling of the first stage is similar to the modeling
in the area instead of hunting. There are other predatory of the first strategy, i.e., the selection of the carcass.
animals living around the Tasmanian Devil, which hunt large Therefore, the first stage of prey selection and attack it
prey and are unable to eat it all. Additionally, these animals is modeled using (6) to (8). In the second strategy, when
may not be able to eat sufficiently from their prey until updating the i0 th Tasmanian devil, the position of other
the Tasmanian devil arrives. In these cases, the Tasmanian population members is assumed as preys location. The k 0 th
devil prefers to feed on these carrions. Tasmanian devil population member is randomly selected as prey, while k is a
behavior in scanning the habitat area to find carrion is natural random number between 1 to N and opposite i. The
similar to the algorithm search process in problem-solving prey selection process is simulated in (6).
space. This Tasmanian devil strategy actually demonstrates
Pi = Xk , i = 1, 2, . . . , N , k ∈ {1, 2, . . . , N |k 6= i}, (6)
the power of TDO exploration in scanning different areas
of the search space to identify the original optimal area. Here, Pi is the selected prey by the ith Tasmanian devil.

19602 VOLUME 10, 2022

M. Dehghani et al.: Tasmanian Devil Optimization: New Bio-Inspired Optimization Algorithm

After determining the prey position, a new position is the position of Tasmanian devils and the objective function.
calculated for the Tasmanian devil. In calculating this new After this, the algorithm enters the next iteration and the
position, if the objective function value of the selected prey TDO population update process continues until the end of the
is better, the Tasmanian devil moves towards it, otherwise algorithm iterations according to equations (3) to (11). TDO
it moves away from that position. Modeling of this process updates and stores the best candidate solution during these
is presented in (7). The new position calculated for the iterations. After the algorithm is fully implemented, TDO
Tasmanian devil replaces the previous position if it improves introduces the best candidate solution as the solution to the
the value of the target function. This step of the second problem. The various steps of TDO are presented in flowchart
strategy is modeled in (8). format in Figure 2 and its pseudocode in Algorithm 1.
(
new,S2 xi,j + r · (pi,j − I · xi,j ), FPi < Fi ; C. COMPUTATIONAL COMPLEXITY
xi,j = (7)
xi,j + r · (xi,j − pi,j ), otherwise, This section analyzes the computational complexity of TDO.
(
Xinew,S2 , Finew,S2 < Fi ; The computational complexity of TDO initialization is equal
Xi = (8) to O(N · m) where N is the number of members of the
Xi , otherwise,
Tasmanian devil population and m is the number of problem
Here, Xinew,S2 is the new status of i0 th Tasmanian based on variables. TDO has a problem-solving process in the number
the second strategy, xi,j new,S2
is its value for the jth variable, of repetitive T . The process of updating population members
new,S2 on their way to the carcass or prey has a computational
Fi is its objective function value, and FPi is its objective
complexity equal to O(N · m · T ). The prey chasing process in
function value of selected prey.
the second strategy has a computational complexity equal to
The main difference between this strategy and the first
O(NS2 · m · T ) where NS2 is the number of Tasmanian demons
strategy is the second stage and the simulation of prey
who have used the second feeding strategy. Thus, the total
chasing. The chase of prey in the vicinity of the attack site is
computational complexity of TDO is equal to O((N · m) ·
similar to the local search of the search space. This Tasmanian
((1 + T ) + (T · NS2 ))).
devil behavior actually demonstrates the TDO’s ability to
exploit to converge to better candidate solutions. In order III. SIMULATION STUDIES AND DISCUSSION
to simulate this chase process, the Tasmanian devil follows In this section, simulation studies of TDO performance
the prey in the neighborhood of the attacked place. The prey in optimization are presented. TDO is employed to solve
chase stage is modeled by the Tasmanian devil using (9) to twenty-three standard benchmark functions, including seven
(11). At this stage, the Tasmanian devil position is considered unimodal functions, six high-dimensional multimodal func-
the center of a neighborhood where the prey chasing process tions, and ten fixed-dimensional unimodal functions [42].
takes place. The radius of this neighborhood indicates the The information of these benchmark functions is presented in
range that the Tasmanian devil follows the prey, which can the Appendix and in Tables 16 to 18. The performance quality
be calculated using (9). Thus, a new position based on the of TDO is compared with eight well-known metaheuristic
chasing process in this neighborhood can be calculated for the algorithms, TSA, MPA, WOA, GWO, TLBO, GSA, PSO, and
Tasmanian devil, which is mathematically simulated in (10). GA. The values of the control parameters of these algorithms
The new calculated position is acceptable to the Tasmanian are specified in Table 1.
devil if it provides a better value for the objective function Each of the competitor algorithms and the proposed
than its previous position. This position update process is TDO is used in twenty independent executions to optimize
simulated for the Tasmanian devil in (11). the benchmark functions, while each execution contains
t 1000 iterations. In presenting the simulation results, ‘‘avg’’
R = 0.01(1 − ), (9)
T is the average of the best obtained candidate solutions and
new
xi,j = xi,j + (2r − 1) · R · xi,j , (10) ‘‘std’’ is the standard deviation of these values.
(
Xi , Fi < Fi ;
new new
Xi = (11) A. EVALUATION OF UNIMODAL TEST FUNCTION (F1-F7)
Xi , otherwise,
The selected unimodal functions F1 to F7 have only one main
where R is the neighborhood radius of the point of attacked optimal solution. This feature has made unimodal functions
location, t is the iteration counter, T is the maximum number suitable for evaluating the exploitation ability of optimization
of iterations, Xinew is the new status of the ith Tasmanian devil algorithms. The optimization results of F1 to F7 functions
new is its value for the jth variable, and
in neighborhood of Xi , xi,j using TDO and eight competitor algorithms are presented
new
Fi is its objective function value. in Table 2. The simulation results show that TDO with
high exploitation power has been able to provide the global
4) REPETITIONS PROCESS, FLOWCHART, AND optimal solution for F6. TDO is also the first best optimizer in
PSEUDO-CODE OF TDO solving F1, F2, F3, F4, F5, and F7. The analysis of the results
When the update of all TDO members is completed, the first of this table shows that TDO has been able to provide much
iteration of the algorithm ends. New values are calculated for more competitive results compared to the eight competitor

VOLUME 10, 2022 19603

 M. Dehghani et al.: Tasmanian Devil Optimization: New Bio-Inspired Optimization Algorithm

FIGURE 2. Flowchart of TDO.

19604 VOLUME 10, 2022

M. Dehghani et al.: Tasmanian Devil Optimization: New Bio-Inspired Optimization Algorithm

FIGURE 3. Boxplot of performance of TDO and eight competitor algorithms in solving test functions.

VOLUME 10, 2022 19605

 M. Dehghani et al.: Tasmanian Devil Optimization: New Bio-Inspired Optimization Algorithm

FIGURE 3. (Continued.) Boxplot of performance of TDO and eight competitor algorithms in solving test functions.

algorithms with high exploitation power, which shows the that TDO with high exploration ability is able to identify the
superiority of TDO. main optimal area in the search space and has a superior
and competitive performance compared to eight competitor
algorithms.
B. EVALUATION OF HIGH-DIMENSIONAL MULTIMODAL
TEST FUNCTION (F8-F13)
The selected high-dimensional multimodal functions have C. EVALUATION OF FIXED-DIMENSIONAL MULTIMODAL
a large number of local optimal solutions. Therefore, TEST FUNCTION (F14-F23)
optimization algorithms must have high exploration power The selected fixed-dimensional multimodal functions have a
in scanning the search space to find the original local small number of variables as well as a small number of local
optimization by passing through local optimal solutions. optimal solutions. These problems challenge the exploration
The implementation results of TDO and eight competitor ability of the optimization algorithms to discover the main
algorithms on F8 to F13 functions are reported in Table 3. optimal region of search space. The optimization results
What is clear from the analysis of this table is that TDO, with obtained from TDO and eight competitor algorithms in F14
its high exploration power, has provided the global optimal to F23 optimization are presented in Table 4. Analysis of the
for F9 and F11 functions. TDO is also the best optimizer in results of this table shows that TDO with its high exploration
solving F8, F10, F12, and F13. The simulation results show power, has provided the global optimal for F14 and F17. TDO

19606 VOLUME 10, 2022

M. Dehghani et al.: Tasmanian Devil Optimization: New Bio-Inspired Optimization Algorithm

Algorithm 1 Pseudo-Code of Proposed TDO Algorithm TABLE 1. Parameter values for the competitor algorithms.
Start TDO.
1. Input the optimization problem information.
2. Set the number of iterations (T ) and the number of
members of the population (N ).
3. Initialization of the position of Tasmanian devils and
evaluation of the objective function.
4. For t = 1:T
6. For i = 1:N
8. If Probability < 0.5, Probability = rand
9. Strategy 1: Feeding by eating carrion
(exploration phase)
10. Select carrion for the ith Tasmanian devil using
Eq. (3).
11. Calculate new status of Tasmanian devil using
Eq. (4).
12. Update the ith Tasmanian devil using (5).
13. else
14. Strategy 2: Feeding by eating prey
(exploitation phase)
15. Stage 1: Prey selection and attacking
16. Select prey for the ith Tasmanian devil using (6).
17. Calculate new status of Tasmanian devil using
Eq. (7).
18. Update the ith Tasmanian devil using (8).
19. Stage 2: Prey chasing
20. Update neighborhood radius using (9).
21. Calculate new status of the ith Tasmanian devil in
neighborhood of Xi using (10).
22. Update the ith Tasmanian devil using (11).
23. end if
24. end for i = 1:N
25. Save the best proposed solution so far.
26. end for t = 1:T
27. Output: The best solution obtained by TDO for given
optimization problem.
End TDO.

has outperformed eight competitor algorithms in solving

F15, F16, and F20. Also, the analysis of ‘‘avg’’ and ‘‘std’’
criteria, indicates the more effective performance of TDO
in optimizing F18, F19, F21, F22, and F23. The simulation
results of F14 to F23 functions show the superiority of TDO
performance in providing optimal solutions compared to
eight competitor algorithms.
The performance of TDO and eight competitor algorithms
in optimizing benchmark functions is presented as a boxplot
in Figure 3. regard, a statistical analysis is presented to examine whether
the superiority of TDO has been significant or not against any
D. STATISTICAL ANALYSIS of the competitor algorithms in this subsection. To provide
Presentation of simulation results using ‘‘avg’’ and ‘‘std’’ statistical analysis on the performance of TDO and eight
criteria provides valuable information on the ability of competitor algorithms, Wilcoxon rank sum test [43] has been
optimization algorithms and their comparison. However, it is used. In this test, a p-value is used to show the significant
always possible, even with the slightest probability, that the superiority of the corresponding algorithm over a competitor
superiority of one algorithm over another is a chance. In this algorithm.

VOLUME 10, 2022 19607

 M. Dehghani et al.: Tasmanian Devil Optimization: New Bio-Inspired Optimization Algorithm

TABLE 2. Optimization results of TDO and competitor algorithms on unimodal test function.

TABLE 3. Optimization results of TDO and competitor algorithms on high-dimensional multimodal test function.

The simulation results obtained from the Wilcoxon rank To analyze the sensitivity to the parameter N , the proposed
sum test are presented in Table 5. What can be deduced from TDO is employed for the population size of Tasmanian devils
the analysis of the results of this test is that in cases where equals to 20, 30, 50, and 100 in solving F1 to F23. The
a p-value is less than 0.05, TDO has a significant statistical simulation results of the sensitivity analysis of TDO to the
superiority over the competitor algorithm. parameter N are presented in Table 5. TDO convergence
curves in solving these functions and for different values of
N are shown in Figure 4. What can be deduced from the
E. SENSITIVITY ANALYSIS analysis of the simulation results is that with the increase
TDO is able to solve optimization problems in a repetition- in the population of Tasmanian devils, the search power of
based process based on search space scans by members TDO has improved and led to a decrease in the values of the
of the population of Tasmanian devils. Thus, the number of objective functions.
population members of Tasmanian devils and the number To analyze the sensitivity to the parameter T , the proposed
of iterations of the algorithm affect the performance of TDO algorithm for different values of T equal to 100, 500,
the TDO. In this subsection, TDO sensitivity analysis to 800, and 1000 is implemented on the benchmark functions F1
parameters N and T is studied. to F23. The results of TDO sensitivity analysis study under

19608 VOLUME 10, 2022

M. Dehghani et al.: Tasmanian Devil Optimization: New Bio-Inspired Optimization Algorithm

TABLE 4. Optimization results of TDO and competitor algorithms on fixed-dimensional multimodal test function.

TABLE 5. p-values obtained from Wilcoxon rank sum test. A. WELDED BEAM DESING OPTIMIZATION PROBLEM
Welded beam design is a minimization problem which its
main purpose is to reduce the fabrication cost of welded
beam [13]. A schematic of this problem is shown in Figure 6.
The optimum values of the design variables and the values
of the objective function using TDO and eight competitor
algorithms are presented in Table 8. TDO provides the best
candidate solution by providing the values of the design vari-
ables equal to (0.205730, 3.470521, 9.036603, 0.205731) and
the corresponding objective function value equal to 1.724901.
The statistical results of the performances of TDO and eight
competitor metaheuristics are presented in Table 9. The
the changes of parameter T are reported in Table 6. The simulation results show that TDO is superior to eight com-
behavior of TDO convergence curves under the influence of petitor algorithms by providing optimal performance. The
parameter T is presented in Figure 5. What is evident from convergence curve behavior of TDO in achieving the optimal
the simulation results of the sensitivity analysis is that the solution for the welded beam design problem is shown
increase in values T has led the algorithm to converge to better in Figure 7.
solutions and reduce the values of the objective functions.
B. PRESSURE VESSEL DESING OPTIMIZATION
IV. TDO APPLICATION FOR ENGINEERING PROBLEM
DESIGN PROBLEMS Pressure vessel design is a minimization problem whose main
The performance of TDO in real-world applications is purpose is to reduce the total cost of material, welding, and
evaluated by optimizing four engineering design optimization forming of a cylindrical vessel [44]. A schematic of this
problems including welded beam design, pressure vessel problem is shown in Figure 8. The implementation results
design, speed reducer design, and tension/compression spring of TDO and eight competitor algorithms in optimizing the
design. pressure vessel design problem are presented in Table 10.

VOLUME 10, 2022 19609

 M. Dehghani et al.: Tasmanian Devil Optimization: New Bio-Inspired Optimization Algorithm

FIGURE 4. Sensitivity analysis of the TDO for the number of population members.

TDO provides the optimal 11 by providing better values for presented in Table 11. The simulation results show the
design variables equal to (0.7780535, 0.3860383, 40.31357, superiority of TDO in solving the pressure vessel design
199.9841) and the corresponding objective function value problem more effectively than eight competitor algorithms.
equal to 5887.1783. The statistical results obtained from The TDO convergence curve to optimize this problem is
the implementation of TDO and eight metaheuristics are shown in Figure 9.

19610 VOLUME 10, 2022

M. Dehghani et al.: Tasmanian Devil Optimization: New Bio-Inspired Optimization Algorithm

FIGURE 5. Sensitivity analysis of the TDO for the maximum number of iterations.

C. SPEED REDUCER DESING OPTIMIZATION PROBLEM design problem are presented in Table 12. TDO has been
Speed reducer design is a minimization problem whose able to provide the optimal solution to this problem with the
main purpose is to reduce the weight of the speed values of the design variables equal to (3.5, 0.7, 17, 7.3, 7.8,
reducer [45], [46]. A schematic of this problem is shown 3.35021, 5.28668) and the corresponding objective function
in Figure 10. The application results of TDO and eight value equal to 2996.3482. The statistical results of the
competitor metaheuristics in optimizing the speed reducer implementation of TDO and eight competitor metaheuristics

VOLUME 10, 2022 19611

 M. Dehghani et al.: Tasmanian Devil Optimization: New Bio-Inspired Optimization Algorithm

TABLE 6. Sensitivity analysis of the TDO for the number of population members.

TABLE 7. Sensitivity analysis of the TDO for the maximum number of iterations.

19612 VOLUME 10, 2022

M. Dehghani et al.: Tasmanian Devil Optimization: New Bio-Inspired Optimization Algorithm

FIGURE 9. Convergence analysis of the TDO for the pressure vessel

design optimization problem.

FIGURE 6. Schematic view of the welded beam design problem.

FIGURE 10. Schematic view of speed reducer design problem.

FIGURE 7. Convergence analysis of the TDO for the welded beam design
optimization problem.

FIGURE 11. Convergence analysis of the TDO for the speed reducer
design optimization problem.

FIGURE 8. Schematic view of pressure vessel design problem.

are presented in Table 13. The simulation results show the

superiority of TDO compared to eight competitor algorithms
in minimizing the objective function of this problem. FIGURE 12. Schematic view of tension/compression spring problem.
The TDO convergence curve during achieving the optimal
solution is shown in Figure 11.
in Figure 12. The values obtained for the design variables
D. TENSION/COMPRESSION SPRING DESING and the objective function of this problem are presented in
OPTIMIZATION PROBLEM Table 14. TDO presents the optimal solution to the problem
Tension/compression spring design is a minimization prob- by providing the values of the design variables equal to
lem whose main purpose is to reduce the tension/compression (0.0518001, 0.359375, 11.1509) and the value of the objec-
spring weight [13]. A schematic of this problem is shown tive function equal to 0.012671024. The statistical results

VOLUME 10, 2022 19613

 M. Dehghani et al.: Tasmanian Devil Optimization: New Bio-Inspired Optimization Algorithm

TABLE 8. Comparison results for the welded beam design problem.

TABLE 9. Statistical results for the welded beam design problem.

TABLE 10. Comparison results for the pressure vessel design problem.

TABLE 11. Statistical results for the pressure vessel design problem.

obtained from the optimization of the tension/compression to eight competitor algorithms in solving this problem. The
spring design problem using TDO and eight competitor convergence curve behavior of TDO in providing the optimal
metaheuristics are presented in Table 15. The simulation solution to the tension/compression spring design problem is
results show that TDO has a superior performance compared shown in Figure 13.

19614 VOLUME 10, 2022

M. Dehghani et al.: Tasmanian Devil Optimization: New Bio-Inspired Optimization Algorithm

TABLE 12. Comparison results for the speed reducer design problem.

TABLE 13. Statistical results for the speed reducer design problem.

TABLE 14. Comparison results for the tension/compression spring design problem.

TABLE 15. Statistical results for the tension/compression spring design problem.

VOLUME 10, 2022 19615

 M. Dehghani et al.: Tasmanian Devil Optimization: New Bio-Inspired Optimization Algorithm

TABLE 16. Unimodal objective functions.

the quality of TDO results, its performance was compared

with eight well-known algorithms, TSA, MPA, WOA, GWO,
TLBO, GSA, PSO, and GA. What was concluded from
the simulation results was that TDO by providing strong
performance and creating the appropriate balance between
exploration and exploitation, is superior than the eight
competitor algorithms and provides far more competitive
optimization results. TDO’s performance in optimizing four
design problems showed TDO’s high ability to solve real-
world optimization problems.
The authors provide perspectives for future studies in
this paper, the main ones being the design of binary and
multi-objective TDO versions. The use of TDO in solving
FIGURE 13. Convergence analysis of the TDO for the
tension/compression spring design optimization problem. optimization problems in various sciences and real-world
problems are other suggestions that open the way for further
studies.
V. CONCLUSION AND FUTURE WORKS
In this paper, a new bio-inspired metaheuristic algorithm APPENDIX A
called Tasmanian Devil Optimization (TDO) was introduced. See Tables 16–18.
The fundamental inspiration of TDO is the Tasmanian devil
feeding behavior in nature, which has two strategies (i) eating APPENDIX B
carrion and (ii) feeding through hunting. TDO mathematical WELDED BEAM DESIGN PROBLEM
modeling was presented along with a description of its steps
Consider X = [x1 , x2 , x3 , x4 ] = [h, l, t, b].
and strategies. The performance of TDO in solving optimiza-
tion problems was tested on twenty-three objective functions Minimize f (x) = 1.10471x12 x2 + 0.04811x3 x4 (14.0 + x2 ).
of unimodal and multimodal types. The optimization results Subject to : g1 (x) = τ (x) − 13600 ≤ 0,
of unimodal functions showed the exploitation ability of TDO g2 (x) = σ (x) − 30000 ≤ 0,
in convergence towards global optimal. The optimization
g3 (x) = x1 − x4 ≤ 0,
results of multimodal functions showed that TDO has a high
exploration ability in the scanning search space, passing local g4 (x) = 0.10471x12 + 0.04811x3 x4 (14 + x2 )
areas, and discovering the main optimal area. To analyze − 5.0 ≤ 0,

19616 VOLUME 10, 2022

M. Dehghani et al.: Tasmanian Devil Optimization: New Bio-Inspired Optimization Algorithm

TABLE 17. High-dimensional multimodal objective functions.

g5 (x) = 0.125 − x1 ≤ 0, With

g6 (x) = δ(x) − 0.25 ≤ 0, 0.1 ≤ x1 , x4 ≤ 2 and 0.1 ≤ x2 , x3 ≤ 10.
g7 (x) = 6000 − pc (x) ≤ 0.
APPENDIX C
where PRESSURE VESSEL DESIGN PROBLEM
Consider X = [x1 , x2 , x3 , x4 ] = [Ts , Th , R, L] .
r
x2
τ (x) = τ 0 + (2τ τ 0 ) + (τ 00 )2 ,
2R Minimize f (x) = 0.6224x1 x3 x4 + 1.778x2 x32
6000 + 3.1661x12 x4 + 19.84x12 x3 .
τ0 = √ ,
2 x1 x2 Subject to : g1 (x) = −x1 + 0.0193x3 ≤ 0,
MR
τ 00 = , g2 (x) = −x2 + 0.00954x3 ≤ 0,
J 
x2  4
M = 6000 14 + , g3 (x) = −πx32 x4 − πx33 + 1296000 ≤ 0,
2 3
s g4 (x) = x4 − 240 ≤ 0.
x22 x1 + x3 2
 
R = + , With
4 2
0 ≤ x1 , 10 ≤ x3 , x4 ≤ 200.
( "  #)
√ x22

x1 + x3 2 x2 ≤ 100, and
J = 2 x1 x2 2 + ,
12 2
APPENDIX D
504000 SPEED REDUCER DESIGN PROBLEM
σ (x) =
x4 x32 Consider X = x1, x2 , x3 , x4 , x5 , x 6 , x7
 

δ (x) =
65856000
, = [b, m, p, l1 , l2 , d1 , d2 ] .
30 · 106 x4 x33


Minimize f (x) = 0.7854x1 x22
q  

x32 x46  s 
3.3333x32 + 14.9334x3 − 43.0934
4.013 30 · 106 36  x3 6
30 · 10 
pc (x) = 1− .  
196 28 4(12 · 106 ) − 1.508x1 x62 + x72 + 7.4777

VOLUME 10, 2022 19617

 M. Dehghani et al.: Tasmanian Devil Optimization: New Bio-Inspired Optimization Algorithm

TABLE 18. Fixed-dimensional multimodal objective functions.

 
× x63 + x73 + 0.7854(x4 x62 + x5 x72 ). − 1 ≤s0,
27 745x5 2
 
1
Subject to : g1 (x) = − 1 ≤ 0, g6 (x) = + 157.5 · 106
x1 x22 x3 85x73 x2 x3
397.5 − 1 ≤ 0,
g2 (x) = − 1 ≤ 0, x2 x3
x1 x22 x3 g7 (x) = − 1 ≤ 0,
40
1.93x43 5x 2
g3 (x) = − 1 ≤ 0, g8 (x) = − 1 ≤ 0,
x2 x3 x64 x1
1.93x53 x1
g4 (x) = − 1 ≤ 0, g9 (x) = − 1 ≤ 0,
12x2
x2 x3 x74
1.5x 6 + 1.9
g10 (x) = − 1 ≤ 0,
s
745x4 2
 
1 x4
g5 (x) = + 16.9 · 106
110x63 x2 x3 1.1x 7 + 1.9
g11 (x) = − 1 ≤ 0.
x5
19618 VOLUME 10, 2022
M. Dehghani et al.: Tasmanian Devil Optimization: New Bio-Inspired Optimization Algorithm

With [14] A. Faramarzi, M. Heidarinejad, S. Mirjalili, and A. H. Gandomi, ‘‘Marine

predators algorithm: A nature-inspired Metaheuristic,’’ Expert Syst. Appl.,
2.6 ≤ x1 ≤ 3.6, 0.7 ≤ x2 ≤ 0.8, 17 ≤ x3 ≤ 28, vol. 152, Aug. 2020, Art. no. 113377.
[15] S. Mirjalili, S. M. Mirjalili, and A. Lewis, ‘‘Grey wolf optimizer,’’ Adv.
7.3 ≤ x4 ≤ 8.3, 7.8 ≤ x5 ≤ 8.3, 2.9 ≤ x6 ≤ 3.9, and Eng. Softw., vol. 69, pp. 46–61, Mar. 2014.
[16] D. Karaboga and B. Basturk, ‘‘Artificial bee colony (ABC) optimization
5 ≤ x7 ≤ 5.5. algorithm for solving constrained optimization problems,’’ in Proc. Int.
Fuzzy Syst. Assoc. World Congr., 2007, pp. 789–798.
APPENDIX E [17] S. Kaur, L. K. Awasthi, A. L. Sangal, and G. Dhiman, ‘‘Tunicate swarm
algorithm: A new bio-inspired based metaheuristic paradigm for global
TENSION/COMPRESSION SPRING DESIGN PROBLEM optimization,’’ Eng. Appl. Artif. Intell., vol. 90, Apr. 2020, Art. no. 103541.
[18] D. Połap and M. Woźniak, ‘‘Red fox optimization algorithm,’’ Expert Syst.
Consider X = [x1 , x2 , x3 ] = [d, D, P] . Appl., vol. 166, Mar. 2021, Art. no. 114107.
Minimize f (x) = (x3 + 2) x2 x12 . [19] S. Z. Koohi, N. A. W. A. Hamid, M. Othman, and G. Ibragimov, ‘‘Raccoon
optimization algorithm,’’ IEEE Access, vol. 7, pp. 5383–5399, 2019.
x23 x3 [20] A. G. Hussien, M. Amin, M. Wang, G. Liang, A. Alsanad, A. Gumaei,
Subject to : g1 (x) = 1 − ≤ 0, and H. Chen, ‘‘Crow search algorithm: Theory, recent advances, and
71785x14 applications,’’ IEEE Access, vol. 8, pp. 173548–173565, 2020.
4x22 − x1 x2 1 [21] R. V. Rao, V. J. Savsani, and D. P. Vakharia, ‘‘Teaching-learning-
g2 (x) = + − 1 ≤ 0, based optimization: A novel method for constrained mechanical design
12566(x2 x13 ) 5108x12 optimization problems,’’ Comput.-Aided Des., vol. 43, no. 3, pp. 303–315,
Mar. 2011.
140.45x1
g3 (x) = 1− ≤ 0, [22] Y. Meraihi, A. B. Gabis, S. Mirjalili, and A. Ramdane-Cherif, ‘‘Grasshop-
x22 x3 per optimization algorithm: Theory, variants, and applications,’’ IEEE
Access, vol. 9, pp. 50001–50024, 2021.
x1 + x2
g4 (x) = − 1 ≤ 0. [23] P. J. Van Laarhoven and E. H. Aarts, ‘‘Simulated annealing,’’ Simu-
1.5 lated Annealing: Theory and Applications. Dordrecht, The Netherlands:
Springer, 1987, pp. 7–15.
With [24] E. Rashedi, H. Nezamabadi-Pour, and S. Saryazdi, ‘‘GSA: A gravitational
search algorithm,’’ J. Inf. Sci., vol. 179, no. 13, pp. 2232–2248, 2009.
0.05 ≤ x1 ≤ 2, 0.25 ≤ x 2 ≤ 1.3 and 2 ≤ x3 ≤ 15. [25] M. Dehghani, Z. Montazeri, A. Dehghani, and A. Seifi, ‘‘Spring search
algorithm: A new meta-heuristic optimization algorithm inspired by
Hooke’s law,’’ in Proc. IEEE 4th Int. Conf. Knowl.-Based Eng. Innov.
ACKNOWLEDGMENT (KBEI), Dec. 2017, pp. 0210–0214.
The authors would like to thank Eva Trojovská for her [26] H. Eskandar, A. Sadollah, A. Bahreininejad, and M. Hamdi, ‘‘Water
significant help with the graphic design of the article. cycle algorithm—A novel Metaheuristic optimization method for solv-
ing constrained engineering optimization problems,’’ Comput. Struct.,
vols. 110–111, pp. 151–166, Nov. 2012.
REFERENCES [27] H. Du, X. Wu, and J. Zhuang, ‘‘Small-world optimization algorithm
[1] J. O. Agushaka and A. E. Ezugwu, ‘‘Advanced arithmetic optimization for function optimization,’’ in Proc. Int. Conf. Natural Comput., 2006,
algorithm for solving mechanical engineering design problems,’’ PLoS pp. 264–273.
ONE, vol. 16, no. 8, Aug. 2021, Art. no. e0255703. [28] A. Hatamlou, ‘‘Black hole: A new heuristic optimization approach for data
[2] S. Mirjalili, ‘‘The ant lion optimizer,’’ Adv. Eng. Softw., vol. 83, pp. 80–98, clustering,’’ Inf. Sci., vol. 222, pp. 175–184, Feb. 2012.
May 2015. [29] Z. Wei, C. Huang, X. Wang, T. Han, and Y. Li, ‘‘Nuclear reaction
[3] R. G. Rakotonirainy and J. H. van Vuuren, ‘‘Improved metaheuristics for optimization: A novel and powerful physics-based algorithm for global
the two-dimensional strip packing problem,’’ Appl. Soft Comput., vol. 92, optimization,’’ IEEE Access, vol. 7, pp. 66084–66109, 2019.
Jul. 2020, Art. no. 106268. [30] S. Mirjalili, S. M. Mirjalili, and A. Hatamlou, ‘‘Multi-verse optimizer:
[4] X. Wu, S. Zhang, W. Xiao, and Y. Yin, ‘‘The exploration/exploitation A nature-inspired algorithm for global optimization,’’ Neural Comput.
tradeoff in whale optimization algorithm,’’ IEEE Access, vol. 7, Appl., vol. 27, no. 2, pp. 495–513, 2016.
pp. 125919–125928, 2019. [31] B. Alatas, ‘‘ACROA: Artificial chemical reaction optimization algo-
[5] S.-H. Liu, M. Mernik, D. Hrnčič, and M. Črepinšek, ‘‘A parameter control rithm for global optimization,’’ Expert Syst. Appl., vol. 38, no. 10,
method of evolutionary algorithms using exploration and exploitation pp. 13170–13180, 2011.
measures with a practical application for fitting Sovova’s mass transfer [32] A. H. Kashan, ‘‘A new metaheuristic for optimization: Optics inspired
model,’’ Appl. Soft Comput., vol. 13, no. 9, pp. 3792–3805, Sep. 2013. optimization (OIO),’’ Comput. Oper. Res., vol. 55, pp. 99–125, Mar. 2015.
[6] F. Zhang, J. Chen, T. Mao, and Z. Wang, ‘‘Feedback interval optimization [33] A. Faramarzi, M. Heidarinejad, B. Stephens, and S. Mirjalili, ‘‘Equilibrium
for MISO LiFi systems,’’ IEEE Access, vol. 9, pp. 136811–136818, 2021. optimizer: A novel optimization algorithm,’’ Knowl.-Based Syst., vol. 191,
[7] K. Fukada, M. Parizy, Y. Tomita, and N. Togawa, ‘‘A three-stage annealing Mar. 2020, Art. no. 105190.
method solving slot-placement problems using an ising machine,’’ IEEE [34] W. Zhao, L. Wang, and Z. Zhang, ‘‘Atom search optimization and its
Access, vol. 9, pp. 134413–134426, 2021. application to solve a hydrogeologic parameter estimation problem,’’
[8] D. E. Goldberg and J. H. Holland, ‘‘Genetic algorithms and machine Knowl.-Based Syst., vol. 163, pp. 283–304, Jan. 2019.
learning,’’ Mach. Learn., vol. 3, nos. 2–3, pp. 95–99, 1988. [35] H. Abedinpourshotorban, S. M. Shamsuddin, Z. Beheshti, and
[9] R. Storn and K. Price, ‘‘Differential evolution—A simple and efficient D. N. A. Jawawi, ‘‘Electromagnetic field optimization: A physics-
heuristic for global optimization over continuous spaces,’’ J. Global inspired metaheuristic optimization algorithm,’’ Swarm Evol. Comput.,
Optim., vol. 11, no. 4, pp. 341–359, 1997. vol. 26, pp. 8–22, Feb. 2016.
[10] S. Hofmeyr and S. Forrest, ‘‘Architecture for an artificial immune system,’’ [36] M. Dehghani, M. Mardaneh, J. Guerrero, O. Malik, and V. Kumar,
Evol. Comput., vol. 8, no. 4, pp. 443–473, Dec. 2000. ‘‘Football game based optimization: An application to solve energy
[11] R. Eberhart and J. Kennedy, ‘‘Particle swarm optimization,’’ in Proc. Int. commitment problem,’’ Int. J. Intell. Eng. Syst., vol. 13, no. 5, pp. 514–523,
Conf. Neural Netw., 1995, pp. 1942–1948. Oct. 2020.
[12] M. Dorigo, V. Maniezzo, and A. Colorni, ‘‘Ant system: Optimization by [37] R. Moghdani and K. Salimifard, ‘‘Volleyball premier league algorithm,’’
a colony of cooperating agents,’’ IEEE Trans. Syst., Man, Cybern. B, Appl. Soft. Comput., vol. 64, pp. 161–185, Mar. 2018.
Cybern., vol. 26, no. 1, pp. 29–41, Feb. 1996. [38] S. Doumari, H. Givi, M. Dehghani, and O. Malik, ‘‘Ring toss game-based
[13] S. Mirjalili and A. Lewis, ‘‘The whale optimization algorithm,’’ Adv. Eng. optimization algorithm for solving various optimization problems,’’ Int.
Softw., vol. 95, pp. 51–67, Feb. 2016. J. Intell. Eng. Syst., vol. 14, no. 3, pp. 545–554, Jun. 2021.

VOLUME 10, 2022 19619

 M. Dehghani et al.: Tasmanian Devil Optimization: New Bio-Inspired Optimization Algorithm

[39] M. Dehghani, Z. Montazeri, H. Givi, J. Guerrero, and G. Dhiman, ‘‘Darts ŠTĚPÁN HUBÁLOVSKÝ received the M.Sc. and
game optimizer: A new optimization technique based on darts game,’’ Int. Ph.D. degrees from the Faculty of Mathematics
J. Intell. Eng. Syst., vol. 13, no. 5, pp. 286–294, Oct. 2020. and Physics, Charles University, Prague, Czech
[40] D. H. Wolper and W. G. Macready, ‘‘No free lunch theorems for Republic, in 1995 and 1998, respectively. In 2012,
optimization,’’ IEEE Trans. Evol. Comput., vol. 1, no. 1, pp. 67–82, he became an Associate Professor at the Faculty
Apr. 1997. of Informatics and Management, University of
[41] O. L. Buchmann and E. R. Guiler, ‘‘Behaviour and ecology of the Hradec Králové, Czech Republic. He is currently
Tasmanian devil, Sarcophilus harrisii,’’ in The Biology of Marsupials.
the Vice Dean at the Faculty of Science, University
London, U.K.: Macmillan, 1977, pp. 155–168.
of Hradec Králové. His research interests include
[42] X. Yao, Y. Liu, and G. Lin, ‘‘Evolutionary programming made faster,’’
IEEE Trans. Evol. Comput., vol. 3, no. 2, pp. 82–102, Jul. 1999. technical cybernetics, computer simulation and
[43] F. Wilcoxon, ‘‘Individual comparisons by ranking methods,’’ in Break- optimization, and big data processing.
throughs in Statistics. New York, NY, USA: Springer-Verlag, 1992,
pp. 196–202.
[44] B. K. Kannan and S. N. Kramer, ‘‘An augmented Lagrange multiplier
based method for mixed integer discrete continuous optimization and
its applications to mechanical design,’’ J. Mech. Des., vol. 116, no. 2,
pp. 405–411, 2008.
[45] A. H. Gandomi and X.-S. Yang, ‘‘Benchmark problems in structural
optimization,’’ in Computational Optimization, Methods and Algorithms.
Berlin, Germany: Springer, 2011, pp. 259–281. PAVEL TROJOVSKÝ received the M.Sc. degree in
[46] E. Mezura-Montes and C. A. C. Coello, ‘‘Useful infeasible solutions in teaching of mathematics, physics, and computer
engineering optimization with evolutionary algorithms,’’ in Proc. Mex. Int. science from the University of Hradec Králové,
Conf. Artif. Intell., 2005, pp. 652–662. Hradec Králové, Czech Republic, in 1989, and the
Ph.D. degree in general questions of mathematics
and computer science from the Faculty of Math-
MOHAMMAD DEHGHANI received the B.S. ematics and Physics, Charles University, Prague,
degree in electrical engineering from the Shahid Czech Republic, in 2001. In 2011, he became
Bahonar University of Kerman, Iran, in 2012, the an Associate Professor of system engineering
M.S. degree in electrical engineering from Shiraz and informatics with the University of Pardubice,
University, Shiraz, Iran, in 2016, and the Ph.D. Pardubice, Czech Republic. He is currently an Associate Professor and the
degree from the Shiraz University of Technology, Vice Dean for Creative Activities at the Faculty of Science, University
Shiraz, in 2020. His current research interests of Hradec Králové. His research interests include number theory and its
include optimization, metaheuristic algorithms, applications in cryptography, applied mathematics, computer simulation and
power systems, and energy commitment. optimization, and big data processing.

19620 VOLUME 10, 2022