Improved Raven Roosting Optimization algorithm (IRRO)
Introduction
Optimization problems have high complexity and are applicable in many fields of science and engineering. As there are many solutions to a problem, it is necessary to find the nearest -optimal solution among the many possibilities [1]. There are many ways to solve such problems, and one method is use of bio-inspired meta-heuristic algorithms. These algorithms are inspired from the behavior of living organisms. They contain a population of organisms that interact with each other and with their environment. And this interaction creates complex group behaviors. Algorithms such as the Particle Swarm Optimization (PSO) [2]; Cat Swarm Optimization (CSO) [3]; Imperialist Competitive [4]; Firefly Algorithm [5] and Cuckoo Optimization [6], are based on bio-inspired imitations. These algorithms try to find the optimal solution to a problem from all the possibilities in reasonable time and with tolerable computational cost [7]. The Raven Roosting Optimization (RRO) [8] is a meta-heuristic algorithm based on imitation that has been proposed recently. This algorithm focuses on the social behavior of ravens and the flow of information between members of the population to find food. In this algorithm, a member who has the best food source is considered as the leader. Part of the population follows the leader to obtain a food source and the other members fly towards their best personal position that's been had so far. But the RRO algorithm has the problem of premature convergence; this term means that the population converged too early at a suboptimal point. Usually, premature convergence occurs because of a weakness in its capacity for exploration (global search). This problem could be resolved under the hypothesis that the population that flies towards the leader is then divided according to groups of weak ravens and greedy ravens. Also, a parameter was used to control the amount of food remaining for the ravens. In order to improve the exploration capability, the average best fitness reduced as the dependent variable. The purpose of this paper was analysis and evaluation of grouping the population and other parameters of the proposed algorithm according to average best fitness, using the standard benchmark functions and comparisons with other bio-inspired metaheuristic algorithms based on imitation. All the experiments performed in this study, were based on the set parameters, and evaluated in comparison with other algorithms in Ref. [8]. The experimental results on 30 test functions how the improvement of the proposed algorithm compared to RRO [8], PSO [2], BA [9], CSO [10], GWO [11] and WOA [12] algorithms. The main contributions of this paper can be summarized as follows: 1) Presenting an Improved Raven Roosting Optimization (IRRO) algorithm in order to resolve the premature convergence of the basic Raven Roosting Optimization (RRO) algorithm. 2) Focusing on the population that follows the leader. The population is divided into weak and greedy raven groups. Increasing the percent of weak ravens reinforces the exploitation ability of IRRO. 3) Controlling the remained amount of food for ravens in their current position by introducing a new parameter that is decreased in each iteration of the algorithm. This lead to increasing the tendency of the ravens to change their location that is analogous to better exploration.
This paper was organized as follows: In Section 2, there is a brief review of the meta-heuristic algorithms and applications. In section 3, the Raven Roosting Optimization (RRO) algorithm is reviewed. Section 4 introduces the Improved Raven Roosting Optimization (IRRO) algorithm. The experiments and results are presented in Section 5. Section 6 discusses on the experiments and results. Finally, conclusions and opportunities for future works are discussed in Section 7.
Section snippets
Related work
There are several methods for solving optimization problems. One of these is the use of bio-inspired meta-heuristic algorithms. Fig. 1 shows the general classification of meta-heuristic algorithms. As seen in Fig. 1, highlighted sections show trends in the field of study. Meta-heuristic algorithms are divided into three categories; evolutionary algorithms, non-bio meta-heuristic algorithms and bio-inspired meta- heuristic algorithms.
Raven Roosting Optimization Algorithm (RRO)
Recently proposed is the bio-inspired metaheuristic algorithm, Raven Roosting Optimization (RRO) [8]. This algorithm focuses on the social behavior of ravens and the flow of information between members of the population in order to find food. Usually, ravens live collectively in groups of 200 to ten thousand in winter and autumn (the non-breeding seasons). These groups are called social sleep or “roost” and are formed near to food sources on trees. The reasons for social sleep are still a
Improved Raven Roosting Optimization (IRRO)
As mentioned in section 3, the roost is considered the center of information, a place where information on food sources is exchanged between members. In this paper, the main focus was structure of the RRO algorithm. The part of the raven population that leaves the roost to find food and follows the leader is divided into two groups, according to weak ravens and greedy ravens. The ravens that cannot find food sources for themselves in a day are determined as weak ravens. Members of this group
Experiment and results
The proposed algorithm was implemented in MATLAB version R2015a. After setting the parameters of the algorithm, nine different versions of the proposed algorithm (IRRO0-IRRO8) was configured based on Table 2. To evaluate the proposed algorithm, 30 standard benchmark functions (CEC2017) [41] were used in two levels of dimensionality (50 and 100). The proposed algorithm has been compared with the RRO [8], PSO [2], BA [9], CSO [10], GWO [11] and WOA [12] algorithms. The settings of parameters and
Discussion
This study presents an extension of the RRO algorithm. The extension focused on the structure of the population that Fly to follow the leader. And this population was divided into two groups according to weak ravens and greedy ravens. The results presented in Section 4 show the parameter “Weak_Population_Percent” indicated the percentage of the group of weak ravens that followed the leader to find the food sources. Setting this parameter affected exploitation of the proposed algorithm and led
Conclusions and future works
The category of metaheuristic algorithms are bio-inspired metaheuristic algorithms based on imitation that have no common memory, and the velocity and movement of particles are used in searching a problem environment. The algorithms are evaluated in terms of congestion speed, local and global search, etc. Raven Roosting Optimization (RRO) algorithm is of this category of metaheuristic algorithms that suffers from premature convergence. This paper introduced an extension of the RRO called
Acknowledgements
The first author would like to acknowledge the contribution of Mahmood Alian and Dr. S Mostapha Kalami Heris for their help and comments.
References (42)
- et al.
Cuckoo search via Lévy flights
Cuckoo optimization algorithm
Appl. Soft Comput. J.
(2011)- et al.
Grey wolf optimizer
Adv. Eng. Softw.
(2014) - et al.
The whale optimization algorithm
Adv. Eng. Softw.
(2016) - et al.
ARO: a new model free optimization algorithm for real time applications inspired by the asexual reproduction
Expert Syst. Appl.
(2011) - et al.
GSA: a gravitational search algorithm
Inf. Sci. (Ny).
(2009) - et al.
Electromagnetic field optimization: a physics-inspired metaheuristic optimization algorithm
Swarm Evol. Comput.
(2016) - et al.
A survey of swarm intelligence for dynamic optimization: algorithms and applications
Swarm Evol. Comput.,
(2017) - et al.
A hybrid algorithm using ant and bee colony optimization for feature selection and classification (AC-ABC Hybrid)
Swarm Evol. Comput.
(October 2017) Future paths for integer programming and links to artificial intelligence
Comput. Oper. Res.
(1986)
ACO based speed control of SRM fed by photovoltaic system
Int. J. Electr. Power Energy Syst.
MPPT control design of PV system supplied SRM using BAT search algorithm
Sustain. Energy, Grids Netw.
Improved harmony algorithm and power loss index for optimal locations and sizing of capacitors in radial distribution systems
Int. J. Electr. Power Energy Syst.
Improved Harmony Algorithm for optimal locations and sizing of capacitors in radial distribution systems
Int. J. Electr. Power Energy Syst.
A survey of bio inspired optimization algorithms
Int. J. Soft Comput. Eng.
Particle swarm optimization, neural networks, 1995
Proc. IEEE Int. Conf.
Cat swarm optimization
Imperialist competitive algorithm: an algorithm for optimization inspired by imperialistic competition
Firefly algorithm, stochastic test functions and design optimization
Int. J. Bio-inspired Comput.
The raven roosting optimisation algorithm
Soft Comput.
A new metaheuristic bat-inspired algorithm
Nicso
Cited by (25)
A comprehensive meta-analysis of emerging swarm intelligent computing techniques and their research trend
2022, Journal of King Saud University - Computer and Information SciencesA powerful meta-heuristic search algorithm for solving global optimization and real-world solar photovoltaic parameter estimation problems
2022, Engineering Applications of Artificial IntelligenceA novel stochastic fractal search algorithm with fitness-Distance balance for global numerical optimization
2021, Swarm and Evolutionary ComputationCitation Excerpt :Base models are the first versions of algorithms to be developed and presented to the literature and include (in chronological order) Genetic Algorithm (GA, 1975) [1], Simulated Annealing (SA, 1987) [2], Particle Swarm Optimization (PSO, 1995) [3], Differential Evaluation (DE, 1997) [4], Ant Colony Optimization (ACO, 1999) [5], Harmony Search (HS, 2001) [6], Artificial Bee Colony (ABC, 2007) [7], Gravitational Search Algorithm (GSA, 2009) [8], Cuckoo Search (CS, 2009) [9], Grey Wolf Optimizer (GWO, 2014) [10], Intreior Search Algorithm (ISA, 2014) [11], Symbiotic Search Algorithms (SOS, 2014) [12], Moth-Flame Optimization (MFO, 2015) [13], Lightning Search Algorithm (LSA, 2015) [14], Stochastic Fractal Search (SFS, 2015) [15], Sine Cosine Algorithm (SCA, 2015) [16], Elephant Herding Optimization (EHO, 2015) [17], Whale Optimization Algorithm (WOA, 2016) [18], Crow Search Algorithm (CSA, 2016) [19], Yin-Yang-Pair Optimization (YYPO, 2016) [20], Electromagnetic Field Optimization (EFO, 2016) [21], Salp Swarm Algorithm (SSA, 2017) [22], Coyote Optimization Algorithm (COA, 2018) [23], Moth Swarm (MS, 2018) [24], Memetic Frog Leaping Algorithm (MFLA, 2018) [25], Emperor Penguin Optimizer (EPO, 2018) [26], Tree Growth Algorithm (TGA, 2018) [27], Earthworm Optimisation Algorithm (EOA, 2018) [28], Butterfly Optimization Algorithm (BOA, 2019) [29], Artificial Electric Field Algorithm (AEFA, 2019) [30], Atom Search Optimization (ASO, 2019) [31], Poor and Rich Optimization (PRO, 2019) [32], Sooty Tern Optimization Algorithm (STOA, 2019) [33], Equilibrium Optimizer (EO, 2019) [34], Monarch Butterfly Optimization (MBO, 2019) [35], and Barnacles Mating Optimizer (BMO, 2020) [36]. Variations are versions of algorithms that are partially modified and redesigned using various methods in order to improve search performance [37–72]. Today, there are dozens, or even hundreds, of variations of some base algorithms.
New Caledonian crow learning algorithm: A new metaheuristic algorithm for solving continuous optimization problems
2020, Applied Soft Computing JournalCitation Excerpt :Some ravens follow the leader to obtain a food source, and the remaining ravens fly toward positions of the best food sources that they have identified thus far. However, like most metaheuristic algorithms, CSA and RRO exhibit premature convergence and get stuck at local optima [29,30]. NC-crows are medium in size, all-black members of the Corvidae family and have relatively large brains [11,12].
Pareto optimality and game theory approach for optimal deployment of DG in radial distribution system to improve techno-economic benefits
2020, Applied Soft Computing JournalFitness-distance balance (FDB): A new selection method for meta-heuristic search algorithms
2020, Knowledge-Based SystemsCitation Excerpt :In the MHS process, what is expected of the selection method is that it can determine the solution candidates that can make the most contribution to the search process in a stable and conscious manner [4,15]. The selection method should effectively determine the direction of the search process and make it easier for search operators [6,16]. This brings up various issues: Do the current selection methods mentioned so far have this function?