International Journal of Advances in Electrical and Electronics Engineering

15

ISSN: 2319-1112

A Novel Classification of Evolutionary Algorithms: A Critical Assessment

Monu Tyagi 1,Souvik Ganguli 2
1,2

M.E. (PSED) Student1, Assistant Professor2 Department of Electrical & Instrumentation Engineering, Thapar University, Patiala E-mail: monutyagi026@yahoo.co.in1, souvik.ganguli@thapar.edu2

Abstractin this paper, a novel classification of evolutionary algorithms has been addressed from the literature available on its various developments. Evolutionary algorithms have been widely used in PID controller tuning for disturbance rejection, in robot and antenna design and in 3D reconstruction problems. Moreover, multi-objective functions are developed in evolutionary algorithms to solve some real-time optimization problems. These applications using evolutionary algorithms have been classified and discussed vividly in this paper. This paper also presents the various components of evolutionary algorithms too.

Index Terms-Dynamic multiple objective Evolutionary Algorithm (DMOEA), parent centric recombination operators (PCX), Evolutionary added design (EAD), Evolutionary automatic synthesis (EAS), univarite marginal distribution algorithm (UMDA).

I. INTRODUCTION volutionary algorithms (EA) are stochastic optimization techniques based on natural evolution and survival of the fittest strategy found in biological organisms. Evolutionary algorithms have been successfully applied to solve complex optimization problem in business, engineering, and science. Some commonly used EAs are Genetic algorithms (GAs), Evolutionary Programming (EP), Evolutionary Strategy (ES) and Differential Evolution (DE). Each of these methods has its own characteristics, strengths and weaknesses. In general, an EA algorithm can generate a set of initial solutions randomly based on the given seed and population size. Afterwards, it will go through evolution operations such as cross-over and mutation before evaluated by the objective function. The wining entity in the population will be selected as the parents (or seed) of the next generation. The optimization iteration continues until the termination criteria are satisfied. Typically, either the evolution process reached users define maximum number of iteration or the improvement in objective function between the two generations converges [4, 13]. A self-tuning PID controller is developed by Teng Fong-Chwee et al. based on pole assignment approach to overcome fractional dead time, constant and known dead time, plus time-varying dead time. This method showed good results in handling dead time processes [1]. T. Back et al. present a overview of Evolutionary Algorithms for parameter optimization [2]. Evolutionary Computation is developed by D.B. Fogel. In this literature covering evolutionary programming, evolution strategies, and genetic algorithms from a perspective of achieving machine intelligence through evolution [3]. Kalyanmoy Deb et al. proposed a generic parent-centric recombination operator (PCX) and a computationally fast population-alteration model called G3 model. The performance of G3 model with PCX operator is found better in consistency and reliability when compared to the differential evolution technique and the quasi-Newton method [5]. Corina Rotar developed a new evolutionary algorithm for multiobjective optimization which is inspired by endocrine system and uses the Pareto non-dominance concept [6]. A robust tuning method for disturbance rejection of PID controller using Evolutionary algorithm is proposed by Dong Hwa Kim et al. which tunes the gain of PID controller using fitness value of immune algorithm [7]. Gregory S. Hornby presented a method for automated antenna design and optimization based on Evolutionary algorithm [8]. A partial 3D reconstruction method based on Evolutionary algorithm is proposed by Monica Perez-Meza et al. for various vision stereo problems. This method can be used in a mobile robot with monocular vision for navigation tasks [9]. Xiufen Zou et al. proposed a new Evolutionary algorithm based on L-optimality for solving multiobjective optimization problems [10]. Exploring new horizons in evolutionary design of robots is

ISSN:2319-1112 /V1N1:15-19 IJAEEE

IJAEEE,Volume1,Number 1 Monu Tyagi et al.

proposed by Stephane Donicieus et al. using Evolutionary algorithm and work in robotics [12]. Kezong Tong. et al. developed a basic evolutionary algorithm. In this author explained all function of evolutionary algorithms [13]. Evolutionary computation: An Overview and Recent Trends proposed by Gunther Raidl. In this Gunther explained recent trends in evolutionary algorithms [14]. Section II discusses the various components of evolutionary algorithms while in Section III the various classifications of EA have illustrated as per the literature available. II. COMPONENTS OF EVOLUTIONARY ALGORITHMS In this section we discuss Evolutionary algorithms. EAs have a number of components, procedure or operators that must be specified in order to define a particular EA [2, 3].

1. Representation (definition of individuals) The first step in defining an EA is to link the real world to the EA world, that is to set up a bridge between the original problem context and the problem solving space where evolution will take place. Object forming possible solutions within the original problem context are referred to as phenotypes, their encoding, the individuals within the EA, are called genotypes. The first design step is commonly called representation. 2. Evaluation function (or fitness function) The role of the evaluation function is to represent the requirements to adapt to. It forms the basis for selection, and thereby it facilitates improvements. Technically, it is a function or procedure that assigns a quality measure to genotypes. Typically, this function is composed from a quality measure in the phenotype space and the inverse representation. 3. Population The role of the population is to hold (the representation of) possible solution. A population is a multi-set of genotypes. The population thus forms the unit of evolution. 4. Parent selection mechanism The role of parent selection or mating selection is to distinguish among individuals based on their quality, in particular, to allow the better individuals to become parents of the next generation. An individual is a parent if it has been selected to undergo variation in order to create offspring. In EA, parent selection is typically probabilistic. Thus, high quality individuals get a higher chance to become parents than those with low quality. 5. Variation operators, recombination and mutation The role of variation operators is to create new individuals from old ones. A binary variation operator is called recombination or crossover. As the names indicate, an operator merges information from two parent genotypes into one or two offspring genotypes. A unary variation operator is commonly called mutation. It is applied to one genotype and delivers a (slightly) modified mutant, the child or offspring of it. 6. Survivor selection mechanism (replacement) The role of survivor selection or environmental selection is to distinguish among individuals based on their quality. In that it is similar to parent selection, but it is used in a different stage of the evolutionary cycle. The survivor selection mechanism is called after having created the offspring of the selected parents.

ISSN:2319-1112 /V1N1:15-19 IJAEEE

17 A Novel Classification of Evolutionary Algorithms: A Critical Assessment

Fig.1- Evolutionary Algorithm Flowchart

III. CLASSIFICATION OF EVOLUTIONARY ALGORITHMS A. Multi-objective Optimization Multiobjectives problems due to their complexity these are suitable for evolutionary approaches. In real world optimization applications, it is often hard to formulate the optimization goal as a scalar function. Typically, there are several criteria or objectives, and not unusually, these objectives stay in conflict with each other. In this classification researcher focus on the study of evolutionary algorithms for solving multiobjective optimization problems with a large number of objectives. In this optimization developed a new dynamic multipleobjective evolutionary algorithm (DMOEA) and some modern algorithms, such as the indicator-based evolutionary algorithm, multiple single objective pareto sampling, and nondominated sorting genetic algorithm, is presented by employing the convergence metric and relative hyper volume metric. Multiobjective optimization another important term used is ENDOCRINE PARADIGM. This approach known as multiobjective optimization based on endocrine paradigm. In this approach researcher propose new evolutionary technique, which is related by the behavior of the endocrine systems and uses pareto non-dominance concept [6]. In evolutionary algorithm populations number are called chromosome and even if they evolve according to the genetic principles (selection, crossover and mutation) and in this classification a supplementary mechanism based on the endocrine paradigm. Endocrine paradigm based on MENDA technique [10].

B. Control P, I and D parameter by Robust tuning for disturbance rejection PID controller is a very important term for industry because PID controller used in almost all field of industry. In this classification researcher focused on control of P, I and D parameter using gain margin and phase margin for tuning an optimal controller that can be operated on robust control. Here P, I and D parameter compared in antibody are randomly allocated during selection process to obtained and optimal gain for robustness [1, 7]. C. For Real parameter optimization In this classification researcher propose generic parent centric recombination operators (PCX) and a steady state, elitepreserving, scalable and computationally fast population alternation model (called G3 ). Performance of

G3 model with the PCX

operator is investigated on three commonly used test problems and is compared with a number of evolutionary and classical optimization algorithms. This classification uses the proposed G3 model with PCX operator for more complex problems and to real-world optimization

ISSN:2319-1112 /V1N1:15-19 IJAEEE

IJAEEE,Volume1,Number 1 Monu Tyagi et al.

problems [5]. D. Design of Robot using EP This classification is also known as evolutionary robotics (ER). There are three main category of evolutionary robotics (ER) presented:1- ER as an automatic parameter tuning system which is the very mature and punctual application and is used to solve real robotics problems. 2- EAD (Evolutionary added design) in this category the main benefit the designer as an efficient tool, equipment to build robotics systems. 3- EAS (evolutionary automatic synthesis) in this category EAS corresponds to the automatic design of mechatronic device in robotic field [11, 12]. E. 3D Reconstruction In this classification researcher focused on construction field using evolutionary algorithm. When reconstruction a scenario, it is necessary to know the structure of the elements present on the scene to have an interpretation. In this work we link 3D scenes reconstruction to evolutionary algorithms through the vision stereo theory. Vision stereo theory gives us an idea about three dimensional characteristics of the world. This theory basically requires two cameras. In this work we employ only a camera, which is translated along a path, capturing images every certain distance, here evolutionary algorithm to partially reconstruct the scene in real time [9]. F. Hybrid evolutionary algorithms In this classification for better result researcher added some specific algorithms with evolutionary. Hybrids of EAs and other metaheuristics such as local search, simulated annealing, tabu search, guided local search, variable neighborhood search, and variable depth search. By this hybrid evolutionary algorithm primary responsibility is to take care of diversification and coverage ao all major area of search space, while the embedded other metaheuristic is mainly responsible for intensification of the search , thus the finetuning of the most promising candidate solutions [14].

Estimation of distribution algorithm This classification samples new solutions from a probability model which characterizes the distribution of promising solutions in the search space at each generation. The simplest form of an EDA is the univarite marginal distribution algorithm (UMDA). A particularly well-known and often applied variant of EDAs are ant colony optimization (ACO) algorithms [14].

G.

Design of automated Antenna This classification is very important for design best antenna using evolutionary algorithms. By this using we save time and labor. Whereas the current practice of designing antennas by hand is severely limited because it is both more require time and labor. Presently this classification used in NASAs space technology. By exploring such a wide range of designs EAs may be able to produce designs of previously unachievable performance. The faster design cycles of an evolutionary approach results in less development costs and allows for an iterative what-if design and test approach for different scenarios [8].

H.

IV. CONCLUSION From the literature it can be concluded multi-objective function is more convenient for solving any optimization problem. In this classification only change in object of any optimization problem and rest of step is same as evolutionary algorithm so this classification is very simple other than the classification. An important advantage of the evolutionary approach to the underlying problem is that in all cases, the MOEA conducted only 1% of the number of evolutions performed by the exhaustive search

ISSN:2319-1112 /V1N1:15-19 IJAEEE

19 A Novel Classification of Evolutionary Algorithms: A Critical Assessment algorithm. A possible direction of future research would be to apply evolution to problems with high dimensionality where the true pareto front cannot be obtained exactly and compare the results with those from other meta-heuristic algorithms. REFERENCES
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] Teng Fong-Chwee., Self tuning PID controllers for dead time process., IEEE Trans., VOL. 35, NO. 1, pp. 119-125, 1988. T. Back and H.P. Schewefel., An overview of Evolutionary Algorithms for parameter optimization., IEEE International conference on E-Commerce (CEC03), VOL. 1, pp. 1-23, 1993. D.B. fogel., Evolutionary Computation, IEEE Conference of Evolutionary Computation., VOL. 1, pp. 193, 1995. J.B Yang, Gradient Protection and Local region search for multiobjective optimization., European journal of operational research., VOL. 112, NO. 2, Pp. 432-459., 1999. Kaly Moy Deb., et al., A Computationally efficient Evolutionary Algorithm for Real parameter optimization., april 11, 2002. Corina rato., A New Evolutionary Algorithm for multiobjective optimization based on endocrine paradigm.,ICTAMI , pp. 271-284, 2003. Dong Hwa Kim, Jae Hoon cho., robust tuning for Disturbance Rejection of PID controller using Evolutionary Algorithm., IEEE Annual meeting of the fuzzy information, pp. 248-253, 2004. Gregory S. Horn by., Automated Antenna Design with Evolutionary Algorithms., American institute of aeronautics and astronutics, Moffett Field, CA 94035, 2006, pp. 1-8. Monica Perez-meza, and Redrigo Montufar-Chaveznava., Partial 3D Reconstruction using Evolutionary algorithms., International Journal of Engineering and applied science , pp. 107-112, 2007. Xiufen Zou, et al., A New evolutionary algorithm for solving many optimization problems., IEEE TRANS. ON SYSTEM< MAN< AND CYBERNETICS-PART B: CYBERNETICS, VOL. 38, NO. 5, pp. 1402-1412, 2008. D. Floreno and C. Claudio Mattiussi., Bio inspired Artificial intelligence: Theories and mehods, and technologies . Ser intelligent Robotics and automous agents., MIT press of artificial intelligence, pp. 335-398, 2008. Stephane Doncieux, et al., Exploring New Horizons in Evolutionary Design of Robots., IROS, pp. 20-29, 2009. Kezong Tong., et al Linear Evolutionary Algorithms cheapter-2 in Evolutionary Algorithms written by Eisuke kita.,pp. 28- 38., 2011. Gunther Raidl., Evolutionary Computation: An overview and Recent Trends., FavoritenstraBe 9-11/1861, 1040 vienna, Austria, pp- 2-8, 2011

ISSN:2319-1112 /V1N1:15-19 IJAEEE