ArticlePublisher preview available

Thermal analysis of brushless DC motor using multiobjective optimization

Wiley
International Transactions on Electrical Energy Systems
Authors:
Kamal Chakkarapani
Kamal Chakkarapani
Kamal Chakkarapani
  • This person is not on ResearchGate, or hasn't claimed this research yet.
Thangavelu Thyagarajan at Anna University, Chennai
Thangavelu Thyagarajan
Kalpana Dharmalingam at Central Electrochemical Research Institute
Kalpana Dharmalingam
Read publisher preview
Download citation
To read the full-text of this research, you can request a copy directly from the authors.

Abstract and Figures

In this article, the design of brushless DC (BLDC) motor is performed using multiobjective optimization algorithm (MOOA) by satisfying multiple objectives. Initially, sensitivity analysis is carried out to find the most influencing parameters that affect the performance of BLDC motor. MOOAs such as Pareto envelope‐based selection algorithm (PESA), Pareto archived evolution strategy (PAES) and nondominated sorting genetic algorithm‐II (NSGA‐II) is employed in the optimal design of the BLDC motor. The proposed MOOA have three objectives namely: output torque maximization, volume minimization, and minimization of total losses. MOOAs are analyzed using performance metrics and qualitative comparison is provided to select the best algorithm. Later, finite element method (FEM) is used to investigate the transient and thermal characterization on the BLDC motor designed using NSGA‐II. The thermal results thus obtained using NSGA‐II for the above motor under different operating conditions is also compared with the existing single objective optimization algorithm. From the comparisons, it is observed that NSGA‐II algorithm significantly outperforms the existing single objective optimization algorithm. Finally, the usefulness of the designed machine based on NSGA‐II is compared with the results obtained from simulation and hardware analysis. In this article, design of brushless DC (BLDC) motor has been optimized using multi objective optimization (MOO) algorithms. The thermal analysis and transient analysis has been carried out on the BLDC designed used nondominated sorted genetic algorithm‐II (NSGA‐II).The usefulness of the designed machine based on NSGA‐II is compared with the results obtained from simulation and hardware analysis.
Figures - available from: International Transactions on Electrical Energy Systems
This content is subject to copyright. Terms and conditions apply.
A preview of this full-text is provided by Wiley.
Learn more
Content available from International Transactions on Electrical Energy Systems 
This content is subject to copyright. Terms and conditions apply.
RESEARCH ARTICLE
Thermal analysis of brushless DC motor using
multiobjective optimization
Kamal Chakkarapani
1
| Thyagarajan Thangavelu
2
| Kalpana Dharmalingam
2
1
Department of Electrical and Electronics
Engineering, Sri Venkateswara College of
Engineering, Chennai, India
2
Department of Instrumentation
Engineering, MIT Campus, Anna
University, Chennai, India
Correspondence
Kamal Chakkarapani, Department of
Electrical and Electronics Engineering, Sri
Venkateswara College of Engineering,
Chennai, India.
Email: kamalc@svce.ac.in
Peer Review
The peer review history for this article is
available at https://publons.com/publon/
10.1002/2050-7038.12546.
Summary
In this article, the design of brushless DC (BLDC) motor is performed using
multiobjective optimization algorithm (MOOA) by satisfying multiple
objectives. Initially, sensitivity analysis is carried out to find the most
influencing parameters that affect the performance of BLDC motor.
MOOAs such as Pareto envelope-based selection algorithm (PESA), Pareto
archived evolution strategy (PAES) and nondominated sorting genetic algo-
rithm-II (NSGA-II) is employed in the optimal design of the BLDC motor.
The proposed MOOA have three objectives namely: output torque maximi-
zation, volume minimization, and minimization of total losses. MOOAs are
analyzed using performance metrics and qualitative comparison is provided
to select the best algorithm. Later, finite element method (FEM) is used to
investigate the transient and thermalcharacterizationontheBLDCmotor
designed using NSGA-II. The thermal results thus obtained using NSGA-II
for the above motor under different operating conditions is also compared
with the existing single objective optimization algorithm. From the compar-
isons, it is observed that NSGA-II algorithm significantly outperforms the
existing single objective optimization algorithm. Finally, the usefulness of
the designed machine based on NSGA-II is compared with the results
obtained from simulation and hardware analysis.
KEYWORDS
BLDC motor, NSGA-II, optimization, performance metrics, thermal analysis
List of Symbols and Abbreviations: BLDC, Brushless DC; DSD, Definite Screening Design; FEA, Finite Element Analysis; FEM, Finite Element
Method; GD, Generational Distance; HV, Hyper Volume; IGD, Inverted Generational Distance; MOOA, Multi Objective Optimization Algorithm;
MOO, Multi Objective Optimization; NSGA-II, Nondominated Sorting Genetic Algorithm-II; PAES, Pareto Archived Evolution Algorithm; PESA,
Pareto Envelope-based Selection Algorithm; SOOA, Single Objective Optimization Algorithm; SOO, Single Objective Optimization; A
g
, Air gap area at
the winding surface; B
g
, Magnetic flux density; B
sy
, Stator core maximum flux density; B
r
, Permanent Magnet Remanance; F
m
, Magneto-motive force
(MMF) of each magnet; J
cu
, Current density; k
l
, Magnetic field leakage; l
m
, Magnet thickness; l
s
, Stator/rotor axial length; l
w
, Winding thickness; l
g
,
Mechanical air gap; l
y
, Stator/ rotor core thickness; P
total
, Total power loss; P
w
, Windage losses; P
b
, Friction of bearing losses; P
cu
, Copper losses; P
h
,
Hysteresis losses; P
e
, Eddy current losses; P, Number of pole pairs; r
r
, Rotor radius; T
em
, Electromagnetic torque; T
ripple
, Ripple Torque; T
min
,
Minimum Torque; T
max
, Maximum Torque; T
avg
, Average Torque; V
t
, Total volume; ℜ, sum of the reluctances of the winding, air gap and magnet
regions for each pole; λ, Stator/rotor axial length; β, Pole-arc per pole pitch ratio; μ
0
, Free space permeability; μ
r1
, Magnet permeability; μ
r2
, Winding
permeability.
Received: 23 March 2020 Revised: 21 May 2020 Accepted: 18 June 2020
DOI: 10.1002/2050-7038.12546
Int Trans Electr Energ Syst. 2020;30:e12546. wileyonlinelibrary.com/journal/etep © 2020 John Wiley & Sons Ltd 1of18
https://doi.org/10.1002/2050-7038.12546
... In general, the performance of the machine is determined by its dimensions. Neural networks give rigorous tools for analysis, prediction, and optimization, and have the potential to revolutionize the way of machine design, resulting in more innovative and effective design solutions [4]. This is accomplished by properly training the neural network. ...
Design of BLDC Motor Using African Vulture Optimization Algorithm Based Trained FNN
Article
Full-text available
  • Apr 2024
Brush-less DC motors with permanent magnets are widely used in robotics, electric vehicles, and other industrial applications. Enhancing the performance of a BLDC motor can indeed be challenging due to the presence of non-linearities and complex design considerations. The present research aims to increase the effectiveness and performance of BLDC motors by using a feed-forward neural network (FNN). It is possible to quickly determine the FNN’s appropriate weight and bias settings by using the cutting-edge African Vulture Optimisation Algorithm (AVOA). Combining biological inspiration, machine learning, and motor technology, it develops a distinctive and promising way to improve the performance of BLDC motors. This study also includes the sensitive analysis of controlling parameters of the AVOA to investigate its effect on the statistical variables of the fitness function. This aids in optimizing the algorithm for better outcomes. The proposed method is then contrasted with various FNN learning algorithms based on Practical Swarm Optimization (PSO), Gravitational Search Algorithm (GSA), and PSO-GSA. Considerations like convergence rate and the capacity to steer clear of local minima are probably evaluated throughout the comparison. According to the results, AVOA-based optimization performs better than the other methods (PSO, GSA, and PSO-GSA) in terms of convergence speed and its ability to avoid becoming stuck in local minima. This suggests that the AVOA is useful for training FNNs to optimize the design parameters of BLDC motors.
View
... Hyperboxes with a lower density of the solution are more likely to be selected for the recombination process. Power systems [169,170], driving assistance system [171], and radial distribution systems [172] are some real-world applications of multi-objective optimization problems that are solved with the use of PESA-II. ...
A review of classical methods and Nature-Inspired Algorithms (NIAs) for optimization problems
Article
Full-text available
  • Oct 2023
Optimization techniques are among the most promising methods to deal with real-world problems, consisting of several objective functions and constraints. Over the decades, many methods have come into existence to solve optimization problems. However, the complexity of these problems is increasing over time. Thereby, it opens up a field of research in developing a robust procedure compatible with such complex optimization problems that provide optimal solutions best suited to the needs of the decision-makers. This review paper presents a survey of the recent use of classical methods and Nature-Inspired Algorithms (NIAs) to solve single and multiple objective problems of optimization in diverse application areas. Moreover, this study briefly describes these widely used solution methods based on the classification of classical approaches and NIAs. Recently published articles based on real-world applications have been included to demonstrate the advantages of each solution technique. In addition, research gaps involving various techniques and future prospects within this field are discussed.
View
... From the comparisons, it is observed that NSGA-II algorithm significantly outperforms the existing single objective optimization algorithm. Finally, the usefulness of the designed machine based on NSGA-II is compared with the results obtained from simulation and hardware analysis [6]. ...
Thermal Analysis of an Electric Motor in an Electric Vehicle
Conference Paper
Full-text available
  • Apr 2023
div class="section abstract"> Parametric thermal analysis of permanent magnet synchronous motor by considering real time driving cycle is presented. As driving motors of electric vehicle, permanent magnet motors exhibit high efficiency and high power density. However, they are susceptible to suffer irreversible demagnetization and insulation failure of coils under severe thermal condition. Therefore, it is essential to precisely predict heat losses and temperature distribution in driving motors under real time driving cycle. The thermal behavior of the driving motor is analyzed by means of thermal parametric methods. Here we are using simulation software to measure the temperature distribution and heat losses in the Driving motor, when an electric motor is operated for a certain amount of time the motor produces heat because of the Heat produced by the motor the performance, efficiency and reliability of the motor is affected. Performing the Analysis method on the Motor that is designed according to the parameters such as power output, Torque, Efficiency, Reliability, and Sustainability. Design Required Cooling System for the Electric Motor for EV application. By Designing an Appropriate Cooling System we can Easily Dissipate the Heat that is produced by the motor and we can also see improvement in performance, Sustainability, Torque Efficiency and other parameters that are affected by the Heat produced by the Electric motor. </div
View
Optimization Techniques for Cogging Torque Reduction and Thermal Characterization in Brushless DC Motor
Article
  • Mar 2024
This paper presents soft computing-based optimization techniques for the cogging torque reduction and thermal characterization by finite element analysis in a permanent magnet brushless DC motor (BLDC). Stator and rotor structure of BLDC motor are optimized to reduce the cogging torque, noise, and vibration by using the design parameters namely: length of magnet, length of air gap and opening in the stator slot which are selected by performing variance-based sensitivity analysis. The proposed method is suitable in the preliminary design phase of the motor to determine the optimal structure to improve the efficiency. The comparison of results obtained using firefly algorithm , ant colony optimization algorithm and Bat algorithm indicate that Firefly-based optimization algorithm is capable of giving improved design parameter output. Cogging torque is created due to the interaction of magnets in the rotor and the stator slot of the motor. Thorough thermal analysis is also conceded out to investigate the thermal characteristics at dissimilar portions of the motor namely: stator core, stator slot, rotor core and permanent magnet at different operating environments in the continuous operation mode. Thermal investigation is required for the various high speed e-vehicle applications. The usefulness of the designed machine simulation is compared with the results obtained from hardware analysis. The outcomes attained from software simulation studies are validated through experimental hardware setup.
View
View
Optimization design of brushless DC motor based on improved JAYA algorithm
Article
Full-text available
  • Mar 2024
Brushless direct current motor is widely used in industrial production because of its simple structure, wide speed range and low noise. To improve the operation efficiency of brushless DC motor and reduce the production and application costs, the optimization of brushless DC motor is analyzed by introducing the JAYA algorithm. This method determines the optimal parameters of a brushless DC motor using the theory of electromagnetic structure parameter selection and efficiency calculation. The population diversity of the JAYA algorithm is improved through an empirical learning strategy, and an adaptive strategy is introduced to balance the development ability and search performance of the algorithm. This ensures population diversity and improves convergence speed. The experiment showcases that the improved JAYA algorithm has a lower rank average in unimodal function operations, demonstrating stronger local development ability and better stability. It exhibits strong search ability in many local optima of multimodal functions. Moreover, the motor's average efficiency after optimization is 94.48%. The algorithm reaches the global optimum after approximately 40 iterations and offers faster convergence speed and higher accuracy. The adaptive JAYA algorithm is stable at around 93% when the number of iterations reaches 90, with a maximum efficiency of 95.3%. It is 5–12 percentage points higher than the other three comparison algorithms. The optimal solution of the motor parameters in the adaptive JAYA algorithm is closest to the theoretical parameter optimization value, meeting both the constraints of variables and the constraints of the model. The stator diameter, tooth magnetic induction, winding current density, air gap magnetic induction, and stator yoke magnetic induction values are 201.5 mm, 1.8 T, 2.049 A/mm², 0.63 T, and 0.91 T, respectively. The research overcomes the problem of parameter optimization in the optimization design of brushless DC motor, improving their economic value of brushless DC motor in industrial production and application.
View
View
Design and Simulation Analysis of Stator Slots for Small Power Permanent Magnet Brushless DC Motors
Article
Full-text available
  • Feb 2023
With the development of power electronics technology, permanent magnet brushless DC motors have developed rapidly and are now widely used in electric vehicles, flywheel energy storage, rail transit, and other applications. The stator slot structure is one of the main factors affecting the performance of the motor. A low-power permanent magnet brushless DC motor was selected as the research object, and the finite element analysis method was used to study the effects of different slot and pole combinations and stator slot types on the cogging torque, reluctance torque, and back electromotive force of the permanent magnet brushless DC motor. The influence of the stator slot structure of the motor on the performance of the motor was analyzed, and the optimal slot-pole combination and stator slot type were determined. The results showed that the cogging torque of the 2-stage 24-slot motor was 14 mN·m, and the reluctance torque was 75 mN·m. The cogging torque and reluctance torque were the smallest, and the back electromotive force waveform was similar to a trapezoidal wave. The motor cogging torque of the pear-shaped round slot was the smallest, with a value of 460 mN·m, and the motor reluctance torque of the pear-shaped trapezoidal slot was the smallest, with a value of 1.2 N·m. The back electromotive force waveforms of the motors with four different stator slot types were similar.
View
MOCHIO: a novel Multi-Objective Coronavirus Herd Immunity Optimization algorithm for solving brushless direct current wheel motor design optimization problem
Article
Full-text available
  • Jan 2022
A prominent and realistic problem in magnetics is the optimal design of a brushless direct current (BLDC) motor. A key challenge is designing a BLDC motor to function efficiently with a minimum cost of materials to achieve maximum efficiency. Recently, a new metaheuristic optimization algorithm called the Coronavirus Herd Immunity Optimizer (CHIO) is reported for solving global optimization problems. The inspiration for this technique derives from the idea of herd immunity as a way of combating the coronavirus pandemic. A variant of CHIO called Multi-Objective Coronavirus Herd Immunity Optimizer (MOCHIO) is proposed in this paper, and it is applied to optimize the BLDC motor design optimization problem. A static penalty constraint handling is introduced to handle the constraints, and a fuzzy-based membership function has been introduced to find the best compromise results. The BLDC motor design problem has two main objectives: minimizing the motor mass and maximizing the efficiency with five constraints and five decision/design variables. First, MOCHIO is tested with benchmark functions and then applied to the BLDC motor design problem. The experimental results are compared with other competitors are presented to confirm the viability and dominance of the MOCHIO. Further, six performance metrics are calculated for all algorithms to assess the performances.
View
View
Multiobjective Design Optimization and Analysis of Magnetic Flux Distribution for Slotless Permanent Magnet Brushless DC Motor using Evolutionary Algorithms
Article
Full-text available
  • Apr 2019
  • J MAGN MAGN MATER
In this paper, Multi-Objective Optimization (MOO) techniques namely: Weighted Sum Method (WSM), Multi Objective Genetic Algorithm (MOGA) and Niched Pareto Genetic Algorithm (NPGA) are proposed for the design optimization and analysis of magnetic flux distribution for a slotless permanent magnet Brushless DC (BLDC) motor. The sensitivity analysis is carried out to identify the design variables of BLDC motor which influence the objective function. The objective functions are conflicting in nature (maximization of output torque and minimization of total volume and losses), hence, are optimized simultaneously by means of MOO algorithms. The proposed MOOs account for four kinds of design variables namely: rotor radius, stator/ rotor axial length, magnet thickness and winding thickness simultaneously to maximize the output torque and to reduce the total volume and total power loss. The conventional MOO such as WSM gives single pareto optimal solution. However, the proposed MOGA and NPGA algorithms create accurate and well-distributed pareto front set with few function evaluations. The performances of the three MOOs are evaluated and compared using the four performance metrics namely: Hyper Volume (HV), Generational Distance (GD), Inverted Generational Distance (IGD) and Spread. From the comparison, it is observed that NPGA gives better results. Using the design parameter obtained from NPGA, the magnetic flux distribution analysis of the BLDC motor is carried out to analyze variation of flux distribution in the different parts of the motor. The results thus obtained are compared with those obtained through Partial differential equation and single objective GA optimization. The detailed thermal analysis is also carried out to analyze the thermal behavior at different parts of the machine for different working conditions in the continuous operation mode and hence, the results obtained are compared with single objective GA optimization. The advantages of MOOs in the design optimization of slotless permanent magnet BLDC motor are highlighted.
View
R-Metric: Evaluating the Performance of Preference-Based Evolutionary Multi-Objective Optimization Using Reference Points
Article
Full-text available
  • Dec 2018
  • IEEE T EVOLUT COMPUT
Measuring the performance of an algorithm for solving multi-objective optimization problem has always been challenging simply due to two conflicting goals, i.e., convergence and diversity of obtained trade-off solutions. There are a number of metrics for evaluating the performance of a multi-objective optimizer that approximates the whole Pareto-optimal front. However, for evaluating the quality of a preferred subset of the whole front, the existing metrics are inadequate. In this paper, we suggest a systematic way to adapt the existing metrics to quantitatively evaluate the performance of a preference-based evolutionary multi-objective optimization algorithm using reference points. The basic idea is to pre-process the preferred solution set according to a multi-criterion decision making approach before using a regular metric for performance assessment. Extensive experiments on several artificial scenarios and benchmark problems fully demonstrate its effectiveness in evaluating the quality of different preferred solution sets with regard to various reference points supplied by a decision maker.
View
A Many-Objective Population Extremal Optimization Algorithm with an Adaptive Hybrid Mutation Operation
Article
  • May 2019
  • INFORM SCIENCES
Many-objective optimization problems abbreviated as MaOPs with more than three objectives have attracted increasing interests due to their widely existing in a variety of real-world applications. This paper presents a novel many-objective population extremal optimization called MaOPEO-HM algorithm for MaOPs by introducing a reference set based many-objective optimization mechanism into a recently developed population extremal optimization framework and designing an adaptive hybrid mutation operation for updating the population. Despite of the successful applications of extremal optimization in different kinds of numerical and engineering optimization problems, it has never been explored to the many-objective optimization domain so far. Because most of the existing many-objective evolutionary algorithms are usually guided by a single mutation operation, which has insufficient ability to exploit the search space of MaOPs and will get stuck at any local efficient front, it is the first attempt to design a novel hybrid mutation scheme in MaOPEO-HM algorithm by combining the advantages of polynomial mutation operator and multi-non-uniform mutation operator effectively. The experiment results for DTLZ test problems with 3, 5, 8, 10, and 15 objectives and WFG test problems with 3, 5, and 8 objectives have demonstrated the superiority of the proposed MaOPEO-HM to five state-of-the-art decomposition-based many-objective evolutionary algorithms including NSGA-III, RVEA, EFR-RR, θ-DEA, and MOEA/DD and two non-decomposition-based algorithms including GrEA and Two_Arch2. Furthermore, the great ability of the designed adaptive hybrid mutation operation incorporated into many-objective population extremal optimization (MaOPEO)has also been illustrated by comparing MaOPEO-HM and two MaOPEO algorithms only based on traditional multi-non-uniform mutation or polynomial mutation for DTLZ problems.
View
Constrained multi-objective population extremal optimization based economic-emission dispatch incorporating renewable energy resources
Article
  • May 2019
  • RENEW ENERG
Economic emission dispatch (EED)problem of an electrical power system can be considered as one of the most popular constrained multi-objective optimization problems to minimize the cost and emission simultaneously subjecting to various constraints. Although many approaches have been presented to deal with this problem, it is still a challenge issue especially when more and more renewable energy sources such as wind power and solar power are incorporated into the system due to their intermittence and uncertainty. To improve the EED performance with those renewable power generations, a constrained multi-objective population extremal optimization algorithm called CMOPEO-EED is proposed by utilizing an advanced constraint handling technique, i.e., the superiority of feasible solution approach. To demonstrate the effectiveness of the proposed method, three versions of a modified IEEE 30-bus and 6-generator system with renewable power generations are considered as the test systems. The comprehensive experimental results and analyses fully validate that the proposed CMOPEO-EED method in this paper outperforms these recently reported single-objective success history based adaptive differential evolutionary algorithm (SHADE)-based EED method and constrained non-dominated sorting genetic algorithm-based EED (CNSGAII-EED)method in terms of cost and emission indices.
View
Integrating Weight Assignment Strategies with NSGA-II for Supporting User Preference Multi-Objective Optimization
Article
  • Nov 2017
  • IEEE T EVOLUT COMPUT
Driven by the needs of several industrial projects on the applications of multi-objective search algorithms, we observed that user preferences must be properly incorporated into optimization objectives. However, existing algorithms usually treat all the objectives with equal priorities and do not provide a mechanism to reflect user preferences. To address this, we propose an extension—User-Preference Multi-Objective Optimization Algorithm (UPMOA), to the most commonly applied, non-dominated sorting genetic algorithm II (NSGA-II) by introducing a user preference indicator p, based on existing weight assignment strategies (e.g., Uniformly Distributed Weights (UDW)). We empirically evaluated UPMOA using four industrial problems from three diverse domains (i.e., Communication, Maritime and Subsea Oil&Gas). We also performed a sensitivity analysis for UPMOA with 625 algorithm parameter settings. To further assess the performance and scalability, 103500 artificial problems were created and evaluated representing 207 sets of user preferences. Results show that the UDW strategy with UPMOA achieves the best performance and UPMOA significantly outperformed other three multi-objective search algorithms, and has the ability to solve problems with a wide range of complexity. We also observed that different parameter settings led to the varied performance of UPMOA, thus suggesting that configuring proper parameters is highly problem-specific.
View
View
Multiobjective Optimization of Switched Reluctance Motors Based on Design of Experiments and Particle Swarm Optimization
Article
  • Sep 2015
This paper proposes a comprehensive framework for multiobjective design optimization of switched reluctance motors (SRMs) based on a combination of the design of experiments and particle swarm optimization (PSO) approaches. First, the definitive screening design was employed to perform sensitivity analyses to identify significant design variables without bias of interaction effects between design variables. Next, optimal third-order response surface (RS) models were constructed based on the Audze–Eglais Latin hypercube design using the selected significant design variables. The constructed optimal RS models consist of only significant regression terms, which were selected by using PSO. Then, a PSO-based multiobjective optimization coupled with the constructed RS models, instead of the finite-element analysis, was performed to generate the Pareto front with a significantly reduced computational cost. A sample SRM design with multiple optimization objectives, i.e., maximizing torque per active mass, maximizing efficiency, and minimizing torque ripple, was conducted to verify the effectiveness of the proposed optimal design framework.
View
Optimization of brushless direct current motor design using an intelligent technique
Article
  • Apr 2015
This paper presents a method for the optimal design of a slotless permanent magnet brushless DC (BLDC) motor with surface mounted magnets using an improved bee algorithm (IBA). The characteristics of the motor are expressed as functions of motor geometries. The objective function is a combination of losses, volume and cost to be minimized simultaneously. This method is based on the capability of swarm-based algorithms in finding the optimal solution. One sample case is used to illustrate the performance of the design approach and optimization technique. The IBA has a better performance and speed of convergence compared with bee algorithm (BA). Simulation results show that the proposed method has a very high/efficient performance. Copyright © 2015 ISA. Published by Elsevier Ltd. All rights reserved.
View
An Evolutionary Many-Objective Optimization Algorithm Using Reference-Point Based Nondominated Sorting Approach, Part II: Handling Constraints and Extending to an Adaptive Approach
Article
  • Aug 2014
  • IEEE T EVOLUT COMPUT
In the precursor paper, a many-objective optimization method (NSGA-III), based on the NSGA-II framework, was suggested and applied to a number of unconstrained test and practical problems with box constraints alone. In this paper, we extend NSGA-III to solve generic constrained many-objective optimization problems. In the process, we also suggest three types of constrained test problems that are scalable to any number of objectives and provide different types of challenges to a many-objective optimizer. A previously suggested MOEA/D algorithm is also extended to solve constrained problems. Results using constrained NSGA-III and constrained MOEA/D show an edge of the former, particularly in solving problems with a large number of objectives. Furthermore, the NSGA-III algorithm is made adaptive in updating and including new reference points on the fly. The resulting adaptive NSGA-III is shown to provide a denser representation of the Pareto-optimal front, compared to the original NSGA-III with an identical computational effort. This, and the original NSGA-III paper, together suggest and amply test a viable evolutionary many-objective optimization algorithm for handling constrained and unconstrained problems. These studies should encourage researchers to use and pay further attention in evolutionary many-objective optimization.
View
An Evolutionary Many-Objective Optimization Algorithm Using Reference-Point-Based Nondominated Sorting Approach, Part I: Solving Problems With Box Constraints
Article
  • Aug 2014
  • IEEE T EVOLUT COMPUT
Having developed multiobjective optimization algorithms using evolutionary optimization methods and demonstrated their niche on various practical problems involving mostly two and three objectives, there is now a growing need for developing evolutionary multiobjective optimization (EMO) algorithms for handling many-objective (having four or more objectives) optimization problems. In this paper, we recognize a few recent efforts and discuss a number of viable directions for developing a potential EMO algorithm for solving many-objective optimization problems. Thereafter, we suggest a reference-point-based many-objective evolutionary algorithm following NSGA-II framework (we call it NSGA-III) that emphasizes population members that are nondominated, yet close to a set of supplied reference points. The proposed NSGA-III is applied to a number of many-objective test problems with three to 15 objectives and compared with two versions of a recently suggested EMO algorithm (MOEA/D). While each of the two MOEA/D methods works well on different classes of problems, the proposed NSGA-III is found to produce satisfactory results on all problems considered in this paper. This paper presents results on unconstrained problems, and the sequel paper considers constrained and other specialties in handling many-objective optimization problems.
View