Source-linked AI summary
Multiobjective Firefly Algorithm for Continuous Optimization
Xin-She Yang
TL;DR
Engineering design requires optimization across conflicting objectives and complex nonlinear constraints, making Pareto-front approximation difficult. The paper formulates MOFA by extending the firefly algorithm, validates it on selected test functions, and applies it to engineering design benchmarks. The reported comparisons suggest that MOFA is an efficient multiobjective optimizer.
Problem
Engineering design optimization requires finding diverse non-dominated solutions under multiple objectives and complex nonlinear constraints.
Method
The paper extends the firefly algorithm into MOFA, using firefly attraction, controlled randomness, and Pareto-oriented search for multiobjective optimization.
Results
The reported comparisons suggest that MOFA is an efficient multiobjective optimizer across selected test functions and engineering design benchmarks.
Takeaways & Limitations
MOFA provides a proposed metaheuristic approach for approximating Pareto fronts in continuous optimization and engineering design problems.
Takeaways & Limitations
The convergence behavior depends on γ, whose practical value is determined by the characteristic distance of the optimization system.
Abstract
from arXiv · showhide
Design problems in industrial engineering often involve a large number of design variables with multiple objectives, under complex nonlinear constraints. The algorithms for multiobjective problems can be significantly different from the methods for single objective optimization. To find the Pareto front and non-dominated set for a nonlinear multiobjective optimization problem may require significant computing effort, even for seemingly simple problems. Metaheuristic algorithms start to show their advantages in dealing with multiobjective optimization. In this paper, we extend the recently developed firefly algorithm to solve multiobjective optimization problems. We validate the proposed approach using a selected subset of test functions and then apply it to solve design optimization benchmarks. We will discuss our results and provide topics for further research.
1 Introduction
Engineering design optimization involves conflicting objectives, nonlinear constraints, uncertainty, and substantial difficulty approximating diverse Pareto fronts. The paper extends the firefly algorithm to multiobjective optimization and evaluates it on test functions and engineering benchmarks.
- 1 Introduction: Engineering design problems combine multiple conflicting objectives with nonlinear constraints shaped by codes, materials, resources, and costs.
- 1 Introduction: Multiobjective optimization seeks a set of non-dominated solutions approximating the Pareto front rather than one unique optimum.
- 1 Introduction: Algorithms must also distribute solutions uniformly and preserve enough diversity to sample the search space efficiently.
- 1 Introduction: Uncertainty in real-world properties creates robust Pareto solution sets from which designers can select suitable options.
- 1 Introduction: The study extends the firefly algorithm into MOFA, validates it on selected multiobjective test functions, and applies it to beam and disc-brake design.
2 Multiobjective Firefly Algorithm
MOFA extends firefly search by combining attraction, randomized movement, Pareto dominance, and diversity mechanisms to approximate multiobjective Pareto fronts directly. Its procedure initializes fireflies, evaluates objectives, preserves non-dominated solutions, and iterates toward a front approximation.
- 2.1 The Basic Firefly Algorithm: Attractiveness decreases with distance according to γ, whose practical range is typically 10^-5 to 10^5 and influences convergence behavior.The characteristic distance is Γ = 1/√γ.
- 2.1 The Basic Firefly Algorithm: The basic firefly algorithm moves each firefly toward a brighter one while adding randomized motion, with randomness typically reduced over iterations.The movement is a random walk biased toward brighter fireflies; Lévy-flight randomization is also possible.
- 2.2 Multiobjective Firefly Algorithm: For multiobjective problems, random normalized weights can form a combined objective, while changing weights each iteration promotes diverse sampling along the Pareto front.
- 2.2 Multiobjective Firefly Algorithm: MOFA instead extends firefly principles to produce the Pareto-optimal front directly, using a population of fireflies and retaining non-dominated solutions across iterations.
- 2.3 Pareto Optimal Front: Pareto dominance identifies solutions no worse in every objective and strictly better in at least one, while the Pareto front is the set of non-dominated solutions.
- 2.3 Pareto Optimal Front: Diverse solution-generation techniques, including Lévy flights, are used to improve approximation of the Pareto front.
3 Numerical Results
MOFA was evaluated on diverse multiobjective test functions and compared with established algorithms. It produced rapidly converging, smooth Pareto-front approximations and better results in almost all five comparison cases.
- Experimental setup: The implementation used detailed parameter studies, with most problems supporting α0 = 0.1 to 0.5, β0 = 0.7 to 1.0, and γ = 1.The studies varied α0, β0, and γ across specified ranges before selecting these practical settings.
- Test functions: MOFA was validated on five test functions spanning convex, non-convex, discontinuous, and more complex Pareto structures.The set included SCH, ZDT1, ZDT2, ZDT3, and LZ functions.
- Evaluation measures: The estimated Pareto fronts and true fronts were compared using analytical true-front forms, including ZDT1’s f2 = 1 −√f1.The absolute distance uses the estimated and true Pareto fronts, while generalized distance provides a relative measure.
- Convergence: MOFA’s generalized distance Dg decreased in an exponential-like manner over the iterations for the test functions.Figure 3 reports this convergence behavior for ZDT1 during the first 1000 iterations.
- Comparison study: Table 2 reports better MOFA results for almost all five comparison cases, with ZDT2 showing the same order but slightly better performance than DEMO.The comparison included established methods such as VEGA, NSGA-II, and MODE.
4 Design Optimization
MOFA is applied to welded-beam and disc-brake engineering benchmarks, generating smooth Pareto fronts and faster convergence while handling constrained multiobjective designs.
- Benchmark problems: The study selects welded-beam and disc-brake benchmarks from established engineering design problems for MOFA evaluation.The welded-beam problem minimizes fabrication cost and end deflection; the disc-brake problem minimizes mass and braking time under design constraints.
- Welded beam design: The welded-beam formulation minimizes fabrication cost and end deflection over four design variables subject to nonlinear constraints and bounds.The variables are welded-area width and length plus beam depth and thickness.
- Welded beam design: 50 non-dominated solutions after 1000 iterations produced a smooth welded-beam Pareto front consistent with prior results, while MOFA converged faster.The comparison is shown in Figures 6 and 7.
- Disc brake design: The disc-brake formulation minimizes overall mass and braking time by selecting disc radii, engaging force, and friction-surface count under torque, pressure, temperature, and length constraints.The design-variable bounds include 55 ≤ r ≤ 80, 75 ≤ R ≤ 110, 1000 ≤ F ≤ 3000, and 2 ≤ s ≤ 20.
- Disc brake design: 50 disc-brake solution points after 1000 iterations formed a smooth Pareto front that was the same or better than prior results, with convergence compared in Figure 9.The broader convergence comparison reports the highest convergence rates for MOFA in both benchmark cases, with exponentially decreasing curves.
- Overall benchmark findings: Across the benchmarks and test functions, simulations suggest that MOFA efficiently handles highly nonlinear problems with complex constraints and diverse Pareto-optimal sets.This conclusion is presented as a simulation-based suggestion rather than a universal guarantee.
5 Conclusions
The paper formulates MOFA for multiobjective optimization, tests it on selected functions, and applies it to industrial design benchmarks. Comparisons suggest that MOFA is efficient, while parameter studies, convergence analysis, hybridization, and discrete extensions remain future work.
- Conclusions: MOFA is formulated from the firefly algorithm and tested on selected test functions before application to industrial engineering design benchmarks.The benchmark applications include design optimization problems in industrial engineering.
- Conclusions: Comparisons with other algorithms suggest that MOFA is an efficient multiobjective optimizer.The conclusion is based on the reported comparative results.
- Future research: Future work includes parametric studies, convergence analysis, hybridization with other algorithms, and formulation of a discrete MOFA.These directions are presented as ways to study parameter ranges, understand mechanisms, improve the algorithm, and design new algorithms.