Multi-objective optimization of a 2-stage spur gearbox using NSGA-II and decision-making methods
- PDF / 2,190,820 Bytes
- 22 Pages / 595.276 x 790.866 pts Page_size
- 107 Downloads / 179 Views
(2020) 42:477
TECHNICAL PAPER
Multi‑objective optimization of a 2‑stage spur gearbox using NSGA‑II and decision‑making methods Edmund S. Maputi1 · Rajesh Arora1 Received: 29 January 2019 / Accepted: 7 August 2020 © The Brazilian Society of Mechanical Sciences and Engineering 2020
Abstract Gears are crucial elements in mechanical systems and contribute to the overall performance of machinery. As such, optimization of gearbox transmission systems remains a challenging problem faced by researchers and designers for numerous years. In the present work, three objectives viz volume, power output and centre-distance are investigated simultaneously. A two-stage spur gearbox design problem is formulated with eight design variables viz. face-width (stage 1), face-width (stage 2), shaft diameter (stage 1), shaft diameter (stage 2), module, pinion teeth number(stage 1) and pinion teeth number (stage 2) considered. Two geometric and three design constraints are formulated. Pareto frontiers are generated using NSGA-II evolutionary algorithm. The Pareto frontier is investigated using decision tools viz. FUZZY, LINMAP and TOPSIS and the best solutions selected using the deviation index. Validation of results was done using previous reports in the literature and geometric modelling software. Variation and sensitivity studies indicate that module, pinion tooth number and face-width variables had a higher influence on volume as compared to power output and centre-distance. The combined optimization of volume, centre-distance and power output reveals key insights for the design of compact gearboxes. Keywords Multi-objective optimization · NSGA-II · Multistage gear · Optimal design · Pareto · Decision-making methods
1 Introduction Gear transmissions are common mechanical elements in automotive, aerospace, marine and manufacturing industries. The reason for this fact relates to the ability gears systems have to vary the rotational speed, torque and power output to suit a variety of applications. Gear design standards and procedures based on experimental data and analytical methods are widely applied in industry today [1]. Early from 1970s, numerous heuristics have been applied to study gear systems with the drawback of substantial resource requirements in terms of time and money [2]. Furthermore, complexity derived by numerous gear types with multiple configurations, applications and capacity negatively impacts product development time. The advancement of technology has led to increased accessibility to computing resources and hence Technical Editor: Wallace Moreira Bessa, D.Sc. * Edmund S. Maputi [email protected] 1
Department of Mechanical Engineering, Amity University Haryana, Gurgaon 122413, India
the introduction of meta-heuristic methods viz. Genetic algorithm (GA), simulated annealing (SA), ant-colony optimization (ACO) and teaching and learning-based optimization (TLBO) [3, 4] came into the picture. Compared to heuristic methods, which are problem-dependent and efficient in local search, meta-heuristic methods achie
Data Loading...
