Optimization of the gear ratios in automatic transmission systems using an artificial neural network and a genetic algorithm

One of the most important tasks in designing an automatic transmission system is to find the gear ratios and the corresponding number of gear teeth. In this paper, an artificial neural network and a genetic algorithm are used for this optimization with regard to an epicyclic gear train. First, MATLA...

Full description

Saved in:
Bibliographic Details
Published in:Proceedings of the Institution of Mechanical Engineers. Part D, Journal of automobile engineering Journal of automobile engineering, 2014-09, Vol.228 (11), p.1338-1343
Main Authors: Shamekhi, Amir H., Bidgoly, Abbas, Noureiny, Ebrahim N.
Format: Article
Language:English
Subjects:
Algorithms
Artificial neural networks
Automatic transmissions
Automotive engineering
Gear ratios
Gear teeth
Genetic algorithms
Mathematical models
Matlab
Neural networks
Optimization
Software
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:One of the most important tasks in designing an automatic transmission system is to find the gear ratios and the corresponding number of gear teeth. In this paper, an artificial neural network and a genetic algorithm are used for this optimization with regard to an epicyclic gear train. First, MATLAB and an artificial neural network are employed to model the system, and the results depict the accuracy of the artificial neural network calculations. Then, using the same software and with the aid of a genetic algorithm, the optimized speed ratios and gear ratios are obtained. It can be seen that a series of gear ratios is produced. Another genetic algorithm was used to calculate the optimized gear ratio and the corresponding number of gear teeth. A Simpson gear train is used to demonstrate the methodology. The proposed model is very accurate and efficient, such that the resulting numbers of optimized gears have an error of less than ±0.3%. This method is much easier and has a lower computation cost than solving the related equations.
ISSN:0954-4070
2041-2991
DOI:10.1177/0954407014528887