Comparison between TAGUCHI Method and Neural Network in Optimization
| Autor: | Ikuo Tanabe, Junnosuke Mizutani, Tomoaki Taira |
|---|---|
| Rok vydání: | 2005 |
| Předmět: |
Engineering
Mathematical optimization Artificial neural network Mathematical model business.industry Mechanical Engineering Value (computer science) Fault (power engineering) Control factor Industrial and Manufacturing Engineering Taguchi methods Mechanics of Materials Coefficient of friction business Control methods |
| Zdroj: | TRANSACTIONS OF THE JAPAN SOCIETY OF MECHANICAL ENGINEERS Series C. 71:2651-2656 |
| ISSN: | 1884-8354 0387-5024 |
| DOI: | 10.1299/kikaic.71.2651 |
| Popis: | Recently, in the various industry fields, TAGUCHI Method is used for quality control. On the other hand, Neural Network is used to find the optimum condition because it has learning ability for non-liner phenomenon. In this study, these quality control methods are compared. At first, the minimum value of the output in mathematical models is found by using TAGUCHI Method and Neural Network. And the advantage and fault of each method are explained from the result. In addition, both methods were improved in order to correct itself faults. Finally, the friction of sliding surface was used for practical evaluation regarding both methods. As that result, the condition for reducing coefficient of friction could be estimated with both methods. It is concluded from the result that (1) TAGUCHI Method was mighter than Neural Network when input data was discontinuance such as physical value using only integral number, (2) The optimum condition of the high precision in TAGUCHI Method can be found by approximation between the levels of control factor, (3) TAGUCHI Method is particularly effective when there are a little measurement data and when high reappearance is required, and (4) Neural Network is effective in founding the optimal condition of high precision from sufficient measurement data. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|