Optimization of cutting parameters while turning Ti-6Al-4 V using response surface methodology and machine learning technique
| Autor: | V. K. Sridhar, G. Prasanthi, A. Kiran Kumar, S. K. Gugulothu, Mulugundam Siva Surya |
|---|---|
| Rok vydání: | 2021 |
| Předmět: |
Mean squared error
Artificial neural network business.industry Titanium alloy Machine learning computer.software_genre Box–Behnken design Industrial and Manufacturing Engineering Root mean square Machining Modeling and Simulation Surface roughness Artificial intelligence Response surface methodology business computer Mathematics |
| Zdroj: | International Journal on Interactive Design and Manufacturing (IJIDeM). 15:453-462 |
| ISSN: | 1955-2505 1955-2513 |
| DOI: | 10.1007/s12008-021-00774-0 |
| Popis: | Titanium alloys have huge applications in the field of aerospace. However, finding the best combination of machining parameters is still a challenge for many researchers. The present work investigates the influence of cutting parameters on surface roughness and material removal rate while turning Ti–6Al–4 V using TiCN coated carbide tool. The effect of input parameters on the output responses is studied using response surface methodology (RSM) and machine learning techniques. The Box Behnken method (L15 array) is selected to design the set of experiments. In this investigation, three different levels of speed, feed, and depth of cut are considered as input parameters. The surface roughness and material removal rate are measured for each experiment, and the output factors are optimized using response surface methodology. The cutting parameters are optimized to obtain the least surface roughness and the highest material removal rate. The ANOVA analysis confirms that speed with 44.62% has the highest contribution for surface roughness and depth of cut with 64.43% has the highest contribution for material removal rate. The Root means square errors (RMSEs) obtained for MRR and surface roughness using an artificial neural network are 0.397 and 0.291, respectively, which shows significantly less error. The lower the RMSE value, the better is the model prediction. The machine learning technique (artificial neural network) exhibited 5.04% and 10.66% errors for surface roughness and MRR, respectively. The percentage error values resulting from the machine learning technique are less when compared to RSM. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full text from SpringerLink