Hybrid grey relational analysis-support vector machine (GR-SVM) in predicting machining surface roughness
Machining process is defined as a process of material removal from a work piece in the form of chips. This process has improved significantly over the years to meet the field requirements. However, a major issue in the process is how to obtain accurate results of the machining performance measuremen...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2013
|
| Subjects: | |
| Online Access: | http://eprints.utm.my/id/eprint/41645/5/AshaniraMatDerisMFSKSM2013.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Machining process is defined as a process of material removal from a work piece in the form of chips. This process has improved significantly over the years to meet the field requirements. However, a major issue in the process is how to obtain accurate results of the machining performance measurement value at optimal point of cutting conditions. Machining performance for surface roughness has been widely discussed by researchers but determining the optimal solution for surface roughness remains as one of the most challenging problem due to the complexity of the modeling process. Thus, this research proposed a hybrid model combining Grey Relational Analysis (GRA) and Support Vector Machine (SVM) to predict surface roughness values for end milling and abrasive water jet (AWJ) machining processes. In the proposed hybrid Grey Relational-Support Vector Machine (GR-SVM) prediction model, GRA acts as a feature selection method in pre-processing process to eliminate irrelevant factors and SVM solves the regression functions in machining problems to determine the surface roughness value. Efficiency of the proposed prediction model was demonstrated by comparing the results of the hybrid model with the experimental data and results of conventional SVM prediction model based on correlation and Root Mean Square Error (RMSE) values. The results showed that the hybrid GR-SVM prediction model presented the most accurate results due to its ability to remove redundant features and irrelevant elements from the experimental datasets. These results have shown that the optimal solution of machining performance can be achieved by using the proposed hybrid GR-SVM prediction model. |
|---|