Classification of stainless steel and mild steel using vibration technique
The production of material in industry must attain some standard such as the standard required by American Society for Testing and Materials (ASTM) International. The requirement of the material standard is important in some crucial field such as aerospace, engineering and automotive. This resear...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
Universiti Malaysia Perlis (UniMAP)
2014
|
| Subjects: | |
| Online Access: | http://dspace.unimap.edu.my:80/dspace/handle/123456789/31301 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The production of material in industry must attain some standard such as the
standard required by American Society for Testing and Materials (ASTM) International.
The requirement of the material standard is important in some crucial field such as
aerospace, engineering and automotive. This research presents a development of a material classification scheme with non-destructive testing on the material to classify the material type. The classification of the material can be useful in post-production verification. Many testing methods have been developed to reach the standard of the material production. The testing of the material mechanical properties using vibration technique could determine the natural frequencies, the damping ratio and mode shapes of the structure. The testing method chose to be implemented in this research is impact hammer testing. Frequency Response Function (FRF) signals obtained from the testing and natural frequencies of the
materials are extracted from FRF signals. In this research, the features considered as the input data for the algorithm training are the natural frequencies of the material and its
amplitude. Later, the input data obtained are classified using Artificial Neural Network
(ANN) with Levenberg-Marquardt Backpropagation and k-Nearest Neighbor (k-NN). Each of the classifier produced a different classification rate depending on the performance of the training input data set. The result from the classification system shows that k-NN is giving the accuracy of 99.69% with the k value of 3. While, Levenberg-Marquardt
Backpropagation is giving the best classification rate of 99.43%. |
|---|