Please use this identifier to cite or link to this item:
http://hdl.handle.net/10603/17391
Title: | Experimental investigation of quality characteristics in macro and micro machining processes using intelligent techniques |
Researcher: | Palani S |
Guide(s): | Natarajan U |
Keywords: | Back-propagation Differential evolution algorithm Machine vision system Macro machining Mechanical Engineering Micro machining |
Upload Date: | 11-Mar-2014 |
University: | Anna University |
Completed Date: | 01/09/2013 |
Abstract: | The technological advances with the ever-growing use of computer newlinecontrolled machine tools which are integrated with machine vision newlinesystem(MVS), have brought up new issues to emphasize the need for more newlineprecise predictive models. In the macro machining operations, a novel attempt has been made newlineto predict the average surface roughness (Ra) of the turned components using newlinea differential evolution algorithm (DEA) based artificial neural network newline(ANN), applied to train feed forward multi-layer perceptron neural networks newline(MLPNN) and compared with the widely used back-propagation (BP) based newlineANN and an adaptive neuro-fuzzy inference system (ANFIS) model. It is newlinefound that the predicted values of Ra are in good agreement with the newlineexperimental values. It is also found that the error percentage is very minimal newlinein DEA based ANN and also observed that the convergence speed for the newlineANN-DEA model is higher than the ANN-BP and ANFIS. The images of milled surface grabbed by the MVS could be newlineextracted using spatial frequency domain two dimensional Fourier transform newline(2D-FT) to get the features of image texture (major peak frequency (F1), newlineprincipal component magnitude squared value (F2), and average gray level newline(Ga)). A new ANFIS is applied to predict Ra. The proposed model is verified newlineand compared by using ANN-feed forward back-propagation (ANN-FFBP) model. It is found that the average prediction error is less than that of ANNFFBP, newlineindicating that the ANFIS is more reliable than the ANN-FFBP. From newlinethe performance of ANN-FFBP and ANFIS models in terms of average newlineabsolute percentage error, it is observed that the ANFIS model outperforms newlineANN. In the second phase of work, a tool-based micro-machining process newlineis employed for automated, noncontact, and flexible prediction of quality newlineresponse characteristics such as Ra, tool wear (TW) and material removal rate newline(MRR) of micro-turned miniaturized parts through a MVS which is integrated newlinewith an ANFIS . newline |
Pagination: | xviii, 177p. |
URI: | http://hdl.handle.net/10603/17391 |
Appears in Departments: | Faculty of Mechanical Engineering |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
01_title.pdf | Attached File | 32.84 kB | Adobe PDF | View/Open |
02_certificates.pdf | 934.71 kB | Adobe PDF | View/Open | |
03_abstract.pdf | 10.6 kB | Adobe PDF | View/Open | |
04_acknowledgement.pdf | 6.41 kB | Adobe PDF | View/Open | |
05_contents.pdf | 30.02 kB | Adobe PDF | View/Open | |
06_chapter 1.pdf | 24.34 kB | Adobe PDF | View/Open | |
07_chapter 2.pdf | 107.77 kB | Adobe PDF | View/Open | |
08_chapter 3.pdf | 2.69 MB | Adobe PDF | View/Open | |
09_chapter 4.pdf | 1.73 MB | Adobe PDF | View/Open | |
10_chapter 5.pdf | 2.05 MB | Adobe PDF | View/Open | |
11_chapter 6.pdf | 2.56 MB | Adobe PDF | View/Open | |
12_chapter 7.pdf | 20.33 kB | Adobe PDF | View/Open | |
13_references.pdf | 32.84 kB | Adobe PDF | View/Open | |
14_publications.pdf | 11.13 kB | Adobe PDF | View/Open | |
15_vitae.pdf | 6.01 kB | Adobe PDF | View/Open |
Items in Shodhganga are licensed under Creative Commons Licence Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0).
Altmetric Badge: