UTHM Institutional Repository

Improved performance of an artificial neural network pattern recognizer using design of experiments technique

Masood, Ibrahim and Zainal Abidin, Nadia Zulikha and Hassan, Adnan (2014) Improved performance of an artificial neural network pattern recognizer using design of experiments technique. In: International Conference on Science and Technology (lCSTlE) 2014, 17-18 September 2014, Pulau Pinang.


Download (121kB)


In the field of bivariate quality control (BQC), artificial neural network (ANN)-based pattern recognition schemes have been developed for monitoring and diagnosis of process variation pattern. In order to achieve efficient recognition performance (RP), proper design of an ANN-based model recognizer is essential. Extensive literature review revealed that the recognizers were mainly designed empirically. In this study, design of experiments technique was utilized for determining the suitable design parameters. Based on raw data input representation for a generalized-model ANN recognizer, four design parameters were investigated, i.e., size of recognition window, number of training data, quality of training data and number of hidden neuron.The process variation patterns were focused on bivariate process mean shifts with linear correlation function, p = 0.1 - 0.9 and mean shifts, p = + 0.75 - 3.00 standard deviations.The findings suggested that improved RP could be achieved by setting specific design parameters into the specialized-model ANN recognizer (RP = 98.85%). This model gave superior result compared to a single generalized-model ANN (RP = 96.5%. This study has provided a new perspective in designing a specialized-model ANN pattern recognition scheme.

Item Type: Conference or Workshop Item (Paper)
Uncontrolled Keywords: artificial neural network; design of experiments; generalized-model recognizer; specialized-model recognizer; bivariate quality control
Subjects: T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7800-8360 Electronics
Divisions: Faculty of Mechanical and Manufacturing Engineering > Department of Manufacturing and Industrial Engineering
Depositing User: Normajihan Abd. Rahman
Date Deposited: 13 Jan 2015 06:44
Last Modified: 13 Jan 2015 06:44
URI: http://eprints.uthm.edu.my/id/eprint/6142
Statistic Details: View Download Statistic

Actions (login required)

View Item View Item


Downloads per month over past year