ARTIFICIAL NEURAL NETWORK BASED CLASSIFICATION OF AUSTENITIC STAINLESS STEEL WELD DEFECTS IN TOFD TECHNIQUE
Journal Title: Indian Journal of Computer Science and Engineering - Year 2011, Vol 2, Issue 6
Abstract
In this paper, an automatic detection system to recognize welding defects based on Time of flight diffraction technique is described. The proposed classification consists in detecting the four types of austenitic stainless steel weld defects and non-defect type. The austenitic stainless steel welds with artificially created defects have been considered. A scan Signals are obtained by conducting TOFD experiment on these weld defects. To improve the efficiency of defect detection, a discrete wavelet transform based denoising was also adopted as a preprocessing technique. Time scale features have been extracted from the denoised TOFD signals and an artificial neural network for weld defect classification was developed. A multi layer feed forward network with BFGS quasi-Newton back propagation has been applied for classification of the signals. The effect of hidden layers on the network was analyzed. The optimum performance function for this network was also found.
Authors and Affiliations
S. Lalithakumari , Dr. B. Sheelarani , Dr. B. Venkatraman
Keywords
Related Articles
- EP ID EP140103
- DOI -
- Views 117
- Downloads 0
ANFIS IN THE CHARACTERIZATION OF FIBROSIS AND CARCINOMA USING LUNG CT IMAGES
The diagnosis of tuberculosis and lung cancer is difficult, as symptoms of both diseases are similar. Due to high TB prevalence and radiological similarities, a large number of lung cancer patients initially get wrongly...
Palmprint recognition using Contourlet Transform Energy Features
Palmprint recognition is an accepted and widely used biometric. Richness of feature and the less cost involved in acquisition make it more reliable and user friendly. The region of interest is extracted from the palmprin...
Efficient Algorithms for Distributed Mutual Exclusion in Mobile Ad-Hoc Network
For distributed mutual exclusion problem in mobile environment we presented two algorithms.Both the algorithms are token based. The first algorithm is providing tokens to the nodes which are in critical section and the s...
IDENTIFYING NETWORK LOAD BALANCING REQUIREMENTS ON HISTORICAL TRAFFIC FLOW USING MACHINE LEARNING APPROACH FOR FLEXIBLE MULTIPATH NETWORKS
Multipath routing uses several paths to distribute traffic from a source to destination. This not only improves network performance but also achieves load balancing and fault tolerance. Though multipath routing is not de...
Computer Profiling Based Model for Investigation
Computer profiling is used for computer forensic analysis, and proposes and elaborates on a novel model for use in computer profiling, the computer profiling object model. The computer profiling object model is an inform...