Paper Presentations: Recent Trends In Neural Network

We will send you updates on Recent Trends In Neural Network and also updates on other new papers uploaded into this site. Click here to Subscribe to Paper Presentations for Engineering Students by Email and recieve daily updates. Abstract: The detection of rail defects is a labor-intensive process in spite of recent advances made with modern rail detection equipment. Such detection systems provide facilities for both the collection and analysis of rail data. While these systems are effective, it is desirable to improve their analysis capabilities in order to both expedite the rail detection process and to increase the accuracy of the rail detection process. The objective of this research investigation was to determine the feasibility of using neural networks to improve the automated detection and classification of rail defects. Since the process of recognizing defective rail depends primarily on the ability to identify irregular patterns in the data and since neural networks are well-known for their ability to detect patterns in data [2, 4, 5, 6], the application of neural networks to this problem was natural.

The basic approach consisted of performing neural network analysis on actual rail data. Union Pacific Railroad (UPRR) rail detector crews collected this data from several different locations. The data was collected and analyzed using UPRR's System 1000 rail detection system from Harsco Track Technologies. The rail detector crews monitored the collection and analysis and performed hand testing as needed. After studying the complexity of the data, it was decided that this pilot study would focus on the identification of defective bolt holes.

The results obtained strongly indicate that neural networks can provide an effective tool for enhancing both the speed and accuracy of the automated identification of defective bolt holes. From investigating transducer outputs obtained from various types of rail defects, it appears that it would not be difficult to extend this approach to other types of rail defects.

Courtesy: N.Santhanam, Adi Parasakthi Engineering College (APEC), Melmaruvathur, Tamilnadu