Concept of an expert system for rail defect classification

The authors propose an initial concept of an expert system that utilizes systematized knowledge derived from periodic ultrasonic rail inspections in the track to classify rail defects. The proposal involves the use of four multi-valued attributes recorded in manual and automatic flaw detection tests: problem identification, defect location, type and direction of defect development, and extended defect classification. Based on these, a decision table was built to allow for preliminary defect categorization, and an algorithm for generating defect codes was proposed.Expert systems are computer tools that allow for the convenient and effective use of accumulated knowledge to make appropriate decisions. They can both replace human experts and offer them (…)

Concept of an expert system for rail defect classification – PDF file for download