A Non-destructive Defect Estimation Method for Metal Pole Based on Machine Learning Approach
Abstract
This paper suggests a non-destructive defect estimation method for metal pole by analyzing its hammering sounds. A machine learning algorithm, that is, Support Vector Machine algorithm is applied for the spectrum distributions obtained from the hammering sounds by using a Fourier Transform. A defect estimation algorithm incorporating this algorithm is actually developed and tested for 200 hammering sounds of metal poles with/without a defect. The number of hammering sounds used for the learning phase is 180 and the remaining 20 are used for test phase. As a result, 100% accuracy rate is attained and this shows the feasibility of machine learning approach.
Keywords
Non-destructive testing, Defect estimation, Machine learning, Support vector machine, Hammering sounds, FFT, Metal pole
DOI
10.12783/dtcse/aics2016/8236
10.12783/dtcse/aics2016/8236
Refbacks
- There are currently no refbacks.