| ABSTRACT: | FUNDAMENTAL STUDY OF FLAW ESTIMATION IN EDDY CURRENT TESTING USING
GENETIC ALGORISM
K. Koyama, H. Hoshikawa
Nihon University, Narashino, Chiba, Japan
In recent years, not only flaw detection but also flaw estimation of high accuracy is demanded in eddy
current non-destructive testing. The eddy current flaw-testing signal (ECT signal) is different depending on
the depth, length, width and shape of the flaw so that the optimization technique, which optimizes flaw
shape minimizes the difference between ECT signal and flaw signal calculated by FEM, is proposed. The
Genetic Algorism (GA), which can efficiently do a global retrieval as an optimization technique, is used. In
the flaw shape estimation using GA, it is necessary to calculate the flaw signal of each shape, so that there is
a problem of long computational times to calculate the flaw signal. The authors propose the technique for
calculating the flaw signal by superposing ECT signal that does not use FEM to calculate the flaw signal of
each flaw shape. In this experimental condition, the authors found linearity between the signal of short
length flaw than the surface coil diameter and the signal of long length flaw than the coil diameter. That is,
ECT signal is calculated by superposing ECT signal of shorter flaw than the surface coil diameter. The
authors have conducted fundamental study of the flaw shape estimation form ECT signal using GA. In this
paper, the proposed technique for calculating the flaw signal by superposing ECT signal is also discussed.
|