---

概要

In this paper, we propose new inverse analysis method for reconstruction of magnetic permeability distribution using 2D magnetostatic analysis with Genetic Algorithm. We use Finite Element Method for 2D magnetostatic analysis, where the array of material number of each element is considered as genetic code. First, we produce genetic codes, i.e., individuals, randomly. We perform the FEM analysis to obtain the distribution of magnetic flux density near the specimen surface with respect to each individual. Subsequently, we calculate the fitness of each individual based on square error between the target value and the calculated value. Then we reproduce the generation by performing genetic operation on each individual stochastically according to its fitness. Finally, by repeatin g same procedure, we obtain the optimum solution, i.e., the individual which has the highest fitness in the final generation. It is found that the proposed method can reconstruct the distribution of magnetic permeability correctly and identify the defect shape.

---

全文掲載のPDFファイルダウンロード

• PDFファイルダウンロード