Method of estimation of current distribution using genetic algorithms with variable-length chromosomes

The problem of determining the source by using field data observed in finite space is called an inverse problem. Recently, the necessity of developing an inverse problem analytic method has become much more important because of its wide applications, especially in the field of electromagnetics. Most of the inverse problems, however, are suffering from ill-posed problems. As a result, they are difficult to solve analytically and uniquely. This paper describes a new analytic method for inverse problems, applied to an estimation of static two-dimensional current distribution in materials, using the observed external magnetic field as practical examples. We also propose an efficient searching strategy by dividing the searching process into two phases and devising the structure of chromosomes. Finally, the feasibility of the proposed method is confirmed by simulations.