Optimal magnetic sensor location for spherical harmonic identification applied to radiated electrical devices

This paper deals with the use of spherical harmonic representation to identify the magnetic multipolar sources associated to a given item, by measuring the near field around it. If the model presents many advantages, its use in an inverse problem approach requires special caution to critical points like the choice of sensor number and location. A method is presented that makes it possible, from a high number of information, to reduce the sensor number to a minimum and to optimize their location with respect to robustness. The robustness issue is solved by using an original method based on a genetic algorithm. A simple but didactic example illustrates the difficulties that can be met and our strategy to overcome them