Neural network approach compared to sensitivity analysis based on finite element technique for optimization of permanent magnet generators

The paper presents the optimization procedure of a permanent magnet generator for a 20 kW wind turbine prototype "Peripheral" neodymium alloy magnet rotor structure has been considered to perform the optimal shape design. A fully connected four layer feedforward neural network has been introduced and compared to a technique based on the finite element method and sensitivity analysis. The considered methods are in very good agreement.