The magnetic-geared generator integrates the magnetic gear and the generator by using the magnetic field modulation technology. It has the characteristics of high power density, high material utilization, and has a wide application prospect. However, compared with the general generator, its structure is relatively complex which makes its design and optimization become more complex. Therefore, a new structure and an optimization method based on orthogonal regression statistics is proposed. The experimental results fully prove the effectiveness of the proposed structure and optimization method.