The asymmetric-paths winding is a novel ac winding with a fractional ratio of the number of poles to the number of paths, which is beneficial for the optimized design of machines. Resulting from the numerous combinations of slot-vectors for path windings, the optimized arrangement with a low asymmetric degree can hardly be achieved by existing methods. Focusing on the optimization for arrangements of this winding, the mathematical optimization model considering two objectives for the percentage of the electromotive force error and the total harmonic distortion is built. In addition, an improved discrete particle swarm optimization approach is proposed and includes a setting method for the initial positions, a discrete operator with chaotic mutation for the velocity update, and an aggregate strategy combined with fuzzy inference. To validate the proposed approach, a winding design of a hydrogenerator is presented combined with the finite element analysis. The results show that the optimized arrangement presents a low asymmetric degree and harmonic. This approach provides a technical support for the arrangement optimization of an asymmetric-paths winding.