The Particle Swarm Optimization Algorithm Based on Dynamic Chaotic Perturbations and its Application to K-Means

The dynamic chaotic perturbations are introduced for the standard particle swarm optimization. In the paper, small disturbances are used when the optimal value changed. The chaotic disturbances within dynamical range of disturbances are used when the optimal value unchanged many times. This not only can reduce the blind search of the chaotic particle swarm algorithm, and can improve the search speed and search efficiency, so that the limited time will be spent on the most effective search. According to the characteristics of different chaotic map, the Tent mapping is used to generate dynamical range of disturbance and the Chebyshev mapping is used to chaotically perturb between the global optimal and the optimal or sub-optimal in individual optimal solution. The algorithm is applied to the K-means algorithm, which can overcome the shortcomings of the local optimum and the sensitive to initial value in the K-means algorithm, can stably acquire the global optimal solution.