To enhance the performance of the particle swarm optimization (PSO), various mutation operators were proposed. But it is hard to select a proper operator in advance in the real word, because the objects are quite different. To incorporate the characteristic of the operators, several known operators were implemented to PSO all together, the results have shown that the performance was enhanced for most functions, but deteriorated for few functions. Besides which the function evaluations increased sharply with the increase of operators. To address the problem, an adaptive operator selection strategy is introduced where the swarm is divided into groups with different probabilities to employ the operators. The probabilities are adjusted adaptively by comparing the effectiveness of the groups to select a proper operator. The simulation results have shown PSO with the strategy provides robust and consistent performance.