In this article, the charged system search (CSS) optimization method is improved to identify the parameters of a non-linear hysteretic Bouc-Wen differential model. The CSS is suitable for those optimization problems involving non-smooth or non-convex domains. Bouc-Wen is a well-established non-linear model which has been used to portray the hysteretic and high non-linear real behavior of numerous physical and mechanical systems. To improve the effectiveness and adaptability of the CSS algorithm, it is combined with sub-optimization mechanism. The obtained results show that the adaptive CSS embodies great robustness and accuracy to be successfully employed in such highly non-linear identification problems.