To the weaknesse of space exploration and exploitation of ions motion algorithms, an improved algorithm is proposed. In the liquid phase of the ions motion algorithm, a diversity feedback search mechanism and a global optimal guidance strategy are embedded in the algorithm structure. Meanwhile, the initialization process in the crystal phase of the optimization algorithm is generated by using the opposition-based learning method, in which the initialization probability is changed by dynamic inertia. The proposed algorithm is tested using the 23 international bench mark functions and compared with some popular meta-heuristics algorithms, furthermore are comprehensively evaluated in terms of the average convergence value, variance, Wilcoxon's sign rank test, convergence success rate and optimal convergence time. The results show the effectiveness of the proposed algorithm.