The spherical evolution (SE) search algorithm has a more novel search style than the common meta-heuristic algorithm. In contrast to the traditional hypercube search model, SE uses a spherical search approach and achieves very good results. Although the SE algorithm is very effective, it suffers from the problem of sometimes slow convergence due to the large search space and the imbalance between the development of the algorithm and the exploration capability. To address this, we innovatively proposed a strategy to intelligently adjust the search space based on the population structure information and improved the SE algorithm (ISE). Review of the effectiveness of our strategy, we compare ISE with SE algorithm and several other well-known meta-heuristic algorithms. Our problems with 30 different solution spaces on the IEEE CEC2017 benchmark serve as the test set for the experiments. The experimental results show that ISE has a significant advantage over other algorithms.