This paper discusses portfolio selection problem when security returns are uncertain variables.Considering the real investment is a multi-period process, we propose multi-period mean-variance and mean semivariance models.Then we give the crisp forms of the multi-period mean-variance model with security returns all being uncertain linear or normal variables respectively.In addition, a hybrid intelligent algorithm based on 99 methods is also designed for the optimization problem in general cases.Finally, two numerical examples are given to illustrate the modeling idea.