Abstract: During the numerical optimization process of cascades for multi-component isotopes separation, there are problems such as integer-real optimization and local optimal solution, which makes the optimization of cascades more difficult. Taking the Q cascade as a model, a numerical optimization method of mixing integer-real optimization is proposed, which can optimize the integer type and the real type at the same time. By introducing cascade flow distribution constraints and a single-objective penalty function containing constraints, the method can effectively avoid the problem of local optimal solutions in the cascade numerical optimization process. The results show that the method can effectively optimize cascades and obtain discrete cascade optimization results close to Q cascade. The optimized discrete cascade total flow and the flow distribution is close to the Q cascade. According to the production and construction of separation cascades, based on the optimized Q cascade, square-off cascades of different shapes are constructed and optimized to approximate Q cascade, which can effectively realize the optimization of the square-off cascade, and reduce cascade construction costs.