Multi-level production scheduling problem is a typical combinatorial optimization problem in manufacturing system, which is traditionally modeled as hierarchical sub-problems and optimized at each sub-problem level. An improved integrate evolutionary algorithm, which can cope with the whole multi-levels' scheduling information at one time and can find the satisfied optimal solution in acceptable computational cost. The new algorithm employs a twin-screw coding strategy, which enhances the efficiency and performance of the algorithm's convergence. Besides, a real three-level production scheduling problem case study is employed to evaluate the performance of the proposed algorithm. Experimental results show that our algorithm has outperformed the existing one.