Because the teaching methods of Chinese teaching in higher vocational colleges are improper and the problems in the teaching process cannot be found in time, the teaching effect is not good and there are many difficulties in the teaching process of teachers. According to actual needs, based on complex evolution and improved neural network algorithms, this paper builds a Chinese online teaching system in high vocational colleges. After communication with users and social research, the model function module is determined, and the system performance is improved through the mapping optimization model to eliminate data transmission problems in the system operation, and the realization process of the system function module is shown in the form of a diagram. In addition, this paper designs a control experiment to verify the performance of the model and uses the traditional algorithm model as a control to start the test teaching and uses statistical methods to compare the teaching effects. The research results show that the teaching system constructed in this paper has certain advantages over the traditional model.