Networked non-linear control systems (NNCS) are studied in this paper. A control scheme based on predictive control is proposed to address the problems introduced by the network. Based on the theory of the linearization of the non-linear process, a predictive control algorithm is constructed to generate a sequence of control predictions, some of which can suitably be selected to control the non-linear process and further compensate for the time delay. Methods are proposed to deal with the selection of control predictions, the control prediction sequence updating, and data disorders processing. The proposed control scheme is able to avoid the synchronization issue by modelling the predictive controller as an event-driven element. The stability of this NNCS has been studied by discussing the stability of an augmented system, the stability conditions of which can be transformed into a series of inequalities. In addition, the control method is validated by both simulations and experiments.