This paper presents a model for freeway traffic flow simulation and prediction. The model uses cellular automation theory to model complex traffic behavior. The advantage of the cellular automata approach is that the roadway to be modeled is quantized into simple homogeneous cells, time is quanitzed into discrete steps, and physical quantities take on a finite set of values. Also, the state of the cells is updated at each discrete timestep by using a vehicle update algorithm that combines a few vehicle motion models, governed by a relatively small set of parameters. Then vehicles just move one or several cells at each discrete timestep according to the self-defined rule. This approach makes the computer operation feasible. At last, the paper puts forward a suppose that if the simulation system is equiped with the self-study system of NN (neural network) module according to the statistical data from the transducer fixed on the freeway, it can predict the traffic status ahead of 10 minutes.