Abstract:Neural networks was established to predict the compressive strength of concrete at different age and special emphasis was put on the strength efficiency of high fly ash content concrete. In order to establish neural networks, two important features were defined. One was connection topology of networks, and the other was the learning rule of networks. From the simulation results, conclusion comes as: (1) The compressive strengh of concrete with high content of fly ash (HFCC) can be predicted successfully by use of the model of neural networks. (2) HFCC can be realized with effective chemical admixture. It will lead to obvious reduction of cement proportion and be good for environment protection.