Abstract: A new dynamic research model for injection water oil fields was presented. The model contains increase production measures and it is adaptive to various random factors together in the process of oil field exploitation. In this paper, using neuron network models of nonlinear multidimensional time series predic-
tion, neuron network predictors for the oil production and water production of oil fields were constructed. Results illustrated that the predictors were up to higher prediction precision. But this method improved the theory of the dynamic prediction of oil production and water production .Finally, an example of the dynamic prediction were given.

Key words: oil field dynamic research, production prediction, time series, nerve network