Application of chaotic theory to oil production rate time series prediction

Time series analysis of oil production rate is performed using chaos theory. And based on phase space reconstruction, oil production rate prediction model was established, through which oil production rate was predicted directly and after smoothing respectively for each well. The analysis indicates that oil production rate shows chaotic behaviors for some wells, for which the average largest Lyaunov Exponent is about 0.063, while for the others, it´s obscure because the minimum embedding dimension couldn´t be obtained through Cao Method. The prediction model provides a reliable result, and the prediction after smoothing works even better. The average relative errors of direct prediction and smoothing-prediction are 13.02% and 1.58% respectively for wells in Tahe Oilfield, which indicates that it´s promising to develop a predictive model based on the previous oil production rate for wells.