一种特高含水期油井含水率多井联合预测方法及系统
By using a cascaded TCN-LSTM structure, a multi-well joint prediction model was constructed, which solved the problem of complex fluctuation characteristics of water cut in oil wells during the ultra-high water cut period, realized multi-well joint modeling, and improved the accuracy and consistency of water cut prediction.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHINA UNIV OF PETROLEUM (EAST CHINA)
- Filing Date
- 2026-05-25
- Publication Date
- 2026-07-17
AI Technical Summary
Existing technologies struggle to achieve multi-well joint modeling in oil well scenarios with extremely high water cut, failing to effectively characterize complex well-level dynamics. Single-well time-series models cannot utilize shared patterns across multiple wells, and multi-well deep learning solutions often prioritize production rather than water cut.
A cascaded TCN-LSTM structure is adopted. By acquiring historical production dynamic data from multiple oil wells, data preprocessing and normalization are performed to construct supervised learning samples. The TCN module is used to extract local fluctuation features, and the LSTM module is used to characterize long-term dependencies. A multi-well joint training set is constructed for model training, and finally the predicted water cut value at the next time step is output.
It enables multi-well joint modeling while maintaining the time sequence of individual wells, effectively handling the non-stationary and nonlinear changes in water cut of individual wells during ultra-high water cut periods, and improving the consistency of the model in depicting the overall trend and its engineering applicability.
Smart Images
Figure CN122242786B_ABST