| The Bohai Oilfield is located in the Bonan Low Uplift Zone of the Bohai Bay. The high structural part of the oilfield is adjacent to the strike - slip fault zone, where faults are relatively well - developed. After several years of development and production, the injection - production connectivity relationship in the high structural part of the oilfield has become relatively complex, and there is a contradiction between the production dynamic characteristics of oil and water wells and the distribution characteristics of static faults. In order to clarify the injection - production connectivity relationship of the reservoir in the high structural part and deepen the understanding of the fault distribution pattern, so as to solve the current problem of poor injection - production effectiveness, this study innovatively introduces a Convolutional Neural Network (CNN) to extract the local features from the historical production data of oil and water wells. These features include liquid production rate, oil production rate, bottomhole flowing pressure, water injection volume, injection pressure, etc., and are used as the input feature sequence of the Bidirectional Gated Recurrent Unit (BiGRU). In the BiGRU, the forward GRU can capture the influence of historical production and injection volumes on the current state, while the backward GRU understands the potential influence of future data on the current production capacity from the information flow from the future to the past, thus fully considering the historical and future development and production information of the oilfield. The global features extracted by the BiGRU are integrated with the local features extracted by the CNN to form comprehensive features that not only contain the local patterns and structural information of the data, but also include the bidirectional dependency relationships and long - and short - term memory information of the time series. These comprehensive features can comprehensively reflect the interaction relationship between oil production wells and water injection wells. By analyzing these comprehensive features, the influence patterns of the water injection operations of water injection wells on parameters such as the liquid production rate and pressure of oil production wells can be identified, and then the injection - production connectivity relationship can be predicted. This method can effectively improve the prediction accuracy of the injection - production connectivity relationship and provide a scientific basis for optimizing the measures of oil and water wells in the oilfield. |
分享按钮