Research On Reservoir Dynamic Monitoring And Early Warning Method Based On Big Data Analysis

Precise grasp of reservoir development dynamics to improve crude oil recovery is an important task for oilfield enterprises.Many experts,scholars and geological researchers have established many models for understanding and explaining reservoir dynamics through seismic imaging,3D simulation and other means in geological structure,sediment distribution,rock properties,etc.,and have achieved some results.In recent years,with the development of big data analysis technology,the research on reservoir dynamics has gradually developed from experience-driven to data-driven direction,and this paper is based on this development trend to carry out the research on reservoir dynamics monitoring and early warning method based on big data analysis.Firstly,after researching and collecting the relevant evaluation indexes for reservoir development,a reservoir dynamic monitoring and early warning index system was constructed based on the correlation analysis method from the actual meaning of each index;then,considering the characteristics of reservoir dynamic big data and the related business requirements of early warning,a reservoir dynamic data of a block in Changqing oilfield was taken as the research object,and a reservoir dynamic monitoring and early warning model based on SVM algorithm was established;For the problems of many parameters in the SVM model and the difficulty of finding the best,the model parameters were optimized using grid search and PSO algorithm respectively,and the reservoir dynamic monitoring and warning model based on PSO-SVM algorithm was established;in addition,the reservoir dynamic monitoring and warning model based on RF algorithm was constructed,and the application and analysis of the above three reservoir dynamic monitoring and warning models were evaluated;finally,based on the big data Finally,based on the big data platform,a cluster was built to import,store and calculate massive data,and a reservoir dynamic monitoring and early warning system integrating dynamic monitoring of indicators,static data query,single-well capacity curve plotting and dynamic fluctuation warning functions was developed in Java language,and tested and verified for the reservoir dynamics of a block in Changqing oilfield under study.The research results show that the model constructed in this thesis has achieved the goal of monitoring and warning the reservoir dynamics in oilfield,and the early warning modeling time is shorter,the consideration factors are more comprehensive,and the practical application effect in oilfield is better,which provides an analysis tool and effective method for oilfield enterprises to carry out comprehensive evaluation of reservoir and accurately grasp the reservoir development dynamics under the background of big data.