Research And Application Of Digital Log Interpretation Method

Log interpretation is an important part of oil-gas exploration and development,which involves the processing,analysis and application of a large amount of log interpretation data,to further characterize oil-gas reservoir information and guide oil-gas exploration and development.The growing demand for oil-gas production has led to higher requirements for the accuracy,efficiency,and cost of log interpretation.With the development of electronic information technology,digital log interpretation methods provide new solutions for log interpretation.The integrity and resolution of log interpretation data,as well as the reasonable analysis and application of data,are crucial to the accuracy,efficiency and cost of log interpretation.Therefore,this thesis aims to establish the digital log interpretation methods and focuses on the problems of missing log interpretation data,low resolution of log interpretation data,and reservoir quality evaluation in the log interpretation.The main research contents are as follows:To address the problem of missing log interpretation data,this thesis proposes a dualattention CNN-LSTM hybrid neural network based on the characteristics of data in the study area to predict log interpretation data.A data pre-processing method based on domain knowledge of logging is also proposed to improve the accuracy of prediction.The corresponding rock mechanics parameters of the entire well profile are further predicted.This method effectively achieves the prediction of log interpretation data and complete the missing log interpretation data.To address the problem of low resolution of log interpretation data,which makes it difficult to identify thin layers,a digital log interpretation data resolution enhancement method based on nuclear logging physical principles is proposed.This method starts from the physics principle of nuclear and analyzes the relationship between real formation physical parameters and measurement values,and then uses the sliding window method and neutron density cross constraints to reduce the uncertainty of the result to obtain accurate high-resolution formation physical parameters.The log interpretation data processed by this method effectively identifies the thin layers contained in the formation.To address the problem of reservoir quality evaluation in log interpretation,this thesis proposes a reservoir quality evaluation method based on nuclear magnetic resonance log interpretation data.The method analyzes the relationship between nuclear magnetic resonance log interpretation data and reservoir quality,and establishes the reservoir quality evaluation standard.On this basis,a low-cost reservoir quality intelligent evaluation method based on conventional log interpretation data is further proposed,and the integrated learning algorithm is introduced to realize the intelligent evaluation of reservoir quality.This method realizes the evaluation of reservoir quality based on conventional log interpretation data,improves the evaluation efficiency,and reduces the interpretation analysis cost.In this thesis,the methods are tested and verified using the actual log interpretation data in the study area.The experimental results prove that the proposed digital log interpretation methods can effectively solve the problems existing in current log interpretation and meet practical application requirements.