Study On Quality Control And Intelligent Early Waring Technology Of Array Induction Logging Data

Qualified logging data is the premise and foundation of logging interpretation.In order to better study the quality of logging data,this paper combines artificial intelligence machine learning algorithm with array induction logging data quality control and intelligent early warning research,and establishes array induction logging data prediction model based on LSTM neural network algorithm.The original array data and synthetic focusing curve data of array induction logging are studied and analyzed respectively to realize the quality control and intelligent early warning of array induction logging data.The following are the main research contents and achievements.First,the array induction logging database is established,including the measured database,the forward simulation database and the forward simulation anomaly database.First,the measured data of 26 Wells were processed by the LEAD software with depth correction,borehole correction and synthetic focusing,and the multi-source heterogeneous fusion was sorted into a unified standard format to form the measured database.Secondly,the three-layer and five-layer models are constructed based on the measured data,and the forward simulation database is generated after numerical calculation.Finally,according to the common abnormal data types in actual logging,three common abnormal data types are preset: curve flatness,curve out of order and frequency out of order,and the corresponding forward simulation anomaly database is established.Secondly,the prediction model of array induction logging data based on LSTM neural network is established,including data preprocessing,LSTM neural network structure design and prediction model establishment.Firstly,according to the corresponding data in array induction logging database,the data cleaning,data standardization and correlation analysis are preprocessed.Secondly,the structure of LSTM neural network is designed by setting LSTM neural network parameters.Finally,the designed neural network structure and processed data are trained by using python programming environment,so as to carry out parameter optimization to complete the establishment of the prediction model.Thirdly,the quality control and intelligent warning of array induction logging data are studied,including the original array data and the synthetic focusing curve data.For the original array data,firstly,the forward simulation data is forecasted and visualized,and the index evaluation is analyzed.Secondly,according to the forward simulation anomaly database,the abnormal data types of the forward simulation original array are identified using the Puta criterion.Finally,the original array data in the actual well is predicted and visually displayed,and the error color warning is used for intelligent warning.In this way,the quality control and intelligent warning of the original array data are realized.The synthetic focusing curve data is studied by using the actual well data,and the processed synthetic focusing curve data are predicted,visualized display and error color warning analysis,so as to realize the quality control and intelligent warning of the array induction logging synthetic focusing curve data.According to the research content of this paper,it is found that by establishing the prediction model of array induction logging data based on LSTM neural network to predict and analyze array induction logging data,the quality control and intelligent early warning of array induction logging data can be effectively carried out,which improves efficiency and reduces labor cost.