Application Of Big Data In Optimization Design Of Oil Well Lifting Technology

The purpose of optimization design of lifting process is to carry out safe production in high yield and high efficiency,this requires the selection of swabbing parameters and swabbing equipment suitable for oil well characteristics,and according to the selected construction plan construction,as far as possible to draw the equipment and oil layer potential.Common artificial lifting methods include rod pump,hydraulic pump,electric submersible pump,screw pump and so on.Different lifting methods have different adaptability to oil reservoir type,development mode and production capacity of oil well.With the deepening of petroleum exploration and development,and the construction of digital oilfield in recent years,the data of oilfield has exploded,and oil informatization has entered the "big data era".At the same time,a large number of historical data with labels and labels can be used to optimize the design of lifting methods.In this paper,the optimization mechanism of the lifting scheme is studied,and the optimization process of the lifting scheme is divided into the primary selection based on the technical adaptability and the final selection based on the technical economic efficiency,moreover,the basic evaluation index set is established,and the corresponding calculation method is given.A deep neural network model was established based on a large number of historical design data of China petroleum A2 project,and the potential connection between the adaptive indexes and the optimization of the lifting scheme is calculated and analyzed.In order to improve the accuracy and efficiency of the algorithm,in the process of data preprocessing,this paper use the weighted CNN method to remove the sample data is redundant,then the improved LOF method was used to eliminate abnormal data,and the improved k-nearest neighbor algorithm is used to complete the data missing value and as the prediction method for the oil production of the design well.In the final selection of the lifting scheme,based on the technical economic efficiency index,the optimal lifting equipment model and parameters were obtained by using the principal component analysis-grey correlation method.The optimization design of lifting scheme is more accurate and reasonable.Finally,using Python as the development language,theano as a deep learning algorithm library,wrote a set of "optimization design software for oil well lifting process based on big data".Based on a large number of historical data,the function of optimization design of lifting method is realized,which provides new ideas and theoretical support for the optimization design of oil well lifting process.The accuracy and efficiency of the method are verified by combining with field practice.