Research And Application Of Optimization Model Of Oil Field Stimulation Measures Based On Data Mining

With the development of most oil fields entering the middle and late stage,necessary oil and gas stimulation measures must be taken to ensure continuous and efficient production of oil fields.Due to the variety of stimulation methods,such as fracturing,acidification,overhaul,and hole refill,the effect and cost of each measure are different,and each stimulation measure applied to each operation area of the oil field has different effects.In order to reasonably plan the stimulation actions for the oilfield,it is necessary to establish a prediction model for the effect of stimulation actions first,and then make specific measures planning based on the predicted results.Based on the neural network algorithm of data mining,this paper studies and applies the index optimal selection model of oilfield stimulation actions.BP and RBF neural networks are used in this paper,aiming at the problems of slow convergence speed of BP neural network and easy falling into local optimum,the ant colony algorithm with heuristic optimization characteristics is used to optimize the BP neural network,and a prediction model of the effect of oilfield stimulation actions based on ACO-BP was established;Aiming at the selection problem of RBF network center,quantum particle swarm optimization algorithm was used to optimize the RBF neural network to obtain the optimal network center,base width and connection weight.A prediction model of oil field stimulation actions based on QPSO-RBF was established.This paper selects the monthly data of oilfield stimulation measures from January 2017 to December 2019 in 11 operating areas of an oil production plant in Changqing Oilfield,and predicts the effect of oilfield stimulation actions through five prediction models,namely BP,ACO-BP,RBF,PSO-RBF and QPSO-RBF.By analyzing and comparing the results of five prediction models,it is shown that the ACO-BP,PSO-RBF,and QPSO-RBF models have a significant improvement in prediction accuracy and stability compared to the BP and RBF neural network models,among them,ACO-BP prediction model has a smaller MSE and MAE,R~2 is closer to 1,it has a good application effect in predicting the effect of oilfield stimulation actions.Based on the above five prediction models,C#is the development language,SQLite database is the back-end database,we designed and implemented an oilfield stimulation actions index optimal selection system,which has become a management system that integrates measure data management,measure effect prediction,and measure planning.It provides a reference plan and scientific basis for oilfield managers to formulate plans for stimulation actions.