Research On Workload Prediction Method Of Oil Well Stimulation

Most oilfield production has entered the middle and late stage,and the prediction of oil well stimulation measure workload is not only related to the future development goal and direction of the oil production engineering system,but also related to the benign and scientific development of the entire oilfield development system in the future,which is an important part of oilfield development planning.In this paper,the main stimulation mechanism of several major oil well measures commonly used in the field is studied.Combined with the main research contents of this paper,the basic index set consisting of the oil well measure effect prediction index,the production increase forecasting indicator and the production increase plan workload evaluation index is established.Firstly,the indicators for the workload forecasting of the stimulation measures are obtained,which are mainly composed of the control indicators and the state indicators.According to the evaluation index system of the workload increase plan,this paper comprehensively uses the ISM interpretation structure model theory and the SVM-FCM feature method to select and optimize the indicators.The prediction of oil increment effect plays a cornerstone role in the whole measure workload prediction.This paper constructs a prediction model composed of three modules: indicator selection subsystem,data preprocessing subsystem and oil increment prediction subsystem.The model uses gray correlation method to calculate the correlation degree of different measures,and obtains the influence parameters of different oil-increasing measures.In the process of pre-processing,the FOMABCLD algorithm and Bayesian theory are used to eliminate the outliers and make up for the missing values.Combined with the double-scale Bayesian formula,the oil-increasing effect is predicted by using a large number of effective historical data under different measure index systems.In the design of measure workload prediction model,both the deterministic,uncertain prediction model and their solutions are constructed,as well as the analysis of the optimal multiple workload scenarios.In the deterministic measure quantity prediction model,a multi-objective programming model considering cost,oil-increasing effect and workload limitation is established.In combination with stochastic optimization theory,a hybrid intelligent solution method is designed to predict the workload effectively.The fuzzy analytic hierarchy process(fahp)is used to realize the optimization of various schemes according to the priority of each index,which is of both theoretical and practical significance.Based on the A5 project database of CNPC,combined with the actual production status of 16 oilfields in CNPC,a software platform for predicting and planning the workload of oil well stimulation measures was developed using Python language.