Research On Steam Flooding Development Effectiveness Prediction And Parameter Optimization Method Based On Intelligent Computation

With the development of the global economy, people want to get more oil. As the available oil resources are gradually reduced along with the continuous exploitation, it is becoming more and more important to exploit the heavy oil reservoir efficiently. As an effective method for heavy oil reservoir, steam drive can get better the exploitation effect. The exploitation of heavy oil reservoir has the characteristics of high investment, high cost and high risk of economy. In order to fully tap the heavy oil thermal recovery potential, to achieve the optimal combination of heavy oil recovery, to make heavy oil recovery more scientific, optimization, the research on the steam flooding development effectiveness prediction and parameter optimization design method has a very important theoretical value and application value.Intelligent computation method is a new technique that has been developed in recent years. It has strong ability to deal with large scale complex nonlinear dynamic systems. Therefore, by utilizing advanced technology of the intelligent calculation method, integrating closely with the steam flooding reservoir actual condition, we research and develop a steam flooding development effectiveness prediction and parameter optimization method based on computational intelligence. In this method, controllable parameters can automatically be adapted to the uncontrollable reservoir properties, the optimum injection production parameters of steam flooding can be obtained, the optimum injection production program can be provided to guide the reasonable compilation of development scheme, realizing scientific management in deep heavy oil steam flooding, completing three aspects harmonization, namely reservoir geology, injection production, on-site management, guiding steam flooding efficient operation, reaching the maximum improvement of the recovery rate.Aiming at some problems appeared in the standard particle swarm optimization algorithm and artificial bee colony algorithm, the improved particle swarm optimization algorithm and the improved artificial bee colony algorithm are proposed in this paper on the basis of in-depth analysis and study of the particle swarm optimization algorithm and artificial bee colony algorithm, and the application strategy of the improved algorithms in the steam flooding development effectiveness prediction model and the parametric optimization technique is studied.The steam flooding development effectiveness prediction method based on intelligent computation is researched and developed in this paper. Based on BP neural network, least squares support vector machine(LSSVM), adaptive mutation chaotic artificial bee colony(AMCABC) algorithm and D-S evidence theory, the BP artificial network prediction model of steam flooding development effectiveness based on AMCABC-BP, the LSSVM prediction model of steam flooding development effectiveness based on AMCABC-LSSVM and the combined prediction model of steam flooding development effectiveness based on D-S evidence theory are established respectively. Taking Liaohe Oilfield as an example, three models and their corresponding learning algorithms are analyzed and simulated in this paper. The results show that the prediction accuracy of the combined prediction model of seam flooding development effectiveness based on D-S evidence theory is higher than other models, and this method is feasible and effective.Aiming at the problems existed in the selection of steam flooding injection parameters, the optimization method of steam flooding injection based on intelligent calculation is developed. By combining steam flooding analytical model and the combined prediction model of steam flooding development effectiveness based on D-S evidence theory, and studying deeply steam flooding injection optimization method, we proposed the single objective steam flooding injection scheme optimization method based on stochastic perturbation particle swarm optimization algorithm, the multi-objective steam flooding injection scheme optimization method based on adaptive dynamic regrouping particle swarm optimization algorithm. The results of an example show that it is feasible by applying intelligence computational methods to steam flooding injection scheme optimization. And it can obtain better results than non intelligent optimization methods. After being compared the two methods, the steam flooding injection scheme optimization method based on multi-objective is more reasonable and effective than the steam flooding injection scheme optimization method based on single objective.A novel steam flooding technology, which change steam injection rate through the way of periodic oscillatory, is proposed. The steam flooding injection scheme optimization model based on oscillating steam injection rate is established. And the model is solved by adopting adaptive mutation chaotic artificial bee colony(AMCABC) algorithm, and the optimum steam flooding injection scheme based on oscillating steam injection rate is obtained. Taking the Qi 40 block of Liaohe Oilfield as an example, the optimization method is verified, and the steam flooding injection scheme optimization method based on oscillating steam injection rate is superior to the other methods. The dynamic optimization and adjustment of steam flooding can be carried out in real time by the steam injection method based on oscillation and variable speed. And it may realize the efficient operation of the steam flooding under the condition of only changing the steam injection process and no additional increase of any investment.