基于量纲分析的优化神经网络模型预测GAGD非混相开发油藏采收率

An optimized neural network prediction model for gas assisted gravity drainage recovery based on dimensional analysis

与传统的注气方式相比,气体辅助重力驱作为新兴发展起来的提高采收率手段以其独特的驱替优势越来越受到国内各大油田的重视。目前国内外有关GAGD非混相开发油藏采收率的预测模型有很多,但模型大多是简单的非线性关系,普遍存在预测精度差的问题。近年来机器学习作为一种新兴手段已经广泛的用于石油工程行业,其中人工神经网络已成为处理复杂非线性回归问题最具潜力的方法。本文基于量纲分析,提出了一种可以有效预测GAGD非混相开发油藏采收率的人工神经网络模型。针对其他文献中鲜有考虑油藏倾角的问题,对量纲参数之一的邦德数利用油藏倾角进行了修正。在此基础上分别利用遗传算法和粒子群算法对模型参数进行优化,得到预测精度最高的优化模型。测试结果表明优化后的预测模型对于数值模拟、物理模拟以及实际油田的采收率预测精度均高于常规的非线性函数预测模型。

Compared with the traditional gas injection method,gas-assisted gravity drainage is an emerging development of enhanced oil recovery.It has attracted more and more attention from domestic oilfields due to its unique displacement advantages.At present,there are many prediction models for the recovery of GAGD immiscible reservoirs at home and abroad,but most of the models are simple nonlinear relationships and their prediction accuracy is poor.In recent years,machine learning has been widely used in the petroleum engineering industry and artificial neural networks have become the most potential method for dealing with complex nonlinear regression problems.Based on dimensional analysis,this paper proposes an artificial neural network model that can effectively predict the recovery of GAGD immiscible development reservoirs.In view of the fact that the dip angle of the reservoir is rarely considered in other studies,the Bond number of one of the dimensional parameters is corrected by the reservoir inclination.On this basis,a genetic algorithm and a particle swarm optimization algorithm are used to optimize the model parameters,and an optimization model with the highest prediction accuracy is obtained.By comparing this with the prediction results of other nonlinear regression models proposed in the literature,it is found that the optimized neural network model has higher precision for numerical simulation,physical simulation and actual oilfield recovery.