基于支持向量机对优势渗流通道识别的研究

Research on Identification of Dominant Seepage Channel Based on Support Vector Machine

长期注水驱动开发,萨中开发区部分油井储层内的孔隙结构发生非均质性变化,形成优势渗流通道,造成注入水长期沿着优势渗流通道无效循环,导致油田生产成本不断攀升,开发效益持续下降,给油田开发剩余油带来巨大压力。优势渗流通道形成后,与同岩性油层相比,具体表现为自然电位幅度升高、微电极曲线幅度下降、深浅侧向曲线幅度严重下降等响应特征,提取测井曲线特征,并对特征进行预处理,完成学习样本、测试样本的准备。由于粒子群算法具有易实现、收敛快、并行性高等优点,采用粒子群算法优化的支持向量机对优势渗流通道识别进行建模,进一步提高支持向量机算法的泛化能力。通过实验结果可知,粒子群支持向量机具有较强的泛化能力,可较为准确地识别优势渗流通道。

Long-term water injection drives development.The pore structure in some oil well reservoirs in the Sazhong Development Zone has changed non-homogeneously,forming a dominant seepage channel,resulting in the ineffective circulation of injected water along the dominant seepage channels for a long time,which leads to the continuous increase of production costs and the continuous decline of development benefits,and brings huge pressure to the remaining oil in the development of the oilfield.After the formation of the dominant seepage channel,compared with the same lithology oil layer,the specific manifestations are response characteristics such as the increase in spontaneous potential amplitude,the decrease in the amplitude of the microelectrode curve,and the severe decrease in the amplitude of the deep and shallow lateral curves.The well logging curve features are extracted and pretreated to complete the preparation of learning samples and test samples.Because the particle swarm algorithm has the advantages of easy implementation,fast convergence and high parallelism,the support vector machine optimized by particle swarm optimization is used to model the identification of dominant seepage channels,and the generalization ability of the support vector machine algorithm is further improved.The experiment shows that the particle swarm support vector machine has strong generalization and can identify the dominant seepage channel more accurately.