论文部分内容阅读
低孔低渗油藏油水关系复杂,测井响应特征不明显,常规的人工经验性油水解释已不能满足实际开发需要。研究采用最小二乘支持向量机分类理论,选取多种相对独立的测井参数对低孔低渗储层流体性质识别分析,以工区试油已证实含油水类型的层位作为训练样本进行训练,建立不同流体性质储层的分类器相应支持向量机和分类面,通过已建立分类器的分类函数,可对待识别的层位进行识别分析。通过对工区的样本学习和预测,并与实际试油资料进行对比,符合率达到91.7%,从而表明,最小二乘支持向量机在油水识别中可获得良好的应用。
The relationship between oil and water in low porosity and low permeability reservoirs is complex and the logging response characteristics are not obvious. Conventional empirical oil and water interpretation can not meet the actual development needs. Based on least-squares support vector machine classification theory, a lot of relatively independent logging parameters were selected to identify the fluid properties of low-porosity and low-permeability reservoirs. Based on the test oil-bearing zones in the work area, The corresponding SVMs and classification surfaces of different classifiers of reservoirs are established, and the classification functions of the classifiers are established to identify and analyze the layers to be identified. Through the sample learning and prediction of work area and comparison with the actual test oil data, the coincidence rate reaches 91.7%, which shows that the least square support vector machine can be applied well in oil-water identification.