粒子群算法在砂砾岩体岩性识别中的应用

Application of Particle Swarm Optimization to Glutenite Lithology Identification

针对砂砾岩地层岩性变化大、非均质性强、常规测井曲线影响因素多、砂砾岩地层地质特征与测井曲线呈现非线性关系等特点,采用支持向量机方法(SVM)对地层岩性进行划分。选用粒子群算法对支持向量机参数进行优化,得到岩性识别模型;根据模型对研究区的30多口井的岩性进行划分,取得良好的地质应用效果。

The glutenite stratum has many characteristics, such as the large lithology variation, strong heterogeneity, many influencing factors of the conventional logging curves, nonlinear relation of the glutenite formation geologic feature sand logging curves and so on. We divide the formation lithology with the support vector machine （SVM） method. The SVM parameters are optimized by the particle swarm optimization （PSO）, then the lithology identification model is achieved. The lithology of more than 30 wells in the studied area is divided by this model with a better result.