Studying the lithology identification method from well logs based on DE-SVM

Identify the rock lithology has important meaning for estimating the reserve of petroleum, adopting proper drilling technology and improving recovery. The lithology identification from well log based on DE-SVM was proposed and studied. After digitization and collection the data of the well logs and cores observation results, the mapping model between well logs and strata lithology is established by support vector machine (SVM), so the strata lithology of wells without rock cores can be automatically gotten by well logs. Because the penal factor c and kernel parameter sigma affect the identification accuracy evidently, the global optimization arithmetic-difference evolutionary (DE) is coupled with SVM to optimize above parameters in order to improve the performance of SVM model. The model theory and algorithm are discussed as well as the true example is calculated, it is stated that the proposed method is feasible and can get satisfied results.