Application of Array Induction Logging and Support Vector Machine to Fluid Identification

Based on the data of high-definition array induction logging(HDIL),this paper first analyzes the effect of cross-plot method for fluid identification and points out the limitations of the conventional logging interpretation methods.Support Vector Machines with optimized parameters obtained by Genetic Algorithms(GA-SVM) is applied to the logging interpretation.Combined with HDIL and conventional logging data,a nonlinear mathematical model of fluid identification is established.By using the GA-SVM model,the correct recognition rate to the 45 training samples is 100% and correct recognition rate to the 21 unknown Samples is 90.48%.Compared with the conventional methods of interpretation,this method can improve identification accuracy of fluid properties.