48 Research on Support Vector Machine Method Using in Gas Well Casing Reliability and Life Analysis Available to Purchase

Eight casing failure forms and 32 influencing factors of casing failure in oil and gas wells are given in this paper. According to the quantitative analysis of the influence degree and the occurrence probability for every influencing factor, the Borda counts of them for eight failure forms are obtained by using Borda method. The risk performance indexes of eight failure forms are derived on the basis of the Borda matrix. The casing life analysis model is established, which is based on the support vector machine method (SVM). The calculating software of casing string life analysis is compiled by Visual Basic computer language basing on the established model. In the calculating software, eight risk performance indexes and the casing residual life are used as input vectors and output vectors respectively. The ideal model parameters of the calculating software are determined by small samples training, which from nineteen casing damage wells. Finally, the software of casing string life analysis in complex well status is completed with the SVM model as predictor. The residual lives of 60 casing damage wells are predicted by the software, whose accuracy and reliability are verified by the comparison between predictive values and actual values. The comparison results show that the analysis model has higher prediction accuracy for long-lived gas well casing.