Abstract: To improve the accuracy of the evaluation of the city buried gas pipeline external corrosion, the main factors affecting pipeline corrosion were identified, an evaluation index set was constructed, combined with the advantages of SVM and RS, a model was built for prediction of external corrosion of city buried gas pipeline. Specific steps were given as follows: sample data collection, sample data preprocessing, attribute reduction algorithm to filtering core index set, and SVM training data and verifying model. A certain city gas pipeline was taken as an example to verify the model. The results show that the results of predicting the corrosion of pipeline level by using RS-SVM model are consistent with the actual results, the traditional multivariant regression method for prediction of pipeline corrosion rate has a 14.1% average relative error, while RS-SVM prediction model of pipeline corrosion rate has a 7.9% average relative error, the accuracy of RS-SVM model is higher than that of the traditional method.

Key words: rough set(RS), support vector machine(SVM), buried gas pipeline, external corrosion, evaluation