Prediction of Tank Bottom Corrosion Classification Based on FSVM

Abstract

The condition of tank bottom corrosion is the determinant factor affecting the safe operation of the tank. Considering the difficulty to shut down and carry out testing on tank floors, This article determines the characteristic parameters of tank bottom corrosion based on tank bottom magnetic flux leakage testing results and the fuzzy support vector machine is established based on the characteristic parameters of tank bottom corrosion. Through the prediction and analysis on 25 tanks in a depot, comparing with the result of magnetic flux leakage test, coincidence rate reaches 100%.The results show that this prediction technique is able to resolve the problem of the prediction of tank bottom corrosion rank, and can be applied in engineering applications.