Skip to Main Content
A novel approach based on support vector regression is proposed to establish a model for prediction of the corrosion rate of the steel. Under different seawater environment, the dataset can be identified into natural subgroups by clustering algorithm, but, in the real world the minor prototypes may be within a small, dense region located at a relatively large distance from any of the major cluster centers, which degrades the prediction performance. In this paper, we present SVR ensembles to address the minor prototypes problem and the combination strategy of hierarchical SVR is investigated. Our experiment results show that the generalization ability of SVR ensembles model consistently surpasses that of SVR by applying the test samples, and indicate that SVR ensembles may be a promising and practical methodology to monitor the seawater corrosion rate of steel.