Pattern recognition on electrochemical noise of an epoxy coating/metal system under marine alternating hydrostatic pressure

摘要

In recent several decades, electrochemical noise (EN) technique has received a growing interest for its non-disturbance and sensitivity to corrosion monitoring. However, choosing an effective method to analyse EN records limits the application of this technique. The previous study on the failure behaviour of coating/metal system under marine alternating hydrostatic pressure (AHP) in our group indicates that electrochemical signals generated by the corrosion of metal can reflect the protective properties of the organic coating. In this research, the EN measurement of an epoxy coating/metal system was performed under AHP, and the pattern recognition (PR) method was utilized for EN analysis to obtain the related corrosion information and identify the different states of the system. The PR procedures, including principal component analysis (PCA), hierarchical agglomerative cluster analysis (HACA) and linear discriminant analysis (LDA), were applied for establishing an evaluation model for the EN records of the coating under AHP. Firstly, according to the PCA results, four statistical parameters were chosen as the descriptors for clustering. Then, using the selected parameters as variables, the cases from different corrosion states were classified by the HACA to five clusters (Fig.1). Based on the cluster results, the corrosion states of the ungrouped data points from the similar failure processes can be distinguished according to the established discriminant functions.