Intelligent embedded system for decision support in pulsed eddy current corrosion detection using Extreme Learning Machine

Abstract

In this work an intelligent electronic system was proposed to generate, acquire and process Pulsed Eddy Currents (PEC) signals for corrosion detection in carbon steel pipes thermally insulated with composite coating. The system includes analog (excitation circuit and data acquisition) and digital (signal processing, feature extraction and decision support) sub-systems. The proposed signal processing chain comprises feature extraction using both discrete Fourier and Wavelet transforms, combined with information compaction by Principal Component Analysis (PCA) and decision support through intelligent classification techniques. Two neural network architectures were considered for classification, the traditional Multi-Layer Perceptron (MLP) and also the Extreme Learning Machine (ELM). The results obtained from thermally insulated carbon steel pipes subject to corrosion under isolation indicate the efficiency of the proposed method. Combining the ELM technique and data compaction it was possible to achieve a fast-training portable intelligent system for PEC evaluation and decision support.