A Strategy for Feature Extraction and Classification with its Application to Process Supervision

This paper proposes a strategy for feature extraction and classification of high dimensional noisy data. Firstly, the moving median filter is used for reducing the effect of noise and outliers in the measurement data. Then, the data is projected to a lower dimension feature space using radial basis function (RBF) network and polygonal line. It is integrated with learning vector quantization (LVQ) network for pattern classification. The method is applied for fault detection, identification and classification in a simulated continuous stirred-tank reactor (CSTR). The result shows that the proposed method is excellent for process supervision of non-linear systems.