Clustering multivariate time-series data

A new methodology for clustering multivariate time-series data is proposed. The new methodology is based on calculating the degree of similarity between multivariate time-series datasets using two similarity factors. One similarity factor is based on principal component analysis and the angles between the principal component subspaces while the other is based on the Mahalanobis distance between the datasets. The standard K-means clustering algorithm is modified to cluster multivariate time-series datasets using similarity factors. Simulation data from two nonlinear dynamic systems: a batch fermentation and a continuous exothermic chemical reactor, are clustered to demonstrate the effectiveness of the proposed technique. Comparisons with existing clustering methods show several advantages of the proposed method. Copyright (c) 2006 John Wiley & Sons, Ltd.