Water quality event detection based on Multivariate empirical mode decomposition

Water quality event detection is critically important to national security and people's health. And Empirical Mode Decomposition (EMD) is a common used method to analyze nonlinear and non-stationary signals. However the standard EMD can just deal with one variation and certain information of the original signals would be missed. In this paper, we proposed a multivariate water quality event detection approach for detecting accidental or intentional water contamination events, thus to improve the detection rate. The proposed approach is based on multivariate empirical mode decomposition (MEMD), which is a novel algorithm to analyze nonlinear and non-stationary signals. A sequence of n-dimension intrinsic mode functions (IMFs)is obtained by the process of MEMD, then the Mahalanobis Distance is used to perform information fusion and the normalized instantaneous energy (NIE) is used to perform anomaly detection.