State-space modal representations for decomposition of multivariate non-stationary signals

Abstract This work introduces a parametric modal decomposition method for multivariate non-stationary signals based on block-diagonal time-dependent state space representation and Kalman filtering/smoothing. Each second-order block is constructed with the real imaginary parts of each mode instantaneous eigenvalues, thus represents single oscillatory component. The identification state/parameter trajectories hyperparameters, constituted by mixing matrix, state, parameter noise covariances, initial conditions, accomplished tailored Expectation-Maximization algorithm. methodology evaluated in numerical example, concerning signal three components, featuring crossings vanishing amplitudes. Codes examples are available https://github.com/ldavendanov/NS-modal-decomposition .