Time-varying modelling of arbitrary non-stationary signals

摘要: A novel approach to time-varying signal modeling is presented reliably identify rational models in arbitrary nonstationary environments. It motivated by the shortcomings of both adaptation methods, which cannot handle nonstationarities, and description techniques, tend use unjustifiable assumptions on observed data. Only limited a priori knowledge about nonstationarity, namely, expected maximum rate change model parameters, necessary estimate these parameters online. The criterion considered constrained-least-squares cost functional incorporates with equal weight all instantaneous errors up time observation. An appropriate algorithm developed its performance discussed illustrate increased confidence estimation results from approach. >