Adaptive filters cascade applied to a frequency identification improvement problem

Abstract

Problem of frequency identification performance improvement for a single-tone sinusoidal signal is solved via construction of an adaptive filters cascade. The cascade consists of adaptive band-pass filters tuned by estimates of the frequency provided by a given identification algorithm. Stability of the cascade is studied and boundedness of trajectories is proven with Lyapunov analysis under certain assumption on identification algorithm. Numerical simulations are given illustrating improved identification performance for different identification algorithms.

An Adaptive-Observer-Based Robust Estimator of Multi-sinusoidal Signals

“…As said above, the case of overparametrization corresponds to a situation in which the number of frequency parameters of the observer, n, is larger than n e , the number of nonzero amplitude sinusoids with unique frequency forming the measured signal 1 (more details of overparametrization in the context of adaptive control can be found in [37]). In this case a minimal realization of the generator of the measured signals is a multi-harmonic oscillator of order 2n e , composed by the collection of exactly n e unique harmonic oscillators [38].…”

“…However, the accuracy drops in the case of time-varying frequencies. In this respect, a number of methods have been conceived to provide online detection of the time-varying amplitude/frequency (see, for example, [1], [2], [3], [4] and references cited therein). Among them, it is worth to recall the adaptive notch-filtering method (ANF) (see [5], [6]) and the Phase-Locked-Loop (PLL) (see, for instance [7], [8]) for their popularity in power and electrical systems, though they are applicable for a single sinusoid only.…”

Improved Transients in Multiple Frequencies Estimation via Dynamic Regressor Extension and Mixing

“…On the other hand, it is well-known that many gradient-based estimation methods can exhibit poor transients even for relatively small number of estimated parameters, the transients can oscillate or even display a peaking phenomena. A method to increase frequency estimation performance with adaptive band-pass filters was proposed in [11] but for a single frequency case only. Thus, the problem of performance improvement for multiple frequencies estimation remains open.…”

Editorial for the special issue on recent advances in adaptive methods for frequency estimation with applications

“…The problem of frequency identification performance improvement for a single-tone sinusoidal signal is solved via construction of an adaptive filters cascade in [7]. The cascade consists of adaptive band-pass filters tuned by estimates of the frequency provided by a given identification algorithm.…”