Time-invariant and time-varying filters versus neural approach applied to DC component estimation in control algorithms of active power filters

摘要

This paper presents an application of digital filters and neural networks to the extraction of a DC signal component. This problem arises, among others, in control of active power filters (APF) used for power quality improvement. Solutions to the basic problem of DC component estimation are well-known and so the difficulty of the task comes rather from the required minimization of the calculation time. It should ensure fast reaction of the control system to load changes. As a result, lower value of the current total harmonic distortion coefficient (THD) and better efficiency of the APF can be obtained.The paper includes propositions of both time-varying and neural filters as well as comparison with the classical approach to the DC component estimation based on infinite impulse response (IIR) low-pass filters. The results obtained by simulations have been presented. The future work will be devoted to digital signal processor implementation of APF control algorithms based on the best solution.