Neural-Networks and Synchronous Reference Frame Applied in the Harmonic Compensation with a Three-Phase Parallel Active Power Filter

Abstract

Abstract. This paper presents an alternative method based on artificial neural network, which is used to obtain the reference currents for harmonic current suppression and reactive power compensation in a shunt active power filter applied to three-phase four-wire system. The neural network consists of a multilayer perceptron, which is trained to estimate the peak amplitude of the load fundamental component currents. The performance of the proposed artificial neural network is evaluated by computer simulation, and it is compared with the method based on synchronous reference frame.

Improving the SRF method to compensate low-order harmonics under nonsinusoidal network voltages

“…This method limited the compensated current THD to 2.07% in the best case. In [11], the performance of the SRF method was improved by neural networks under sinusoidal load terminal voltages, and the compensated current THD and transient response were limited to 1.45% and 1.5 cycles, respectively. In [12], a conventional LPF was replaced with a wavelet-based LPF in the p-q method to improve the time response of the APF.…”

Performance improvement of shunt active power filter by a new wavelet-based low-pass filter: Theory and implementation

“…In this paper, the compensated current THD has been limited to 2.7% in the best case, and the transient response time is equal to two cycles. In Campanho et al (2012), performance of the d-q method has been improved by neural networks under sinusoidal load terminal voltages, and the THD of the compensated current and transient response have been limited to 1.45% and 1.5 cycles, respectively. In Patidarl and Sing (2008), a fast-acting l/z controller for the shunt active power filter (SAPF) is proposed.…”