A Unified Artificial Neural Network Architecture for Active Power Filters

Abstract

In this paper, an efficient and reliable neural active power filter (APF) to estimate and compensate for harmonic distortions from an AC line is proposed. The proposed filter is completely based on Adaline neural networks which are organized in different independent blocks. We introduce a neural method based on Adalines for the online extraction of the voltage components to recover a balanced and equilibrated voltage system, and three different methods for harmonic filtering. These three methods efficiently separate the fundamental harmonic from the distortion harmonics of the measured currents. According to either the Instantaneous Power Theory or to the Fourier series analysis of the currents, each of these methods are based on a specific decomposition. The original decomposition of the currents or of the powers then allows defining the architecture and the inputs of Adaline neural networks. Different learning schemes are then used to control the inverter to inject elaborated reference currents in the power system. Results obtained by simulation and their real-time validation in experiments are presented to compare the compensation methods. By their learning capabilities, artificial neural networks are able to take into account time-varying parameters, and thus appreciably improve the performance of traditional compensating methods. The effectiveness of the algorithms is demonstrated in their application to harmonics compensation in power systems.Index Terms-Active power filter (APF), adaptive control, artificial neural networks (ANNs), harmonics, selective compensation, three-phase electric system.

Neural harmonic detection approaches for FPGA area efficient implementation

“…3, the three-monophase method works directly with the distorted current i L (t) and identifies fundamental component i Lf (t), and thus the harmonics i Lh (t) for each phase with two Adalines. We can see in [7], the detailed equations characterizing the Adalines. For the first one, multiplying the load current by sinωt, we obtain the expression (3)     …”

“…Beyond their simplicity they cannot easily take noises and errors into account, and they need incompressible time-delay for convergence. The most powerful technique is with no doubt the Instantaneous Power Theory (IPT) [2], [5][6][7], proposed by H. Akagi. However, the original p-q formulation derived from this theory is not available for single phase compensation.…”

“…It is also reported in [10], that synchronous frame adaptive linear neural networks used to control a T-type APF achieves good performance with better dynamic response and lower Total Harmonic Distortions (THD) in source-side currents. For its part, Ould Abdeslam et al went further in the work discussed in [7] while proposing a unified I neural approach of the entire adaptive harmonics compensation system. The approach generalized the use of a single and simplified type of neural network called Adaline Neural Network (ANN) [11].…”

“…an improved version of the three-monophase method [7] and another one called Two-Phase Flow introduced in [16]. In section III, those two approaches are compared to three other methods according to their modeling structure, their harmonic compensation performance and hardware resource consumption during implementation on a medium size FPGA target.…”

Neural Adaptive Kalman Filter for Sensorless Vector Control of Induction Motor

“…It is an "intelligent" technique based on the algorithms of ANN. We propose to determine PI parameters by using ADALINE (ADAptive LInear NEuron), this method is motivated by the need of the simplicity and flexibility in ANN (it should adapt only one weight); the main advantage of this technique according to its Simplicity algorithmic comparing with other similar methods [18].…”