Inderscience Publishers

Neural network-based shunt active filter with direct current control for power quality conditioning

0
- By: , ,

Courtesy of Inderscience Publishers

In this paper, a novel phase locked loop less control algorithm for harmonic and reactive power compensation with power factor improvement using three-phase shunt active power filter under non-ideal mains voltage conditions is presented. The performance of APF mainly depends upon how quickly and accurately compensation signal are estimated. The artificial neural network trained with conventional compensator data, can deliver compensation signals more accurately at varied load condition. The non-sinusoidal mains voltage problem is resolved by using neural network-based control system which estimates the reference compensation currents by extracting fundamental frequency components. The compensation process is based on sensing source currents only, without sensing source voltage and load currents, which gives more simplicity in control with less number of sensors. The proposed compensation approach has been first evaluated in MATLAB/Simulink. The simulated results are validated experimentally by developing a laboratory prototype using a floating-point DSP TMS320F28335 to show the effectiveness of the controller.

Keywords: power quality, harmonics, active power filter, APF, PI controller, artificial neural network, hysteresis current controller, power electronics

Customer comments

No comments were found for Neural network-based shunt active filter with direct current control for power quality conditioning. Be the first to comment!