Convolutional Neural Network-based harmonic mitigation technique for an adaptive shunt active power filter
Owing to the use of nonlinear loads in the distribution side, there are power quality issues such as voltage swell/sag, harmonics, flickers, voltage imbalance, and outage. The harmonics in power system affect the quality of power and hence a suitable methodology is vital to mitigate the harmonics an...
Guardado en:
| Autores principales: | , , , |
|---|---|
| Formato: | article |
| Lenguaje: | EN |
| Publicado: |
Taylor & Francis Group
2021
|
| Materias: | |
| Acceso en línea: | https://doaj.org/article/ed4b2f57321d4bcbad91e418e977f4fc |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| Sumario: | Owing to the use of nonlinear loads in the distribution side, there are power quality issues such as voltage swell/sag, harmonics, flickers, voltage imbalance, and outage. The harmonics in power system affect the quality of power and hence a suitable methodology is vital to mitigate the harmonics and compensation of reactive power. In this paper, CNN (Convolutional Neural Network)-based harmonic mitigation is performed. A 5-level cascaded H-bridge inverter is employed as a shunt active filter in which the reference current is generated by the SRF theory, incorporating CNN for harmonic extraction. The DC-link potential across capacitor is retained by means of ANN (Artificial Neural Network) controller whose behaviour is compared with a proportional controller as well as FLC. The gating pulse for the cascaded inverter is generated by means of PWM generator incorporated with Hysteresis Current Controller (HCC). By this control strategy, the harmonics in the current and voltage get mitigated; subsequently, the reactive power compensation is achieved with unity power factor. By implementing the five-level inverter, the THD and the settling time are minimized. The performance of the system is analysed using MATLAB for nonlinear load and the hardware is implemented with FPGA Spartan 6E. The THD of 0.93% is accomplished in simulation and 1.4% in the hardware execution. |
|---|