Sunflower-based butterfly optimization algorithm with enhanced RNN for the harmonics elimination in multilevel inverter
Abstract Multilevel inversion describes a power conversion technique that reduces Total Harmonic Distortion (THD) by gradually increasing the output voltage and approaching a sine wave. The fundamental goal of Multi Level Inverters (MLIs) is to produce an approximate sinusoidal voltage from many lev...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Springer
2025-07-01
|
| Series: | Discover Applied Sciences |
| Subjects: | |
| Online Access: | https://doi.org/10.1007/s42452-025-07475-3 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Abstract Multilevel inversion describes a power conversion technique that reduces Total Harmonic Distortion (THD) by gradually increasing the output voltage and approaching a sine wave. The fundamental goal of Multi Level Inverters (MLIs) is to produce an approximate sinusoidal voltage from many levels of dc voltages, which are typically supplied from capacitor voltage sources that convert DC input voltage to AC output voltage. A key goal is to obtain a pure sinusoidal waveform at the output of the Multi Level Inverter (MLI). In a cascaded MLI, the Selective Harmonic Elimination (SHE) and Pulse Width Modulation (PWM) approach is employed to mitigate lower harmonics by maintaining the needed fundamental voltage. To determine Switching Angles (SA), an objective function is generated from the SHE problem. In this paper, the Sunflower based– Butterfly Optimization Algorithm (SF-BOA) is presented as a method for evaluating transcendental nonlinear equations using an MLI in a SHE approaches. SF-BOA’s optimized switching angle is used for 11-level three-phase PWM control using the Cascaded H Bridge architecture for harmonic reduction of the entire modulation index. Although Artificial Intelligence (AI) systems can effectively solve a non-linear transcendental equation, their time consumption together with the convergence capability differs. Enhanced Recurrent Neural Network (ERNN) shows a kind of recurrent neural network in which the hidden neurons are tweaked by SF-BOA with the goal of minimizing THD. As per the simulation data, the SF-BOA approach is much appropriate and suitable than other compared algorithms like Harris Hawks Optimization (HHO), Whale optimization algorithm, Marine Predator Algorithm (MPA), Multi Group Marine Predator Algorithm (MGMPA). |
|---|---|
| ISSN: | 3004-9261 |