| Summary: | 碩士 === 國立中正大學 === 電機工程研究所 === 107 === This thesis applies the taguchi fractional order particle swarm optimization (TFPSO) with a 2kW series buck-boost converter and TI control circuit, which is self-developed and has functions of buck and boost, as the maximum power tracker (MPPT) of the solar photovoltaic system combine with solar power prediction. No matter under ideal environmental conditions or partial shading condition(PSC), the converter can operate at maximum power point. We train parameters to be the best for MPPT on computer simulation by using Taguchi method. To verify its performance, we conducted experiment base on single- peak power curve, double-peak power curve, triple-peak power curve, quadruple-peak power curve, insolation variations, and temperature variations. Results show that the proposed TFPSO has better performance then FPSO.
Considering that 2kW polycrystalline solar photovoltaic panels are prone to aging problems, Therefore, the use of convolutional neural networks (CNN) for solar power prediction, and Compare and analyze the ideal power and predicted power.
Keywords：Taguchi fractional order particle swarm optimization, CNN, MPPT
|
|---|
