Research on Staged Harmonic Prediction and Regulation Strategy Based on Deep Learning Model

Under the background of the country's vigorous development of clean energy, China has put forward the strategic decision of “carbon peak, carbon neutral”. Distributed photovoltaic (PV) power generation has gradually entered a high boom development stage, and after years of development, the proportion of PV power generation system is getting bigger and bigger, which also leads to a variety of power electronic devices are put into the micro grid use, which really improves the efficiency of the microgrid work at the same time, but also makes the harmonic problem become more and more serious. In order to reduce the harmonic problems caused by the stability of the distribution network is affected by the problem, this paper through the different time periods under the photovoltaic grid-connected harmonic problems generated in the study. Firstly, the required harmonics are extracted using IP-IQ harmonic detection method, secondly, the harmonics are predicted and analyzed at different times by using different deep learning modules. And finally the current harmonics filtering in the grid is accomplished by eliminating the harmonics in the grid through an active filter (APF).