•  
  •  
 

Keywords

direct current transmission project; converter station; convolutional neural network; particle swarm optimization‑bat algorithm; reactive power control

Abstract

As the scale of direct current (DC) transmission projects continues to expand, the alternating current (AC)-DC interconnection system has brought certain challenges to the transient voltage recovery capability of the power grid. In order to reduce frequent operations of tap changers of converter transformers caused by voltage fluctuations in converter stations, an optimization method for reactive power control of DC converter stations based on multi-channel fusion and multi-scale dynamic adaptive residual learning (MC-MSDARL) and particle swarm optimization-bat algorithm (PSO-BA) is proposed. Firstly, research is conducted on the transient overvoltage characteristics of the converter stations, and the impact of AC filters, reactive power compensation equipment, and phase-shifting cameras on transient overvoltage is analyzed. Then, the multi-scale dynamic adaptive residual convolution method is used to dynamically update the size of the convolution kernel, improve the model’s learning ability, map the relationship between the operating state of the DC system and voltage stability, and construct a transient voltage stability prediction model. Finally, an optimization model for reactive power control of DC converter stations is established to reduce voltage fluctuations and network losses, and the PSO-BA is used to solve the model. PSASP is employed to build a DC power grid for simulation verification, and the experimental results show that the proposed method improves the transient voltage stability capability and effectively reduces frequent operations of tap changers of converter transformers.

DOI

10.19781/j.issn.1673-9140.2024.06.005

First Page

43

Last Page

52

Download

Share

COinS