Scholarlink

With the increasing popularity of distributed power generation, the distributed resources on the power system user side are random, autonomous and private, which makes it impossible to upload the operating power data to the upper management organization in time and accurately, which affects the optimal dispatching construction of the distribution network. This paper proposes a model based on virtual aggregation for similar situations. We design some types of aggregation model according to resource response characteristics. We apply the Multivariate Variational Model Decomposition-Fast Independent Component Analysis (MVMD-FastICA) method. This is used to process the parameter of bus power to get the power curves of different aggregations. In addition, we use a neural network to screen meteorological conditions, dates and electricity prices that affect the power generation of distributed resources. We develop a forecasting model using the advanced short-term neural network, which optimizes the sparrow search algorithm promotes active power forecasting of virtual aggregation. And ensures accurate and forward-looking prediction.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Power Prediction of Distributed Resource Aggregation Model Based on Optimized Neural Network

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Yi Zhao,
  • Xiaoming Zhang,
  • Yuhang Sun,
  • Wenbin Cao,
  • Ming Yu,
  • Xiaoyi Qian

摘要

With the increasing popularity of distributed power generation, the distributed resources on the power system user side are random, autonomous and private, which makes it impossible to upload the operating power data to the upper management organization in time and accurately, which affects the optimal dispatching construction of the distribution network. This paper proposes a model based on virtual aggregation for similar situations. We design some types of aggregation model according to resource response characteristics. We apply the Multivariate Variational Model Decomposition-Fast Independent Component Analysis (MVMD-FastICA) method. This is used to process the parameter of bus power to get the power curves of different aggregations. In addition, we use a neural network to screen meteorological conditions, dates and electricity prices that affect the power generation of distributed resources. We develop a forecasting model using the advanced short-term neural network, which optimizes the sparrow search algorithm promotes active power forecasting of virtual aggregation. And ensures accurate and forward-looking prediction.