Research on a new power system achievement recommendation method based on attention mechanism

With the increasing complexity of the power system, accurate prediction has become crucial for industry scheduling and operation. Traditional methods are inadequate in addressing nonlinear challenges, while deep learning combined with attention mechanisms provides higher accuracy. This article proposes a new performance recommendation method for power systems based on attention mechanism, which integrates convolutional neural network (CNN) and gated recurrent unit (GRU) to achieve short-term load forecasting. By introducing attention mechanism, the model can focus on key data features, extract the core part of time series information, reduce noise, and improve learning efficiency and interpretability. In addition, this article also explores the combination of Bayesian Personalized Sorting (BPR) and attention mechanism for the design and implementation of group recommendation algorithms, considering group project interaction, user project interaction, and group composition. The experimental results show that the proposed ANBGR (Attention based Neural Bayesian Group Recommendation) algorithm has significantly improved recommendation accuracy and normalized cumulative loss compared to other recommendation algorithms.