Context-aware Hosting Capacity Analysis and Forecasting of Distribution Networks with Distributed Energy Resources

In modern power systems, integrating distributed energy resources (DER) such as solar photovoltaic systems, electric vehicles, and energy storage systems necessitates accurately forecasting network hosting capacity and estimating network parameters. Traditional deep learning models frequently fail to consider the dynamic and temporal relationships of network and DER variables on contextual factors, resulting in inaccurate predictions of network hosting capacity. This paper introduces the Context-aware Hosting Capacity Forecasting Model (Ca-HCFM), which enhances the forget gate output with contextual inputs, effectively capturing the dynamic nature of the network and DER variables and improving the accuracy of hosting capacity forecasts. The effectiveness of $\mathbf{C a}-\mathrm{HCFM}$ is tested against various standard models, including Support Vector Regression (SVR), Long-Short Term Memory (LSTM), Convolution LSTM (ConvLSTM), Bidirectional LSTM (BiLSTM), Stack LSTM (StackLSTM), and Gated Recurrent Unit (GRU). The results demonstrate that Ca-HCFM outperforms these baseline models, offering engineers and network operators a robust tool for the sustainable and efficient integration of DER into distribution networks.