Hybrid LSTM-GWO Approach for Optimal Hosting Capacity of Solar PV in Distribution System

The integration of Solar Photovoltaic (SPV) systems in distribution systems is increasing rapidly due to rising environmental concerns and advancement in technology. However, with increase in penetration of SPV systems, certain negative impacts are observed in distribution systems. The Hosting Capacity (HC) assessment enables the SPV system interconnection without violating any operational limits. In this paper, a hybrid approach is proposed for SPV profile forecasting using Long Short-Term Memory (LSTM) model and optimizing HC of distribution systems with Grey Wolf Optimizer (GWO) metaheuristic technique. For accurate forecasting of solar irradiance profile, LSTM model can effectively capture the sequential patterns in solar irradiance. GWO is gradient-free technique which makes it suitable for distribution network problems with nonlinear, non-convex characteristics. The proposed methodology is applied on IEEE 33-bus system and the results suggests that the proposed LSTM model is fast and accurate in predicting solar irradiance profile. Further, it is observed that the GWO based optimal placement provides higher SPV HC than placement by choosing strategic locations.