Comparative Analysis of Hybrid ML Models for Solar Power Forecasting in Bangladesh

The study deals with the evaluation of the hybrid machine learning model KNN-SVM for forecasting solar power in Bangladesh. Furthermore, the results are compared with that achieved from the individual models: K-Nearest Neighbor, Support Vector Machine, and Long Short-Term Memory. In this regard, the hybrid mode performed better, obtaining the minimum RMSE of 0.0066 both for HTGSR and HTPEG. While LSTM had the edge for sequence prediction, nonlinear patterns were handled by the KNN-SVM hybrid, where instance-based learning is balanced against margin optimization. This hybrid approach fits well with solar energy forecasting for improved energy management and integration of renewable energy into the national grid in Bangladesh for the purpose of sustainable energy planning.