Hybrid Deep Learning Approach for Improver Solar Power Forecasting

Solar power plants are becoming more prevalent all over the world. The output of solar power plants is uncertain due to the environment. Predicting solar radiation is therefore an essential for these plants. So, this work proposes a convolution neural network and bi-direction long short-term memory based hybrid approach for accurate mid-term radiation prediction. Different previously proposed DL models are contrasted with the proposed hybrid DL model. Additionally, distribution errors like low dimension errors are taken into consideration when assessing the expected solar radiation's distribution. The results confirms the validity of the proposed hybrid approach and further improves the accuracy of individual deep learning models.