Deep learning in renewable energy forecasting: A cross-dataset evaluation of temporal and spatial models

Unpredictability of renewable energy sources coupled with the complexity of those methods used for various purposes in this area calls for the development of robust methods such as deep learning (DL) models within the renewable energy domain. Given the nonlinear relationships among variables in renewable energy datasets, DL models are preferred over traditional machine learning (ML) models because they can effectively capture and model complex interactions between variables. This research aims to identify the factors responsible for the accuracy of DL techniques, such as sampling, stationarity, linearity, and hyperparameter optimization for different algorithms. The proposed DL framework compares various methods and alternative training/test ratios. Seven ML methods, such as long short-term memory (LSTM), stacked LSTM, convolutional neural network (CNN), CNN–LSTM, deep neural network (DNN), multilayer perceptron (MLP), and encoder–decoder (ED), were evaluated on two different datasets. The first dataset contains the weather and power generation data. It encompasses two distinct datasets, hourly energy demand data and hourly weather data in Spain, while the second dataset includes power output generated by the photovoltaic panels at 12 locations. This study deploys regularization approaches, including early stopping, neuron dropping, and L2 regularization, to reduce the overfitting problem associated with DL models. The LSTM and MLP models show superior performance. Their validation data exhibit exceptionally low root mean square error values. The results demonstrate that DL models consistently outperform traditional ARIMA models, with validation RMSE values ranging from 0.038 to 0.074, compared to ARIMA's much higher values ranging from 0.154 to 0.191. Specifically, stacked LSTM achieved the lowest RMSE of 0.037, while ED reached 0.043 at higher validation splits. In the second dataset, the regularized time-distributed MLP outperformed others with RMSE values as low as 0.121, confirming the effectiveness of regularization.