ABSTRACT
Accurate prediction of solar irradiance holds significant value for renewable energy usage and power grid management. However, traditional forecasting methods often overlook the time dependence of solar irradiance sequences and the varying importance of different influencing factors. To address this issue, this study proposes a dual-path information fusion and twin attention-driven solar irradiance forecasting model. The proposed framework comprises three components: a residual attention temporal convolution block (RACB), a dual-path information fusion module (DIFM), and a twin self-attention module (TSAM). These components collectively enhance the performance of multi-step solar irradiance forecasting. First, the RACB is designed to enable the network to adaptively learn important features while suppressing irrelevant ones. Second, the DIFM is implemented to reinforce the model's robustness against input data variations and integrate multi-scale features. Lastly, the TSAM is introduced to extract long-term temporal dependencies from the sequence and facilitate multi-step prediction. In the solar irradiance forecasting experiments, the proposed model is compared with six benchmark models across four datasets. In the one-step predictions, the average performance metrics RMSE, MAE, and MAPE of the four datasets decreased within the ranges of 0.463-2.390 W/m2, 0.439-2.005 W/m2, and 1.3-9.2%, respectively. Additionally, the average R2 value across the four datasets increased by 0.008 to 0.059. The experimental results indicate that the model proposed in this study exhibits enhanced accuracy and robustness in predictive performance, making it a reliable alternative for solar irradiance forecasting.
ABSTRACT
Fault alarm time lag is one of the difficulties in fault diagnosis of wind turbine generators (WTGs), and the existing methods are insufficient to achieve accurate and rapid fault diagnosis of WTGs, and the operation and maintenance costs of WTGs are too high. To invent a new method for fast and accurate fault diagnosis of WTGs, this study constructs a stacking integration model based on the machine learning algorithms light gradient boosting machine (LightGBM), extreme gradient boosting (XGBoost), and stochastic gradient descent regressor (SGDRegressor) using publicly available datasets from Energias De Portugal (EDP). This model is automatically tuned for hyperparameters during training using Bayesian tuning, and the coefficient of determination (R2) and root mean square error (RMSE) were used to evaluate the model to determine its applicability and accuracy. The fitted residuals of the test set were calculated, the Pauta criterion (3σ) and the temporal sliding window were applied, and a final adaptive threshold method for accurate fault diagnosis and alarming was created. The model validation results show that the adaptive threshold method proposed in this study is better than the fixed threshold for diagnosis, and the alarm times for the GENERATOR fault type, GENERATOR_BEARING fault type, and TRANSFORMER fault type are 1.5 h, 5.8 h, and 3 h earlier, respectively.