Hourly Power Consumption Forecasting Using RobustSTL and TCN

Power consumption forecasting is a crucial need for power management to achieve sustainable energy. The demand increasing over time, while the of possesses challenges with nonlinearity patterns and various noise in datasets. To this end, paper proposes RobustSTL temporal convolutional network (TCN) model forecast hourly consumption. Through RobustSTL, instead standard STL, decomposition method can extract time series data despite containing dynamic patterns, noise, burstiness. trend, seasonality, remainder components obtained from operation enhance prediction accuracy by providing significant information dataset. These are then used as input TCN applying deep learning forecasting. employing dilated causal convolutions residual blocks long-term outperforms recurrent networks studies. assess proposed model, conducts comparison experiment between counterpart models. result shows that grasp rules historical related Our overcomes schemes MAPE, MAE, RMSE metrics. Additionally, obtains best results precision, recall, F1-score values. also indicates predicted fit pattern actual data.