Optimizing short-term load forecasting performance is a challenge due to the non-linearity and randomness of electrical load, as well as the variability of system operating patterns. Existing methods often fail to consider how to effectively combine their complementary advantages and fail to fully capture the internal information in the load sequence, leading to a decrease in accuracy. To achieve accurate and efficient short-term load forecasting, this study proposes a novel power grid load forecasting model that integrates Convolutional Neural Networks (CNNs), Long Short-Term Memory Networks (LSTMs), Multi-Head Self-Attention Mechanism (MHSA), Global Attention Mechanism (GAM), and Channel Attention Mechanism (CAM) to achieve efficient and precise short-term load forecasting. This model aims to address the issue in traditional methods where complex temporal features and important information in power grid load data are not fully captured. Firstly, the CNN module is used to extract high-dimensional spatial features from the load data, and a pooling layer is applied to reduce dimensionality while retaining key information. Then, the Multi-Head Self-Attention mechanism is employed to model the long-range dependencies of the sequence data, enhancing the ability to extract temporal features. Next, the LSTM layer further captures the time dependencies in the load sequence. Subsequently, the Global Attention mechanism helps the model focus more on the most relevant parts of the input sequence, improving the model’s performance and generalization ability. The Channel Attention module is then applied to weight different feature channels, highlighting important information and reducing redundancy. Finally, the flattened output layer produces the forecast results. Experimental validation shows that the proposed CNN-MHSA-LSTM-GAM-CAM model outperforms existing mainstream methods in terms of load forecasting accuracy, providing effective support for the optimized scheduling of smart grids.

Accurate urban $P{M}_{2.5}$ forecasting serves a crucial function in air pollution warning and human health monitoring. Recently, deep learning techniques have been widely employed for urban $P{M}_{2.5}$ forecasting. Unfortunately, two problems exist: (1) Most techniques are focused on training and prediction on a central cloud. As the number of monitoring sites grows and the data explodes, handling a large amount of data on the central cloud can cause tremendous computational pressures and increase the risk of data leakages. (2) Existing methods lack an adaptive layer to capture the varying impacts of different external factors (e.g., weather conditions, temperature, and wind speed). In this paper, a federated deep learning network (FedDeep) is developed for edge-assisted multi-urban $P{M}_{2.5}$ forecasting. First, we assign each urban region to an edge cloud server (ECS). An external spatio-temporal network (ESTNet) is then deployed on each ECS. Data from different urban regions are uploaded to the corresponding ECS for training, which avoids processing all the data on the central cloud and effectively alleviates computational pressure and data leakage issues. Second, in ESTNet, we develop a gating fusion layer to adaptively fuse external factors to improve prediction accuracy. Finally, we adopted $P{M}_{2.5}$ data collected from air quality monitoring sites in 13 prefecture-level cities, Jiangsu Province for validation. The experimental results proved that FedDeep outperformed the advanced baselines in terms of prediction accuracy and model efficiency.