Accurately predicting multi-energy loads is essential for optimizing the dispatch and economic operation of integrated energy systems (IES). However, existing multi-energy load forecasting methods have two main limitations: (1) they fail to consider the complex correlations between multi-energy loads and auxiliary features; (2) single time-scale feature extraction methods can result in the loss of critical temporal feature information. Therefore, multi-energy load forecasting remains a challenging task. To overcome these limitations, this paper proposes a novel multi-energy load forecasting method based on a multi-scale spatio-temporal graph neural network (MS-STGNN). Specifically, the proposed continuous graph learning module quantifies the correlations between multi-energy loads and auxiliary features using an adjacency matrix, while the graph convolution module aggregates feature information among neighboring nodes through the same matrix to improve the correlations between multi-energy loads and auxiliary features. The model’s robustness is further enhanced by the feature attention module. Moreover, to mitigate temporal feature information loss, we develop a multi-scale convolution module that uses filters of various sizes to extract multi-dimensional temporal features of different time steps. Extensive experiments show that the MS-STGNN method has higher prediction accuracy and better generalization ability than existing methods on the IES dataset at the Tempe campus of Arizona State University. The code is publicly available at https://github.com/nuist-cs/MS-STGNN .

Accurate forecasting of multienergy loads is essential for designing, operating, scheduling, and managing integrated energy systems (IESs). Recent research suggests that transformer models have the potential to improve long-sequence predictions. However, existing transformer models often emphasize capturing temporal dependencies while neglecting crucial dependencies among different variables necessary for multienergy load forecasting. Moreover, transformer models encounter challenges related to quadratic time complexity and significant memory usage, which hinder their direct applicability to tasks involving long-sequence, multienergy load forecasting. To tackle these challenges, we propose a model called DTformer and apply it to the task of multihorizon, multienergy load forecasting in IES. Within DTformer, we employ patch embedding to convert the input multienergy load sequences into a 3-D vector array, preserving both temporal and variable information. Subsequently, we propose the temporal top windowed attention (TWA) module and the dual variable attention module to handle extended temporal dependencies and intervariable dependencies. Importantly, the computational complexity and memory requirements of the TWA model are regulated at a level of $O(N^\frac{4}{3})$ . Through extensive experimentation, we found that our DTformer surpasses baseline models in terms of performance using the IES dataset sourced from Arizona State University's Tempe campus.

Driven by economic incentives, illegal electricity consumers pose significant threats to the economic and security aspects of the power system by illicitly accessing or manipulating electrical resources. With the widespread adoption of Advanced Metering Infrastructure (AMI), researchers have turned to leveraging smart meter data for electricity theft detection. However, existing models rely on methods that model a single electricity load curve and cannot capture the temporal dependencies, periodicity, and underlying features between electricity consumption cycles. This paper proposes a dynamic generation algorithm to address these issues to construct a topological structure between periodic nodes, updated during training. Subsequently, residual graph convolution operations extract temporal and spatial dependencies among nodes. Additionally, to address the issue of model instability caused by scarce theft data, we employ the SMOTE (Synthetic Minority Over-sampling Technique) oversampling technique and enhance overall classification performance by modifying class weights in the loss function. We trained this network architecture on the real SGCC (State Grid Corporation of China) dataset, and the results demonstrate its superiority over other mainstream existing models.

Predicting photovoltaic (PV) power generation is a crucial task in the field of clean energy. Achieving high-accuracy PV power prediction requires addressing two challenges in current deep learning methods: (1) In photovoltaic power generation prediction, traditional deep learning methods often generate predictions for long sequences one by one, significantly impacting the efficiency of model predictions. As the scale of photovoltaic power stations expands and the demand for predictions increases, this sequential prediction approach may lead to slow prediction speeds, making it difficult to meet real-time prediction requirements. (2) Feature extraction is a crucial step in photovoltaic power generation prediction. However, traditional feature extraction methods often focus solely on surface features, and fail to capture the inherent relationships between various influencing factors in photovoltaic power generation data, such as light intensity, temperature, and more. To overcome these limitations, this paper proposes a mid-term PV power prediction model that combines Graph Convolutional Network (GCN) and Informer models. This fusion model leverages the multi-output capability of the Informer model to ensure the timely generation of predictions for long sequences. Additionally, it harnesses the feature extraction ability of the GCN model from nodes, utilizing graph convolutional modules to extract feature information from the ‘query’ and ‘key’ components within the attention mechanism. This approach provides more reliable feature information for mid-term PV power prediction, thereby ensuring the accuracy of long sequence predictions. Results demonstrate that the GCN–Informer model significantly reduces prediction errors while improving the precision of power generation forecasting compared to the original Informer model. Overall, this research enhances the prediction accuracy of PV power generation and contributes to advancing the field of clean energy.

Future transportation is advancing in the direction of intelligent transportation systems, where an essential part is vehicle and pedestrian detection. Due to the complex urban traffic environment, vehicles and pedestrians in road monitoring have different forms of occlusion problems, resulting in the missed detection of objects. We design an improved you only look once version 7 (YOLOv7) tiny algorithm for vehicle and pedestrian detection under occlusion, with the following four main improvements. In order to locate the object more accurately, $1 \times 1$ convolution and identity connection are added to the $3 \times 3$ convolution, and convolution reparameterization is used to enhance the inference speed of the network model. In view of the complex road background and more interference, the coordinate attention was added to the connection part of backbone and neck to enhance the network's capacity to detect the object and lessen interference from other targets. At the same time, before being sent to the detection head, global attention mechanism is added to improve the accuracy of model detection by capturing three-dimensional features. Considering the issue of imbalanced training samples, we propose focal complete intersection over union (CIOU) loss instead of CIOU loss to become the bounding box regression loss, so that the regression process attention to high-quality anchor boxes. Experiments show that the improved YOLOv7 tiny algorithm achieves 82.2% map @ 0.5 in pattern analysis, statistical modelling and computational learning visual object classes dataset, which is 2.8% higher than before the improvement. The performance of map @ 0.5:0.95 is 5.2% better than the previous improvement. The proposed improved algorithm can availably to detect partial occlusion objects.

With the ongoing advancement of electric power Internet of Things (IoT), traditional power inspection methods face challenges such as low efficiency and high risk. Unmanned aerial vehicles (UAVs) have emerged as a more efficient solution for inspecting power facilities due to their high maneuverability, excellent line-of-sight communication capabilities, and strong adaptability. However, UAVs typically grapple with limited computational power and energy resources, which constrain their effectiveness in handling computationally intensive and latency-sensitive inspection tasks. In response to this issue, we propose a UAV task offloading strategy based on deep reinforcement learning (DRL), which is designed for power inspection scenarios consisting of mobile edge computing (MEC) servers and multiple UAVs. Firstly, we propose an innovative UAV-Edge server collaborative computing architecture to fully exploit the mobility of UAVs and the high-performance computing capabilities of MEC servers. Secondly, we established a computational model concerning energy consumption and task processing latency in the UAV power inspection system, enhancing our understanding of the trade-offs involved in UAV offloading strategies. Finally, we formalize the task offloading problem as a multi-objective optimization issue and simultaneously model it as a Markov Decision Process (MDP). Subsequently, we proposed a task offloading algorithm based on a Deep Deterministic Policy Gradient (OTDDPG) to obtain the optimal task offloading strategy for UAVs. The simulation results demonstrated that this approach outperforms baseline methods with significant improvements in task processing latency and energy consumption.