The increasing integration of electric vehicles (EVs) and photovoltaic (PV) generation introduces significant uncertainty into modern distribution grids. This paper presents a dual-stage, AI-driven framework for resilient energy management that combines machine learning-based forecasting with generative modelling for scenario-based control. The first stage uses a hybrid forecasting architecture: long short-term memory (LSTM) and convolutional LSTM (ConvLSTM) models for EV demand prediction and eXtreme gradient boosting (XGBoost) for PV generation forecasting. The second stage employs a generative adversarial network (GAN) to produce realistic EV and PV scenarios, capturing both typical variability and a wide range of operating conditions. The framework is validated on a modified IEEE 33-bus distribution system with integrated EV charging and stationary storage. Results show that the dual-model forecasting approach achieves high accuracy across diverse temporal patterns, while GAN-based scenario generation improves the adaptability of control decisions. Scenario-based optimisation enhances performance under uncertainty, especially at high-variance nodes, and offers greater flexibility than deterministic control in balancing energy cost and demand satisfaction.

Renewable energy sources (RES) such as solar photovoltaic (PV) are employed in electric vehicle (EV) charging stations to promote sustainable transportation and reduce the load on the AC grid. This paper introduces a modular power converter topology that interfaces with solar PV, EV batteries, and the AC grid. The proposed topology supports EV battery charging in two modes: DC/DC (from PV modules) and AC/DC (from the AC grid). Additionally, during peak hours, EV batteries can function as energy storage units (ESUs) in vehicle-to-grid (V2G) mode. A central element of the proposed topology is the submodule (SM), which is implemented as a single-stage isolated bidirectional Cuk-based converter. This converter is chosen for its exceptional features, including high efficiency and inherent power factor correction (PFC) due to continuous input and output currents. The paper first details the operating modes of the Cuk-based SM. To improve performance, an extra switching state is added in the AC/DC (rectifier) mode, enabling the second-order harmonic from the AC grid to be stored within the Cuk SM instead of being passed to the battery. Additionally, power losses at the SM level are examined, and the effectiveness of the proposed topology is demonstrated through experimental results.

This article investigates an energy management problem in the microgrid by scheduling heating ventilation air conditioning (HVAC) and battery energy storage system (BESS) with a distributed algorithm. A multilayer energy management architecture is presented at a system-level to co-optimize the HVAC-BESS by taking into account solar energy forecasts. A surplus-based consensus algorithm is proposed to solve the optimization problem, where the local power mismatch is introduced as a surplus term, and the HVAC-BESS can thus be coscheduled to maximize renewable energy efficiency at the peak generation time. A set of the convex cost functions are formulated to minimize the HVAC's user dissatisfaction degree and alleviate power loss during the BESS operation. The goal is to collectively minimize the total energy cost in a distributed manner, subject to individual load constraints and power balance constraints. It is theoretically proved that a global convergence of the proposed algorithm is achieved provided that the directed network is strongly connected. The results from a number of case studies are promising, demonstrating the effectiveness and robustness of the algorithm under practical scenarios.

Anomaly detection in multivariate time series data is challenging due to complex temporal and feature correlations. This paper proposes a novel unsupervised multi-scale stacked spatial-temporal graph attention network for multivariate time series anomaly detection (STGAT-MAD). The core of our framework is to coherently capture the feature and temporal correlations among multivariate time-series data by stackable STGAT networks. Meanwhile, a multi-scale input network is exploited to capture the temporal correlations in different time-scales. Besides, a new dataset derived from a real-world wind farm is built and released for multivariate time series anomaly detection. Experiments on the proprietary dataset and three public datasets show that our method significantly outperforms existing baseline approaches, and provides interpretability for anomaly location.

Recently, unmanned aerial vehicle (UAV) swarm has been advocated to provide diverse data-centric services including data relay, content caching and computing task offloading in vehicular networks due to their flexibility and conveniences. Since only offloading computing tasks to edge computing devices (ECDs) can not meet the real-time demand of vehicles in peak traffic flow, this paper proposes to combine edge computing and UAV swarm for cooperative task offloading in vehicular networks. Specifically, we first design a cooperative task offloading framework that vehicles' computing tasks can be executed locally, offloaded to UAV swarm, or offloaded to ECDs. Then, the selection of offloading strategy is formulated as a mixed integer nonlinear programming problem, the object of which is to maximize the utility of the vehicle. To solve the problem, we further decompose the original problem into two subproblems: minimizing the completion time when offloading to UAV swarm and optimizing the computing resources when offloading to ECD. For offloading to UAV swarm, the computing task will be split into multiple subtasks that are offloaded to different UAVs simultaneously for parallel computing. A Q-learning based iterative algorithm is proposed to minimize the computing task's completion time by equalizing the completion time of its subtasks assigned to each UAV. For offloading to ECDs, a gradient descent algorithm is used to optimally allocate computing resources for offloaded tasks. Extensive simulations are lastly conducted to demonstrate that the proposed scheme can significantly improve the utility of vehicles compared with conventional schemes.

Condition Monitoring of photovoltaic systems plays an important role in maintenance interventions due to its ability to solve problems of loss of energy production revenue. Nowadays, machine learning-based failure diagnosis is becoming increasingly growing as an alternative to various difficult physical-based interpretations and the main pile foundation for condition monitoring. As a result, several methods with different learning paradigms (e.g. deep learning, transfer learning, reinforcement learning, ensemble learning, etc.) have been used to address different condition monitoring issues. Therefore, the aim of this paper is at least, to shed light on the most relevant work that has been done so far in the field of photovoltaic systems machine learning-based condition monitoring.

Predictions of renewable energy (RE) generation and electricity load are critical to smart grid operation. However, the prediction task remains challenging due to the intermittent and chaotic character of RE sources, and the diverse user behavior and power consumers. This article presents a novel method for the prediction of RE generation and electricity load using improved stacked gated recurrent unit-recurrent neural network (GRU-RNN) for both univariate and multivariate scenarios. First, multiple sensitive monitoring parameters or historical electricity consumption data are selected according to the correlation analysis to form the input data. Second, a stacked GRU-RNN using a simplified GRU is constructed with improved training algorithm based on AdaGrad and adjustable momentum. The modified GRU-RNN structure and improved training method enhance training efficiency and robustness. Third, the stacked GRU-RNN is used to establish an accurate mapping between the selected variables and RE generation or electricity load due to its self-feedback connections and improved training mechanism. The proposed method is verified by using two experiments: prediction of wind power generation using multiple weather parameters and prediction of electricity load with historical energy consumption data. The experimental results demonstrate that the proposed method outperforms state-of-the-art methods of machine learning or deep learning in achieving an accurate energy prediction for effective smart grid operation.

Distributed generation (DG) systems comprising of renewable energy generation technologies will play a significantly increasing role for future power systems. One of the key concerns for deployment of DG systems is specifically related to their availability and reliability, particularly when operating in a harsh environment. Condition monitoring (CM) can meet the requirement but has been challenged by huge amount of data to be processed especially in real time in order to reveal healthy conditions of the system. In this paper, an optimal sensor selection method based on principal component analysis (PCA) is proposed for condition monitoring of a DG system oriented to wind turbines. The proposed method is examined with both simulation data from PSCAD/EMTDC and SCADA data of an operational wind farm in the time, frequency, and time-frequency domains. The results have shown that the proposed technique could reduce the number of sensors whilst still maintaining sufficient information to assess the system's conditions.