This article presents a soft actor-critic (SAC)-based control method for the anti-sway problem of port crane systems under complex working conditions. First, considering the uncertainty of environmental disturbances, a distributed set-membership (SM) estimation strategy with multisensor is proposed to obtain the state information of the crane. Moreover, a fusion method considering kernel density estimation (KDE) and information entropy (IE) is proposed to dynamically assess sensor reliability and obtain the refined global state estimation. Second, an SAC anti-sway control framework with SM global estimation is developed for optimizing control parameters amid system constraints. The control strategy steers the system toward tracking the target state by designing a reward function, achieving optimized control adapted to dynamic disturbances, and improving antisway efficiency. Finally, the effectiveness and superiority of the proposed approach are validated by performance analysis.

This work proposes a distributed zonotope set-membership filtering (DZSMF) to mitigate non-line-of-sight (NLOS)-induced uncertainties while ensuring the computational efficiency for robust real-time localization. First, a zonotope set-membership filtering is designed for received signal strength (RSSI)-based localization, in which process noise, measurement noise, and higher order Taylor-series residuals are jointly confined within a zonotope ensemble. The filter gain is analytically obtained by minimizing the Fradius of the posterior error zonotope. Second, a compact zonotope refinement mechanism is established by intersecting three zonotopes generated from each anchor-node group's iterative updates. These locally optimized zonotopes are subsequently integrated through a fully distributed, hierarchical information fusion strategy to produce a global state estimate accompanied by rigorous, easy-to-interpret uncertainty intervals. DZSMF uses observable sensor signals and the deployment and selection of multianchor node groups to alleviate NLOS faced in positioning. Finally, experimental results validate the accuracy and effectiveness of the proposed method. The present study represents a substantial improvement over initial estimates, with a 90.38% enhancement in performance. In comparison to existing methods, it demonstrates a 17.93%, 48.95%, 41.82%, and 1.99% improvement in terms of accuracy when evaluated against the extended kalman filtering (EKF), classical ellipsoidal set membership filtering (CESMF), interval ellipsoidal set membership filtering (IESMF), nonlinear dual set membership filtering (NDSMF) algorithms, respectively.

Smart grids have been widely studied for their ability to improve efficiency and reliability, with prior research focusing on demand response, secure data sharing, and distributed optimization. However, existing approaches often address privacy protection and energy efficiency separately, leaving a gap in simultaneously achieving both within a scalable and incentive-compatible framework. To address this challenge, this work proposes a blockchain-based energy conservation mechanism that integrates an adaptive weighted average consensus scheme, optimized via Non-Dominated Sorting Genetic Algorithm II, with a dynamic incentive contract supported by cryptocurrency rewards. The mechanism allows prosumers to validate transactions by exchanging only the percentage of power change, thereby preserving privacy, while the incentive design motivates active participation in energy scheduling. Simulation results show that the proposed approach reduces average daily energy consumption per prosumer by 7.5 kWh (30-node case) and 8.8 kWh (300-node case), and decreases 24-hour weighted electricity costs by up to 8.66%. These findings highlight the effectiveness of the mechanism in achieving measurable energy and cost savings.

This paper investigates a novel set-membership global estimation method for a class of discrete time-varying systems with unknown-but-bounded noises. Firstly, in order to improve the accuracy of the state estimation, a multi-sensor network structure is deployed for the considered system, in which the adjacent sensors can communicate with each other. Considering the different sampling rates of multiple sensors, a resampling strategy is proposed to transform the asynchronous sampling system into a synchronous one. Additionally, the unknown-but-bounded noises and system parameter variations are considered. A novel distributed set-membership filter with parameter variation is designed, and the optimal local state estimation ellipsoid is obtained by developing a convex optimization method. Subsequently, a soft actor-critic algorithm based on reinforcement learning is proposed, which fuses all local estimation ellipsoids from each sensor to obtain global estimation results, providing a solution for the considered system. Finally, a port crane system is employed for performance analysis to verify the feasibility and effectiveness of the proposed method. Note to Practitioners-This work addresses the problem of global state estimation in time-varying multi-sensor networks, which is a key component of modern intelligent systems that provides the foundation for monitoring and control. In practice, asynchronous sampling and unknown-but-bounded noises may reduce the reliability of multi-sensor estimation, and it is difficult to determine appropriate fusion weights for local estimations. In this work, we introduce a resampling strategy to synchronize data, design distributed set-membership filters to obtain local ellipsoidal estimations, and develop a global fusion strategy based on deep reinforcement learning. The learning framework is implemented through an offline training-online deployment scheme, enabling adaptive fusion while accounting for sensor performance differences. The proposed approach has been applied to a crane system for monitoring, showing effective performance. This work offers a promising tool for monitoring in industries such as intelligent manufacturing, autonomous driving, and robotics, and also provides a foundation for precise system control.

This paper investigates the problem of interval trajectory tracking control for an automated guided vehicle in the presence of unknown-but-bounded noises. A trajectory tracking control method is proposed based on set-membership global estimation considering road boundary constraints. Firstly, a group of distributed set-membership estimators based on sensor network topology is designed to obtain local state estimation ellipsoids. Secondly, a radial intersection boundary approximation strategy is proposed to obtain the approximate intersection ellipsoid of the local ellipsoids as the global state estimation ellipsoid. This strategy reduces the computational complexity while yielding a more accurate global estimation. In addition, in order to realize accurate interval trajectory tracking, an ellipsoid constraint applicable to the set-membership framework is proposed to equivalently model the interval constraint, and an interval trajectory tracking controller is designed based on the global estimation results. Simulation results demonstrate that the proposed method achieves interval trajectory tracking efficiently and performs well.

Developing efficient and reliable renewable energy sources (RESs) to reduce the exploitation of traditional power generation is essential for achieving net zero strategy (NZS) in a sustainable and affordable way. However, due to their nature, many RESs are geographically distributed; for instance, wind power stations are typically built along coastlines to harness sea wind energy, and photovoltaic power panels maybe distributed over residential buildings. This geographical distribution naturally leads to the term distributed energy resources (DERs). To efficiently utilize and coordinate DERs, the concept of networked microgrids (NMGs) has emerged. Compared to single microgrids (MGs), NMGs can make full use of DERs by enabling energy exchange and coordinated control among interconnected MGs. However, the interconnectedness also increases the complexity, uncertainty, and potential vulnerability of NMGs to faults and cyberattacks. This review aims at summarizing the latest effective methodologies to secure the distributed secondary control layer, focusing on the impact of attacks on controllers, sensors, and communication channels. Additionally, this article outlines future research directions that could enhance the reliability, resilience and flexibility of NMGs under cyberattacks, thereby guiding efforts to achieve a secure NZS.

Reliable operation of net-zero microgrids requires dispatch strategies that remain sustainable under both probabilistic and worst-case variations of renewable generation and demand. Deterministic optimisation assumes perfect forecasts and collapses under real PV fluctuations or load uncertainty. Stochastic optimisation enhances adaptability by modelling probabilistic deviations, whereas robust optimisation guarantees feasibility under the most adverse conditions. Achieving true reliability, therefore, requires integrating both perspectives within a unified, uncertainty-aware methodology. This study develops a reliability-oriented optimisation and evaluation methodology that unifies deterministic, robust, stochastic, and stochastic-robust dispatch paradigms within a networked net-zero microgrid (N-NZMG) comprising photovoltaic (PV) generation, battery energy storage, and flexible loads. A continuous linear programming formulation ensures tractable computation while enforcing battery, network, and daily net-zero constraints. To enable fair comparison across uncertainty treatments, a normalised multi-criteria ranking scheme is introduced to convert heterogeneous performance indicators into unified dimensionless scores. Simulation results under Australian solar conditions demonstrate that deterministic dispatch maintains ideal performance only under perfect forecasts but rapidly deteriorates when PV output fluctuates. The robust strategy maintains feasibility across all irradiance levels while increasing battery cycling by about 17%. The stochastic formulation effectively adapts to probabilistic PV fluctuations and load uncertainty, maintaining the net-zero balance within 10(-4) kWh and limiting curtailment to below 0.05%. The hybrid stochastic-robust case yields results comparable to those of the robust case, with enhanced resilience under severe variability. The proposed unified methodology advances the concept of reliable net-zero operation by showing how probabilistic adaptability and worst-case protection can jointly sustain microgrid performance during fluctuations in renewable and demand sources. It offers a structured and scalable pathway toward reliability-driven optimisation of renewable-dominated energy systems.

Decarbonising electricity systems is central to net-zero ambitions, placing increasing importance on localised, scalable solutions. Achieving energy neutrality at household and community levels offers a practical pathway, yet the operation of rooftop PV-battery microgrids remains challenged by variability, tariff structures, and system constraints. This paper proposes a Forecast-based, Daily-updated, Multi-objective Optimisation (FDMO) framework for the operation of Net Zero Microgrids (NZMGs) under real-world conditions. The method introduces three key innovations: (i) strict enforcement of daily net-zero energy balance, (ii) forecast-driven rolling optimisation with embedded error correction, and (iii) a multi-objective formulation that jointly optimises grid exchange, battery dispatch, and PV curtailment using tariff-aligned penalty signals. The framework simultaneously determines optimal import, export, storage, and curtailment decisions while ensuring daily neutrality without requiring retroactive adjustments. Application to representative surplus and deficit scenarios demonstrates that FDMO consistently prioritises storage utilisation over curtailment, strategically coordinates grid interactions, and maintains feasibility under operational limits. In high-deficit conditions, grid imports are constrained to 3.46 kWh while delivering 7.86 kWh from storage; under high-surplus conditions, the system exports 6.28 kWh, stores 12.30 kWh, and limits curtailment to 3.07 kWh. The results highlight the effectiveness of the proposed framework as a robust and practical dispatch strategy for solar-dominated residential microgrids, enabling cost-efficient operation, policy compliance, and reliable grid interaction. While demonstrated using Queensland-specific data and tariffs, the formulation is fully transferable across jurisdictions through the substitution of local forecasts and market signals.

The paper presents a game theory optimisation strategy for Microgrid (MG) energy management system (EMS) of a single AC residential system. This EMS integrates a Genetic Algorithm (GA) with Multi-Objective Particle Swarm Optimisation (MOPSO), enhanced by Stackelberg game theory to achieve robustness and cost-effectiveness. The proposed method leverages a leader-follower structure for game theory strategy where the GA governs dynamic grid pricing, and MOPSO optimises operational scheduling of the cost and environmental objectives. Results demonstrate superior trade-offs between operating cost and emissions compared to stand-alone meta-heuristic algorithms.

Urban energy systems are increasingly challenged by the growing integration of renewable energy and the rising demand for electric vehicle (EV) charging infrastructure. This study presents the design and optimisation of a hybrid microgrid system integrating photovoltaic (PV) systems, battery energy storage systems (BESS), and EV charging stations for urban environments. The system configuration is optimised using HOMER Pro software to minimise Net Present Cost (NPC) while maximising renewable energy utilisation under varying EV charging load conditions. Real-world data, including solar irradiance profiles and time-of-use (TOU) electricity pricing are integrated to ensure practical applicability. Key findings indicate that the system effectively adapts to increasing EV charging de-mands by scaling PV capacity, BESS throughput, and grid power usage. Sensitivity analyses highlight the significant impact of load variability and electricity pricing on economic performance, demonstrating the robustness of the proposed microgrid design across diverse scenarios. This work underscores the potential of hybrid microgrids to mitigate the challenges associated with large-scale EV adoption, reduce reliance on grid power, and contribute to decarbonisation efforts. The findings provide a valuable reference for developing cost-effective and sustainable energy solutions in urban settings.

In isolated networked microgrid with high penetration of solar and wind-based generation, maintaining system stability and achieving optimal dynamic performance poses significant challenges due to reduced mechanical inertia traditionally provided by synchronous generators. This paper introduces a novel Resilient Back Propagation Bayesian Neural network-based virtual inertia control strategy to enhance frequency response and overall stability of the considered network microgrid. By leveraging robust control techniques, the proposed approach provides virtual inertia to respond effectively to varying system conditions and disturbances, improving system robustness and minimising the isolated networked microgrid’s tie-line power and frequency deviations. Comprehensive simulations, including case studies under varying disturbance conditions such as load fluctuations and renewable energy variations, prove the efficacy of the introduced virtual inertia control strategy. The proposed method outperforms conventional proportional integral derivative, and artificial bee colony-optimised proportional integral derivative control techniques in key dynamic performance metrics. It achieves the lowest integral time absolute error of 18.3912, compared to 30.8946 for proportional integral derivative and 20.8212 for optimised proportional integral derivative control techniques, demonstrating superior frequency response. Additionally, it achieves a mean square error of 4.0897e-07, significantly lower than 4.239e-06 for neural network-based fractional order proportional integral derivative control and 3.072e-06 for feed-forward neural network, confirming improved accuracy. Results indicate improved dynamic performance metrics, including faster frequency stabilization and reduced overshoot with minimal tie-line power and frequency deviation, compared to proportional integral derivative and optimal control techniques. The strategy’s adaptability and computational efficiency offer practical benefits for real-world implementation, contributing to more resilient and efficient microgrid operations.

Microgrid systems with high penetration of renewable energy resources often experience significant frequency instability due to low system inertia, especially during contingency events such as sudden changes in load or generation. To address these challenges, this work proposes a data-driven intelligent control approach integrated with an auxiliary virtual inertia control framework to enhance frequency regulation in isolated microgrids, ensuring high effective inertia during contingency conditions. Here, the data-driven frequency control framework dynamically adjusts the control parameters for enhanced frequency stability by reducing frequency deviation, while virtual inertia improves system inertia during load disturbances and high penetration of renewable energy resources. A theoretical framework for virtual inertia control is formulated, followed by performance evaluations through simulations using the considered isolated microgrid in MATLAB/Simulink. The simulated results show substantial improvements in frequency stabilisation, reduced oscillations, and faster recovery times. The proposed approach represents a significant advancement in intelligent control systems for improvement in microgrid frequency stability.

Managing the current high penetration of renewable energy sources globally poses a significant challenge due to the distributed and diverse nature of generation components. To maximize real-time power generation, these resources need to perform optimally. Digital twin technology offers a comprehensive framework for managerial support by replicating grid features in a digital environment. This research creates a digital twin of the microgrid to optimize power generation, focusing on computational efficiency and self-healing control. The framework is tested in a laboratory microgrid, with modeling performed using a polynomial regression algorithm. Optimization is achieved through a gradient descent algorithm, and the self-healing model is implemented using a logistic regression algorithm. Real-time data extracted from the microgrid drives this process. The results can be utilized for predictive analysis before deploying a microgrid or to optimize generation in existing systems using the digital twin model. Even though the research focuses on a single microgrid unit, it introduces a framework proposal for extensively distributed microgrids integrating multiple renewable energy sources.

In this paper, a new structure of a multiport DCDC converter based on a switched-inductor-capacitor (SLC) cell has been proposed with high voltage gain characteristics. In the proposed topology, power can flow from the input port to the output and battery ports. The SLC cell significantly improves the output voltage gain of the proposed converter and maintains a quadratic voltage gain during the operation. Based on two different duty cycles to control the power of multiple ports, three different modes of operation are studied for the proposed work. The steady-state modeling has been performed to analyze the operations of the proposed converter. Compared to the conventional topology, the voltage gain is 2 times higher at a 50% duty cycle in the proposed topology. Hence, a wider range of duty cycle operations becomes available for the battery charging/discharging port. Finally, the simulation studies have been performed in MATLAB/Simulink software, which show inline relation with the mathematical modeling of the proposed converter.

This paper investigates the dynamic performance of a grid-connected electric vehicle (EV) charging microgrid controlled by a Virtual Synchronous Generator (VSG) under multiple disturbance scenarios. The study benchmarks VSG control against conventional PQ control, focusing on frequency and voltage stability under grid frequency deviations and staged load fluctuations. Simulation results demonstrate that VSG control significantly enhances system resilience by emulating inertia and damping characteristics of synchronous generators. Compared to PQ control, VSG reduces frequency deviation by $\text{1 5 - 5 0 \%}$ and voltage deviation by $\text{2 0}-\text{1 0 0} \%$, while also limiting overshoot and improving recovery time. The findings confirm that VSG control provides a robust and flexible solution for stabilising EV charging microgrids, and support its practical deployment in next-generation renewable-integrated power systems.

This paper investigates the global state estimation problem under the influence of unknown-but-bounded process and measurement noises in complex networks. To improve state estimation accuracy and reliability, this paper first designs a distributed set-membership estimation scheme based on absolute and relative measurements. This scheme aims to obtain local state estimation ellipsoidal sets for the considered network nodes subject to unknown-but-bounded noises, which utilizes the relative measurement information among sensor nodes to enhance the accuracy of these local estimations. In addition, considering the accuracy requirements, an ellipsoid-density-based fusion method is proposed to fuse all the local estimation ellipsoids into a global state estimation ellipsoid, thereby improving the accuracy and confidence of the global state estimation. Finally, the effectiveness of the proposed method is verified by simulation examples, demonstrating its superiority for the global state estimation problem.

This paper explores the design, optimisation, and techno-economic assessment of an AC residential Microgrid (MG) for sustainable energy application. The proposed multi objective optimisation strategy aimed to optimise energy management systems integrating renewable energy sources (RES), energy storage systems (ESS), and advanced power management. Using HOMER Pro software for sizing mechanism the study investigates multiple Microgrid frameworks with different tariffs to find the optimal residential MG size determining lowest Net present cost (NPC) with optimal setup. Furthermore, a hybrid optimisation model using the optimal sizing, combining Genetic Algorithm (GA) and Multi-Objective Particle Swarm optimisation (MOPSO), this study enhance energy reliability, reduce greenhouse gas emissions, and ensure cost-effectiveness. A Monte Carlo Simulation (MCL) is incorporated to model uncertainties in the MG PV and load system to test the robustness of the proposed strategy. A comparative analysis of PSO, GA, and Grey Wolf Optimisation (GWO) highlights the superiority of the hybrid GA-MOPSO optimiser in balancing system cost, reliability, and environmental impact. The integrated optimiser outperforms conventional methods by enhancing energy storage efficiency, optimising grid interaction, and reducing dependence on peak-hour supply. The resulting Pareto front demonstrates improved solution diversity and convergence. These findings underscore the effectiveness of integrated evolutionary strategies in advancing cost-efficient and sustainable MG operations.

The virtual synchronous generator (VSG) exhibits prominent features such as communication-less control in microgrids due to its virtual inertia support. However, basic VSG control may lead to active power oscillations, causing improper transient power delivery and frequency instability. This paper proposes disturbance-triggered VSG-based adaptive virtual inertia control techniques for battery energy storage systems (BESS) to enhance the frequency stability of microgrids. The proposed method uniquely supports operation with fixed, variable, and induction motor (IM) loads, a capability not achieved by other approaches. It delivers the best frequency nadir of 49.98 Hz, outperforming existing VSG strategies (49.91–49.93 Hz) and other technologies, and achieves the lowest maximum overshoot of 50.01 Hz, ensuring improved better transient performance. Additionally, the proposed control significantly enhances RoCoF performance by 46% compared to basic VSG topologies and incorporates adaptive inertia and damping, which are crucial for real-time grid stability. Simulations were conducted on modified IEEE 4- and 9-bus microgrid models in MATLAB/Simulink. Simulation experiments revealed a settling time of 0.12 seconds, significantly shorter than other methods (0.2–0.3 seconds), demonstrating faster frequency stabilization. Various disturbance scenarios confirmed the effectiveness of the BESS control strategy in providing additional virtual inertia, ensuring robust frequency support and superior power microgrid stability compared to conventional strategies.

In this paper, the zonotopic set-membership state estimation (SMSE) problem for nonlinear systems is investigated. To handle the nonlinear dynamics of the system, a semi-infinite programming (SIP) scheme based on zonotope analysis is established and solved. The SIP scheme aims to obtain a tight zonotope to enclose the nonlinear dynamics of the system. Subsequently, the robustness performance of the estimation error system is analyzed based on the L performance. Sufficient design conditions are derived to guarantee that the state estimation results satisfy the preset L performance index. Finally, illustrative examples are given to demonstrate the effectiveness of the proposed method. (c) 2025 Published by Elsevier Ltd.

In recent years, the global shift toward sustainable energy sources, such as photovoltaics (PV) has accelerated alongside the electrification of transportation. This transition is primarily driven by the urgent need to mitigate global warming and reduce greenhouse gas emissions associated with fossil fuels and internal combustion engine vehicles. However, the intermittent nature of PV power presents challenges for energy management in EV charging stations, making accurate day-ahead forecasting essential. This paper proposes a Physics-Informed Neural Network (PINN)-based PV forecasting method that integrates historical data with the physical principles of PV systems. Case studies across four seasons show that the proposed method achieves the lowest Root Mean Squared Error (RMSE) in three out of four seasons, with RMSEs ranging from 0.1065 to 0.1366. Furthermore, comparative analysis shows that the PINN model surpasses traditional forecasting benchmarks, including the Persistence model, ANN, LSTM, RNN, and others-establishing itself as a promising and robust approach for future PV forecasting research.