This paper investigates the structural controllability of complex networks with periodic switching topologies. First, several graph transformations that preserve structural controllability are demonstrated. Based on the n-walk theory, a criterion is derived that determines structural controllability by analyzing only the joint graph within a single period. The theoretical results are illustrated with three examples.

Although many neural network (NN) adaptive controllers have been proposed to deal with cooperation of nonlinear multiagent systems (MASs), it is still unknown how to achieve asymptotical cooperative goals over a general directed topology. A main challenge is the coupling of nonlinearities learning and cooperative control. Within this context, a novel class of adaptive controllers based on an NN-based cooperative modified state observer (CMSO) is proposed, where the CMSO can approximate unknown nonlinearities so that nonlinearities learning is decoupled into local tracking control under the proposed framework. It is proven that the controllers can achieve asymptotic consensus if the topology has a directed spanning tree. Note that both nonsmooth controllers and smooth controllers are proposed, where smooth controllers can avoid chattering, which may be induced by nonsmooth ones. Finally, a simulation over multiple-robot systems is given to validate the theoretical results.

The increasing adoption of home and transport electrification has significantly increased the overall electricity consumption, leading to larger uncertainties in energy scheduling and management. Accurate electricity forecasting is crucial for energy management and efficiency. Traditional electricity consumption forecasting methods mainly focus on building either a separate model for each customer or a single model for all customers, which may lead to low accuracy as they often fail to capture shared electricity usage patterns among forecasting problems. In this study, we propose a temporal and cross-task co-learning (TCTCL) framework to simultaneously predict the electricity consumption of a large number of customers. TCTCL integrates both temporal representation learning (TRL) and cross-task representation learning (CTRL). As electricity consumption forecasting is inherently data-driven, TCTCL is flexible and can accommodate various deep learning based predictors depending on the specific scenario. Considering the promising performance of the recurrent neural network (RNN), we employ it as the predictor for all tasks in this study. TRL involves training an independent RNN model for each task (e.g., customer or a group of customers), ensuring each model is specialized for its respective task. CTRL enhances forecasting accuracy by capturing cross-task patterns through sharing and reusing model information across customers, i.e., transferring RNN architectures of source tasks to the target forecasting task. The amount of RNN model information from source tasks reused for each target task is regulated by knowledge coefficients, which are optimized by a designed gradient descent-based approach, making it applicable to large-scale electricity consumption networks. The superiority of TCTCL is demonstrated by comparison with several state-of-the-art time series models. We further validate the scalability and applicability of TCTCL on a large-scale power system consisting of 217 customers by integrating it with clustering techniques.

This paper introduces a first-order dynamical system for constrained optimisation problems in which global exponential convergence is achieved through an adaptive penalty mechanism. By treating the penalty parameter as a dynamic state variable, the proposed framework enforces feasibility and improves convergence performance compared with fixed-penalty approaches. Building on this structure, we further propose a modified system that guarantees fixed-time convergence, ensuring convergence within a uniform finite time independent of initial conditions. Further more, the proposed systems achieve global exponential or fixed-time convergence without imposing strong convexity on the original objective function, a common assumption in the existing literature. The effectiveness and practical advantages of the fixed-time convergent dynamical systems are demonstrated through numerical simulations.

The decarbonization of the transport sector and the transition toward net zero energy systems are accelerating the adoption of electric vehicles and rooftop photovoltaic systems in residential buildings. These technologies can significantly reshape household electricity demand and influence the operation of electrified homes. Understanding their combined impact on residential load profiles is essential for effective building energy management and planning of distributed energy resources. This study investigates the effects of electric vehicle ownership and rooftop photovoltaic adoption on residential electricity consumption using smart meter data from 2868 households in Victoria, Australia. Households are categorized into four groups based on the presence of electric vehicles and rooftop photovoltaic systems. Clustering analysis is used to identify representative weekday and weekend load patterns across different seasonal temperature ranges. The results show that households with electric vehicles exhibit substantially higher evening electricity demand, with average peak loads increasing by approximately 35 to 45 percent compared with households without electric vehicles. Rooftop photovoltaic adoption significantly reduces daytime net demand, lowering midday grid electricity use by up to 50 percent during high solar generation periods. Households combining electric vehicles and photovoltaic systems display distinct load characteristics, including reduced daytime grid dependence and increased overnight charging demand. Seasonal analysis indicates that temperature conditions influence the magnitude and timing of these patterns, particularly during extreme weather periods. Economic assessment under current tariff structures shows that rooftop photovoltaic systems can offset 30 to 40 percent of the additional electricity costs associated with electric vehicle charging. These findings provide empirical evidence on how electrification technologies reshape residential energy use and highlight the importance of coordinated strategies that integrate electric vehicles and distributed solar generation in residential buildings.

This paper studies the distributed secure state estimation problem in cyber-physical systems under sparse sensor attacks, by employing an adaptive dynamic gain modulation strategy. Unlike most existing attack detection and isolation-based strategies, a dual-layer collaborative state estimation framework based on attack suppression is established. The framework integrates decentralized and distributed observers, thereby decoupling attack suppression from residual feedback. By analyzing the dynamic behaviors of state deviations between the two observer layers, a universal dynamic suppression gain is derived to guarantee the robustness against sparse attacks. Furthermore, a distributed adaptive dynamic gain modulation strategy is introduced to avoid centralized processing, which enhances the applicability of the attack suppression mechanism while preserving the estimation accuracy. Finally, a numerical simulation is presented to demonstrate the effectiveness of the proposed framework.

In federated learning, improving communication efficiency is a critical challenge, especially under partial participation and biased compression. Many existing approaches rely on unbiased compression or strong assumptions, such as the bounded gradient assumption, which are often difficult to satisfy in practice. In this paper, we propose a novel federated learning algorithm named EF21-MP (EF21 with Momentum and Partial Participation), which combines biased compression with partial participation and stochastic gradient descent. Furthermore, it incorporates momentum and EF21 to reduce variance from stochastic gradient descent and biased compression. It achieves convergence for nonconvex optimization under standard smoothness and bounded variance conditions, without relying on any bounded gradient assumptions, and could support for batch-free training. The numerical results demonstrate that EF21-MP consistently outperforms the existing baselines.

This article proposes an adaptive sliding-mode repetitive control (ASMRC) for a servomotor to track time-varying periodic signals at a fixed sampling period. The proposed control has several advantages, including high accuracy with time-varying periods, robustness analysis, and low sensitivity to fluctuations in the transient period. A noncausal Lagrange fractional-delay filter is employed whose coefficients are updated online according to the time-varying period to achieve high gain at the fundamental and its harmonics, enabling high-precision tracking. The sliding-mode control is integrated with the RC to guarantee robustness against model uncertainties, disturbances, and interpolation errors, enhancing the RC characteristics. A real-time period estimator adjusts the fractional delay, providing a fast response to abrupt period changes. A Lyapunov-based analysis establishes closed-loop stability and quasi-sliding behavior. Comparative hardware experiments on a servomotor demonstrate superior steady-state accuracy, robustness, and period adaptivity.

Data centers play a pivotal role in supporting digital transformation. However, they are among the most energy-intensive infrastructures as a result of increasing global computing load over the years. As data center power consumption continues to increase, examining their dynamic performance and interaction with the electrical grid is essential to enhance efficiency and ensure grid stability. This paper presents a comprehensive dynamic model of a 10 MW data center, incorporating 44,250 servers with a tandem hybrid heating ventilation and air-conditioning (HVAC) system to simulate real-world data center under varying workloads. The developed model is validated using experimental data that demonstrate high accuracy in predicting server power consumption, heat generation, and dynamic temperature. Furthermore, to verify the fidelity of the model to operate in dynamic power systems, the data center model is connected to WSCC 9-bus test system to analyze the impact of various grid disturbances on the data center operations, such as load variations and three-phase short-circuit faults. The model can be scaled up and is valuable for improving the accuracy of dynamic stability analysis of power systems under high integration of data center load. In addition, the model is proposed to provide a practical framework for assessing compliance with fault ride-through (FRT) requirements and enabling demand response participation.

Composite federated learning offers a general framework for solving machine learning problems with additional regularization terms. However, existing methods often face significant limitations: many require clients to perform computationally expensive proximal operations, and their performance is frequently vulnerable to data heterogeneity. To overcome these challenges, we propose a novel composite federated learning algorithm called FedCanon, designed to solve the optimization problems comprising a possibly non-convex loss function and a weakly convex, potentially non-smooth regularization term. By decoupling proximal mappings from local updates, FedCanon requires only a single proximal evaluation on the server per iteration, thereby reducing the overall proximal computation cost. Concurrently, it integrates control variables into local updates to mitigate the client drift arising from data heterogeneity. The entire architecture avoids the complex subproblems of primal-dual alternatives. The theoretical analysis provides the first rigorous convergence guarantees for this proximal-skipping framework in the general non-convex setting. It establishes that FedCanon achieves a sublinear convergence rate, and a linear rate under the Polyak-Łojasiewicz condition, without the restrictive bounded heterogeneity assumption. Extensive experiments demonstrate that FedCanon outperforms the state-of-the-art methods in terms of both accuracy and computational efficiency, particularly under heterogeneous data distributions.

The widespread adoption of photovoltaic (PV), electric vehicles (EVs), and stationary energy storage systems (ESS) in households increases system complexity while simultaneously offering new opportunities for energy regulation. However, effectively coordinating these resources under uncertainties remains challenging. This paper proposes a novel home energy management framework based on deep reinforcement learning (DRL) that can jointly minimise energy expenditure and battery degradation while guaranteeing occupant comfort and EV charging requirements. Distinct from existing studies, we explicitly account for the heterogeneous degradation characteristics of stationary and EV batteries in the optimisation, alongside stochastic user behaviour regarding arrival time, departure time, and driving distance. The energy scheduling problem is formulated as a constrained Markov decision process (CMDP) and solved using a Lagrangian soft actor-critic (SAC) algorithm. This approach enables the agent to learn optimal control policies that enforce physical constraints, including indoor temperature bounds and target EV state of charge upon departure, despite stochastic uncertainties. Numerical simulations over a one-year horizon demonstrate the effectiveness of the proposed framework in satisfying physical constraints while eliminating thermal oscillations and achieving significant economic benefits. Specifically, the method reduces the cumulative operating cost substantially compared to two standard rule-based baselines while simultaneously decreasing battery degradation costs by 8.44

Fast convergence, chattering reduction, and analytical derivation of the sliding time are key problems in sliding-mode control (SMC). In this article, a novel sliding surface, referred to as the logarithmic terminal sliding-mode (LnTSM) manifold, is proposed with an analytical solution for the sliding time. Then, a smooth controller is developed using the super-twisting algorithm and a barrier function technique. In the proposed scheme, the shared gain of two adaptive laws is dynamically tuned by a deep reinforcement learning-based agent to simultaneously ensure transient performance and steady-state precision. The stability of the proposed control scheme is proved via the Lyapunov theorem. Numerical simulations and experiments are subsequently provided to verify the above superiority. The results demonstrate that the proposed method exhibits significantly superior performance compared to other methods and has high value in practical applications.

This study proposes an accelerated iterative learning control scheme using a fractional high-order update rule (FHUR) to improve the convergence rate for linear time-invariant systems. High- and low-order power update terms are used to handle large- and small-tracking errors, respectively, thereby accelerating convergence. Two learning mechanisms are proposed and shown to be optimal among various learning gain selections. The inherent nonlinearity in the FHUR poses significant challenges for the convergence analysis. To address this, a disturbed composite nonlinear mapping method is introduced. Using this method, the tracking errors are proven to converge either to an invariant set or to a set of limit cycles, depending on the underlying learning mechanism. Any desired tracking precision can be achieved by adjusting the parameters in the FHUR. Numerical simulations confirm that the FHUR presents a promising alternative to the commonly used proportional-type update rule for achieving accelerated convergence.

Power distribution networks, incorporating electric vehicles (EVs) and battery energy storage systems (BESSs), can provide valuable flexibility to the upstream grid. This article proposes a new mechanism for modeling and optimizing two-way flexibility exchange between the distribution system operator (DSO) and flexible loads, aiming to minimize the DSO's total cost while satisfying the flexibility requests of the upstream market operator. The DSO first solicits the participation of flexible loads, including EVs and BESSs, which can either accept or reject the request. Considering the agreed state of charge of participating resources, a flexibility market is then formulated, incorporating the user contribution index and Karush-Kuhn-Tucker conditions. The proposed mechanism is tested under various load conditions, price tariffs, and EV penetration levels. The results demonstrate significant cost reductions for DSOs, as they purchase less energy from the upstream market operator compared to scenarios without flexibility management.

As power systems continue to expand and electricity demand grows rapidly, efficient multi-area economic dispatch (MAED) becomes increasingly important for reducing operating costs. However, practical considerations such as transmission losses lead to non-convex coupling constraints in MAED. This substantially increases the computational difficulty of the problem. To address these challenges, this paper proposes a stepwise cooperative reinforcement learning (RL) framework for MAED, consisting of a power-exchange module and a power-distribution module operating at different algorithmic time scales. The power-exchange module optimizes inter-area power exchanges using deep hysteretic RL combined with a finite-time average consensus protocol. The power-distribution module determines intra-area generation dispatch via communication-assisted RL with quadratic function approximation and distributed projection optimization. By first determining inter-area exchanges and then performing local dispatch, the proposed framework decomposes the original non-convex MAED problem into a sequence of tractable subproblems and enforces operational constraints by construction. Simulation studies demonstrate that the proposed method achieves lower total operating cost than conventional optimization baselines and existing RL-based approaches, while maintaining strict feasibility under non-convex coupling constraints.

Data centers are emerging as one of the fastest-growing electricity consumers worldwide due to the rapid expansion of cloud computing, artificial intelligence (AI), and digital services. The large-scale integration of AI data centers into electric power systems introduces significant challenges for grid planning, operation, stability, power quality, and compliance with evolving grid codes. Modern data centers are characterized by power-electronic-dominated infrastructures, including high-density computing platforms, advanced cooling systems, on-site renewable energy and energy storage resources, all of which exhibit dynamic behaviors distinct from conventional passive loads. Consequently, their increasing integration necessitates the development and application of appropriate modeling frameworks to accurately assess grid impacts and enable effective control and coordination strategies. This papers covers grid-integrated data centers, with a focus on their electrical and cooling architectures, and associated modeling approaches. In addition, modeling of critical data center components, interconnection requirements, and key integration challenges are examined. Finally, emerging opportunities for data centers to provide frequency regulation and flexibility services are discussed, outlining future research directions toward reliable, efficient, and grid-interactive data center integration.

This paper investigates a class of distributed optimal dispatch problems in smart grid power management, with the objective of minimizing total generation cost while satisfying certain safety constraints. First, the optimal power dispatch for a single time slot is examined, and a distributed primal-dual algorithm is developed to achieve the dispatch goal, even when the initial total generation does not match total demand. Subsequently, the algorithm is extended to address dynamic dispatch problems, which incorporate transmission line flow constraints. This extension accounts for varying power demands over multiple time slots, while accommodating the power change rate limitations. Finally, software simulations and hardware-in-the-loop tests are conducted to validate the proposed algorithms.

This article presents a two-degree-of-freedom sliding mode control with event-triggered feedback to guarantee robust output tracking for an uncertain plant against matched disturbances. The proposed controller comprises an event-based feedback loop for the plant measurements and the analog feedthrough path for the tracking (reference) signal, which is generated by an exogenous system. Here, the event-triggering mechanism uses only the feedback signals, unlike the existing works, and thus, it retains the two-degree-of-freedom architecture in the controller implementation. The main advantage of the proposed event-based controller is that the event mechanism generates a sparse sampling sequence due to the use of only feedback signals. It is seen that the solvability of the tracking problem is reduced to that of the sliding mode regulator equations. Indeed, it is established under this assumption that for any bounded initial conditions, the proposed sliding mode controller guarantees robust output tracking for the closed-loop system with arbitrary accuracy. Finally, the simulation results are presented to demonstrate the output tracking by using the proposed methodology.

The rapid integration of electric vehicles (EVs) and solar photovoltaic (PV) systems introduces significant uncertainty and operational stress in low-voltage (LV) distribution networks, often leading to voltage violations at high penetration levels. Residential battery energy storage systems (BESSs), when operated within a virtual power plant (VPP) framework, play an important role in supporting voltage regulation while improving consumer economic benefits. However, BESS scheduling remains a challenge due to the conflicting objectives of network safety and cost minimization, particularly under realistic operating conditions. This paper proposes an adaptive weighted multi-objective optimization (AW-MOO) framework. It explicitly prioritizes voltage violation mitigation (network impact) while subsequently minimizing consumer cost. Owing to the multimodal nature of the optimization landscape, where multiple scheduling solutions can yield same network-level performance, Pareto-based multi-objective optimization methods struggle to identify economically optimal solutions. To address this issue, the proposed framework reformulates the problem into a single-objective optimization with an adaptively tuned weight that dynamically balances network and economic objectives. AW-MOO is implemented by a constrained particle swarm optimization algorithm, where the weight is adaptively adjusted during the evolutionary process. AW-MOO is validated on a real-world LV distribution network with 108 residential consumers. Its superiority is demonstrated through comparisons with constant weighting strategies and a no-BESS case. The results show that AW-MOO can eliminate voltage violations and reduce consumer costs by more than 25%, highlighting its effectiveness in network and economic benefits.