This article mainly focuses on the issue of event-triggered secure load frequency control (LFC) for multi-area power systems under denial-of-service (DoS) attacks based on input-output data. In contrast to prior results, the issues of controller design and security analysis are solved by considering the characteristics of the DoS attacks, instead of employing average dwell time approach. By analyzing the interaction between attack patterns and event-triggered mechanisms, the valid and invalid durations within attack periods that actually impact the system are defined. Based on which, an attack compensation recovery strategy is proposed within DoS attack intervals under the model-free adaptive control framework. Then, the data-driven controller and parameter estimation are reconstructed using the last successfully triggered signal before an attack occurs, effectively mitigating the impact of communication blockages. Furthermore, a sufficient condition for ensuring the boundedness of tracking errors is derived through recursive analysis, which directly constrain the ratio of valid to invalid time during attack periods. Finally, simulations are provided to validate the effectiveness of the proposed method for multi-area power systems.
As electric power networks (EPNs) and district heating networks (DHNs) become increasingly interconnected, coordinated state estimation (CSE) plays a key role in integrated electricity and heating systems (IEHSs). In DHNs, the operating conditions involve both hydraulic processes and thermal processes, where not only temperatures but also mass flow may vary. To address the resulting variable mass flow (VM) thermal dynamics, a CSE method for IEHSs based on an improved thermal circuit method (TCM-VM) is proposed in this paper. The TCM-VM is developed through coordinate transformation and heat-loss correction, which allows pipeline thermal dynamics under variable mass flow conditions to be modeled in the frequency domain within a transformed coordinate system. Moreover, the EPN states, DHN hydraulics, and network topology constraints are established in the time domain. The CSE problem is addressed using the minimum information loss (MIL) framework to combine electrical and thermal measurements with different sampling intervals. The proposed method is validated under both constant and variable mass flow conditions, and the effects of the measurement configurations, window lengths, and noise levels are assessed. The effectiveness of the proposed method is evaluated through simulations on both a small-scale system and a large-scale IEHS.
The large-scale deployment of smart meters in power grids necessitates accurate prediction of failure rates to enable preventive maintenance. However, existing prediction models are often limited by insufficient multi-factor modeling and weak interpretability. To address these limitations, we propose a Bayesian network (BN)-enhanced Cox proportional hazards model (CPHM) for failure prediction of grouped smart meters. Multiple source covariates, including operational lifetimes, verification records, production batches, deployment areas, and component parameters, are jointly incorporated. Smart meters are first grouped by production batch and deployment area to capture intra-group homogeneity and inter-group heterogeneity in failure behavior. A global BN is then trained on the entire dataset to model probabilistic dependencies among covariates and quantify their risk contributions, thereby enhancing interpretability. Specifically, the conventional Cox linear predictor is replaced by a BN-enhanced risk function that incorporates conditional dependencies among covariates. Based on the learned BN, group-specific CPHMs with nonparametric baseline hazard functions are constructed to characterize distinct failure dynamics across groups. Validation on operational data from 38,238 smart meters, spanning 24 production batches and 358 deployment areas, shows that the proposed method achieves a C-index of 0.95, an R2 of 0.90, and a mean time-dependent AUC of 0.97, outperforming state-of-the-art survival models. The results demonstrate strong capability in handling censored data and heterogeneous failure characteristics.
This paper proposes a secure distributed control framework for wind turbine systems under intermittent denial-of-service attacks. First, a Taylor-compensated soft-switching composite extended state observer is developed to reconstruct unmeasurable states and structurally isolate the estimation dynamics from abrupt measurement injections upon measurement-channel recovery. By algebraically resolving mode-dependent gains via a scalar-parameterized polytopic formulation, this architecture accommodates affine drivetrain uncertainties. Consequently, it bridges measurement losses without high-frequency noise amplification and attenuates transition-induced estimation transients upon measurement-channel recovery, mitigating transient torsional shocks on the mechanical drivetrain. Furthermore, a dynamically expandable prescribed performance control strategy is established to ensure resilient tracking fidelity. By actively accommodating attack-induced error inflations via a repulsive fractional barrier within a finite admissible relaxation budget, this dynamic mechanism preserves prescribed-performance feasibility while keeping the adaptive performance envelope bounded. Theoretical analyses prove the semi-global uniform ultimate boundedness of all closed-loop signals, and semi-physical emulations validate the secure tracking and smooth state recovery capabilities of the proposed approach.
The adoption of electric vehicles (EVs) results in a stronger integration between energy and transportation systems. In this work, we develop scalable optimization models to co-design heterogenous hydrogen and battery EV fleets and the supporting energy infrastructure (e.g., distributed energy resources, and fast-charging and hydrogen-fueling station) to efficiently meet transportation demands. A novel integer-clustering formulation is used to model large-scale EV operation at the fleet level while maintaining accurate individual vehicle dispatch, which significantly improves the computation efficiency with guaranteed performance relative to the true individual vehicle formulation. Furthermore, we extend the integer-clustering formulation to capture EV battery degradation by providing an exact closed-form reformulation of the nonlinear and iterative rainflow algorithm for cycle counting, which retains the MILP form. We apply the optimization model to a real-world case study of the public bus network in Boston using real trip schedule data and technoeconomic parameters. Realistic insights are provided into the future evolution of coupled electricity-transportation-hydrogen systems, including the effects of electricity price structure, hydrogen fuel cost, carbon emission constraint, temperature effects on EV range, and distribution system upgrade cost.
