Accurate battery capacity forecasting is crucial for ensuring safe operation and effective maintenance scheduling. However, capacity prediction remains challenging due to the complex, nonlinear degradation processes influenced by diverse operational conditions and usage patterns. Existing operational condition analysis methods either treat voltage, current, and temperature independently, losing cross-variable coupling effects, or aggregate exposure durations without preserving temporal ordering, discarding transition dynamics that influence degradation pathways. This work addresses both limitations through a transition-aware encoding method that discretizes measurements into joint operational bins, tracking the sequence of transitions between bins and preserving both coupled effects and temporal dynamics. An encoder–decoder neural network processes these compact representations to generate capacity forecasts over extended horizons. Based on experimental data from lithium–iron–phosphate (LFP) cells undergoing nonlinear degradation, the proposed transition-aware encoding forecasts absolute capacity with a mean absolute percentage error of 1.68% and captures cycle-to-cycle capacity variation to within 0.16%, while simultaneously compressing raw time-series data by 94.3%. Compared to methods that discard temporal ordering or treat measurements independently, the proposed approach reduces worst-case capacity prediction errors by more than 50%.

In this paper, we address two practical challenges of distributed learning in multi-agent network systems, namely personalization and resilience. Personalization is the need of heterogeneous agents to learn local models tailored to their own data and tasks, while still generalizing well; on the other hand, the learning process must be resilient to cyberattacks or anomalous training data to avoid disruption. Motivated by a conceptual affinity between these two requirements, we devise a distributed learning algorithm that combines distributed gradient descent and the Friedkin-Johnsen model of opinion dynamics to fulfill both of them. We quantify its convergence speed and the neighborhood that contains the final learned models, which can be easily controlled by tuning the algorithm parameters to enforce a more personalized/resilient behavior. We numerically showcase the effectiveness of our algorithm on synthetic and real-world distributed learning tasks, where it achieves high global accuracy both for personalized models and with malicious agents compared to standard strategies.

Over-actuated systems, namely systems with more inputs than outputs, can increase control performance, yet are susceptible to model-based undetectable attacks if the actuator channel is compromised. In this paper, we show how implementing a sparse actuator schedule can introduce security by rendering these attacks ineffective. We formulate a novel methodology whereby a periodic sparse schedule, implemented at the actuators, secures the system by ensuring that a malicious adversary cannot exploit actuator redundancy to deploy undetectable attacks. The schedule is designed offline and repeats periodically at the actuators, so that the adversary is constrained to only tamper with the active actuators. We devise a degeneracyaware greedy selection procedure with low computational complexity to design an actuator schedule that renders undetectable attacks ineffective, whilst simultaneously providing relatively small performance degradation. We illustrate the effectiveness of our approach using a reference tracking model predictive controller on the IEEE-39 bus power network employing the designed sparse schedule.

This paper examines the properties of output-redundant systems, that is, systems possessing a larger number of outputs than inputs, through the lense of the geometric approach of Wonham et al. We begin by formulating a simple output allocation synthesis problem, which involves “concealing” input information from a malicious eavesdropper having access to the system output, while still allowing for a legitimate user to reconstruct it. It is shown that the solvability of this problem requires the availability of a redundant set of outputs. This very problem is instrumental to unveiling the fundamental geometric properties of output-redundant systems, which form the basis for our subsequent constructions and results. As a direct application, we demonstrate how output allocation can be employed to effectively protect the input information from certain output eavesdroppers with guaranteed results.

This article addresses sequential Bayesian filtering for nonlinear and stochastic dynamical systems. We extend a Galerkin-approach that was previously used for the prediction of non-Gaussian probability density functions, to incorporate linear and non-linear measurement updates. The proposed method results in a linear pipeline of prediction and update steps, which are computed as sparse matrix operations on the finite-dimensional coefficient vector. The performance of our approach is demonstrated in numerical experiments for nonlinear dynamical 2D- and 4D-systems, using results of a standard particle filter as reference, both in terms of accuracy and computational expenses.

High penetration of wind energy is pushing wind farms (WFs) to offer grid support capabilities, such as active power tracking. One of the main challenges in active power tracking for WFs is the interaction of wind turbines (WTs) through their wakes. This reduces the available wind in downstream WTs, leading them to saturation, while also affecting structural loading. With the increasing number of WTs in individual WFs, the computational and communication complexity of implementing centralized control architectures grows, posing challenges for real-world applications. In this article, we present a novel distributed control approach for active power tracking for WFs, namely multirate consensus-based distributed control (MCDC). The MCDC is designed to ensure that tracking errors caused by WT saturation are equally compensated throughout the WF, while only requiring local information exchanges between WTs. Furthermore, the proposed controller ensures that WT aerodynamic loading is balanced across the WF in a distributed manner. Finally, the overall power reference is distributed via a leader–follower consensus algorithm, resulting in a fully distributed approach. Our control approach facilitates the WF modularity and sparsity, which reduces the costs associated with control design and its applicability. Throughout this article, we demonstrate the effectiveness of the proposed MCDC through high-fidelity simulations, presenting performance comparable to the centralized control.

Developing accurate models for batteries, capturing ageing effects and nonlinear behaviors, is critical for the development of efficient and effective performance. Due to the inherent difficulties in developing physics-based models, data-driven techniques have been gaining popularity. However, most machine learning methods are black boxes, lacking interpretability and requiring large amounts of labeled data. In this paper, we propose a physics-informed encoder–decoder model that learns from unlabeled data to separate slow-changing battery states, such as state of charge (SOC) and state of health (SOH), from fast transient responses, thereby increasing interpretability compared to conventional methods. By integrating physics-informed loss functions and modified architectures, we map the encoder output to quantifiable battery states, without needing explicit SOC and SOH labels. Our proposed approach is validated on a lithium-ion battery ageing dataset capturing dynamic discharge profiles that aim to mimic electric vehicle driving profiles. The model is trained and validated on sparse intermittent cycles (6 %–7 % of all cycles), accurately estimating SOC and SOH while providing accurate multistep ahead voltage predictions across single and multiple-cell based training scenarios.

The wireless communication used by vehicles in collaborative vehicle platoons is vulnerable to cyber-attacks, which threaten their safe operation. To address this issue we propose a topology-switching coalitional model predictive control (MPC) method based on a reduced order unknown input observer which detects and isolates the cyber-attacks, so that the attacked communication links can be disabled by means of a topology switch. Also, the MPC controller is designed to guarantee robustness against undetected attacks and the increase of uncertainty derived from disabling communication links. The proposed control method also conforms to a relaxed string stability condition and is guaranteed to be safe against crashes.

With the increasing share of renewable energy, concerns regarding ensuring power system stability are ever more relevant and have been accompanied by discussions to address this yet unsolved issue. Nonetheless, enhancing sparsity and increasing generation capacity by overplanting wind turbines not only mitigates the stability problem but also accelerates the transition from fossil fuel to renewable energy sources. With the high penetration of wind energy, there will be a paradigm shift from maximizing energy extraction to generating energy on demand. In this panorama, a cooperative wind farm control may strengthen the stability of the wind power plant through compensation strategies. Still, large-scale farms raise relevant control issues regarding computation effort and information sharing, such as topology constraints and communication overhead. Here, we contribute by presenting a multi-rate distributed control strategy based on average consensus. This strategy involves estimating the power-tracking errors at a fast sampling rate and executing local control actions that collaboratively mitigate these errors over an extended sampling period. This approach achieves performance comparable to that of the resource-intensive centralized approach. The reliability is therefore enhanced by improving the power regulation while reaching modularity and sparsity inside the farm.

We describe a new variant of zero dynamics attack (ZDA), what we call a switched ZDA, targeting linear time-invariant (LTI) sampled-data systems with non-uniform sampling. Specifically, we consider continuous-time systems and construct attacks that exploit the unstable sampling zeros resulting from a zero-order hold (ZOH) mechanism. These attacks can be constructed by strong adversaries who have knowledge of the plant dynamics, with the additional requirement that they can determine the next sampling instant. We provide sufficient conditions when cyber-physical systems are vulnerable to switched ZDAs, and prove that these attacks can be disruptive while remaining stealthy. We also provide two possible countermeasures that make switched ZDAs ineffective. The first countermeasure revolves around creating a mismatch between the next sampling instant as predicted by the adversary and the true one, which makes the switched ZDAs no longer stealthy. The second countermeasure relies on increasing the inter-sample times such that the system no longer contains unstable sampling zeros, making the switched ZDA no longer disruptive. We demonstrate the vulnerability of sampled-data systems with non-uniform sampling to switched ZDAs in several illustrative examples, and exemplify the effectiveness of the proposed countermeasures.

Model-based fault detection identifies anomalies by comparing a system's output with the prediction from a model. Although such a technique can be very powerful, it may suffer from the computational complexity of its underlying models, especially for large systems. An alternative approach that circumvents this cost increase uses barrier functions, which abstract the system's behaviour into a single value. In this paper, we propose a fault detection mechanism via output-based barrier functions, that does not require to estimate the full state, copes with noisy processes, and is tailored to safety-critical faults as given by a user-defined safe region. We leverage such a mechanism by introducing so-called p-fault tolerant sets, which guarantee that a faulty system requires at least p time steps before reaching any unsafe state. Our approach is validated through numerical experiments on two systems with linear and nonlinear dynamics, along with the classic three-tank model.