The design of physics-augmented neural networks (PANNs) for the purposes of constitutive modeling has received considerable attention as of late for a variety of material behaviors. Here, we revisit the classical framework of isotropic incompressible hyperelasticity in light of recent advances in the study of constitutive inequalities. We show that polyconvexity implies true-stress-true-strain monotonicity for a large class of incompressible strain-energy functions. The resulting elastic law obeys the physically reasonable Legendre-Hadamard (or ellipticity) condition as well as the notion of increasing stress with increasing strain. These results then inform the architecture of four distinct PANNs which are subsequently calibrated to three different sets of experimental data each. We show that different PANN parametrizations - satisfying the same constitutive constraints a priori - have varying approximation power for the description of material behavior. Moreover, even when distinct parametrizations perform comparatively well within the calibration regime, they show pronounced differences in extrapolation. This observation motivates a critical discussion about the predictive power of PANNs which also has implications for the modeling of more complex material behavior by virtue of neural networks.
This study investigates response time as a behavioral indicator related to listening effort (LE) for evaluating speech enhancement (SE) systems. English and Norwegian intelligibility matrix tests were conducted within a single-task paradigm that incorporated click-time recording (logging the precise time of all participant clicks), enabling simultaneous estimation of speech intelligibility and LE-related temporal behavior. Three temporal proxy measures for LE were examined—time per stimulus, reaction time, and word click time—across a broad range of input signal-to-noise ratios (SNRs) and for both discriminative and generative enhancement approaches. Time per stimulus showed an inverted-U pattern across SNRs, whereas reaction time and word click time exhibited monotonic behavior, providing more directly interpretable metrics for comparative evaluation. Analyses of pairwise SNR comparisons revealed that increases in our LE-related temporal measures at higher SNRs precede measurable intelligibility declines, suggesting that these temporal metrics can be more sensitive than intelligibility in this regime. Overall, the proposed framework—where LE-related measurements remain unknown to participants—offers a comprehensive and nuanced behavioral tool for SE evaluation, complementing intelligibility particularly under realistic, moderate-to-high SNR conditions.
Accurate production forecasts are essential for the integration of renewable energy sources into the power grid. This paper illustrates how to obtain probabilistic forecasts of wind power generation using gradient boosting trees and an ensemble of weather forecasts. To this end, we perform a comparative analysis across three state-of-the-art probabilistic prediction methods-conformalized quantile regression, natural gradient boosting and conditional diffusion models-all of which can be combined with tree-based machine learning. The methods are validated using four years of data for all Belgian offshore wind farms. We benchmark the models against the power curve and a calibrated wake model as well as a probabilistic method using stochastic variational Gaussian process regression. The tree-based models significantly reduce the mean absolute error in comparison to the deterministic baselines. Additionally, all three methods outperform the Gaussian process baseline in probabilistic skill, while two out of the three also improve point forecast accuracy. The conditional diffusion model attains the best performance, with improvements of 5% in mean absolute error and 12% in continuous rank probability score compared to the probabilistic baseline. Last, the results indicate an average improvement in point forecast accuracy of 17% by using an ensemble of weather forecasts instead of a single provider.
Maintaining the operational continuity of critical infrastructures during natural hazards is essential for supporting societal well-being and economic development. This paper introduces the RECINAT framework, as a proactive approach for assessing and enhancing the technical and economic infrastructure resilience against natural hazard-triggered technological events. The study specifically focuses on distribution power systems in the Nordic region, which face significant vulnerabilities from rainfall-induced landslides. This study utilizes an indicator-based methodology to quantify resilience through a hazard-preventive pathway with metrics such as normalized electrical energy not supplied (NEENS), interruption of nodes (ION), and the cost of energy not supplied (CENS). Integration of distributed energy resources configured as virtual power plants (VPPs) is proposed for system resilience enhancement. An enhanced IEEE 33-bus distribution test system is adapted to represent a regional grid in Southwest Norway as the case study. The strategic deployment of these VPPs is formulated as a multi-objective optimization problem and solved using the nondominated sorting Genetic Algorithm II (NSGA-II) as the targets of simultaneous minimization of NEENS, CENS, total VPP capacity, and active power losses. Numerical results demonstrate that optimized VPP deployment reduces economic losses from service interruptions and improves the overall redundancy resilience index. The proposed framework can support resilience-oriented operational planning and long-term investment decision-making for distribution networks exposed to increasing climate-related hazards.
With the increasing deployment of Cyber-Physical Systems (CPS), the risk landscape in chemical industrial parks has progressively shifted from the physical domain to the cyber domain. This paradigm shift enables cyber adversaries to manipulate physical processes, thereby inducing catastrophic accidents. Traditional Industrial Control Systems (ICS) defense methods often isolate cyber vulnerability analysis from physical consequence analysis, failing to address this cross-domain threat. To bridge this gap, a hierarchical risk transmission model is established based on a three-layer architecture of chemical park CPS, spanning the information, cyber-physical, and physical layers. Attack graphs are utilized to quantify cross-layer attack paths, and Bayesian networks are employed to infer the severity of physical consequences. A coupling index is proposed to prioritize node defense resources under limited budgets. Furthermore, an attack-defense game model solved by a greedy algorithm is developed to realize a honeypot deployment scheme. The framework is validated using both a real cascaded water tank experimental platform and the non-linear Tennessee Eastman Process benchmark. Results demonstrate that the proposed method effectively intercepts unknown attacks originating from the information layer, thereby preventing catastrophic physical intrusions.
