Data-driven materials science has realized a new paradigm by integrating materials domain knowledge and machine-learning (ML) techniques. However, ML-based research has often overlooked the inherent limitation in predicting unknown data: extrapolative performance, especially when dealing with small-scale experimental datasets. Here, we present a comprehensive benchmark for assessing extrapolative performance across 12 organic molecular properties. Our large-scale benchmark reveals that conventional ML models exhibit remarkable performance degradation beyond the training distribution of property range and molecular structures, particularly for small-data properties. To address this challenge, we introduce a quantum-mechanical (QM) descriptor dataset, called QMex, and an interactive linear regression (ILR), which incorporates interaction terms between QM descriptors and categorical information pertaining to molecular structures. The QMex-based ILR achieved state-of-the-art extrapolative performance while preserving its interpretability. Our benchmark results, QMex dataset, and proposed model serve as valuable assets for improving extrapolative predictions with small experimental datasets and for the discovery of novel materials/molecules that surpass existing candidates.

As the number of electric vehicles (EV), a new category of clean and environmentally friendly transportation means, increases yearly, which not only puts forward new requirements for the modernization of the infrastructure of transportation system, but also demands that the impact on power system be given significant consideration. The increasing of EV utilization in transportation sector and renewable energy integration, such as, PV panel, in local power grid exacerbates system frequency fluctuation. Regulation of system frequency deviation by EV battery storage charging/discharging is a crucial issue for the time being. In this article, a power grid model with a high proportion of PV power is constructed, numerical simulation using time series data is carried out to observe the contribution of the charging/discharging control of aggregated EV group to vehicle to grid (V2G) system frequency regulation. By this way, constructive suggestions are given for the development of charging stations and EV charging/discharging plan, so that the transportation system and power system can be better coordinated and integrated.

Electric field measurement based on electric-field-induced second-harmonic generation (E-FISHG) of gas molecules is a promising method that enables noninvasive field measurement in gases. Although the probing laser beam is focused on the target area by a lens, the signal detected by the E-FISHG method is affected by the electric field along the probing laser path. The influencing path spans many tens of times the Rayleigh length of the front and rear of the focused spot. The E-FISHG signal has been frequently calibrated under uniform electric fields. However, this can yield a significantly incorrect calibrated value if the difference between the electric-field profiles in the measurement and calibration is not considered. In this paper, we propose a method to calibrate and furthermore restore the electric-field profile along a probing optical path by measuring a series of E-FISHG signals by changing the focusing spot position. The electric-field profile along the probing laser path is successfully restored from the measured E-FISHG signal sequence, which is verified by an experiment.

Surface charge accumulation on an insulating spacer is a critical issue that causes electric field distortion and flashover voltage reduction in DC-GIS. In the operating environment, high-voltage electrode is heated by Joule heat, and the surface charge accumulation is strongly affected by the temperature. This study reports the temperature dependence of surface charge accumulation on DC-GIS insulating spacer and its impact on electric field distortion. Surface charge density distribution is measured on a downsized epoxy spacer under isothermal conditions at room temperature, 40 °C, and 70 °C. Special care is taken to realize a low-moisture and low-dust environment. We obtain the saturation tendency of homo-charge accumulation with less random charges and its temperature dependence. Homo-charge accumulation develops more rapidly and more extensively on the spacer at the higher temperatures. Our results indicate that the surface charge accumulation cannot be explained by a conventional resistive field calculation, and that an explanation based on charge trapping is reasonable. They also show that the homo-charge accumulation significantly contributes to the electric field enhancement after the polarity reversal in our experimental system, especially at high temperatures.

Epoxy resin is widely used as electrical insulation material and its electrical degradation is significantly influenced by ionic conduction. However, the theoretical understanding of ionic transfer in epoxy is limited. In this study, sodium ion (Na+) and chloride ion (Cl-) mobility in epoxy resin are computed using molecular dynamics (MD) simulation, and their electric field dependence and temperature dependence are discussed. The calculated ionic mobility is consistent with the experimental results. The ionic mobility and its Arrhenius activation energy of Cl-are respectively larger and smaller than those of Na+. The MD results also suggest that the effect of free volume on ionic conduction is small. To further analyze the ion transport mechanism, the interaction energy between the ion and surrounding epoxy molecules when the ion is at a stable “trap” site and a “barrier” site for ion is calculated. The results show that the potential barrier for ion transportation in epoxy resin is smaller for Cl-than Na+, and hence Cl-ion mobility is larger than Na+. Furthermore, the effect of water in epoxy on ionic mobility is computed. The computed ionic mobility is larger for the water-containing system. This is because hydration inhibits the bonding between the ions and epoxy molecules. A comparison is also conducted between the MD computed mobility and the mobility calculated by classical ionic conduction theory. This study reveals the relationship between the microscopic interactions of the ion and surrounding molecules and the macroscopic ionic mobility in epoxy.