Data-driven diagnosis of high temperature PEM fuel cells based on the electrochemical impedance spectroscopy: Robustness improvement and evaluation

Utilizing machine learning techniques for data-driven diagnosis of high temperature PEM fuel cells is beneficial and meaningful to the system durability. Nevertheless, ensuring the robustness of diagnosis remains a critical and challenging task in real application. To enhance the robustness of diagnosis and achieve a more thorough evaluation of diagnostic performance, a robust diagnostic procedure based on electrochemical impedance spectroscopy (EIS) and a new method for evaluation of the diagnosis robustness was proposed and investigated in this work. To improve the diagnosis robustness: (1) the degradation mechanism of different faults in the high temperature PEM fuel cell was first analyzed via the distribution of relaxation time of EIS to determine the equivalent circuit model (ECM) with better interpretability, simplicity and accuracy;(2) the feature extraction was implemented on the identified parameters of the ECM and extra attention was paid to distinguishing between the long-term normal degradation and other faults;(3) a Siamese Network was adopted to get features with higher robustness in a new embedding. The diagnosis was conducted using 6 classic classification algorithms—support vector machine (SVM), K-nearest neighbor (KNN), logistic regression (LR), decision tree (DT), random forest (RF), and Naive Bayes employing a dataset comprising a total of 1935 collected EIS. To evaluate the robustness of trained models: (1) different levels of errors were added to the features for performance evaluation;(2) a robustness coefficient (Roubust_C) was defined for a quantified and explicit evaluation of the diagnosis robustness. The diagnostic models employing the proposed feature extraction method can not only achieve the higher performance of around 100% but also higher robustness for diagnosis models. Despite the initial performance being similar, the KNN demonstrated a superior robustness after feature selection and re-embedding by triplet-loss method, which suggests the necessity of robustness evaluation for the machine learning models and the effectiveness of the defined robustness coefficient. This work hopes to give new insights to the robust diagnosis of high temperature PEM fuel cells and more comprehensive performance evaluation of the data-driven method for diagnostic application.

Direct measurement of electrocaloric effect in P(VDF-TrFE-CFE)film using infrared imaging

Poly(vinylidene fluoride-trifluoroethylene-chlorofluoroethylene)P(VDF-TrFE-CFE)is a relaxor ferroelectric polymer,which exhibits a temperature-independent electrocaloric effect at room temperature.In this work,the electrocaloric effect in P(VDF-TrFE-CFE)film was directly analysed using infrared imaging.P(VDF-TrFE-CFE)64.8%/27.4%/7.8%(in mole)film of(15±1)mm thickness was deposited on polyethylene naphthalate substrate.Direct ECE of P(VDF-TrFE-CFE)film was measured from 15 to 35
C at different electric fields.A maximum adiabatic temperature change(DTad)of 3.58 K was measured during the cooling cycle at a field of 100 V/mm at 30
C.Finite element analysis of temperature dissipation through the sample estimated that the actual temperature change within P(VDF-TrFE-CFE)film was 4.3 K.Despite the thermal mass of the substrate,a substantial ECE was observed in P(VDF-TrFE-CFE)films.This electrocaloric terpolymer composition could be of interest for electrocaloric cooling applications.

Tailoring properties of hybrid perovskites by domain-width engineering with charged walls

Charged ferroelectric domain walls are fascinating electrical topological defects that can exhibit unusual properties.Here,in the search for novel phenomena,we perform and analyze first-principles calculations to investigate the effect of domain width on properties of domains with charged walls in the photovoltaic material consisting of methylammonium lead iodide hybrid perovskite.We report that such domains are stable and have rather low domain wall energy for any investigated width(that is,up to 13 lattice constants).Increasing the domain width first linearly decreases the electronic band gap from≃1.4 eV to about zero(which therefore provides an efficient band-gap engineering),before the system undergoes an insulator-to-metal transition and then remains metallic(with both the tail-to-tail and head-to-head domain walls being conductive)for the largest widths.All these results can be understood in terms of:(i)components of polarization along the normal of the domain walls being small in magnitude;(ii)an internal electric field that is basically independent of the domain width;and(iii)rather negligible charge transfer between walls.These findings deepen the knowledge of charged ferroelectric domain walls and can further broaden their potential for applications,particularly in the context of halide perovskites for photovoltaics.

Interfacial charge transfer and photocatalytic activity in a reverse designed Bi_(2)O_(3)/TiO_(2)core-shell

In this study,the electronic and photocatalytic properties of core-shell heterojunctions photocatalysts with reversible configuration of TiO_(2)and Bi_(2)O_(3)layers were studied.The core-shell nanostructure,obtained by efficient control of the sol-gel polymerization and impregnation method of variable precursors of semiconductors,makes it possible to study selectively the role of the interfacial charge transfer in each configuration.The morphological,optical,and chemical composition of the core-shell nanostructures were characterized by high-resolution transmission electron microscopy,UV-visible spectroscopy and X-ray photoelectron spectroscopy.The results show the formation of homogenous TiO_(2)anatase and Bi_(2)O_(3)layers with a thickness of around 10 and 8 nm,respectively.The interfacial charge carrier dynamic was tracked using time resolved microwave conductivity and transition photocurrent density.The charge transfer,their density,and lifetime were found to rely on the layout layers in the core-shell nanostructure.In optimal core-shell design,Bi_(2)O_(3)collects holes from TiO_(2),leaving electrons free to react and increase by 5 times the photocatalytic efficiency toward H2 generation.This study provides new insight into the importance of the design and elaboration of optimal heterojunction based on the photocatalyst system to improve the photocatalytic activity.

Creating multiferroic and conductive domain walls in common ferroelastic compounds

Domain walls in ferroelectrics and ferroelastics often present peculiar functional properties,offering an intriguing route toward the design of nano-devices.Here we use first-principles simulations to illustrate an approach for engineering such walls,working with representative ferroelastic perovskites LaGaO_(3) and CaTiO_(3)(insulating,non-magnetic,non-polar).We show that a wide range of substitutional dopants can be used to create long-range-ordered structures confined within the walls of these compounds,yielding functional interfaces with tailor-made properties.We thus identify clear-cut strategies to produce metallic walls within an insulating matrix.Further,we find ways to create magnetic walls that also display ferroelectric order(proper or improper),thus providing an original route to obtain magnetoelectric multiferroics.Given the recent developments on the preparation of high-density domain structures in perovskite films,our results suggest a definite path toward new functional nano-materials.

Ultra-fast interpretable machine-learning potentials

All-atom dynamics simulations are an indispensable quantitative tool in physics,chemistry,and materials science,but large systems and long simulation times remain challenging due to the trade-off between computational efficiency and predictive accuracy.To address this challenge,we combine effective two-and three-body potentials in a cubic B-spline basis with regularized linear regression to obtain machine-learning potentials that are physically interpretable,sufficiently accurate for applications,as fast as the fastest traditional empirical potentials,and two to four orders of magnitude faster than state-of-the-art machine-learning potentials.For data from empirical potentials,we demonstrate the exact retrieval of the potential.For data from density functional theory,the predicted energies,forces,and derived properties,including phonon spectra,elastic constants,and melting points,closely match those of the reference method.The introduced potentials might contribute towards accurate all-atom dynamics simulations of large atomistic systems over long-time scales.

Giant multiphononic effects in a perovskite oxide

Perovskite oxides offer tremendous potential for applications in information storage and energy conversion,owing to a subtle interplay between their spin,charge,orbital and lattice degrees of freedom.Here,we further expand the possible range of perovskite oxides operation towards the fields of thermal management and thermal computing by exploiting an exceptional synergy between different ferroic orders.We propose dynamical control of the heat flow in a distinctive family of perovskite oxides obtained via the application of small electric(~10 kV/cm)and/or magnetic(~1 T)fields.Based on first-principles simulations,we predict a relative heat conductivity variation of~100%in SrMnO_(3) thin films near room temperature resulting from a phase transition that involves huge changes in both the magnetization and electric polarization.The disclosed giant multiphononic effects are fundamentally caused by anharmonic spin-phonon couplings that strongly influence the mean lifetime of phonons.

Semantics for linking data from 4D BIM to digital collaborative support

Synchronous collaboration sessions within the context of 4D BIM position construction professionals into a complex socio—technical system.This system includes hardware,software,people,and broader community aspects.This article strictly focuses on the ontology representation of synchronous collaboration sessions with collocated collective decision-making.The model is designed by considering various 4D BIM model uses while a digital multiuser touch table facilitates the collaboration between actors.The outlined ontological model aims to improve interoperability and to move toward a knowledge-driven,smart-built environment paradigm.A knowledge engineering methodology is outlined,by virtue of which the semantics of the presented model are defined and discussed.Concepts from nearby knowledge fields,especially from the Industry Foundation Classes,are reused.Several examples on querying the knowledge base according to the project meeting requirements are outlined to demonstrate the benefits of using the model.Although 4D BIM model data can be imported by using standard formats,capturing data about the social context remains a challenge in the future.This is expected to change the ontology model structure by considering user ergonomics,data modeling requirements,as well as technical implementation constraints.

The reinforcement learning method for occupant behavior in building control:A review

Occupant behavior in buildings has been considered the major source of uncertainty for assessing energy con-sumption and building performance.Modeling frameworks are usually built to accomplish a certain task,but the stochasticity of the occupant makes it difficult to apply that experience to a similar but distinct environment.For complex and dynamic environments,the development of smart devices and computing power makes intelligent control methods for occupant behaviors more viable.It is expected that they will make a substantial contribution to reducing global energy consumption.Among these control techniques,the reinforcement learning(RL)method seems distinctive and applicable.The success of the reinforcement learning method in many artificial intelligence applications has given an explicit indication of how this method might be used to model and adjust occupant behavior in building control.Fruitful algorithms complement each other and guarantee the quality of the opti-mization.However,the examination of occupant behavior based on reinforcement learning methodologies is not well established.The way that occupant interacts with the RL agent is still unclear.This study briefly reviews the empirical applications using reinforcement learning,how they have contributed to shaping the modeling paradigms and how they might suggest a future research direction.