Machine learning(ML)is recognized as a potent tool for the inverse design of environmental functional material,particularly for complex entities like biochar-based catalysts(BCs).Thus,the tailored BCs can have a disti...Machine learning(ML)is recognized as a potent tool for the inverse design of environmental functional material,particularly for complex entities like biochar-based catalysts(BCs).Thus,the tailored BCs can have a distinct ability to trigger the nonradical pathway in advance oxidation processes(AOPs),promising a stable,rapid and selective degradation of persistent contaminants.However,due to the inherent“black box”nature and limitations of input features,results and conclusions derived from ML may not always be intuitively understood or comprehensively validated.To tackle this challenge,we linked the front-point interpretable analysis approaches with back-point density functional theory(DFT)calculations to form a chained learning strategy for deeper sight into the intrinsic activation mechanism of BCs in AOPs.At the front point,we conducted an easy-to-interpret meta-analysis to validate two strategies for enhancing nonradical pathways by increasing oxygen content and specific surface area(SSA),and prepared oxidized biochar(OBC500)and SSA-increased biochar(SBC900)by controlling pyrolysis conditions and modification methods.Subsequently,experimental results showed that OBC500 and SBC900 had distinct dominant degradation pathways for 1O2 generation and electron transfer,respectively.Finally,at the end point,DFT calculations revealed their active sites and degradation mechanisms.This chained learning strategy elucidates fundamental principles for BC inverse design and showcases the exceptional capacity to integrate computational techniques to accelerate catalyst inverse design.展开更多
In this study,an inverse design framework was established to find lightweight honeycomb structures(HCSs)with high impact resistance.The hybrid HCS,composed of re-entrant(RE)and elliptical annular re-entrant(EARE)honey...In this study,an inverse design framework was established to find lightweight honeycomb structures(HCSs)with high impact resistance.The hybrid HCS,composed of re-entrant(RE)and elliptical annular re-entrant(EARE)honeycomb cells,was created by constructing arrangement matrices to achieve structural lightweight.The machine learning(ML)framework consisted of a neural network(NN)forward regression model for predicting impact resistance and a multi-objective optimization algorithm for generating high-performance designs.The surrogate of the local design space was initially realized by establishing the NN in the small sample dataset,and the active learning strategy was used to continuously extended the local optimal design until the model converged in the global space.The results indicated that the active learning strategy significantly improved the inference capability of the NN model in unknown design domains.By guiding the iteration direction of the optimization algorithm,lightweight designs with high impact resistance were identified.The energy absorption capacity of the optimal design reached 94.98%of the EARE honeycomb,while the initial peak stress and mass decreased by 28.85%and 19.91%,respectively.Furthermore,Shapley Additive Explanations(SHAP)for global explanation of the NN indicated a strong correlation between the arrangement mode of HCS and its impact resistance.By reducing the stiffness of the cells at the top boundary of the structure,the initial impact damage sustained by the structure can be significantly improved.Overall,this study proposed a general lightweight design method for array structures under impact loads,which is beneficial for the widespread application of honeycomb-based protective structures.展开更多
In recent years,the use of deep learning to replace traditional numerical methods for electromagnetic propagation has shown tremendous potential in the rapid design of photonic devices.However,most research on deep le...In recent years,the use of deep learning to replace traditional numerical methods for electromagnetic propagation has shown tremendous potential in the rapid design of photonic devices.However,most research on deep learning has focused on single-layer grating couplers,and the accuracy of multi-layer grating couplers has not yet reached a high level.This paper proposes and demonstrates a novel deep learning network-assisted strategy for inverse design.The network model is based on a multi-layer perceptron(MLP)and incorporates convolutional neural networks(CNNs)and transformers.Through the stacking of multiple layers,it achieves a high-precision design for both multi-layer and single-layer raster couplers with various functionalities.The deep learning network exhibits exceptionally high predictive accuracy,with an average absolute error across the full wavelength range of 1300–1700 nm being only 0.17%,and an even lower predictive absolute error below 0.09%at the specific wavelength of 1550 nm.By combining the deep learning network with the genetic algorithm,we can efficiently design grating couplers that perform different functions.Simulation results indicate that the designed single-wavelength grating couplers achieve coupling efficiencies exceeding 80%at central wavelengths of 1550 nm and 1310 nm.The performance of designed dual-wavelength and broadband grating couplers also reaches high industry standards.Furthermore,the network structure and inverse design method are highly scalable and can be applied not only to multi-layer grating couplers but also directly to the prediction and design of single-layer grating couplers,providing a new perspective for the innovative development of photonic devices.展开更多
Finding materials with specific properties is a hot topic in materials science.Traditional materials design relies on empirical and trial-and-error methods,requiring extensive experiments and time,resulting in high co...Finding materials with specific properties is a hot topic in materials science.Traditional materials design relies on empirical and trial-and-error methods,requiring extensive experiments and time,resulting in high costs.With the development of physics,statistics,computer science,and other fields,machine learning offers opportunities for systematically discovering new materials.Especially through machine learning-based inverse design,machine learning algorithms analyze the mapping relationships between materials and their properties to find materials with desired properties.This paper first outlines the basic concepts of materials inverse design and the challenges faced by machine learning-based approaches to materials inverse design.Then,three main inverse design methods—exploration-based,model-based,and optimization-based—are analyzed in the context of different application scenarios.Finally,the applications of inverse design methods in alloys,optical materials,and acoustic materials are elaborated on,and the prospects for materials inverse design are discussed.The authors hope to accelerate the discovery of new materials and provide new possibilities for advancing materials science and innovative design methods.展开更多
Flowfield inverse design can obtain the desired flow and contour with high design efficiency,short design cycle,and small modification need.In this study,the Euler equations are formulated in the stream-function coord...Flowfield inverse design can obtain the desired flow and contour with high design efficiency,short design cycle,and small modification need.In this study,the Euler equations are formulated in the stream-function coordinates and combined with the given boundary conditions to derive a gridless space-marching method for the inverse design of subsonic,transonic,and supersonic flowfields.Designers can prescribe the flow parameters along the reference streamline to design flowfields and aerodynamic contours.The method is validated by the theoretical transonic solution,computational fluid dynamics,and experimental data,respectively.The method supports the fabrication of a Mach 2.0 single expansion tunnel.The calibration data agree well with the prescribed pressure distribution.The method is successfully applied to inverse design of contractions,nozzles,and asymmetric channels.Compared to classical analytic contractions,the contractions designed by the space-marching method provide a more accurate transonic flow.Compared to the classical Sivells’nozzle,the nozzle designed by the space-marching method provides a smaller workload,a more flexible velocity distribution,a 20%reduction in length,and an equally uniform flow.Additionally,the space-marching method is applied to design the asymmetric channels under various Mach numbers.These asymmetric channels perfectly eliminate Mach waves,achieving the shock-free flow turning and high flow uniformity.These results validate the feasibility of the space-marching method,making it a good candidate for the inverse design of subsonic,transonic,and supersonic internal flowfields and aerodynamic contours.展开更多
Traditional inverse neural network(INN)approaches for inverse design typically require auxiliary feedforward networks,leading to increased computational complexity and architectural dependencies.This study introduces ...Traditional inverse neural network(INN)approaches for inverse design typically require auxiliary feedforward networks,leading to increased computational complexity and architectural dependencies.This study introduces a standalone INN methodology that eliminates the need for feedforward networks while maintaining high reconstruction accuracy.The approach integrates Principal Component Analysis(PCA)and Partial Least Squares(PLS)for optimized feature space learning,enabling the standalone INN to effectively capture bidirectionalmappings between geometric parameters and mechanical properties.Validation using established numerical datasets demonstrates that the standalone INN architecture achieves reconstruction accuracy equal or better than traditional tandem approaches while completely eliminating the workload and training time required for Feedforward Neural Networks(FNN).These findings contribute to AI methodology development by proving that standalone invertible architectures can achieve comparable performance to complex hybrid systems with significantly improved computational efficiency.展开更多
The discovery of new superconducting materials,particularly those exhibiting high critical temperature(Tc),has been a vibrant area of study within the field of condensed matter physics.Conventional approaches primaril...The discovery of new superconducting materials,particularly those exhibiting high critical temperature(Tc),has been a vibrant area of study within the field of condensed matter physics.Conventional approaches primarily rely on physical intuition to search for potential superconductors within the existing databases.However,the known materials only scratch the surface of the extensive array of possibilities within the realm of materials.展开更多
Research on detonation has traditionally focused on forward solutions,with limited attention to inverse design methods,which has significantly hindered the development of detonation engines.In this paper,the Method of...Research on detonation has traditionally focused on forward solutions,with limited attention to inverse design methods,which has significantly hindered the development of detonation engines.In this paper,the Method of characteristics for Curved-Detonation(MOCD)is proposed to enable the inverse design of detonation waves.MOCD is based on the Method of Curved-shock Characteristics(MOCC)and integrates higher-order aerodynamic parameters from Curved Detonation Equations(CDE),allowing the calculation of the wedge angle given specific wave angle.The effectiveness of MOCD is validated using both oblique and curved detonation waves with single-step and detailed chemical reactions.Various applications demonstrate the ability to meet the inverse design requirements of detonation engines.For example,inverse design for given wave angles can optimize engine thrust and prevent Mach reflections.Additionally,inverse design schemes tailored to incoming flow conditions,such as varying Mach numbers and equivalence ratios,enhance the feasibility of detonation engines.Applying the method to given post-wave aerodynamic parameters enables more precise engine design,which is crucial for improving propulsion performance and effective thermal protection.In summary,the advantages of MOCD include not only performing a fast solution of the detonation flow field,but also allowing the inverse design of the detonation wave.展开更多
Reverse design of highly GeO2-doped silica optical fibers with broadband and flat dispersion profiles is proposed using a neural network(NN) combined with a particle swarm optimization(PSO) algorithm.Firstly,the NN mo...Reverse design of highly GeO2-doped silica optical fibers with broadband and flat dispersion profiles is proposed using a neural network(NN) combined with a particle swarm optimization(PSO) algorithm.Firstly,the NN model designed to predict optical fiber dispersion is trained with an appropriate choice of hyperparameters,achieving a root mean square error(RMSE) of 9.47×10-7on the test dataset,with a determination coefficient(R2) of 0.999.Secondly,the NN is combined with the PSO algorithm for the inverse design of dispersion-flattened optical fibers.To expand the search space and avoid particles becoming trapped in local optimal solutions,the PSO algorithm incorporates adaptive inertia weight updating and a simulated annealing algorithm.Finally,by using a suitable fitness function,the designed fibers exhibit flat group velocity dispersion(GVD) profiles at 1 400—2 400 nm,where the GVD fluctuations and minimum absolute GVD values are below 18 ps·nm-1·km-1and 7 ps·nm-1·km-1,respectively.展开更多
Structural colors based on metasurfaces have very promising applications in areas such as optical image encryption and color printing.Herein,we propose a deep learning-enabled reverse design of polarization-selective ...Structural colors based on metasurfaces have very promising applications in areas such as optical image encryption and color printing.Herein,we propose a deep learning-enabled reverse design of polarization-selective structural color based on coding metasurface.In this study,the long short-term memory(LSTM)neural network is presented to enable the forward and inverse mapping between coding metasurface structure and corresponding color.The results show that the method can achieve 98%accuracy for the forward prediction of color and 93%accuracy for the inverse design of the structure.Moreover,a cascaded architecture is adopted to train the inverse neural network model,which can solve the nonuniqueness problem of the polarization-selective color reverse design.This study provides a new path for the application and development of structural colors.展开更多
The traditional forward design process of metasurface optical filters is computationally costly and time-consuming;therefore,inverse design based on deep learning(DL)can help accelerate the process.We propose the glob...The traditional forward design process of metasurface optical filters is computationally costly and time-consuming;therefore,inverse design based on deep learning(DL)can help accelerate the process.We propose the globaland local-spectrum-aware transformer(GLSaT),a DL model that concerns the intrinsic correlations within the spectral sequences,compensating the drawbacks of current networks that only focus on structure-to-spectrum mappings.With both interand intra-fragment attention mechanisms implemented,the GLSaT achieves 32.9%higher accuracy than fully connected networks in our reflection tests.It also demonstrates an inherent balance between predictive precision and computational efficiency,outperforming alternative architectures.Furthermore,our extensive experimental validations demonstrate its generalization capability across diverse metasurface functionalities.The GLSaT architecture shows great potential for enhancing the efficiency of data-driven metasurface inverse design in the future.展开更多
In order to develop a generic framework capable of designing novel amorphous alloys with selected target properties,a predictor−corrector inverse design scheme(PCIDS)consisting of a predictor module and a corrector mo...In order to develop a generic framework capable of designing novel amorphous alloys with selected target properties,a predictor−corrector inverse design scheme(PCIDS)consisting of a predictor module and a corrector module was presented.A high-precision forward prediction model based on deep neural networks was developed to implement these two parts.Of utmost importance,domain knowledge-guided inverse design networks(DKIDNs)and regular inverse design networks(RIDNs)were also developed.The forward prediction model possesses a coefficient of determination(R^(2))of 0.990 for the shear modulus and 0.986 for the bulk modulus on the testing set.Furthermore,the DKIDNs model exhibits superior performance compared to the RIDNs model.It is finally demonstrated that PCIDS can efficiently predict amorphous alloy compositions with the required target properties.展开更多
We proposed and demonstrated the ultra-compact 1310/1550 nm wavelength multiplexer/demultiplexer assisted by subwavelength grating(SWG)using particle swarm optimization(PSO)algorithm in silicon-on-insulator(SOI)platfo...We proposed and demonstrated the ultra-compact 1310/1550 nm wavelength multiplexer/demultiplexer assisted by subwavelength grating(SWG)using particle swarm optimization(PSO)algorithm in silicon-on-insulator(SOI)platform.Through the self-imaging effect of multimode interference(MMI)coupler,the demultiplexing function for 1310 nm and 1550 nm wavelengths is implemented.After that,three parallel SWG-based slots are inserted into the MMI section so that the effective refractive index of the modes can be engineered and thus the beat length can be adjusted.Importantly,these three SWG slots significantly reduce the length of the device,which is much shorter than the length of traditional MMI-based wavelength demultiplexers.Ultimately,by using the PSO algorithm,the equivalent refractive index and width of the SWG in a certain range are optimized to achieve the best performance of the wavelength demultiplexer.It has been verified that the device footprint is only 2×30.68μm^(2),and 1 dB bandwidths of larger than 120 nm are acquired at 1310 nm and 1550 nm wavelengths.Meanwhile,the transmitted spectrum shows that the insertion loss(IL)values are below 0.47 dB at both wavelengths when the extinction ratio(ER)values are above 12.65 dB.This inverse design approach has been proved to be efficient in increasing bandwidth and reducing device length.展开更多
Terahertz(THz)metamaterials,with their exceptional ability to precisely manipulate the phase,amplitude,polarization and orbital angular momentum(OAM)of electromagnetic waves,have demonstrated significant application p...Terahertz(THz)metamaterials,with their exceptional ability to precisely manipulate the phase,amplitude,polarization and orbital angular momentum(OAM)of electromagnetic waves,have demonstrated significant application potential across a wide range of fields.However,traditional design methodologies often rely on extensive parameter sweeps,making it challenging to address the increasingly complex and diverse application requirements.Recently,the integration of artificial intelligence(AI)techniques,particularly deep learning and optimization algorithms,has introduced new approaches for the design of THz metamaterials.This paper reviews the fundamental principles of THz metamaterials and their intelligent design methodologies,with a particular focus on the advancements in AI-driven inverse design of THz metamaterials.The AI-driven inverse design process allows for the creation of THz metamaterials with desired properties by working backward from the unit structures and array configurations of THz metamaterials,thereby accelerating the design process and reducing both computational resources and time.It examines the critical role of AI in improving both the functionality and design efficiency of THz metamaterials.Finally,we outline future research directions and technological challenges,with the goal of providing valuable insights and guidance for ongoing and future investigations.展开更多
Conventional pit excavation engineering methods often struggle to manage the complex deformation patterns associated with asymmetric excavations,resulting in significant safety risks and increased project costs.These ...Conventional pit excavation engineering methods often struggle to manage the complex deformation patterns associated with asymmetric excavations,resulting in significant safety risks and increased project costs.These challenges highlight the need for more precise and efficient design methodologies to ensure structural stability and economic feasibility.This research proposes an innovative automatic optimization inverse design method(AOIDM)that integrates an enhanced genetic algorithm(EGA)with a multiobjective optimization model.By combining advanced computational techniques with engineering principles,this approach improves search efficiency by 30%and enhances deformation control accuracy by 25%.Additionally,the approach exhibits potential for reducing carbon emissions to align with sustainable engineering goals.The effectiveness of this approach was validated through comprehensive data analysis and practical case studies,demonstrating its ability to optimize retaining structure designs under complex asymmetric loading conditions.This research establishes a new standard for precision and efficiency in automated excavation design,with accompanying improvements in safety and cost-effectiveness.Furthermore,it lays the foundation for future geotechnical engineering advancements,offering a robust solution to one of the most challenging aspects of modern excavation projects.展开更多
Liquid crystal elastomers(LCEs)are advanced materials characterized by their rubber-like hyperelasticity and liquid crystal phase transitions,offering exceptional mechanical properties.The development of smart mechani...Liquid crystal elastomers(LCEs)are advanced materials characterized by their rubber-like hyperelasticity and liquid crystal phase transitions,offering exceptional mechanical properties.The development of smart mechanical metamaterials(SMMs)from LCEs expands the potential for controlling mechanical responses and achieving nonlinear behaviors not possible with traditional metamaterials.However,the challenge lies in managing the interplay between nonlinear material responses and structural complexity,making the inverse design of LCE-based SMMs exceptionally demanding.In this paper,we introduce a design framework for LCE smart mechanical metamaterials that leverages neural networks and evolution strategies(ES)to optimize designs with nonlinear mechanical responses.Our approach involves constructing a flexible,unit-cell-based metamaterial model that integrates the soft elastic behavior and thermo-mechanical coupling of LCEs.The combination of microscopic liquid crystal molecule rotation and macroscopic block rotation enables highly tunable and nonlinear mechanical behaviors,of which the precise inverse design of stress-stretch responses is obtained via neural networks combined with ES.In addition,stimuli responses in the liquid crystal elastomers enable real-time adaptability and achieve tailored stress plateaus that are not possible with traditional metamaterials.Our findings provide new pathways in the design and optimization of advanced materials in flexible electronic devices,intelligent actuators,and systems for energy absorption and dissipation.展开更多
Chiral metamaterials are manmade structures with extraordinary mechanical properties derived from their special geometric design instead of chemical composition.To make the mechanical deformation programmable,the non-...Chiral metamaterials are manmade structures with extraordinary mechanical properties derived from their special geometric design instead of chemical composition.To make the mechanical deformation programmable,the non-uniform rational B-spline(NURBS)curves are taken to replace the traditional ligament boundaries of the chiral structure.The Neural networks are innovatively inserted into the calculation of mechanical properties of the chiral structure instead of finite element methods to improve computational efficiency.For the problem of finding structure configuration with specified mechanical properties,such as Young’s modulus,Poisson’s ratio or deformation,an inverse design method using the Neural network-based proxy model is proposed to build the relationship between mechanical properties and geometric configuration.To satisfy some more complex deformation requirements,a non-homogeneous inverse design method is proposed and verified through simulation and experiments.Numerical and test results reveal the high computational efficiency and accuracy of the proposed method in the design of chiral metamaterials.展开更多
To capture the nonlinear dynamics and gain evolution in chirped pulse amplification(CPA)systems,the split-step Fourier method and the fourth-order Runge–Kutta method are integrated to iteratively address the generali...To capture the nonlinear dynamics and gain evolution in chirped pulse amplification(CPA)systems,the split-step Fourier method and the fourth-order Runge–Kutta method are integrated to iteratively address the generalized nonlinear Schrödinger equation and the rate equations.However,this approach is burdened by substantial computational demands,resulting in significant time expenditures.In the context of intelligent laser optimization and inverse design,the necessity for numerous simulations further exacerbates this issue,highlighting the need for fast and accurate simulation methodologies.Here,we introduce an end-to-end model augmented with active learning(E2E-AL)with decent generalization through different dedicated embedding methods over various parameters.On an identical computational platform,the artificial intelligence–driven model is 2000 times faster than the conventional simulation method.Benefiting from the active learning strategy,the E2E-AL model achieves decent precision with only two-thirds of the training samples compared with the case without such a strategy.Furthermore,we demonstrate a multi-objective inverse design of the CPA systems enabled by the E2E-AL model.The E2E-AL framework manifests the potential of becoming a standard approach for the rapid and accurate modeling of ultrafast lasers and is readily extended to simulate other complex systems.展开更多
Polymer flooding is a widely used technique in enhanced oil recovery (EOR),but its effectiveness is often hindered by the poor viscosity retention of conventional polymers like hydrolyzed polyacrylamide (HPAM) under h...Polymer flooding is a widely used technique in enhanced oil recovery (EOR),but its effectiveness is often hindered by the poor viscosity retention of conventional polymers like hydrolyzed polyacrylamide (HPAM) under high-salinity conditions.Although recent advances in molecular engineering have concentrated on modifying polymer architecture and functional groups to address this issue,the complex interplay among polymer topology,charge distribution and hydrophilic-hydrophobic balance renders rational molecular design challenging.In this work,we present an AI-driven inverse design framework that directly maps target viscosity performance back to optimal molecular structures.Guided by practical molecular design strategies,the topological features (grafting density,side-chain length) and functional group-related features(copolymerization ratio,hydrophilic-hydrophobic balance) are encoded into a multidimensional design space.By integrating dissipative particle dynamics simulations with particle swarm algorithm,the framework efficiently explores the design space and identifies non-intuitive,high-performing polymer structure.The optimized polymer achieves a 12%enhancement in viscosity,attributed to the synergistic effect of electrostatic chain extension and hydrophobic aggregation.This study demonstrates the promise of AI-guided inverse design for developing next-generation EOR polymers and provides a generalizable approach for the discovery of functional soft materials.展开更多
The integrated waveguide polarizer is essential for photonic integrated circuits,and various designs of waveguide polarizers have been developed.As the demand for dense photonic integration increases rapidly,new strat...The integrated waveguide polarizer is essential for photonic integrated circuits,and various designs of waveguide polarizers have been developed.As the demand for dense photonic integration increases rapidly,new strategies to minimize the device size are needed.In this paper,we have inversely designed an integrated transverse electric pass(TE-pass)polarizer with a footprint of 2.88μm×2.88μm,which is the smallest footprint ever achieved.A direct binary search algorithm is used to inversely design the device for maximizing the transverse electric(TE)transmission while minimizing transverse magnetic(TM)transmission.Finally,the inverse-designed device provides an average insertion loss of 0.99 dB and an average extinction ratio of 33 dB over a wavelength range of 100 nm.展开更多
基金supported by Project of National and Local Joint Engineering Research Center for Biomass Energy Development and Utilization(Harbin Institute of Technology,No.2021A004).
文摘Machine learning(ML)is recognized as a potent tool for the inverse design of environmental functional material,particularly for complex entities like biochar-based catalysts(BCs).Thus,the tailored BCs can have a distinct ability to trigger the nonradical pathway in advance oxidation processes(AOPs),promising a stable,rapid and selective degradation of persistent contaminants.However,due to the inherent“black box”nature and limitations of input features,results and conclusions derived from ML may not always be intuitively understood or comprehensively validated.To tackle this challenge,we linked the front-point interpretable analysis approaches with back-point density functional theory(DFT)calculations to form a chained learning strategy for deeper sight into the intrinsic activation mechanism of BCs in AOPs.At the front point,we conducted an easy-to-interpret meta-analysis to validate two strategies for enhancing nonradical pathways by increasing oxygen content and specific surface area(SSA),and prepared oxidized biochar(OBC500)and SSA-increased biochar(SBC900)by controlling pyrolysis conditions and modification methods.Subsequently,experimental results showed that OBC500 and SBC900 had distinct dominant degradation pathways for 1O2 generation and electron transfer,respectively.Finally,at the end point,DFT calculations revealed their active sites and degradation mechanisms.This chained learning strategy elucidates fundamental principles for BC inverse design and showcases the exceptional capacity to integrate computational techniques to accelerate catalyst inverse design.
基金the financial supports from National Key R&D Program for Young Scientists of China(Grant No.2022YFC3080900)National Natural Science Foundation of China(Grant No.52374181)+1 种基金BIT Research and Innovation Promoting Project(Grant No.2024YCXZ017)supported by Science and Technology Innovation Program of Beijing institute of technology under Grant No.2022CX01025。
文摘In this study,an inverse design framework was established to find lightweight honeycomb structures(HCSs)with high impact resistance.The hybrid HCS,composed of re-entrant(RE)and elliptical annular re-entrant(EARE)honeycomb cells,was created by constructing arrangement matrices to achieve structural lightweight.The machine learning(ML)framework consisted of a neural network(NN)forward regression model for predicting impact resistance and a multi-objective optimization algorithm for generating high-performance designs.The surrogate of the local design space was initially realized by establishing the NN in the small sample dataset,and the active learning strategy was used to continuously extended the local optimal design until the model converged in the global space.The results indicated that the active learning strategy significantly improved the inference capability of the NN model in unknown design domains.By guiding the iteration direction of the optimization algorithm,lightweight designs with high impact resistance were identified.The energy absorption capacity of the optimal design reached 94.98%of the EARE honeycomb,while the initial peak stress and mass decreased by 28.85%and 19.91%,respectively.Furthermore,Shapley Additive Explanations(SHAP)for global explanation of the NN indicated a strong correlation between the arrangement mode of HCS and its impact resistance.By reducing the stiffness of the cells at the top boundary of the structure,the initial impact damage sustained by the structure can be significantly improved.Overall,this study proposed a general lightweight design method for array structures under impact loads,which is beneficial for the widespread application of honeycomb-based protective structures.
基金sponsored by the National Key Scientific Instrument and Equipment Development Projects of China(Grant No.62027823)the National Natural Science Foun-dation of China(Grant No.61775048).
文摘In recent years,the use of deep learning to replace traditional numerical methods for electromagnetic propagation has shown tremendous potential in the rapid design of photonic devices.However,most research on deep learning has focused on single-layer grating couplers,and the accuracy of multi-layer grating couplers has not yet reached a high level.This paper proposes and demonstrates a novel deep learning network-assisted strategy for inverse design.The network model is based on a multi-layer perceptron(MLP)and incorporates convolutional neural networks(CNNs)and transformers.Through the stacking of multiple layers,it achieves a high-precision design for both multi-layer and single-layer raster couplers with various functionalities.The deep learning network exhibits exceptionally high predictive accuracy,with an average absolute error across the full wavelength range of 1300–1700 nm being only 0.17%,and an even lower predictive absolute error below 0.09%at the specific wavelength of 1550 nm.By combining the deep learning network with the genetic algorithm,we can efficiently design grating couplers that perform different functions.Simulation results indicate that the designed single-wavelength grating couplers achieve coupling efficiencies exceeding 80%at central wavelengths of 1550 nm and 1310 nm.The performance of designed dual-wavelength and broadband grating couplers also reaches high industry standards.Furthermore,the network structure and inverse design method are highly scalable and can be applied not only to multi-layer grating couplers but also directly to the prediction and design of single-layer grating couplers,providing a new perspective for the innovative development of photonic devices.
基金funded by theNationalNatural Science Foundation of China(52061020)Major Science and Technology Projects in Yunnan Province(202302AG050009)Yunnan Fundamental Research Projects(202301AV070003).
文摘Finding materials with specific properties is a hot topic in materials science.Traditional materials design relies on empirical and trial-and-error methods,requiring extensive experiments and time,resulting in high costs.With the development of physics,statistics,computer science,and other fields,machine learning offers opportunities for systematically discovering new materials.Especially through machine learning-based inverse design,machine learning algorithms analyze the mapping relationships between materials and their properties to find materials with desired properties.This paper first outlines the basic concepts of materials inverse design and the challenges faced by machine learning-based approaches to materials inverse design.Then,three main inverse design methods—exploration-based,model-based,and optimization-based—are analyzed in the context of different application scenarios.Finally,the applications of inverse design methods in alloys,optical materials,and acoustic materials are elaborated on,and the prospects for materials inverse design are discussed.The authors hope to accelerate the discovery of new materials and provide new possibilities for advancing materials science and innovative design methods.
基金supported by the National Key Research and Development Program of China(No.2019YFA0405300)the National Natural Science Foundation of China(No.12272405).
文摘Flowfield inverse design can obtain the desired flow and contour with high design efficiency,short design cycle,and small modification need.In this study,the Euler equations are formulated in the stream-function coordinates and combined with the given boundary conditions to derive a gridless space-marching method for the inverse design of subsonic,transonic,and supersonic flowfields.Designers can prescribe the flow parameters along the reference streamline to design flowfields and aerodynamic contours.The method is validated by the theoretical transonic solution,computational fluid dynamics,and experimental data,respectively.The method supports the fabrication of a Mach 2.0 single expansion tunnel.The calibration data agree well with the prescribed pressure distribution.The method is successfully applied to inverse design of contractions,nozzles,and asymmetric channels.Compared to classical analytic contractions,the contractions designed by the space-marching method provide a more accurate transonic flow.Compared to the classical Sivells’nozzle,the nozzle designed by the space-marching method provides a smaller workload,a more flexible velocity distribution,a 20%reduction in length,and an equally uniform flow.Additionally,the space-marching method is applied to design the asymmetric channels under various Mach numbers.These asymmetric channels perfectly eliminate Mach waves,achieving the shock-free flow turning and high flow uniformity.These results validate the feasibility of the space-marching method,making it a good candidate for the inverse design of subsonic,transonic,and supersonic internal flowfields and aerodynamic contours.
基金funding by the Deutsche Forschungsgemeinschaft under Germany’s Excellence Strategy within the Cluster of Excellence PhoenixD(EXC 2122,Project ID 390833453).
文摘Traditional inverse neural network(INN)approaches for inverse design typically require auxiliary feedforward networks,leading to increased computational complexity and architectural dependencies.This study introduces a standalone INN methodology that eliminates the need for feedforward networks while maintaining high reconstruction accuracy.The approach integrates Principal Component Analysis(PCA)and Partial Least Squares(PLS)for optimized feature space learning,enabling the standalone INN to effectively capture bidirectionalmappings between geometric parameters and mechanical properties.Validation using established numerical datasets demonstrates that the standalone INN architecture achieves reconstruction accuracy equal or better than traditional tandem approaches while completely eliminating the workload and training time required for Feedforward Neural Networks(FNN).These findings contribute to AI methodology development by proving that standalone invertible architectures can achieve comparable performance to complex hybrid systems with significantly improved computational efficiency.
基金financially supported by the National Natural Science Foundation of China(Grant Nos.62476278,12434009,and 12204533)the National Key R&D Program of China(Grant No.2024YFA1408601)the Innovation Program for Quantum Science and Technology(Grant No.2021ZD0302402)。
文摘The discovery of new superconducting materials,particularly those exhibiting high critical temperature(Tc),has been a vibrant area of study within the field of condensed matter physics.Conventional approaches primarily rely on physical intuition to search for potential superconductors within the existing databases.However,the known materials only scratch the surface of the extensive array of possibilities within the realm of materials.
基金the support of the National Natural Science Foundation of China,China(Nos.U20A2069,U21B6003,and 12302389)the Advanced Aero-Power Innovation Workstation,China(No.HKCX2024-01-017)。
文摘Research on detonation has traditionally focused on forward solutions,with limited attention to inverse design methods,which has significantly hindered the development of detonation engines.In this paper,the Method of characteristics for Curved-Detonation(MOCD)is proposed to enable the inverse design of detonation waves.MOCD is based on the Method of Curved-shock Characteristics(MOCC)and integrates higher-order aerodynamic parameters from Curved Detonation Equations(CDE),allowing the calculation of the wedge angle given specific wave angle.The effectiveness of MOCD is validated using both oblique and curved detonation waves with single-step and detailed chemical reactions.Various applications demonstrate the ability to meet the inverse design requirements of detonation engines.For example,inverse design for given wave angles can optimize engine thrust and prevent Mach reflections.Additionally,inverse design schemes tailored to incoming flow conditions,such as varying Mach numbers and equivalence ratios,enhance the feasibility of detonation engines.Applying the method to given post-wave aerodynamic parameters enables more precise engine design,which is crucial for improving propulsion performance and effective thermal protection.In summary,the advantages of MOCD include not only performing a fast solution of the detonation flow field,but also allowing the inverse design of the detonation wave.
基金supported by the Fundamental Research Funds for the Central Universities (No.2024JBZY021)the National Natural Science Foundation of China (No.61575018)。
文摘Reverse design of highly GeO2-doped silica optical fibers with broadband and flat dispersion profiles is proposed using a neural network(NN) combined with a particle swarm optimization(PSO) algorithm.Firstly,the NN model designed to predict optical fiber dispersion is trained with an appropriate choice of hyperparameters,achieving a root mean square error(RMSE) of 9.47×10-7on the test dataset,with a determination coefficient(R2) of 0.999.Secondly,the NN is combined with the PSO algorithm for the inverse design of dispersion-flattened optical fibers.To expand the search space and avoid particles becoming trapped in local optimal solutions,the PSO algorithm incorporates adaptive inertia weight updating and a simulated annealing algorithm.Finally,by using a suitable fitness function,the designed fibers exhibit flat group velocity dispersion(GVD) profiles at 1 400—2 400 nm,where the GVD fluctuations and minimum absolute GVD values are below 18 ps·nm-1·km-1and 7 ps·nm-1·km-1,respectively.
基金supported by the National Natural Science Foundation of China(Grant Nos.62375137 and 62175114).
文摘Structural colors based on metasurfaces have very promising applications in areas such as optical image encryption and color printing.Herein,we propose a deep learning-enabled reverse design of polarization-selective structural color based on coding metasurface.In this study,the long short-term memory(LSTM)neural network is presented to enable the forward and inverse mapping between coding metasurface structure and corresponding color.The results show that the method can achieve 98%accuracy for the forward prediction of color and 93%accuracy for the inverse design of the structure.Moreover,a cascaded architecture is adopted to train the inverse neural network model,which can solve the nonuniqueness problem of the polarization-selective color reverse design.This study provides a new path for the application and development of structural colors.
基金supported by the National Natural Science Foundation of China(Grant No.12204541)the Science and Technology Innovation Program of Hunan Province(Grant No.2021RC3083)the High-level Talents Programs of the National University of Defense Technology.
文摘The traditional forward design process of metasurface optical filters is computationally costly and time-consuming;therefore,inverse design based on deep learning(DL)can help accelerate the process.We propose the globaland local-spectrum-aware transformer(GLSaT),a DL model that concerns the intrinsic correlations within the spectral sequences,compensating the drawbacks of current networks that only focus on structure-to-spectrum mappings.With both interand intra-fragment attention mechanisms implemented,the GLSaT achieves 32.9%higher accuracy than fully connected networks in our reflection tests.It also demonstrates an inherent balance between predictive precision and computational efficiency,outperforming alternative architectures.Furthermore,our extensive experimental validations demonstrate its generalization capability across diverse metasurface functionalities.The GLSaT architecture shows great potential for enhancing the efficiency of data-driven metasurface inverse design in the future.
基金supported by the National Natural Science Foundation of China(No.52471184)the Science and Technology Major Project of Hunan Province,China(No.2019GK1012)+1 种基金the Postgraduate Scientific Research Innovation Project of Xiangtan University,China(No.XDCX2023Y174)the Postgraduate Scientific Research Innovation Project of Xiangtan University,China(No.XDCX2023Y173).
文摘In order to develop a generic framework capable of designing novel amorphous alloys with selected target properties,a predictor−corrector inverse design scheme(PCIDS)consisting of a predictor module and a corrector module was presented.A high-precision forward prediction model based on deep neural networks was developed to implement these two parts.Of utmost importance,domain knowledge-guided inverse design networks(DKIDNs)and regular inverse design networks(RIDNs)were also developed.The forward prediction model possesses a coefficient of determination(R^(2))of 0.990 for the shear modulus and 0.986 for the bulk modulus on the testing set.Furthermore,the DKIDNs model exhibits superior performance compared to the RIDNs model.It is finally demonstrated that PCIDS can efficiently predict amorphous alloy compositions with the required target properties.
基金supported by the National Natural Science Foundation of China(No.61505160)the Innovation Capability Support Program of Shaanxi(No.2018KJXX-042)+2 种基金the Natural Science Basic Research Program of Shaanxi(No.2019JM-084)the State Key Laboratory of Transient Optics and Photonics(No.SKLST202108)the Graduate Innovation and Practical Ability Training Project of Xi’an Shiyou University(No.YCS22213190)。
文摘We proposed and demonstrated the ultra-compact 1310/1550 nm wavelength multiplexer/demultiplexer assisted by subwavelength grating(SWG)using particle swarm optimization(PSO)algorithm in silicon-on-insulator(SOI)platform.Through the self-imaging effect of multimode interference(MMI)coupler,the demultiplexing function for 1310 nm and 1550 nm wavelengths is implemented.After that,three parallel SWG-based slots are inserted into the MMI section so that the effective refractive index of the modes can be engineered and thus the beat length can be adjusted.Importantly,these three SWG slots significantly reduce the length of the device,which is much shorter than the length of traditional MMI-based wavelength demultiplexers.Ultimately,by using the PSO algorithm,the equivalent refractive index and width of the SWG in a certain range are optimized to achieve the best performance of the wavelength demultiplexer.It has been verified that the device footprint is only 2×30.68μm^(2),and 1 dB bandwidths of larger than 120 nm are acquired at 1310 nm and 1550 nm wavelengths.Meanwhile,the transmitted spectrum shows that the insertion loss(IL)values are below 0.47 dB at both wavelengths when the extinction ratio(ER)values are above 12.65 dB.This inverse design approach has been proved to be efficient in increasing bandwidth and reducing device length.
基金supported by the National Key R and D Program of China(No.2022YFF0604801)the National Natural Science Foundation of China(Nos.62271056,62171186,62201037)+3 种基金the Technology Innovation Center of Infrared Remote Sensing Metrology Technology of State Administration for Market Regulation(No.AKYKF2423)the Beijing Natural Science Foundation of China-Haidian Original Innovation Joint Fund(No.L222042)the Open Research Fund of State Key Laboratory of Millimeter Waves(No.K202326)the 111 Project of China(No.B14010).
文摘Terahertz(THz)metamaterials,with their exceptional ability to precisely manipulate the phase,amplitude,polarization and orbital angular momentum(OAM)of electromagnetic waves,have demonstrated significant application potential across a wide range of fields.However,traditional design methodologies often rely on extensive parameter sweeps,making it challenging to address the increasingly complex and diverse application requirements.Recently,the integration of artificial intelligence(AI)techniques,particularly deep learning and optimization algorithms,has introduced new approaches for the design of THz metamaterials.This paper reviews the fundamental principles of THz metamaterials and their intelligent design methodologies,with a particular focus on the advancements in AI-driven inverse design of THz metamaterials.The AI-driven inverse design process allows for the creation of THz metamaterials with desired properties by working backward from the unit structures and array configurations of THz metamaterials,thereby accelerating the design process and reducing both computational resources and time.It examines the critical role of AI in improving both the functionality and design efficiency of THz metamaterials.Finally,we outline future research directions and technological challenges,with the goal of providing valuable insights and guidance for ongoing and future investigations.
基金supported by the National Key R&D Program of China(Grant No.2023YFC3009400)the National Natural Science Foundation of China(Grant Nos.52238009 and 52208344).
文摘Conventional pit excavation engineering methods often struggle to manage the complex deformation patterns associated with asymmetric excavations,resulting in significant safety risks and increased project costs.These challenges highlight the need for more precise and efficient design methodologies to ensure structural stability and economic feasibility.This research proposes an innovative automatic optimization inverse design method(AOIDM)that integrates an enhanced genetic algorithm(EGA)with a multiobjective optimization model.By combining advanced computational techniques with engineering principles,this approach improves search efficiency by 30%and enhances deformation control accuracy by 25%.Additionally,the approach exhibits potential for reducing carbon emissions to align with sustainable engineering goals.The effectiveness of this approach was validated through comprehensive data analysis and practical case studies,demonstrating its ability to optimize retaining structure designs under complex asymmetric loading conditions.This research establishes a new standard for precision and efficiency in automated excavation design,with accompanying improvements in safety and cost-effectiveness.Furthermore,it lays the foundation for future geotechnical engineering advancements,offering a robust solution to one of the most challenging aspects of modern excavation projects.
基金supported by the National Natural Science Foundation of China(Grant Nos.12322207,12202120 and T2293720/T2293722)the Shenzhen Science and Technology Program,China(Grant No.JCYJ20220531095210022)+1 种基金the Fundamental Research Funds for the Central Universities(Grant No.HIT.OCEF.2022037)financial support by the National Key Research and Development Program of China(Grant No.2023YFB3812500)。
文摘Liquid crystal elastomers(LCEs)are advanced materials characterized by their rubber-like hyperelasticity and liquid crystal phase transitions,offering exceptional mechanical properties.The development of smart mechanical metamaterials(SMMs)from LCEs expands the potential for controlling mechanical responses and achieving nonlinear behaviors not possible with traditional metamaterials.However,the challenge lies in managing the interplay between nonlinear material responses and structural complexity,making the inverse design of LCE-based SMMs exceptionally demanding.In this paper,we introduce a design framework for LCE smart mechanical metamaterials that leverages neural networks and evolution strategies(ES)to optimize designs with nonlinear mechanical responses.Our approach involves constructing a flexible,unit-cell-based metamaterial model that integrates the soft elastic behavior and thermo-mechanical coupling of LCEs.The combination of microscopic liquid crystal molecule rotation and macroscopic block rotation enables highly tunable and nonlinear mechanical behaviors,of which the precise inverse design of stress-stretch responses is obtained via neural networks combined with ES.In addition,stimuli responses in the liquid crystal elastomers enable real-time adaptability and achieve tailored stress plateaus that are not possible with traditional metamaterials.Our findings provide new pathways in the design and optimization of advanced materials in flexible electronic devices,intelligent actuators,and systems for energy absorption and dissipation.
基金supported by the National Natural Science Foundation of China(grant numbers 11972287 and 12072266)the State Key Laboratory of Structural Analysis,Optimization and CAE Software for Industrial Equipment(GZ23106)+1 种基金the National Key Laboratory of Aircraft Configuration Design(No.2023-JCJQ-LB-070)the Fundamental Research Funds for the Central Universities.
文摘Chiral metamaterials are manmade structures with extraordinary mechanical properties derived from their special geometric design instead of chemical composition.To make the mechanical deformation programmable,the non-uniform rational B-spline(NURBS)curves are taken to replace the traditional ligament boundaries of the chiral structure.The Neural networks are innovatively inserted into the calculation of mechanical properties of the chiral structure instead of finite element methods to improve computational efficiency.For the problem of finding structure configuration with specified mechanical properties,such as Young’s modulus,Poisson’s ratio or deformation,an inverse design method using the Neural network-based proxy model is proposed to build the relationship between mechanical properties and geometric configuration.To satisfy some more complex deformation requirements,a non-homogeneous inverse design method is proposed and verified through simulation and experiments.Numerical and test results reveal the high computational efficiency and accuracy of the proposed method in the design of chiral metamaterials.
基金supported by the National Natural Science Foundation of China(Grant Nos.62227821,62025503,and 62205199).
文摘To capture the nonlinear dynamics and gain evolution in chirped pulse amplification(CPA)systems,the split-step Fourier method and the fourth-order Runge–Kutta method are integrated to iteratively address the generalized nonlinear Schrödinger equation and the rate equations.However,this approach is burdened by substantial computational demands,resulting in significant time expenditures.In the context of intelligent laser optimization and inverse design,the necessity for numerous simulations further exacerbates this issue,highlighting the need for fast and accurate simulation methodologies.Here,we introduce an end-to-end model augmented with active learning(E2E-AL)with decent generalization through different dedicated embedding methods over various parameters.On an identical computational platform,the artificial intelligence–driven model is 2000 times faster than the conventional simulation method.Benefiting from the active learning strategy,the E2E-AL model achieves decent precision with only two-thirds of the training samples compared with the case without such a strategy.Furthermore,we demonstrate a multi-objective inverse design of the CPA systems enabled by the E2E-AL model.The E2E-AL framework manifests the potential of becoming a standard approach for the rapid and accurate modeling of ultrafast lasers and is readily extended to simulate other complex systems.
基金supported by the Key Technologies R&D Program of China National Offshore Oil Corporation(No.KJGG2021-0504).
文摘Polymer flooding is a widely used technique in enhanced oil recovery (EOR),but its effectiveness is often hindered by the poor viscosity retention of conventional polymers like hydrolyzed polyacrylamide (HPAM) under high-salinity conditions.Although recent advances in molecular engineering have concentrated on modifying polymer architecture and functional groups to address this issue,the complex interplay among polymer topology,charge distribution and hydrophilic-hydrophobic balance renders rational molecular design challenging.In this work,we present an AI-driven inverse design framework that directly maps target viscosity performance back to optimal molecular structures.Guided by practical molecular design strategies,the topological features (grafting density,side-chain length) and functional group-related features(copolymerization ratio,hydrophilic-hydrophobic balance) are encoded into a multidimensional design space.By integrating dissipative particle dynamics simulations with particle swarm algorithm,the framework efficiently explores the design space and identifies non-intuitive,high-performing polymer structure.The optimized polymer achieves a 12%enhancement in viscosity,attributed to the synergistic effect of electrostatic chain extension and hydrophobic aggregation.This study demonstrates the promise of AI-guided inverse design for developing next-generation EOR polymers and provides a generalizable approach for the discovery of functional soft materials.
基金supported by the National Natural Science Foundation of China(Nos.62175076,62105028,62475085)the Natural Science Foundation of Hubei Province of China(Nos.2024AFA016,2024AFB612)the Open Project Program of Hubei Optical Fundamental Research Center.
文摘The integrated waveguide polarizer is essential for photonic integrated circuits,and various designs of waveguide polarizers have been developed.As the demand for dense photonic integration increases rapidly,new strategies to minimize the device size are needed.In this paper,we have inversely designed an integrated transverse electric pass(TE-pass)polarizer with a footprint of 2.88μm×2.88μm,which is the smallest footprint ever achieved.A direct binary search algorithm is used to inversely design the device for maximizing the transverse electric(TE)transmission while minimizing transverse magnetic(TM)transmission.Finally,the inverse-designed device provides an average insertion loss of 0.99 dB and an average extinction ratio of 33 dB over a wavelength range of 100 nm.
