As the amplitude of the unsteady flow oscillation is large or large changes occur in the mean background flow such as limit cycle oscillation,the traditional proper orthogonal decomposition reduced order model based o...As the amplitude of the unsteady flow oscillation is large or large changes occur in the mean background flow such as limit cycle oscillation,the traditional proper orthogonal decomposition reduced order model based on linearized time or frequency domain small disturbance solvers can not capture the main nonlinear features.A new nonlinear reduced order model based on the dynamically nonlinear flow equation was investigated.The nonlinear second order snapshot equation in the time domain for proper orthogonal decomposition basis construction was obtained from the Taylor series expansion of the flow solver.The NLR 7301 airfoil configuration and Goland+ wing/store aeroelastic model were used to validate the capability and efficiency of the new nonlinear reduced order model.The simulation results indicate that the proposed new reduced order model can capture the limit cycle oscillation of aeroelastic system very well,while the traditional proper orthogonal decomposition reduced order model will lose effectiveness.展开更多
An equivalent source-load MTDC system including DC voltage control units,power control units and interconnected DC lines is considered in this paper,which can be regarded as a generic structure of low-voltage DC micro...An equivalent source-load MTDC system including DC voltage control units,power control units and interconnected DC lines is considered in this paper,which can be regarded as a generic structure of low-voltage DC microgrids,mediumvoltage DC distribution systems or HVDC transmission systems with a common DC bus.A reduced-order model is proposed with a circuit structure of a resistor,inductor and capacitor in parallel for dynamic stability analysis of the system in DC voltage control timescale.The relationship between control parameters and physical parameters of the equivalent circuit can be found,which provides an intuitive insight into the physical meaning of control parameters.Employing this model,a second-order characteristic equation is further derived to investigate system dynamic stability mechanisms in an analytical approach.As a result,the system oscillation frequency and damping are characterized in a straight forward manner,and the role of electrical and control parameters and different system-level control strategies in system dynamic stability in DC voltage control timescale is defined.The effectiveness of the proposed reduced-order model and the correctness of the theoretical analysis are verified by simulation based on PSCAD/EMTDC and an experiment based on a hardware low-voltage MTDC system platform.展开更多
For dynamic stability analysis and instability mechanism understanding of multi-converter medium voltage DC power systems with droop-based double-loop control,an advanced system-level model reduction method is propose...For dynamic stability analysis and instability mechanism understanding of multi-converter medium voltage DC power systems with droop-based double-loop control,an advanced system-level model reduction method is proposed.With this method,mathematical relationships of control parameters(e.g.,current and voltage control parameters)between the system and its equivalent reduced-order model are established.First,open-loop and closed-loop equivalent reduced-order models of current control loop considering dynamic interaction among converters are established.An instability mechanism(e.g.,unreasonable current control parameters)of the system can be revealed intuitively.Theoretical guidance for adjustment of current control parameters can also be given.Then,considering dynamic interaction of current control among converters,open-loop and closed-loop equivalent reduced-order models of voltage control loop are established.Oscillation frequency and damping factor of DC bus voltage in a wide oscillation frequency range(e.g.,10–50 Hz)can be evaluated accurately.More importantly,accuracy of advanced system-level model reduction method is not compromised,even for MVDC power systems with inconsistent control parameters and different number of converters.Finally,experiments in RT-BOX hardware-in-the-loop experimental platform are conducted to validate the advanced system-level model reduction method.展开更多
Based on Recursive Radial Basis Function(RRBF)neural network,the Reduced Order Model(ROM)of compressor cascade was established to meet the urgent demand of highly efficient prediction of unsteady aerodynamics performa...Based on Recursive Radial Basis Function(RRBF)neural network,the Reduced Order Model(ROM)of compressor cascade was established to meet the urgent demand of highly efficient prediction of unsteady aerodynamics performance of turbomachinery.One novel ROM called ASA-RRBF model based on Adaptive Simulated Annealing(ASA)algorithm was developed to enhance the generalization ability of the unsteady ROM.The ROM was verified by predicting the unsteady aerodynamics performance of a highly-loaded compressor cascade.The results show that the RRBF model has higher accuracy in identification of the dimensionless total pressure and dimensionless static pressure of compressor cascade under nonlinear and unsteady conditions,and the model behaves higher stability and computational efficiency.However,for the strong nonlinear characteristics of aerodynamic parameters,the RRBF model presents lower accuracy.Additionally,the RRBF model predicts with a large error in the identification of aerodynamic parameters under linear and unsteady conditions.For ASA-RRBF,by introducing a small-amplitude and highfrequency sinusoidal signal as validation sample,the width of the basis function of the RRBF model is optimized to improve the generalization ability of the ROM under linear unsteady conditions.Besides,this model improves the predicting accuracy of dimensionless static pressure which has strong nonlinear characteristics.The ASA-RRBF model has higher prediction accuracy than RRBF model without significantly increasing the total time consumption.This novel model can predict the linear hysteresis of dimensionless static pressure happened in the harmonic condition,but it cannot accurately predict the beat frequency of dimensionless total pressure.展开更多
A Non-Intrusive Reduced-Order Model(NIROM)based on Proper Orthogonal Decomposition(POD)has been proposed for predicting the flow fields of transonic airfoils with geometry parameters.To provide a better reduced-order ...A Non-Intrusive Reduced-Order Model(NIROM)based on Proper Orthogonal Decomposition(POD)has been proposed for predicting the flow fields of transonic airfoils with geometry parameters.To provide a better reduced-order subspace to approximate the real flow field,a domain decomposition method has been used to separate the hard-to-predict regions from the full field and POD has been adopted in the regions individually.An Artificial Neural Network(ANN)has replaced the Radial Basis Function(RBF)to interpolate the coefficients of the POD modes,aiming at improving the approximation accuracy of the NIROM for non-samples.When predicting the flow fields of transonic airfoils,the proposed NIROM has demonstrated a high performance.展开更多
The design of a functional observer and reduced-order observer with internal delay for linear singular timedelay systems with unknown inputs is discussed. The sufficient conditions of the existence of observers, which...The design of a functional observer and reduced-order observer with internal delay for linear singular timedelay systems with unknown inputs is discussed. The sufficient conditions of the existence of observers, which are normal linear time-delay systems, and the corresponding design steps are presented via linear matrix inequality（LMI）. Moreover, the observer-based feedback stabilizing controller is obtained. Three examples are given to show the effectiveness of the proposed methods.展开更多
Reacting particle systems play an important role in many industrial applications,for example biomass drying or the manufacturing of pharmaceuticals.The numerical modeling and simulation of such systems is therefore of...Reacting particle systems play an important role in many industrial applications,for example biomass drying or the manufacturing of pharmaceuticals.The numerical modeling and simulation of such systems is therefore of great importance for an efficient,reliable,and environmentally sustainable operation of the processes.The complex thermodynamical,chemical,and flow processes that take place in the particles are a particular challenge in a simulation.Furthermore,typically a large number of particles is involved,rendering an explicit treatment of individual ones impossible in a reactor-level simulation.One approach for overcoming this challenge is to compute effective,physical parameters from single-particle,high-resolution simulations.This can be combined with model reduction methods if the dynamical behaviour of particles must be captured.Pore network models with their unrivaled resolution have thereby been used successfully as high-resolution models,for instance to obtain the macroscopic diffusion coeffcient of drying.Both parameter identification and model reduction have recently gained new impetus by the dramatic progress made in machine learning in the last decade.We report results on the use of neural networks for parameter identification and model reduction based on three-dimensional pore network models(PNM).We believe that our results provide a powerful complement to existing methodologies for reactor-level simulations with many thermally-thick particles.展开更多
The modeling of dynamic stall aerodynamics is essential to stall flutter, due to the flow separation in a large-amplitude pitching oscillation process. A newly neural network based Reduced Order Model(ROM) framework f...The modeling of dynamic stall aerodynamics is essential to stall flutter, due to the flow separation in a large-amplitude pitching oscillation process. A newly neural network based Reduced Order Model(ROM) framework for predicting the aerodynamic forces of an airfoil undergoing large-amplitude pitching oscillation at various velocities is presented in this work. First, the dynamic stall aerodynamics is calculated by solving RANS equations and the transitional SST-γ model. Afterwards, the stall flutter bifurcation behavior is calculated by the above CFD solver coupled with structural dynamic equation. The critical flutter speed and limit-cycle oscillation amplitudes are consistent with those obtained by experiments. A newly multi-layer Gated Recurrent Unit(GRU) neural network based ROM is constructed to accelerate the calculation of aerodynamic forces. The training and validation process are carried out upon the unsteady aerodynamic data obtained by the proposed CFD method. The well-trained ROM is then coupled with the structure equation at a specific velocity, the Limit-Cycle Oscillation(LCO) of stall flutter under this flow condition is predicted precisely and more quickly. In order to predict both the critical flutter velocity and LCO amplitudes after bifurcation at different velocities, a new ROM with GRU neural network considering the variation of flow velocities is developed. The stall flutter results predicted by ROM agree well with the CFD ones at different velocities. Finally, a brief sensitivity analysis of two structural parameters of ROM is carried out. It infers the potential of the presented modeling method to depict the nonlinearity of dynamic stall and stall flutter phenomenon.展开更多
This paper is devoted to application of the Reduced-Order Model(ROM)based on Volterra series to prediction of lift and drag forces due to airfoil periodic translation in transonic flow region.When there is large ampli...This paper is devoted to application of the Reduced-Order Model(ROM)based on Volterra series to prediction of lift and drag forces due to airfoil periodic translation in transonic flow region.When there is large amplitude oscillation of the relative Mach number,as appeared in helicopter rotor movement in forward flight,the conventional Volterra ROM is found to be unsatisfactory.To cover such applications,a matched Volterra ROM,inspired from previous multistep nonlinear indicial response method based on Duhamel integration,is thus considered,in which the step motions are defined inside a number of equal intervals with both positive and negative step motions to match the airfoil forward and backward movement,and the kernel functions are constructed independently at each interval.It shows that,at least for the translation movement considered,this matched Volterra ROM greatly improves the accuracy of prediction.Moreover,the matched Volterra ROM,with the total number of step motions and thus the computational cost close to those of the conventional Volterra ROM method,has the additional advantage that the same set of kernels can match various translation motions with different starting conditions so the kernels can be predesigned without knowing the specific motion of airfoil.展开更多
An improved Reduced-Order Model(ROM)is proposed based on a flow-solution preprocessing operation and a fast sampling strategy to efficiently and accurately predict ionized hypersonic flows.This ROM is generated in low...An improved Reduced-Order Model(ROM)is proposed based on a flow-solution preprocessing operation and a fast sampling strategy to efficiently and accurately predict ionized hypersonic flows.This ROM is generated in low-dimensional space by performing the Proper Orthogonal Decomposition(POD)on snapshots and is coupled with the Radial Basis Function(RBF)to achieve fast prediction speed.However,due to the disparate scales in the ionized flow field,the conventional ROM usually generates spurious negative errors.Here,this issue is addressed by performing flow-solution preprocessing in logarithmic space to improve the conventional ROM.Then,extra orthogonal polynomials are introduced in the RBF interpolation to achieve additional improvement of the prediction accuracy.In addition,to construct high-efficiency snapshots,a trajectory-constrained adaptive sampling strategy based on convex hull optimization is developed.To evaluate the performance of the proposed fast prediction method,two hypersonic vehicles with classic configurations,i.e.a wave-rider and a reentry capsule,are used to validate the proposed method.Both two cases show that the proposed fast prediction method has high accuracy near the vehicle surface and the free-stream region where the flow field is smooth.Compared with the conventional ROM prediction,the prediction results are significantly improved by the proposed method around the discontinuities,e.g.the shock wave and the ionized layer.As a result,the proposed fast prediction method reduces the error of the conventional ROM by at least 45%,with a speedup of approximately 2.0×105compared to the Computational Fluid Dynamic(CFD)simulations.These test cases demonstrate that the method developed here is efficient and accurate for predicting ionized hypersonic flows.展开更多
Many important problems in science and engineering require solving the so-called parametric partial differential equations(PDEs),i.e.,PDEs with different physical parameters,boundary conditions,shapes of computational...Many important problems in science and engineering require solving the so-called parametric partial differential equations(PDEs),i.e.,PDEs with different physical parameters,boundary conditions,shapes of computational domains,etc.Typical reduced order modeling techniques accelerate the solution of the parametric PDEs by projecting them onto a linear trial manifold constructed in the ofline stage.These methods often need a predefined mesh as well as a series of precomputed solution snapshots,and may struggle to balance between the efficiency and accuracy due to the limitation of the linear ansatz.Utilizing the nonlinear representation of neural networks(NNs),we propose the Meta-Auto-Decoder(MAD)to construct a nonlinear trial manifold,whose best possible performance is measured theoretically by the decoder width.Based on the meta-learning concept,the trial manifold can be learned in a mesh-free and unsupervised way during the pre-training stage.Fast adaptation to new(possibly heterogeneous)PDE parameters is enabled by searching on this trial manifold,and optionally fine-tuning the trial manifold at the same time.Extensive numerical experiments show that the MAD method exhibits a faster convergence speed without losing the accuracy than other deep learning-based methods.展开更多
The efficient dynamic modeling and vibration transfer analysis of a fluid-delivering branch pipeline(FDBP)are essential for analyzing vibration coupling effects and implementing vibration reduction optimization.Theref...The efficient dynamic modeling and vibration transfer analysis of a fluid-delivering branch pipeline(FDBP)are essential for analyzing vibration coupling effects and implementing vibration reduction optimization.Therefore,this study proposes a reduced-order dynamic modeling method suitable for FDBPs and then analyzes the vibration transfer characteristics.For the modeling method,the finite element method and absorbing transfer matrix method(ATMM)are integrated,considering the fluid–structure coupling effect and fluid disturbances.The dual-domain dynamic substructure method is developed to perform the reduced-order modeling of FDBP,and ATMM is adopted to reduce the matrix order when solving fluid disturbances.Furthermore,the modeling method is validated by experiments on an H-shaped branch pipeline.Finally,transient and steady-state vibration transfer analyses of FDBP are performed,and the effects of branch locations on natural characteristics and vibration transfer behavior are analyzed.Results show that transient vibration transfer represents the transfer and conversion of the kinematic,strain,and damping energies,while steady-state vibration transfer characteristics are related to the vibration mode.In addition,multiple-order mode exchanges are triggered when branch locations vary in frequency-shift regions,and the mode-exchange regions are also the transformation ones for vibration transfer patterns.展开更多
Accurate and efficient online parameter identification and state estimation are crucial for leveraging digital twin simulations to optimize the operation of near-carbon-free nuclear energy systems.In previous studies,...Accurate and efficient online parameter identification and state estimation are crucial for leveraging digital twin simulations to optimize the operation of near-carbon-free nuclear energy systems.In previous studies,we developed a reactor operation digital twin(RODT).However,non-differentiabilities and discontinuities arise when employing machine learning-based surrogate forward models,challenging traditional gradient-based inverse methods and their variants.This study investigated deterministic and metaheuristic algorithms and developed hybrid algorithms to address these issues.An efficient modular RODT software framework that incorporates these methods into its post-evaluation module is presented for comprehensive comparison.The methods were rigorously assessed based on convergence profiles,stability with respect to noise,and computational performance.The numerical results show that the hybrid KNNLHS algorithm excels in real-time online applications,balancing accuracy and efficiency with a prediction error rate of only 1%and processing times of less than 0.1 s.Contrastingly,algorithms such as FSA,DE,and ADE,although slightly slower(approximately 1 s),demonstrated higher accuracy with a 0.3%relative L_2 error,which advances RODT methodologies to harness machine learning and system modeling for improved reactor monitoring,systematic diagnosis of off-normal events,and lifetime management strategies.The developed modular software and novel optimization methods presented offer pathways to realize the full potential of RODT for transforming energy engineering practices.展开更多
基金supported by the National Natural Science Foundation of China (Grant No. 10902082)New Faculty Research Foundation of Xi’an Jiaotong University
文摘As the amplitude of the unsteady flow oscillation is large or large changes occur in the mean background flow such as limit cycle oscillation,the traditional proper orthogonal decomposition reduced order model based on linearized time or frequency domain small disturbance solvers can not capture the main nonlinear features.A new nonlinear reduced order model based on the dynamically nonlinear flow equation was investigated.The nonlinear second order snapshot equation in the time domain for proper orthogonal decomposition basis construction was obtained from the Taylor series expansion of the flow solver.The NLR 7301 airfoil configuration and Goland+ wing/store aeroelastic model were used to validate the capability and efficiency of the new nonlinear reduced order model.The simulation results indicate that the proposed new reduced order model can capture the limit cycle oscillation of aeroelastic system very well,while the traditional proper orthogonal decomposition reduced order model will lose effectiveness.
基金This work was supported in part by the National Natural Science Foundation of China under Grant No.51977142.
文摘An equivalent source-load MTDC system including DC voltage control units,power control units and interconnected DC lines is considered in this paper,which can be regarded as a generic structure of low-voltage DC microgrids,mediumvoltage DC distribution systems or HVDC transmission systems with a common DC bus.A reduced-order model is proposed with a circuit structure of a resistor,inductor and capacitor in parallel for dynamic stability analysis of the system in DC voltage control timescale.The relationship between control parameters and physical parameters of the equivalent circuit can be found,which provides an intuitive insight into the physical meaning of control parameters.Employing this model,a second-order characteristic equation is further derived to investigate system dynamic stability mechanisms in an analytical approach.As a result,the system oscillation frequency and damping are characterized in a straight forward manner,and the role of electrical and control parameters and different system-level control strategies in system dynamic stability in DC voltage control timescale is defined.The effectiveness of the proposed reduced-order model and the correctness of the theoretical analysis are verified by simulation based on PSCAD/EMTDC and an experiment based on a hardware low-voltage MTDC system platform.
基金supported by the National Key Research and Development Program of China(2020YFB1506800)the China Postdoctoral Science Foundation(2021M692378)the National Natural Science Foundation of China(51977142).
文摘For dynamic stability analysis and instability mechanism understanding of multi-converter medium voltage DC power systems with droop-based double-loop control,an advanced system-level model reduction method is proposed.With this method,mathematical relationships of control parameters(e.g.,current and voltage control parameters)between the system and its equivalent reduced-order model are established.First,open-loop and closed-loop equivalent reduced-order models of current control loop considering dynamic interaction among converters are established.An instability mechanism(e.g.,unreasonable current control parameters)of the system can be revealed intuitively.Theoretical guidance for adjustment of current control parameters can also be given.Then,considering dynamic interaction of current control among converters,open-loop and closed-loop equivalent reduced-order models of voltage control loop are established.Oscillation frequency and damping factor of DC bus voltage in a wide oscillation frequency range(e.g.,10–50 Hz)can be evaluated accurately.More importantly,accuracy of advanced system-level model reduction method is not compromised,even for MVDC power systems with inconsistent control parameters and different number of converters.Finally,experiments in RT-BOX hardware-in-the-loop experimental platform are conducted to validate the advanced system-level model reduction method.
基金co-National Science and Technology Major Project(No.2017-II-0009-0023)Innovation Guidance Support Project for Taicang Top Research Institutes(No.TC2019DYDS09)。
文摘Based on Recursive Radial Basis Function(RRBF)neural network,the Reduced Order Model(ROM)of compressor cascade was established to meet the urgent demand of highly efficient prediction of unsteady aerodynamics performance of turbomachinery.One novel ROM called ASA-RRBF model based on Adaptive Simulated Annealing(ASA)algorithm was developed to enhance the generalization ability of the unsteady ROM.The ROM was verified by predicting the unsteady aerodynamics performance of a highly-loaded compressor cascade.The results show that the RRBF model has higher accuracy in identification of the dimensionless total pressure and dimensionless static pressure of compressor cascade under nonlinear and unsteady conditions,and the model behaves higher stability and computational efficiency.However,for the strong nonlinear characteristics of aerodynamic parameters,the RRBF model presents lower accuracy.Additionally,the RRBF model predicts with a large error in the identification of aerodynamic parameters under linear and unsteady conditions.For ASA-RRBF,by introducing a small-amplitude and highfrequency sinusoidal signal as validation sample,the width of the basis function of the RRBF model is optimized to improve the generalization ability of the ROM under linear unsteady conditions.Besides,this model improves the predicting accuracy of dimensionless static pressure which has strong nonlinear characteristics.The ASA-RRBF model has higher prediction accuracy than RRBF model without significantly increasing the total time consumption.This novel model can predict the linear hysteresis of dimensionless static pressure happened in the harmonic condition,but it cannot accurately predict the beat frequency of dimensionless total pressure.
基金supported by the National Natural Science Foundation of China(No.11802245).
文摘A Non-Intrusive Reduced-Order Model(NIROM)based on Proper Orthogonal Decomposition(POD)has been proposed for predicting the flow fields of transonic airfoils with geometry parameters.To provide a better reduced-order subspace to approximate the real flow field,a domain decomposition method has been used to separate the hard-to-predict regions from the full field and POD has been adopted in the regions individually.An Artificial Neural Network(ANN)has replaced the Radial Basis Function(RBF)to interpolate the coefficients of the POD modes,aiming at improving the approximation accuracy of the NIROM for non-samples.When predicting the flow fields of transonic airfoils,the proposed NIROM has demonstrated a high performance.
基金the National Natural Science Foundation of China (No. 50477042)the Ph.D. Programs Foundation of Ministry of Education of China (No. 20040422052 )the National Natural Science Foundation of Shandong Province (No.Z2004G04)
文摘The design of a functional observer and reduced-order observer with internal delay for linear singular timedelay systems with unknown inputs is discussed. The sufficient conditions of the existence of observers, which are normal linear time-delay systems, and the corresponding design steps are presented via linear matrix inequality（LMI）. Moreover, the observer-based feedback stabilizing controller is obtained. Three examples are given to show the effectiveness of the proposed methods.
基金Funded by the Deutsche Forschungsgemeinschaft(DFG,German Research Foundation)-Project-ID 422037413-TRR 287.
文摘Reacting particle systems play an important role in many industrial applications,for example biomass drying or the manufacturing of pharmaceuticals.The numerical modeling and simulation of such systems is therefore of great importance for an efficient,reliable,and environmentally sustainable operation of the processes.The complex thermodynamical,chemical,and flow processes that take place in the particles are a particular challenge in a simulation.Furthermore,typically a large number of particles is involved,rendering an explicit treatment of individual ones impossible in a reactor-level simulation.One approach for overcoming this challenge is to compute effective,physical parameters from single-particle,high-resolution simulations.This can be combined with model reduction methods if the dynamical behaviour of particles must be captured.Pore network models with their unrivaled resolution have thereby been used successfully as high-resolution models,for instance to obtain the macroscopic diffusion coeffcient of drying.Both parameter identification and model reduction have recently gained new impetus by the dramatic progress made in machine learning in the last decade.We report results on the use of neural networks for parameter identification and model reduction based on three-dimensional pore network models(PNM).We believe that our results provide a powerful complement to existing methodologies for reactor-level simulations with many thermally-thick particles.
基金supported by the National Natural Science Foundation of China(No.11672018).
文摘The modeling of dynamic stall aerodynamics is essential to stall flutter, due to the flow separation in a large-amplitude pitching oscillation process. A newly neural network based Reduced Order Model(ROM) framework for predicting the aerodynamic forces of an airfoil undergoing large-amplitude pitching oscillation at various velocities is presented in this work. First, the dynamic stall aerodynamics is calculated by solving RANS equations and the transitional SST-γ model. Afterwards, the stall flutter bifurcation behavior is calculated by the above CFD solver coupled with structural dynamic equation. The critical flutter speed and limit-cycle oscillation amplitudes are consistent with those obtained by experiments. A newly multi-layer Gated Recurrent Unit(GRU) neural network based ROM is constructed to accelerate the calculation of aerodynamic forces. The training and validation process are carried out upon the unsteady aerodynamic data obtained by the proposed CFD method. The well-trained ROM is then coupled with the structure equation at a specific velocity, the Limit-Cycle Oscillation(LCO) of stall flutter under this flow condition is predicted precisely and more quickly. In order to predict both the critical flutter velocity and LCO amplitudes after bifurcation at different velocities, a new ROM with GRU neural network considering the variation of flow velocities is developed. The stall flutter results predicted by ROM agree well with the CFD ones at different velocities. Finally, a brief sensitivity analysis of two structural parameters of ROM is carried out. It infers the potential of the presented modeling method to depict the nonlinearity of dynamic stall and stall flutter phenomenon.
文摘This paper is devoted to application of the Reduced-Order Model(ROM)based on Volterra series to prediction of lift and drag forces due to airfoil periodic translation in transonic flow region.When there is large amplitude oscillation of the relative Mach number,as appeared in helicopter rotor movement in forward flight,the conventional Volterra ROM is found to be unsatisfactory.To cover such applications,a matched Volterra ROM,inspired from previous multistep nonlinear indicial response method based on Duhamel integration,is thus considered,in which the step motions are defined inside a number of equal intervals with both positive and negative step motions to match the airfoil forward and backward movement,and the kernel functions are constructed independently at each interval.It shows that,at least for the translation movement considered,this matched Volterra ROM greatly improves the accuracy of prediction.Moreover,the matched Volterra ROM,with the total number of step motions and thus the computational cost close to those of the conventional Volterra ROM method,has the additional advantage that the same set of kernels can match various translation motions with different starting conditions so the kernels can be predesigned without knowing the specific motion of airfoil.
基金supported by the National Natural Science Foundation of China(Nos.11902271 and 91952203)the Fundamental Research Funds for the Central Universities of China(No.G2019KY05102)111 project on“Aircraft Complex Flows and the Control”of China(No.B17037)。
文摘An improved Reduced-Order Model(ROM)is proposed based on a flow-solution preprocessing operation and a fast sampling strategy to efficiently and accurately predict ionized hypersonic flows.This ROM is generated in low-dimensional space by performing the Proper Orthogonal Decomposition(POD)on snapshots and is coupled with the Radial Basis Function(RBF)to achieve fast prediction speed.However,due to the disparate scales in the ionized flow field,the conventional ROM usually generates spurious negative errors.Here,this issue is addressed by performing flow-solution preprocessing in logarithmic space to improve the conventional ROM.Then,extra orthogonal polynomials are introduced in the RBF interpolation to achieve additional improvement of the prediction accuracy.In addition,to construct high-efficiency snapshots,a trajectory-constrained adaptive sampling strategy based on convex hull optimization is developed.To evaluate the performance of the proposed fast prediction method,two hypersonic vehicles with classic configurations,i.e.a wave-rider and a reentry capsule,are used to validate the proposed method.Both two cases show that the proposed fast prediction method has high accuracy near the vehicle surface and the free-stream region where the flow field is smooth.Compared with the conventional ROM prediction,the prediction results are significantly improved by the proposed method around the discontinuities,e.g.the shock wave and the ionized layer.As a result,the proposed fast prediction method reduces the error of the conventional ROM by at least 45%,with a speedup of approximately 2.0×105compared to the Computational Fluid Dynamic(CFD)simulations.These test cases demonstrate that the method developed here is efficient and accurate for predicting ionized hypersonic flows.
基金supported by the National Key R&D Program of China under Grant No.2021ZD0110400.
文摘Many important problems in science and engineering require solving the so-called parametric partial differential equations(PDEs),i.e.,PDEs with different physical parameters,boundary conditions,shapes of computational domains,etc.Typical reduced order modeling techniques accelerate the solution of the parametric PDEs by projecting them onto a linear trial manifold constructed in the ofline stage.These methods often need a predefined mesh as well as a series of precomputed solution snapshots,and may struggle to balance between the efficiency and accuracy due to the limitation of the linear ansatz.Utilizing the nonlinear representation of neural networks(NNs),we propose the Meta-Auto-Decoder(MAD)to construct a nonlinear trial manifold,whose best possible performance is measured theoretically by the decoder width.Based on the meta-learning concept,the trial manifold can be learned in a mesh-free and unsupervised way during the pre-training stage.Fast adaptation to new(possibly heterogeneous)PDE parameters is enabled by searching on this trial manifold,and optionally fine-tuning the trial manifold at the same time.Extensive numerical experiments show that the MAD method exhibits a faster convergence speed without losing the accuracy than other deep learning-based methods.
基金supported by the Fundamental Research Funds for the Central Universities(Grant No.N2403006)the National Science and Technology Major Project,China(Grant No.J2019-I-0008-0008).
文摘The efficient dynamic modeling and vibration transfer analysis of a fluid-delivering branch pipeline(FDBP)are essential for analyzing vibration coupling effects and implementing vibration reduction optimization.Therefore,this study proposes a reduced-order dynamic modeling method suitable for FDBPs and then analyzes the vibration transfer characteristics.For the modeling method,the finite element method and absorbing transfer matrix method(ATMM)are integrated,considering the fluid–structure coupling effect and fluid disturbances.The dual-domain dynamic substructure method is developed to perform the reduced-order modeling of FDBP,and ATMM is adopted to reduce the matrix order when solving fluid disturbances.Furthermore,the modeling method is validated by experiments on an H-shaped branch pipeline.Finally,transient and steady-state vibration transfer analyses of FDBP are performed,and the effects of branch locations on natural characteristics and vibration transfer behavior are analyzed.Results show that transient vibration transfer represents the transfer and conversion of the kinematic,strain,and damping energies,while steady-state vibration transfer characteristics are related to the vibration mode.In addition,multiple-order mode exchanges are triggered when branch locations vary in frequency-shift regions,and the mode-exchange regions are also the transformation ones for vibration transfer patterns.
基金supported by the Natural Science Foundation of Shanghai(No.23ZR1429300)Innovation Funds of CNNC(Lingchuang Fund,Contract No.CNNC-LCKY-202234)the Project of the Nuclear Power Technology Innovation Center of Science Technology and Industry(No.HDLCXZX-2023-HD-039-02)。
文摘Accurate and efficient online parameter identification and state estimation are crucial for leveraging digital twin simulations to optimize the operation of near-carbon-free nuclear energy systems.In previous studies,we developed a reactor operation digital twin(RODT).However,non-differentiabilities and discontinuities arise when employing machine learning-based surrogate forward models,challenging traditional gradient-based inverse methods and their variants.This study investigated deterministic and metaheuristic algorithms and developed hybrid algorithms to address these issues.An efficient modular RODT software framework that incorporates these methods into its post-evaluation module is presented for comprehensive comparison.The methods were rigorously assessed based on convergence profiles,stability with respect to noise,and computational performance.The numerical results show that the hybrid KNNLHS algorithm excels in real-time online applications,balancing accuracy and efficiency with a prediction error rate of only 1%and processing times of less than 0.1 s.Contrastingly,algorithms such as FSA,DE,and ADE,although slightly slower(approximately 1 s),demonstrated higher accuracy with a 0.3%relative L_2 error,which advances RODT methodologies to harness machine learning and system modeling for improved reactor monitoring,systematic diagnosis of off-normal events,and lifetime management strategies.The developed modular software and novel optimization methods presented offer pathways to realize the full potential of RODT for transforming energy engineering practices.
