Performance of physical-informed neural network (PINN) for the key parameter inference in Langmuir turbulence parameterization scheme

The Stokes production coefficient(E_(6))constitutes a critical parameter within the Mellor-Yamada type(MY-type)Langmuir turbulence(LT)parameterization schemes,significantly affecting the simulation of turbulent kinetic energy,turbulent length scale,and vertical diffusivity coefficient for turbulent kinetic energy in the upper ocean.However,the accurate determination of its value remains a pressing scientific challenge.This study adopted an innovative approach by leveraging deep learning technology to address this challenge of inferring the E_(6).Through the integration of the information of the turbulent length scale equation into a physical-informed neural network(PINN),we achieved an accurate and physically meaningful inference of E_(6).Multiple cases were examined to assess the feasibility of PINN in this task,revealing that under optimal settings,the average mean squared error of the E_(6) inference was only 0.01,attesting to the effectiveness of PINN.The optimal hyperparameter combination was identified using the Tanh activation function,along with a spatiotemporal sampling interval of 1 s and 0.1 m.This resulted in a substantial reduction in the average bias of the E_(6) inference,ranging from O(10^(1))to O(10^(2))times compared with other combinations.This study underscores the potential application of PINN in intricate marine environments,offering a novel and efficient method for optimizing MY-type LT parameterization schemes.

Multi-scale physics-informed neural networks for solving high Reynolds number boundary layer flows based on matched asymptotic expansions

Multi-scale system remains a classical scientific problem in fluid dynamics,biology,etc.In the present study,a scheme of multi-scale Physics-informed neural networks is proposed to solve the boundary layer flow at high Reynolds numbers without any data.The flow is divided into several regions with different scales based on Prandtl's boundary theory.Different regions are solved with governing equations in different scales.The method of matched asymptotic expansions is used to make the flow field continuously.A flow on a semi infinite flat plate at a high Reynolds number is considered a multi-scale problem because the boundary layer scale is much smaller than the outer flow scale.The results are compared with the reference numerical solutions,which show that the msPINNs can solve the multi-scale problem of the boundary layer in high Reynolds number flows.This scheme can be developed for more multi-scale problems in the future.

Prediction of Porous Media Fluid Flow with Spatial Heterogeneity Using Criss-Cross Physics-Informed Convolutional Neural Networks

Recent advances in deep neural networks have shed new light on physics,engineering,and scientific computing.Reconciling the data-centered viewpoint with physical simulation is one of the research hotspots.The physicsinformedneural network(PINN)is currently the most general framework,which is more popular due to theconvenience of constructing NNs and excellent generalization ability.The automatic differentiation(AD)-basedPINN model is suitable for the homogeneous scientific problem;however,it is unclear how AD can enforce fluxcontinuity across boundaries between cells of different properties where spatial heterogeneity is represented bygrid cells with different physical properties.In this work,we propose a criss-cross physics-informed convolutionalneural network(CC-PINN)learning architecture,aiming to learn the solution of parametric PDEs with spatialheterogeneity of physical properties.To achieve the seamless enforcement of flux continuity and integration ofphysicalmeaning into CNN,a predefined 2D convolutional layer is proposed to accurately express transmissibilitybetween adjacent cells.The efficacy of the proposedmethodwas evaluated through predictions of several petroleumreservoir problems with spatial heterogeneity and compared against state-of-the-art(PINN)through numericalanalysis as a benchmark,which demonstrated the superiority of the proposed method over the PINN.

A transfer learning enhanced physics-informed neural network for parameter identification in soft materials

Soft materials,with the sensitivity to various external stimuli,exhibit high flexibility and stretchability.Accurate prediction of their mechanical behaviors requires advanced hyperelastic constitutive models incorporating multiple parameters.However,identifying multiple parameters under complex deformations remains a challenge,especially with limited observed data.In this study,we develop a physics-informed neural network(PINN)framework to identify material parameters and predict mechanical fields,focusing on compressible Neo-Hookean materials and hydrogels.To improve accuracy,we utilize scaling techniques to normalize network outputs and material parameters.This framework effectively solves forward and inverse problems,extrapolating continuous mechanical fields from sparse boundary data and identifying unknown mechanical properties.We explore different approaches for imposing boundary conditions(BCs)to assess their impacts on accuracy.To enhance efficiency and generalization,we propose a transfer learning enhanced PINN(TL-PINN),allowing pre-trained networks to quickly adapt to new scenarios.The TL-PINN significantly reduces computational costs while maintaining accuracy.This work holds promise in addressing practical challenges in soft material science,and provides insights into soft material mechanics with state-of-the-art experimental methods.

A physics-informed neural network for simulation of finite deformation in hyperelastic-magnetic coupling problems

Recently,numerous studies have demonstrated that the physics-informed neural network(PINN)can effectively and accurately resolve hyperelastic finite deformation problems.In this paper,a PINN framework for tackling hyperelastic-magnetic coupling problems is proposed.Since the solution space consists of two-phase domains,two separate networks are constructed to independently predict the solution for each phase region.In addition,a conscious point allocation strategy is incorporated to enhance the prediction precision of the PINN in regions characterized by sharp gradients.With the developed framework,the magnetic fields and deformation fields of magnetorheological elastomers(MREs)are solved under the control of hyperelastic-magnetic coupling equations.Illustrative examples are provided and contrasted with the reference results to validate the predictive accuracy of the proposed framework.Moreover,the advantages of the proposed framework in solving hyperelastic-magnetic coupling problems are validated,particularly in handling small data sets,as well as its ability in swiftly and precisely forecasting magnetostrictive motion.

Chien-physics-informed neural networks for solving singularly perturbed boundary-layer problems

A physics-informed neural network(PINN)is a powerful tool for solving differential equations in solid and fluid mechanics.However,it suffers from singularly perturbed boundary-layer problems in which there exist sharp changes caused by a small perturbation parameter multiplying the highest-order derivatives.In this paper,we introduce Chien's composite expansion method into PINNs,and propose a novel architecture for the PINNs,namely,the Chien-PINN(C-PINN)method.This novel PINN method is validated by singularly perturbed differential equations,and successfully solves the wellknown thin plate bending problems.In particular,no cumbersome matching conditions are needed for the C-PINN method,compared with the previous studies based on matched asymptotic expansions.

Radiative heat transfer analysis of a concave porous fin under the local thermal non-equilibrium condition:application of the clique polynomial method and physics-informed neural networks

The heat transfer through a concave permeable fin is analyzed by the local thermal non-equilibrium(LTNE)model.The governing dimensional temperature equations for the solid and fluid phases of the porous extended surface are modeled,and then are nondimensionalized by suitable dimensionless terms.Further,the obtained nondimensional equations are solved by the clique polynomial method(CPM).The effects of several dimensionless parameters on the fin's thermal profiles are shown by graphical illustrations.Additionally,the current study implements deep neural structures to solve physics-governed coupled equations,and the best-suited hyperparameters are attained by comparison with various network combinations.The results of the CPM and physicsinformed neural network(PINN)exhibit good agreement,signifying that both methods effectively solve the thermal modeling problem.

Physics-informed neural networks with residual/gradient-based adaptive sampling methods for solving partial differential equations with sharp solutions

We consider solving the forward and inverse partial differential equations(PDEs)which have sharp solutions with physics-informed neural networks(PINNs)in this work.In particular,to better capture the sharpness of the solution,we propose the adaptive sampling methods(ASMs)based on the residual and the gradient of the solution.We first present a residual only-based ASM denoted by ASMⅠ.In this approach,we first train the neural network using a small number of residual points and divide the computational domain into a certain number of sub-domains,then we add new residual points in the sub-domain which has the largest mean absolute value of the residual,and those points which have the largest absolute values of the residual in this sub-domain as new residual points.We further develop a second type of ASM(denoted by ASMⅡ)based on both the residual and the gradient of the solution due to the fact that only the residual may not be able to efficiently capture the sharpness of the solution.The procedure of ASMⅡis almost the same as that of ASMⅠ,and we add new residual points which have not only large residuals but also large gradients.To demonstrate the effectiveness of the present methods,we use both ASMⅠand ASMⅡto solve a number of PDEs,including the Burger equation,the compressible Euler equation,the Poisson equation over an Lshape domain as well as the high-dimensional Poisson equation.It has been shown from the numerical results that the sharp solutions can be well approximated by using either ASMⅠor ASMⅡ,and both methods deliver much more accurate solutions than the original PINNs with the same number of residual points.Moreover,the ASMⅡalgorithm has better performance in terms of accuracy,efficiency,and stability compared with the ASMⅠalgorithm.This means that the gradient of the solution improves the stability and efficiency of the adaptive sampling procedure as well as the accuracy of the solution.Furthermore,we also employ the similar adaptive sampling technique for the data points of boundary conditions(BCs)if the sharpness of the solution is near the boundary.The result of the L-shape Poisson problem indicates that the present method can significantly improve the efficiency,stability,and accuracy.

Effective data sampling strategies and boundary condition constraints of physics-informed neural networks for identifying material properties in solid mechanics

Material identification is critical for understanding the relationship between mechanical properties and the associated mechanical functions.However,material identification is a challenging task,especially when the characteristic of the material is highly nonlinear in nature,as is common in biological tissue.In this work,we identify unknown material properties in continuum solid mechanics via physics-informed neural networks(PINNs).To improve the accuracy and efficiency of PINNs,we develop efficient strategies to nonuniformly sample observational data.We also investigate different approaches to enforce Dirichlet-type boundary conditions(BCs)as soft or hard constraints.Finally,we apply the proposed methods to a diverse set of time-dependent and time-independent solid mechanic examples that span linear elastic and hyperelastic material space.The estimated material parameters achieve relative errors of less than 1%.As such,this work is relevant to diverse applications,including optimizing structural integrity and developing novel materials.

Physics-informed neural network-based petroleum reservoir simulation with sparse data using domain decomposition

Recent advances in deep learning have expanded new possibilities for fluid flow simulation in petroleum reservoirs.However,the predominant approach in existing research is to train neural networks using high-fidelity numerical simulation data.This presents a significant challenge because the sole source of authentic wellbore production data for training is sparse.In response to this challenge,this work introduces a novel architecture called physics-informed neural network based on domain decomposition(PINN-DD),aiming to effectively utilize the sparse production data of wells for reservoir simulation with large-scale systems.To harness the capabilities of physics-informed neural networks(PINNs)in handling small-scale spatial-temporal domain while addressing the challenges of large-scale systems with sparse labeled data,the computational domain is divided into two distinct sub-domains:the well-containing and the well-free sub-domain.Moreover,the two sub-domains and the interface are rigorously constrained by the governing equations,data matching,and boundary conditions.The accuracy of the proposed method is evaluated on two problems,and its performance is compared against state-of-the-art PINNs through numerical analysis as a benchmark.The results demonstrate the superiority of PINN-DD in handling large-scale reservoir simulation with limited data and show its potential to outperform conventional PINNs in such scenarios.

Physics-Informed Deep Neural Network for Bearing Prognosis with Multisensory Signals

Prognosis of bearing is critical to improve the safety,reliability,and availability of machinery systems,which provides the health condition assessment and determines how long the machine would work before failure occurs by predicting the remaining useful life(RUL).In order to overcome the drawback of pure data-driven methods and predict RUL accurately,a novel physics-informed deep neural network,named degradation consistency recurrent neural network,is proposed for RUL prediction by integrating the natural degradation knowledge of mechanical components.The degradation is monotonic over the whole life of bearings,which is characterized by temperature signals.To incorporate the knowledge of monotonic degradation,a positive increment recurrence relationship is introduced to keep the monotonicity.Thus,the proposed model is relatively well understood and capable to keep the learning process consistent with physical degradation.The effectiveness and merit of the RUL prediction using the proposed method are demonstrated through vibration signals collected from a set of run-to-failure tests.

An artificial viscosity augmented physics-informed neural network for incompressible flow

Physics-informed neural networks(PINNs)are proved methods that are effective in solving some strongly nonlinear partial differential equations(PDEs),e.g.,Navier-Stokes equations,with a small amount of boundary or interior data.However,the feasibility of applying PINNs to the flow at moderate or high Reynolds numbers has rarely been reported.The present paper proposes an artificial viscosity(AV)-based PINN for solving the forward and inverse flow problems.Specifically,the AV used in PINNs is inspired by the entropy viscosity method developed in conventional computational fluid dynamics(CFD)to stabilize the simulation of flow at high Reynolds numbers.The newly developed PINN is used to solve the forward problem of the two-dimensional steady cavity flow at Re=1000 and the inverse problem derived from two-dimensional film boiling.The results show that the AV augmented PINN can solve both problems with good accuracy and substantially reduce the inference errors in the forward problem.

Failure-Informed Adaptive Sampling for PINNs,Part II:Combining with Re-sampling and Subset Simulation

This is the second part of our series works on failure-informed adaptive sampling for physic-informed neural networks(PINNs).In our previous work(SIAM J.Sci.Comput.45:A1971–A1994),we have presented an adaptive sampling framework by using the failure probability as the posterior error indicator,where the truncated Gaussian model has been adopted for estimating the indicator.Here,we present two extensions of that work.The first extension consists in combining with a re-sampling technique,so that the new algorithm can maintain a constant training size.This is achieved through a cosine-annealing,which gradually transforms the sampling of collocation points from uniform to adaptive via the training progress.The second extension is to present the subset simulation(SS)algorithm as the posterior model(instead of the truncated Gaussian model)for estimating the error indicator,which can more effectively estimate the failure probability and generate new effective training points in the failure region.We investigate the performance of the new approach using several challenging problems,and numerical experiments demonstrate a significant improvement over the original algorithm.

基于PINN的复合材料自动铺放轨迹整体规划

自动纤维铺放能有效地提高复材构件的制造效率和质量。为满足复材构件的力学性能要求及铺放质量要求,在给定曲面目标域内生成铺放轨迹时需要同时考虑转弯半径、纤维角偏差以及轨迹间距等工艺指标。现有铺放轨迹规划方法大多在对基准轨迹进行优化后,通过路径密化生成铺放轨迹。这仅能保证所生成的轨迹满足单一要求,难以整体满足多个优化目标。为实现多优化目标下的复合材料自动铺放轨迹整体规划,本文将轨迹规划问题转换成为目标域内的泛函优化问题,利用内嵌物理知识神经网络(Physics-informed neural network,PINN)实现目标函数的求解,并提取目标函数的等值线作为轨迹规划的结果。相较于现有策略,本文提出的方法能整体兼顾轨迹的方向性、可铺性以及间隙质量,为实现先进复合材料自动铺放轨迹整体规划提供新思路。

Variational inference in neural functional prior using normalizing flows: application to differential equation and operator learning problems

Physics-informed deep learning has recently emerged as an effective tool for leveraging both observational data and available physical laws.Physics-informed neural networks(PINNs)and deep operator networks(DeepONets)are two such models.The former encodes the physical laws via the automatic differentiation,while the latter learns the hidden physics from data.Generally,the noisy and limited observational data as well as the over-parameterization in neural networks(NNs)result in uncertainty in predictions from deep learning models.In paper“MENG,X.,YANG,L.,MAO,Z.,FERRANDIS,J.D.,and KARNIADAKIS,G.E.Learning functional priors and posteriors from data and physics.Journal of Computational Physics,457,111073(2022)”,a Bayesian framework based on the generative adversarial networks(GANs)has been proposed as a unified model to quantify uncertainties in predictions of PINNs as well as DeepONets.Specifically,the proposed approach in“MENG,X.,YANG,L.,MAO,Z.,FERRANDIS,J.D.,and KARNIADAKIS,G.E.Learning functional priors and posteriors from data and physics.Journal of Computational Physics,457,111073(2022)”has two stages:(i)prior learning,and(ii)posterior estimation.At the first stage,the GANs are utilized to learn a functional prior either from a prescribed function distribution,e.g.,the Gaussian process,or from historical data and available physics.At the second stage,the Hamiltonian Monte Carlo(HMC)method is utilized to estimate the posterior in the latent space of GANs.However,the vanilla HMC does not support the mini-batch training,which limits its applications in problems with big data.In the present work,we propose to use the normalizing flow(NF)models in the context of variational inference(VI),which naturally enables the mini-batch training,as the alternative to HMC for posterior estimation in the latent space of GANs.A series of numerical experiments,including a nonlinear differential equation problem and a 100-dimensional(100D)Darcy problem,are conducted to demonstrate that the NFs with full-/mini-batch training are able to achieve similar accuracy as the“gold rule”HMC.Moreover,the mini-batch training of NF makes it a promising tool for quantifying uncertainty in solving the high-dimensional partial differential equation(PDE)problems with big data.

基于多域物理信息神经网络的复合地层隧道掘进地表沉降预测

复合地层中盾构掘进诱发地表沉降的准确预测是隧道工程安全建设与施工决策的关键问题。基于隧道施工诱发地层变形机制构建隧道收敛变形与掘进位置的联系,并将其耦合至深度神经网络(deep neural network,简称DNN)框架,建立了预测盾构掘进诱发地层变形的物理信息神经网络(physics-informed neural network,简称PINN)模型。针对隧道上覆多个地层的地质特征,提出了多域物理信息神经网络(multi-physics-informed neural network,简称MPINN)模型,实现了在统一的框架内对不同地层的物理信息分区域表达。结果表明:MPINN模型高度还原了有限差分法的计算结果,可以准确预测复合地层中隧道开挖诱发的地表沉降;由于融入了物理机制,MPINN模型对隧道施工诱发地表沉降的问题具有普适性,可应用于不同地质和几何条件下隧道诱发地表沉降的预测;基于工程实测数据,提出的MPINN模型准确预测了监测断面的地表沉降曲线,可为复合地层下盾构掘进过程中地表沉降的预测预警提供参考。

基于PINNs的高度非线性Richards入渗模型研究

针对具有高度非线性系数的非饱和土Richards入渗模型,利用物理信息神经网络(PINNs,physics-informed neural networks)进行求解,并通过有限差分方法对网络预测结果进行验证,发现PINNs预测结果与有限差分预测结果基本吻合;再研究超参数对PINNs误差的影响,确定训练集大小、网络层数等因素对PINNs训练集及测试集误差的影响,在合理的超参数调整下,PINNs预测模型在高度非线性入渗模型中表现出良好的训练效果。该计算方法可广泛应用于热传导、水汽迁移及应力平衡等机场工程问题求解。

基于物理信息神经网络的牵引变流器直流支撑电容参数辨识方法

为了解决车载牵引变流系统直流支撑电容器故障预测问题,该文提出一种基于物理信息神经网络的直流支撑电容器参数辨识方法。该方法只需要利用直流环节预充电过程的直流支撑电容器两端电压及采样频率,无需拟合曲线,无需严格对齐时间轴就可以获得较为准确的电容参数辨识结果。与此同时,为了克服在采集数据时因条件所限造成的数据量稀疏与分布不均问题,该文利用循环一致性生成对抗网络算法增强数据,使该方法可以适用于同一拓扑下宽范围电容区间的电容容值预测,降低了模型训练要求。实验结果表明:在正常条件下,该方法的辨识相对误差约在1%以下,并且降低采样频率能够缓解信噪比对该方法的影响。该方法为解决直流支撑电容参数辨识问题提供了新思路。

基于密集残差物理信息神经网络的各向异性旅行时计算方法

针对目前利用物理信息神经网络计算旅行时只是应用在各向同性介质上、在远离震源时误差较大和效率低等问题,而有限差分法、试射法和弯曲法等方法在多震源、高密度网格上计算成本高等问题,提出一种密集残差物理信息神经网络计算各向异性介质旅行时的方法。首先推导了各向异性因式分解后的程函方程作为损失函数项;其次引入局部自适应反正切函数为激活函数和L-BFGS-B(Limited-memory Broyden-Fletcher-Goldfarb-Shanno-B)作为优化器;最后在网络中采用分段式训练的方式,先训练深层密集残差网络,然后冻结其参数,再训练具有物理意义的浅层密集残差网络,从而评估网络得到旅行时。实验结果表明,所提方法在均匀速度模型下的旅行时最大绝对误差达到了0.0158μs,其他速度模型下平均绝对误差平均下降了两个数量级,在效率方面也平均提高了1倍,明显优于快速扫描法。

基于物理信息神经网络的混凝土破坏准则深度学习研究

混凝土破坏准则是工程结构设计和安全性评估的重要依据。结合一种新的深度学习框架--基于物理信息的深度学习神经网络,将混凝土破坏准则函数方程作为物理约束条件用来构造损失函数对应表征项,增加输入输出之间的物理信息驱动,更全面地反映各种因素之间的内在联系。利用大量试验数据,对深度学习模型进行训练,建立更为准确、适用性更广、更具泛化能力的混凝土破坏准则模型。结果表明:采用的物理信息深度学习神经网络模型,对混凝土破坏准则表达形式和参数有较好的优化识别能力和泛化能力,为规范修订、工程设计以及有限元数值模拟分析评估等提供参考。