<div style="text-align:justify;"> In order to speed up the global optimization-based mesh smoothing, an enhanced steepest descent method is presented in the paper. Numerical experiment results show tha...<div style="text-align:justify;"> In order to speed up the global optimization-based mesh smoothing, an enhanced steepest descent method is presented in the paper. Numerical experiment results show that the method performs better than the steepest descent method in the global smoothing. We also presented a physically-based interpretation to explain why the method works better than the steepest descent method. </div>展开更多
This paper investigates traffic flow of connected and automated vehicles at lane drop on two-lane highway. We evaluate and compare performance of an optimization-based control algorithm(OCA) with that of a heuristic r...This paper investigates traffic flow of connected and automated vehicles at lane drop on two-lane highway. We evaluate and compare performance of an optimization-based control algorithm(OCA) with that of a heuristic rules-based algorithm(HRA). In the OCA, the average speed of each vehicle is maximized. In the HRA, virtual vehicle and restriction of the command acceleration caused by the virtual vehicle are introduced. It is found that(i) capacity under the HRA(denoted as C_(H)) is smaller than capacity under the OCA;(ii) the travel delay is always smaller under the OCA, but driving is always much more comfortable under the HRA;(iii) when the inflow rate is smaller than C_(H), the HRA outperforms the OCA with respect to the fuel consumption and the monetary cost;(iv) when the inflow rate is larger than C_(H), the HRA initially performs better with respect to the fuel consumption and the monetary cost, but the OCA would become better after certain time. The spatiotemporal pattern and speed profile of traffic flow are presented, which explains the reason underlying the different performance. The study is expected to help for better understanding of the two different types of algorithm.展开更多
The objective of this paper is to present a robust safety-critical control system based on the active disturbance rejection control approach, designed to guarantee safety even in the presence of model inaccuracies, un...The objective of this paper is to present a robust safety-critical control system based on the active disturbance rejection control approach, designed to guarantee safety even in the presence of model inaccuracies, unknown dynamics, and external disturbances. The proposed method combines control barrier functions and control Lyapunov functions with a nonlinear extended state observer to produce a robust and safe control strategy for dynamic systems subject to uncertainties and disturbances. This control strategy employs an optimization-based control, supported by the disturbance estimation from a nonlinear extended state observer. Using a quadratic programming algorithm, the controller computes an optimal, stable, and safe control action at each sampling instant. The effectiveness of the proposed approach is demonstrated through numerical simulations of a safety-critical interconnected adaptive cruise control system.展开更多
The commonly used method for estimating crack opening displacement(COD)is based on analytical models derived from strain transferring.However,when large background noise exists in distributed fiber optic sensing(DFOS)...The commonly used method for estimating crack opening displacement(COD)is based on analytical models derived from strain transferring.However,when large background noise exists in distributed fiber optic sensing(DFOS)data,estimating COD through an analytical model is very difficult even if the DFOS data have been denoised.To address this challenge,this study proposes a machine learning(ML)-based methodology to complete rock's COD estimation from establishment of a dataset with one-to-one correspondence between strain sequence and COD to the optimization of ML models.The Bayesian optimization is used via the Hyperopt Python library to determine the appropriate hyper-parameters of four ML models.To ensure that the best hyper-parameters will not be missing,the configuration space in Hyperopt is specified by probability distribution.The four models are trained using DFOS data with minimal noise while being examined on datasets with different noise levels to test their anti-noise robustness.The proposed models are compared each other in terms of goodness of fit and mean squared error.The results show that the Bayesian optimization-based random forest is promising to estimate the COD of rock using noisy DFOS data.展开更多
The pedestrians can only avoid collisions passively under the action of forces during simulations using the social force model, which may lead to unnatural behaviors. This paper proposes an optimization-based model fo...The pedestrians can only avoid collisions passively under the action of forces during simulations using the social force model, which may lead to unnatural behaviors. This paper proposes an optimization-based model for the avoidance of collisions, where the social repulsive force is removed in favor of a search for the quickest path to destination in the pedestrian＇s vision field. In this way, the behaviors of pedestrians are governed by changing their desired walking direction and desired speed. By combining the critical factors of pedestrian movement, such as positions of the exit and obstacles and velocities of the neighbors, the choice of desired velocity has been rendered to a discrete optimization problem. Therefore,it is the self-driven force that leads pedestrians to a free path rather than the repulsive force, which means the pedestrians can actively avoid collisions. The new model is verified by comparing with the fundamental diagram and actual data. The simulation results of individual avoidance trajectories and crowd avoidance behaviors demonstrate the reasonability of the proposed model.展开更多
In computational fluid dynamics(CFD),mesh-smoothing methods are widely used to refine the mesh quality for achieving high-precision numerical simulations.Specifically,optimization-based smoothing is used for high-qual...In computational fluid dynamics(CFD),mesh-smoothing methods are widely used to refine the mesh quality for achieving high-precision numerical simulations.Specifically,optimization-based smoothing is used for high-quality mesh smoothing,but it incurs significant computational overhead.Pioneer works have improved its smoothing efficiency by adopting supervised learning to learn smoothing methods from high-quality meshes.However,they pose difficulties in smoothing the mesh nodes with varying degrees and require data augmentation to address the node input sequence problem.Additionally,the required labeled high-quality meshes further limit the applicability of the proposed method.In this paper,we present graph-based smoothing mesh net(GMSNet),a lightweight neural network model for intelligent mesh smoothing.GMSNet adopts graph neural networks(GNNs)to extract features of the node’s neighbors and outputs the optimal node position.During smoothing,we also introduce a fault-tolerance mechanism to prevent GMSNet from generating negative volume elements.With a lightweight model,GMSNet can effectively smooth mesh nodes with varying degrees and remain unaffected by the order of input data.A novel loss function,MetricLoss,is developed to eliminate the need for high-quality meshes,which provides stable and rapid convergence during training.We compare GMSNet with commonly used mesh-smoothing methods on two-dimensional(2D)triangle meshes.Experimental results show that GMSNet achieves outstanding mesh-smoothing performances with 5%of the model parameters compared to the previous model,but offers a speedup of 13.56 times over the optimization-based smoothing.展开更多
文摘<div style="text-align:justify;"> In order to speed up the global optimization-based mesh smoothing, an enhanced steepest descent method is presented in the paper. Numerical experiment results show that the method performs better than the steepest descent method in the global smoothing. We also presented a physically-based interpretation to explain why the method works better than the steepest descent method. </div>
基金Project supported in part by the Fundamental Research Funds for the Central Universities (Grant No.2021JBZ107)the National Natural Science Foundation of China (Grant Nos.72288101 and 71931002)。
文摘This paper investigates traffic flow of connected and automated vehicles at lane drop on two-lane highway. We evaluate and compare performance of an optimization-based control algorithm(OCA) with that of a heuristic rules-based algorithm(HRA). In the OCA, the average speed of each vehicle is maximized. In the HRA, virtual vehicle and restriction of the command acceleration caused by the virtual vehicle are introduced. It is found that(i) capacity under the HRA(denoted as C_(H)) is smaller than capacity under the OCA;(ii) the travel delay is always smaller under the OCA, but driving is always much more comfortable under the HRA;(iii) when the inflow rate is smaller than C_(H), the HRA outperforms the OCA with respect to the fuel consumption and the monetary cost;(iv) when the inflow rate is larger than C_(H), the HRA initially performs better with respect to the fuel consumption and the monetary cost, but the OCA would become better after certain time. The spatiotemporal pattern and speed profile of traffic flow are presented, which explains the reason underlying the different performance. The study is expected to help for better understanding of the two different types of algorithm.
基金supported by the Fondo para el Primer Proyecto of the Comitépara el Desarrollo de la Investigación(CODI)at the Universidad de Antioquia(Grant Number PRV2024-78509)。
文摘The objective of this paper is to present a robust safety-critical control system based on the active disturbance rejection control approach, designed to guarantee safety even in the presence of model inaccuracies, unknown dynamics, and external disturbances. The proposed method combines control barrier functions and control Lyapunov functions with a nonlinear extended state observer to produce a robust and safe control strategy for dynamic systems subject to uncertainties and disturbances. This control strategy employs an optimization-based control, supported by the disturbance estimation from a nonlinear extended state observer. Using a quadratic programming algorithm, the controller computes an optimal, stable, and safe control action at each sampling instant. The effectiveness of the proposed approach is demonstrated through numerical simulations of a safety-critical interconnected adaptive cruise control system.
基金The Young Scientists Fund of the National Natural Science Foundation of China(Grant No.42407250)the Fund from Research Centre for Resources Engineering towards Carbon Neutrality(RCRE)of The Hong Kong Polytechnic University(Grant No.No.1-BBEM)the Fund from Natural Science Foundation of Jiangsu Province(Grant No.BK20241211)。
文摘The commonly used method for estimating crack opening displacement(COD)is based on analytical models derived from strain transferring.However,when large background noise exists in distributed fiber optic sensing(DFOS)data,estimating COD through an analytical model is very difficult even if the DFOS data have been denoised.To address this challenge,this study proposes a machine learning(ML)-based methodology to complete rock's COD estimation from establishment of a dataset with one-to-one correspondence between strain sequence and COD to the optimization of ML models.The Bayesian optimization is used via the Hyperopt Python library to determine the appropriate hyper-parameters of four ML models.To ensure that the best hyper-parameters will not be missing,the configuration space in Hyperopt is specified by probability distribution.The four models are trained using DFOS data with minimal noise while being examined on datasets with different noise levels to test their anti-noise robustness.The proposed models are compared each other in terms of goodness of fit and mean squared error.The results show that the Bayesian optimization-based random forest is promising to estimate the COD of rock using noisy DFOS data.
基金supported by the National Natural Science Foundation of China(Grant Nos.61233001 and 61322307)the Fundamental Research Funds for Central Universities of China(Grant No.2013JBZ007)
文摘The pedestrians can only avoid collisions passively under the action of forces during simulations using the social force model, which may lead to unnatural behaviors. This paper proposes an optimization-based model for the avoidance of collisions, where the social repulsive force is removed in favor of a search for the quickest path to destination in the pedestrian＇s vision field. In this way, the behaviors of pedestrians are governed by changing their desired walking direction and desired speed. By combining the critical factors of pedestrian movement, such as positions of the exit and obstacles and velocities of the neighbors, the choice of desired velocity has been rendered to a discrete optimization problem. Therefore,it is the self-driven force that leads pedestrians to a free path rather than the repulsive force, which means the pedestrians can actively avoid collisions. The new model is verified by comparing with the fundamental diagram and actual data. The simulation results of individual avoidance trajectories and crowd avoidance behaviors demonstrate the reasonability of the proposed model.
基金supported by the National Key Research and Development Program of China(No.2021YFB0300101)the Youth Foundation of National University of Defense Technology,China(No.ZK2023-11)the National Natural Science Foundation of China(No.12102467)。
文摘In computational fluid dynamics(CFD),mesh-smoothing methods are widely used to refine the mesh quality for achieving high-precision numerical simulations.Specifically,optimization-based smoothing is used for high-quality mesh smoothing,but it incurs significant computational overhead.Pioneer works have improved its smoothing efficiency by adopting supervised learning to learn smoothing methods from high-quality meshes.However,they pose difficulties in smoothing the mesh nodes with varying degrees and require data augmentation to address the node input sequence problem.Additionally,the required labeled high-quality meshes further limit the applicability of the proposed method.In this paper,we present graph-based smoothing mesh net(GMSNet),a lightweight neural network model for intelligent mesh smoothing.GMSNet adopts graph neural networks(GNNs)to extract features of the node’s neighbors and outputs the optimal node position.During smoothing,we also introduce a fault-tolerance mechanism to prevent GMSNet from generating negative volume elements.With a lightweight model,GMSNet can effectively smooth mesh nodes with varying degrees and remain unaffected by the order of input data.A novel loss function,MetricLoss,is developed to eliminate the need for high-quality meshes,which provides stable and rapid convergence during training.We compare GMSNet with commonly used mesh-smoothing methods on two-dimensional(2D)triangle meshes.Experimental results show that GMSNet achieves outstanding mesh-smoothing performances with 5%of the model parameters compared to the previous model,but offers a speedup of 13.56 times over the optimization-based smoothing.
