Neutron-induced fission is an important research object in basic science.Moreover,its product yield data are an indispensable nuclear data basis in nuclear engineering and technology.The fission yield tensor decomposi...Neutron-induced fission is an important research object in basic science.Moreover,its product yield data are an indispensable nuclear data basis in nuclear engineering and technology.The fission yield tensor decomposition(FYTD)model has been developed and used to evaluate the independent fission product yield.In general,fission yield data are verified by the direct comparison of experimental and evaluated data.However,such direct comparison cannot reflect the impact of the evaluated data on application scenarios,such as reactor transport-burnup simulation.Therefore,this study applies the evaluated fission yield data in transport-burnup simulation to verify their accuracy and possibility of application.Herein,the evaluated yield data of235U and239Pu are applied in the transport-burnup simulation of a pressurized water reactor(PWR)and sodium-cooled fast reactor(SFR)for verification.During the reactor operation stage,the errors in pin-cell reactivity caused by the evaluated fission yield do not exceed 500 and 200 pcm for the PWR and SFR,respectively.The errors in decay heat and135Xe and149Sm concentrations during the short-term shutdown of the PWR are all less than 1%;the errors in decay heat and activity of the spent fuel of the PWR and SFR during the temporary storage stage are all less than 2%.For the PWR,the errors in important nuclide concentrations in spent fuel,such as90Sr,137Cs,85Kr,and99Tc,are all less than 6%,and a larger error of 37%is observed on129I.For the SFR,the concentration errors of ten important nuclides in spent fuel are all less than 16%.A comparison of various aspects reveals that the transport-burnup simulation results using the FYTD model evaluation have little difference compared with the reference results using ENDF/B-Ⅷ.0 data.This proves that the evaluation of the FYTD model may have application value in reactor physical analysis.展开更多
Particle entrainment is an inevitable phenomenon in pipeline systems,especially during the development and extraction phases of oil and gas wells.Accurately predicting the critical velocity for particle transport is a...Particle entrainment is an inevitable phenomenon in pipeline systems,especially during the development and extraction phases of oil and gas wells.Accurately predicting the critical velocity for particle transport is a key focus for implementing effective sand control management.This work presents a semi-supervised learning–deep hybrid kernel extreme learning machine(SSL-DHKELM)model for predicting the critical velocity,which integrates multiple machine learning theories including the deep learning approach,which is adept at advanced feature extraction.Meanwhile,the SSL framework enhances the model's capabilities when data availability is limited.An improved slime mould algorithm is also employed to optimize the model's hyperparameters.The proposed model has high accuracy on both the sample dataset and out-of-sample data.When trained with only 10%of the data,the model's error still did not increase significantly.Additionally,this model achieves superior predictive accuracy compared to existing mechanistic models,demonstrating its impressive performance and robustness.展开更多
We live in an era of rapid urbanization as many cities are experiencing an unprecedented rate of population growth and congestion.Public transport is playing an increasingly important role in urban mobility with a nee...We live in an era of rapid urbanization as many cities are experiencing an unprecedented rate of population growth and congestion.Public transport is playing an increasingly important role in urban mobility with a need to move people and goods efficiently around the city.With such pressures on existing public transportation systems,this paper investigates the opportunities to use social media to more effectively engage with citizens and customers using such services.This research forms a case study of the use of passively collected forms of big data in cities-focusing on Sydney,Australia.Firstly,it examines social media data(Tweets)related to public transport performance.Secondly,it joins this to longitudinal big data-delay information continuously broadcast by the network over a year,thus forming hundreds of millions of data artifacts.Topics,tones,and sentiment are modeled using machine learning and Natural Language Processing(NLP)techniques.These resulting data,and models,are compared to opinions derived from a citizen survey among users.The validity of such data and models versus the intentions of users,in the context of systems that monitor and improve transport performance,are discussed.As such,key recommendations for developing Smart Cities were formed in an applied research context based on these data and techniques.展开更多
Charge transfer and transport properties are crucial in the photophysical process of exciton dissociation and recombination at the donor/acceptor(D/A)interface.Herein,machine learning(ML)is applied to predict the char...Charge transfer and transport properties are crucial in the photophysical process of exciton dissociation and recombination at the donor/acceptor(D/A)interface.Herein,machine learning(ML)is applied to predict the charge transfer state energy(ECT)and identify the relationship between ECT and intermolecular packing structures sampled from molecular dynamics(MD)simulations on fullerene-and non-fullerene-based systems with different D/A ratios(RDA),oligomer sizes,and D/A pairs.The gradient boosting regression(GBR)exhibits satisfactory performance(r=0.96)in predicting ECT withπ-packing related features,aggregation extent,backbone of donor,and energy levels of frontier molecular orbitals.The charge transport property affected byπ-packing with different RDA has also been investigated by space-charge-limited current(SCLC)measurement and MD simulations.The SCLC results indicate an improved hole transport of non-fullerene system PM6/Y6 with RDA of 1.2:1 in comparison with the 1:1 counterpart,which is mainly attributed to the bridge role of donor unit in Y6.The reduced energetic disorder is correlated with the improved miscibility of polymer with RDA increased from 1:1 to 1.2:1.The morphology-related features are also applicable to other complicated systems,such as perovskite solar cells,to bridge the gap between device performance and microscopic packing structures.展开更多
为高效精确地预测无信号环形交叉口机动车与非机动车的交通冲突,提出了基于遗传算法优化的BP神经网络(genetic algorithm and back propagation,GA-BP)和支持向量回归(support vector regression,SVR)的组合预测模型(SVR-GA-BP)。通过...为高效精确地预测无信号环形交叉口机动车与非机动车的交通冲突,提出了基于遗传算法优化的BP神经网络(genetic algorithm and back propagation,GA-BP)和支持向量回归(support vector regression,SVR)的组合预测模型(SVR-GA-BP)。通过无人机采集混合交通流高清视频,利用视频识别软件Tracker提取机非交通冲突轨迹数据,以距离碰撞时间(time to collision,TTC)为判别指标,确定机非冲突严重程度。基于偏相关性分析确定交通量、平均速度、大车比例等为机非交通冲突的显著影响因素,选取均方根误差(root mean squared error,RMSE)、平均绝对误差(mean absolute error,MAE)等五种评价指标对SVR模型、BP神经网络、SVR-GA-BP模型的预测值进行精度分析。结果表明,组合模型在一般冲突预测中精度为97.1%,相比SVR和BP神经网络分别提高6.9%和2.5%,在严重冲突预测中精度为96.1%,相比SVR和BP神经网络分别提高7.3%和5.1%。可见SVR-GA-BP组合模型能够有效预测无信号环形交叉口的机非冲突且精度最高,可为同类型交叉口的安全评价提供借鉴。展开更多
Traffic congestion is one of the main challenges in transportation engineering. It directly impactsthe economy by increasing travel time and affecting the environment by excessive fuel consumptionand emission. Road ro...Traffic congestion is one of the main challenges in transportation engineering. It directly impactsthe economy by increasing travel time and affecting the environment by excessive fuel consumptionand emission. Road route recommendation to overcome the congestion by alternativeroute suggestions has gained high importance. The existing route recommendation systems areproposed using the reinforcement learning algorithm (Q-learning). The techniques suggestedin this paper are state-action-reward-state-action (SARSA) algorithm and dynamic programming(DP) to guide the commuters to reach the destination with an optimal solution. The algorithmconsiders travel time, cost, flexibility, and traffic intensity as the user preference attributes torecommend an optimal route. The recommended system is implemented by building a roadnetwork graph. We assign values to each user preference attribute along the edges, which cantake high(1) or low(0) values. By considering these values, the system recommends the route.The proposed system performance is evaluated based on computation time, cumulative reward,and accuracy. The results show that DP outperforms the SARSA algorithm.展开更多
Image classification is a core field in the research area of image proces-sing and computer vision in which vehicle classification is a critical domain.The purpose of vehicle categorization is to formulate a compact s...Image classification is a core field in the research area of image proces-sing and computer vision in which vehicle classification is a critical domain.The purpose of vehicle categorization is to formulate a compact system to assist in real-world problems and applications such as security,traffic analysis,and self-driving and autonomous vehicles.The recent revolution in the field of machine learning and artificial intelligence has provided an immense amount of support for image processing related problems and has overtaken the conventional,and handcrafted means of solving image analysis problems.In this paper,a combina-tion of pre-trained CNN GoogleNet and a nature-inspired problem optimization scheme,particle swarm optimization(PSO),was employed for autonomous vehi-cle classification.The model was trained on a vehicle image dataset obtained from Kaggle that has been suitably augmented.The trained model was classified using several classifiers;however,the Cubic SVM(CSVM)classifier was found to out-perform the others in both time consumption and accuracy(94.8%).The results obtained from empirical evaluations and statistical tests reveal that the model itself has shown to outperform the other related models not only in terms of accu-racy(94.8%)but also in terms of training time(82.7 s)and speed prediction(380 obs/sec).展开更多
Process-based reactive transport modeling(RTM)integrates thermodynamic and kinetically controlled fluid-rock interactions with fluid flow through porous media in the subsurface and surface environment.RTM is usually c...Process-based reactive transport modeling(RTM)integrates thermodynamic and kinetically controlled fluid-rock interactions with fluid flow through porous media in the subsurface and surface environment.RTM is usually conducted through numerical programs based on the first principle of physical processes.However,the calculation for complex chemical reactions in most available programs is an iterative process,where each iteration is in general computationally intensive.A workflow of neural networkbased surrogate model as a proxy for process-based reactive transport simulation is established in this study.The workflow includes(1)base case RTM design,(2)development of training experiments,(3)surrogate model construction based on machine learning,(4)surrogate model validation,and(5)prediction with the calibrated model.The training experiments for surrogate modeling are generated and run prior to the predictions using RTM.The results show that the predictions from the surrogate model agree well with those from processes-based RTM but with a significantly reduced computational time.The well-trained surrogate model is especially useful when a large number of realizations are required,such as the sensitivity analysis or model calibration,which can significantly reduce the computational time compared to that required by RTM.The benefits are(1)it automatizes the experimental design during the sensitivity analysis to get sufficient numbers and coverage of the training cases;(2)it parallelizes the calculations of RTM training cases during the sensitivity analysis to reduce the simulation time;(3)it uses the neural network algorithm to rank the sensitivity of the parameters and to search the optimal solution for model calibration.展开更多
In recent years, (big) data science has emerged as the “fourth paradigm” in physical science research. Data-driven techniques, e.g. machine learning, are advantageous in dealing with problems of high-dimensional fea...In recent years, (big) data science has emerged as the “fourth paradigm” in physical science research. Data-driven techniques, e.g. machine learning, are advantageous in dealing with problems of high-dimensional features and complex mappings between quantities, which are otherwise of great difficulty or huge cost with other scientific paradigms. In the past five years or so, there has been a rapid growth of machine learning-assisted research on thermal transport. In this perspective, we review the recent progress in the intersection between machine learning and thermal transport, where machine learning methods generally serve as surrogate models for predicting the thermal transport properties, or as tools for designing structures for the desired thermal properties and exploring thermal transport mechanisms. We provide perspectives about the advantages of machine learning methods in comparison to the physics-based methods for studying thermal transport properties. We also discuss how to improve the accuracy of predictive analytics and efficiency of structural optimization, to provide guidance for better utilizing machine learning-based methods to advance thermal transport research. Finally, we identify several outstanding challenges in this active area as well as opportunities for future developments,including developing machine learning methods suitable for small datasets, discovering effective physics-based descriptors, generating dataset from experiments and validating machine learning results with experiments, and making breakthroughs via discovering new physics.展开更多
基金the National Natural Science Foundation of China(Nos.11875328,12075327 and 12105170)the Key Laboratory of Nuclear Data foundation(No.JCKY2022201C157)+1 种基金the Fundamental Research Funds for the Central Universities,Sun Yat-sen University(No.22lgqb39)the Open Project of Guangxi Key Laboratory of Nuclear Physics and Nuclear Technology(No.NLK2020-02).
文摘Neutron-induced fission is an important research object in basic science.Moreover,its product yield data are an indispensable nuclear data basis in nuclear engineering and technology.The fission yield tensor decomposition(FYTD)model has been developed and used to evaluate the independent fission product yield.In general,fission yield data are verified by the direct comparison of experimental and evaluated data.However,such direct comparison cannot reflect the impact of the evaluated data on application scenarios,such as reactor transport-burnup simulation.Therefore,this study applies the evaluated fission yield data in transport-burnup simulation to verify their accuracy and possibility of application.Herein,the evaluated yield data of235U and239Pu are applied in the transport-burnup simulation of a pressurized water reactor(PWR)and sodium-cooled fast reactor(SFR)for verification.During the reactor operation stage,the errors in pin-cell reactivity caused by the evaluated fission yield do not exceed 500 and 200 pcm for the PWR and SFR,respectively.The errors in decay heat and135Xe and149Sm concentrations during the short-term shutdown of the PWR are all less than 1%;the errors in decay heat and activity of the spent fuel of the PWR and SFR during the temporary storage stage are all less than 2%.For the PWR,the errors in important nuclide concentrations in spent fuel,such as90Sr,137Cs,85Kr,and99Tc,are all less than 6%,and a larger error of 37%is observed on129I.For the SFR,the concentration errors of ten important nuclides in spent fuel are all less than 16%.A comparison of various aspects reveals that the transport-burnup simulation results using the FYTD model evaluation have little difference compared with the reference results using ENDF/B-Ⅷ.0 data.This proves that the evaluation of the FYTD model may have application value in reactor physical analysis.
基金the National Natural Science Foundation of China(52074220,52304008)the Natural Science Basic Research Program of Shaanxi Province(2022JC-37,2023-JC-QN-0403,2024JC-YBQN-0381)the China Postdoctoral Science Foundation(2023MD734223).
文摘Particle entrainment is an inevitable phenomenon in pipeline systems,especially during the development and extraction phases of oil and gas wells.Accurately predicting the critical velocity for particle transport is a key focus for implementing effective sand control management.This work presents a semi-supervised learning–deep hybrid kernel extreme learning machine(SSL-DHKELM)model for predicting the critical velocity,which integrates multiple machine learning theories including the deep learning approach,which is adept at advanced feature extraction.Meanwhile,the SSL framework enhances the model's capabilities when data availability is limited.An improved slime mould algorithm is also employed to optimize the model's hyperparameters.The proposed model has high accuracy on both the sample dataset and out-of-sample data.When trained with only 10%of the data,the model's error still did not increase significantly.Additionally,this model achieves superior predictive accuracy compared to existing mechanistic models,demonstrating its impressive performance and robustness.
文摘We live in an era of rapid urbanization as many cities are experiencing an unprecedented rate of population growth and congestion.Public transport is playing an increasingly important role in urban mobility with a need to move people and goods efficiently around the city.With such pressures on existing public transportation systems,this paper investigates the opportunities to use social media to more effectively engage with citizens and customers using such services.This research forms a case study of the use of passively collected forms of big data in cities-focusing on Sydney,Australia.Firstly,it examines social media data(Tweets)related to public transport performance.Secondly,it joins this to longitudinal big data-delay information continuously broadcast by the network over a year,thus forming hundreds of millions of data artifacts.Topics,tones,and sentiment are modeled using machine learning and Natural Language Processing(NLP)techniques.These resulting data,and models,are compared to opinions derived from a citizen survey among users.The validity of such data and models versus the intentions of users,in the context of systems that monitor and improve transport performance,are discussed.As such,key recommendations for developing Smart Cities were formed in an applied research context based on these data and techniques.
基金supported by the National Natural Science Foundation of China(Nos.22033004 and 21873045).
文摘Charge transfer and transport properties are crucial in the photophysical process of exciton dissociation and recombination at the donor/acceptor(D/A)interface.Herein,machine learning(ML)is applied to predict the charge transfer state energy(ECT)and identify the relationship between ECT and intermolecular packing structures sampled from molecular dynamics(MD)simulations on fullerene-and non-fullerene-based systems with different D/A ratios(RDA),oligomer sizes,and D/A pairs.The gradient boosting regression(GBR)exhibits satisfactory performance(r=0.96)in predicting ECT withπ-packing related features,aggregation extent,backbone of donor,and energy levels of frontier molecular orbitals.The charge transport property affected byπ-packing with different RDA has also been investigated by space-charge-limited current(SCLC)measurement and MD simulations.The SCLC results indicate an improved hole transport of non-fullerene system PM6/Y6 with RDA of 1.2:1 in comparison with the 1:1 counterpart,which is mainly attributed to the bridge role of donor unit in Y6.The reduced energetic disorder is correlated with the improved miscibility of polymer with RDA increased from 1:1 to 1.2:1.The morphology-related features are also applicable to other complicated systems,such as perovskite solar cells,to bridge the gap between device performance and microscopic packing structures.
文摘为高效精确地预测无信号环形交叉口机动车与非机动车的交通冲突,提出了基于遗传算法优化的BP神经网络(genetic algorithm and back propagation,GA-BP)和支持向量回归(support vector regression,SVR)的组合预测模型(SVR-GA-BP)。通过无人机采集混合交通流高清视频,利用视频识别软件Tracker提取机非交通冲突轨迹数据,以距离碰撞时间(time to collision,TTC)为判别指标,确定机非冲突严重程度。基于偏相关性分析确定交通量、平均速度、大车比例等为机非交通冲突的显著影响因素,选取均方根误差(root mean squared error,RMSE)、平均绝对误差(mean absolute error,MAE)等五种评价指标对SVR模型、BP神经网络、SVR-GA-BP模型的预测值进行精度分析。结果表明,组合模型在一般冲突预测中精度为97.1%,相比SVR和BP神经网络分别提高6.9%和2.5%,在严重冲突预测中精度为96.1%,相比SVR和BP神经网络分别提高7.3%和5.1%。可见SVR-GA-BP组合模型能够有效预测无信号环形交叉口的机非冲突且精度最高,可为同类型交叉口的安全评价提供借鉴。
文摘Traffic congestion is one of the main challenges in transportation engineering. It directly impactsthe economy by increasing travel time and affecting the environment by excessive fuel consumptionand emission. Road route recommendation to overcome the congestion by alternativeroute suggestions has gained high importance. The existing route recommendation systems areproposed using the reinforcement learning algorithm (Q-learning). The techniques suggestedin this paper are state-action-reward-state-action (SARSA) algorithm and dynamic programming(DP) to guide the commuters to reach the destination with an optimal solution. The algorithmconsiders travel time, cost, flexibility, and traffic intensity as the user preference attributes torecommend an optimal route. The recommended system is implemented by building a roadnetwork graph. We assign values to each user preference attribute along the edges, which cantake high(1) or low(0) values. By considering these values, the system recommends the route.The proposed system performance is evaluated based on computation time, cumulative reward,and accuracy. The results show that DP outperforms the SARSA algorithm.
文摘Image classification is a core field in the research area of image proces-sing and computer vision in which vehicle classification is a critical domain.The purpose of vehicle categorization is to formulate a compact system to assist in real-world problems and applications such as security,traffic analysis,and self-driving and autonomous vehicles.The recent revolution in the field of machine learning and artificial intelligence has provided an immense amount of support for image processing related problems and has overtaken the conventional,and handcrafted means of solving image analysis problems.In this paper,a combina-tion of pre-trained CNN GoogleNet and a nature-inspired problem optimization scheme,particle swarm optimization(PSO),was employed for autonomous vehi-cle classification.The model was trained on a vehicle image dataset obtained from Kaggle that has been suitably augmented.The trained model was classified using several classifiers;however,the Cubic SVM(CSVM)classifier was found to out-perform the others in both time consumption and accuracy(94.8%).The results obtained from empirical evaluations and statistical tests reveal that the model itself has shown to outperform the other related models not only in terms of accu-racy(94.8%)but also in terms of training time(82.7 s)and speed prediction(380 obs/sec).
文摘Process-based reactive transport modeling(RTM)integrates thermodynamic and kinetically controlled fluid-rock interactions with fluid flow through porous media in the subsurface and surface environment.RTM is usually conducted through numerical programs based on the first principle of physical processes.However,the calculation for complex chemical reactions in most available programs is an iterative process,where each iteration is in general computationally intensive.A workflow of neural networkbased surrogate model as a proxy for process-based reactive transport simulation is established in this study.The workflow includes(1)base case RTM design,(2)development of training experiments,(3)surrogate model construction based on machine learning,(4)surrogate model validation,and(5)prediction with the calibrated model.The training experiments for surrogate modeling are generated and run prior to the predictions using RTM.The results show that the predictions from the surrogate model agree well with those from processes-based RTM but with a significantly reduced computational time.The well-trained surrogate model is especially useful when a large number of realizations are required,such as the sensitivity analysis or model calibration,which can significantly reduce the computational time compared to that required by RTM.The benefits are(1)it automatizes the experimental design during the sensitivity analysis to get sufficient numbers and coverage of the training cases;(2)it parallelizes the calculations of RTM training cases during the sensitivity analysis to reduce the simulation time;(3)it uses the neural network algorithm to rank the sensitivity of the parameters and to search the optimal solution for model calibration.
基金support by the National Natural Science Foundation of China(No.52122606).
文摘In recent years, (big) data science has emerged as the “fourth paradigm” in physical science research. Data-driven techniques, e.g. machine learning, are advantageous in dealing with problems of high-dimensional features and complex mappings between quantities, which are otherwise of great difficulty or huge cost with other scientific paradigms. In the past five years or so, there has been a rapid growth of machine learning-assisted research on thermal transport. In this perspective, we review the recent progress in the intersection between machine learning and thermal transport, where machine learning methods generally serve as surrogate models for predicting the thermal transport properties, or as tools for designing structures for the desired thermal properties and exploring thermal transport mechanisms. We provide perspectives about the advantages of machine learning methods in comparison to the physics-based methods for studying thermal transport properties. We also discuss how to improve the accuracy of predictive analytics and efficiency of structural optimization, to provide guidance for better utilizing machine learning-based methods to advance thermal transport research. Finally, we identify several outstanding challenges in this active area as well as opportunities for future developments,including developing machine learning methods suitable for small datasets, discovering effective physics-based descriptors, generating dataset from experiments and validating machine learning results with experiments, and making breakthroughs via discovering new physics.
