To address the problems of low accuracy by the CONWEP model and poor efficiency by the Coupled Eulerian-Lagrangian(CEL)method in predicting close-range air blast loads of cylindrical charges,a neural network-based sim...To address the problems of low accuracy by the CONWEP model and poor efficiency by the Coupled Eulerian-Lagrangian(CEL)method in predicting close-range air blast loads of cylindrical charges,a neural network-based simulation(NNS)method with higher accuracy and better efficiency was proposed.The NNS method consisted of three main steps.First,the parameters of blast loads,including the peak pressures and impulses of cylindrical charges with different aspect ratios(L/D)at different stand-off distances and incident angles were obtained by two-dimensional numerical simulations.Subsequently,incident shape factors of cylindrical charges with arbitrary aspect ratios were predicted by a neural network.Finally,reflected shape factors were derived and implemented into the subroutine of the ABAQUS code to modify the CONWEP model,including modifications of impulse and overpressure.The reliability of the proposed NNS method was verified by related experimental results.Remarkable accuracy improvement was acquired by the proposed NNS method compared with the unmodified CONWEP model.Moreover,huge efficiency superiority was obtained by the proposed NNS method compared with the CEL method.The proposed NNS method showed good accuracy when the scaled distance was greater than 0.2 m/kg^(1/3).It should be noted that there is no need to generate a new dataset again since the blast loads satisfy the similarity law,and the proposed NNS method can be directly used to simulate the blast loads generated by different cylindrical charges.The proposed NNS method with high efficiency and accuracy can be used as an effective method to analyze the dynamic response of structures under blast loads,and it has significant application prospects in designing protective structures.展开更多
The topology structure of the artificial neural network is an intelligent control model,which is used for the intelligent vehicle control system and household sweeping robot.When setting the intelligent control system...The topology structure of the artificial neural network is an intelligent control model,which is used for the intelligent vehicle control system and household sweeping robot.When setting the intelligent control system,the connection point of each network is regarded as a neuron in the nervous system,and each connection point has input and output functions.Only when the input of nodes reaches a certain threshold can the output function of nodes be stimulated.Using the networking mode of the artificial neural network model,the mobile node can output in multiple directions.If the input direction of a certain path is the same as that of other nodes,it can choose to avoid and choose another path.The weighted value of each path between nodes is different,which means that the influence of the front node on the current node varies.The control method based on the artificial neural network model can be applied to vehicle control,household sweeping robots,and other fields,and a relatively optimized scheme can be obtained from the aspect of time and energy consumption.展开更多
Numerical simulation plays an important role in the dynamic analysis of multibody system.With the rapid development of computer science,the numerical solution technology has been further developed.Recently,data-driven...Numerical simulation plays an important role in the dynamic analysis of multibody system.With the rapid development of computer science,the numerical solution technology has been further developed.Recently,data-driven method has become a very popular computing method.However,due to lack of necessary mechanism information of the traditional pure data-driven methods based on neural network,its numerical accuracy cannot be guaranteed for strong nonlinear system.Therefore,this work proposes a mechanism-data hybrid-driven strategy for solving nonlinear multibody system based on physics-informed neural network to overcome the limitation of traditional data-driven methods.The strategy proposed in this paper introduces scaling coefficients to introduce the dynamic model of multibody system into neural network,ensuring that the training results of neural network conform to the mechanics principle of the system,thereby ensuring the good reliability of the data-driven method.Finally,the stability,generalization ability and numerical accuracy of the proposed method are discussed and analyzed using three typical multibody systems,and the constrained default situations can be controlled within the range of 10^(-2)-10^(-4).展开更多
Physics-informed neural networks(PINNs)have shown considerable promise for performing numerical simulations in fluid mechanics.They provide mesh-free,end-to-end approaches by embedding physical laws into their loss fu...Physics-informed neural networks(PINNs)have shown considerable promise for performing numerical simulations in fluid mechanics.They provide mesh-free,end-to-end approaches by embedding physical laws into their loss functions.However,when addressing complex flow problems,PINNs still face some challenges such as activation saturation and vanishing gradients in deep network training,leading to slow convergence and insufficient prediction accuracy.We present physics-informed neural networks incorporating lattice Boltzmann method optimized by tanh robust weight initialization(T-PINN-LBM)to address these challenges.This approach fuses the mesoscopic lattice Boltzmann model with the automatic differentiation framework of PINNs.It also implements a tanh robust weight initialization method derived from fixed point analysis.This model effectively mitigates activation and gradient decay in deep networks,improving convergence speed and data efficiency in multiscale flow simulations.We validate the effectiveness of the model on the classical arithmetic example of lid-driven cavity flow.Compared to the traditional Xavier initialized PINN and PINN-LBM,T-PINNLBM reduces the mean absolute error(MAE)by one order of magnitude at the same network depth and maintains stable convergence in deeper networks.The results demonstrate that this model can accurately capture complex flow structures without prior data,providing a new feasible pathway for data-free driven fluid simulation.展开更多
The trajectory tracking control problem is addressed for autonomous underwater vehicle(AUV) in marine environ?ment, with presence of the influence of the uncertain factors including ocean current disturbance, dynamic ...The trajectory tracking control problem is addressed for autonomous underwater vehicle(AUV) in marine environ?ment, with presence of the influence of the uncertain factors including ocean current disturbance, dynamic modeling uncertainty, and thrust model errors. To improve the trajectory tracking accuracy of AUV, an adaptive backstepping terminal sliding mode control based on recurrent neural networks(RNN) is proposed. Firstly, considering the inaccu?rate of thrust model of thruster, a Taylor’s polynomial is used to obtain the thrust model errors. And then, the dynamic modeling uncertainty and thrust model errors are combined into the system model uncertainty(SMU) of AUV; through the RNN, the SMU and ocean current disturbance are classified, approximated online. Finally, the weights of RNN and other control parameters are adjusted online based on the backstepping terminal sliding mode controller. In addition, a chattering?reduction method is proposed based on sigmoid function. In chattering?reduction method, the sigmoid function is used to realize the continuity of the sliding mode switching function, and the sliding mode switching gain is adjusted online based on the exponential form of the sliding mode function. Based on the Lyapu?nov theory and Barbalat’s lemma, it is theoretically proved that the AUV trajectory tracking error can quickly converge to zero in the finite time. This research proposes a trajectory tracking control method of AUV, which can e ectively achieve high?precision trajectory tracking control of AUV under the influence of the uncertain factors. The feasibility and e ectiveness of the proposed method is demonstrated with trajectory tracking simulations and pool?experi?ments of AUV.展开更多
Static load tests are an important means of supervising and detecting a crane＇s lift capacity. Due to space restrictions, however, there are difficulties and potential danger when testing large bridge cranes. To solv...Static load tests are an important means of supervising and detecting a crane＇s lift capacity. Due to space restrictions, however, there are difficulties and potential danger when testing large bridge cranes. To solve the loading problems of large-tonnage cranes during testing, an equivalency test is proposed based on the similarity theory and BP neural networks. The maximum stress and displacement of a large bridge crane is tested in small loads, combined with the training neural network of a similar structure crane through stress and displacement data which is collected by a physics simulation progressively loaded to a static load test load within the material scope of work. The maximum stress and displacement of a crane under a static load test load can be predicted through the relationship of stress, displacement, and load. By measuring the stress and displacement of small tonnage weights, the stress and displacement of large loads can be predicted, such as the maximum load capacity, which is 1.25 times the rated capacity. Experimental study shows that the load reduction test method can reflect the lift capacity of large bridge cranes. The load shedding predictive analysis for Sanxia 1200 t bridge crane test data indicates that when the load is 1.25 times the rated lifting capacity, the predicted displacement and actual displacement error is zero. The method solves the problem that lifting capacities are difficult to obtain and testing accidents are easily possible when 1.25 times related weight loads are tested for large tonnage cranes.展开更多
The flash points of organic compounds were estimated using a hybrid method that includes a simple group contribution method (GCM) implemented in an artificial neural network (ANN) with particle swarm optimization (PSO...The flash points of organic compounds were estimated using a hybrid method that includes a simple group contribution method (GCM) implemented in an artificial neural network (ANN) with particle swarm optimization (PSO). Different topologies of a multilayer neural network were studied and the optimum architecture was determined. Property data of 350 compounds were used for training the network. To discriminate different substances the molecular structures defined by the concept of the classical group contribution method were given as input variables. The capabilities of the network were tested with 155 substances not considered in the training step. The study shows that the proposed GCM+ANN+PSO method represent an excellent alternative for the estimation of flash points of organic compounds with acceptable accuracy (AARD = 1.8%; AAE = 6.2 K).展开更多
Feedforward multi layer neural networks have very strong mapping capability that is based on the non linearity of the activation function, however, the non linearity of the activation function can cause the multiple ...Feedforward multi layer neural networks have very strong mapping capability that is based on the non linearity of the activation function, however, the non linearity of the activation function can cause the multiple local minima on the learning error surfaces, which affect the learning rate and solving optimal weights. This paper proposes a learning method linearizing non linearity of the activation function and discusses its merits and demerits theoretically.展开更多
Ship collision on bridge is a dynamic process featured by high nonlinearity and instantaneity. Calculating ship-bridge collision force typically involves either the use of design-specification-stipulated equivalent st...Ship collision on bridge is a dynamic process featured by high nonlinearity and instantaneity. Calculating ship-bridge collision force typically involves either the use of design-specification-stipulated equivalent static load, or the use of finite element method (FEM) which is more time-consuming and requires supercomputing resources. In this paper, we proposed an alternative approach that combines FEM with artificial neural network (ANN). The radial basis function neural network (RBFNN) employed for calculating the impact force in consideration of ship-bridge collision mechanics. With ship velocity and mass as the input vectors and ship collision force as the output vector, the neural networks for different network parameters are trained by the learning samples obtained from finite element simulation results. The error analyses of the learning and testing samples show that the proposed RBFNN is accurate enough to calculate ship-bridge collision force. The input-output relationship obtained by the RBFNN is essentially consistent with the typical empirical formulae. Finally, a special toolbox is developed for calculation efficiency in application using MATLAB software.展开更多
A new analytical method using Back-Propagation (BP) artificial neural network and kinetic spectrophotometry for simultaneous determination of iron and magnesium in tap water, the Yellow River water and seawater is est...A new analytical method using Back-Propagation (BP) artificial neural network and kinetic spectrophotometry for simultaneous determination of iron and magnesium in tap water, the Yellow River water and seawater is established. By conditional experiments, the optimum analytical conditions and parameters are obtained. Levenberg-Marquart (L-M) algorithm is used for calculation in BP neural network. The topological structure of three-layer BP ANN network architecture is chosen as 15-16-2 (nodes). The initial value of gradient coefficient μ is fixed at 0.001 and the increase factor and reduction factor of μ take the default values of the system. The data are processed by computers with our own programs written in MATLAB 7.0. The relative standard deviation of the calculated results for iron and manganese is 2.30% and 2.67% respectively. The results of standard addition method show that for the tap water, the recoveries of iron and manganese are in the ranges of 98.0%-104.3% and 96.5%-104.5%, and the RSD is in the range of 0.23%-0.98%; for the Yellow River water (Lijin district of Shandong Province), the recoveries of iron and manganese are in the ranges of 96.0%-101.0% and 98.7%-104.2%, and the RSD is in the range of 0.13%-2.52%; for the seawater in Qingdao offshore, the recoveries of iron and manganese are in the ranges of 95.3%-104.8% and 95.3%-104.7%, and the RSD is in the range of 0.14%-2.66%. It is found that 21 common cations and anions do not interfere with the determination of iron and manganese under the optimum experimental conditions. This method exhibits good reproducibility and high accuracy in the determination of iron and manganese and can be used for the simultaneous determination of iron and manganese in tap water and natural water. By using the established ANN- catalytic spectrophotometric method, the iron and manganese concentrations of the surface seawater at 11 sites in Qingdao offshore are determined and the level distribution maps of iron and manganese are drawn.展开更多
In this paper, the method based on uniform design and neural network is proposed to model the complex system. In order to express the system characteristics all round, uniform design method is used to choose the model...In this paper, the method based on uniform design and neural network is proposed to model the complex system. In order to express the system characteristics all round, uniform design method is used to choose the modeling samples and obtain the overall information of the system;for the purpose of modeling the system or its characteristics, the artificial neural network is used to construct the model. Experiment indicates that this method can model the complex system effectively.展开更多
Topography can strongly affect ground motion,and studies of the quantification of hill surfaces’topographic effect are relatively rare.In this paper,a new quantitative seismic topographic effect prediction method bas...Topography can strongly affect ground motion,and studies of the quantification of hill surfaces’topographic effect are relatively rare.In this paper,a new quantitative seismic topographic effect prediction method based upon the BP neural network algorithm and three-dimensional finite element method(FEM)was developed.The FEM simulation results were compared with seismic records and the results show that the PGA and response spectra have a tendency to increase with increasing elevation,but the correlation between PGA amplification factors and slope is not obvious for low hills.New BP neural network models were established for the prediction of amplification factors of PGA and response spectra.Two kinds of input variables’combinations which are convenient to achieve are proposed in this paper for the prediction of amplification factors of PGA and response spectra,respectively.The absolute values of prediction errors can be mostly within 0.1 for PGA amplification factors,and they can be mostly within 0.2 for response spectra’s amplification factors.One input variables’combination can achieve better prediction performance while the other one has better expandability of the predictive region.Particularly,the BP models only employ one hidden layer with about a hundred nodes,which makes it efficient for training.展开更多
This paper presents an artificial neural network（ANN）-based response surface method that can be used to predict the failure probability of c-φslopes with spatially variable soil.In this method,the Latin hypercube s...This paper presents an artificial neural network（ANN）-based response surface method that can be used to predict the failure probability of c-φslopes with spatially variable soil.In this method,the Latin hypercube sampling technique is adopted to generate input datasets for establishing an ANN model;the random finite element method is then utilized to calculate the corresponding output datasets considering the spatial variability of soil properties;and finally,an ANN model is trained to construct the response surface of failure probability and obtain an approximate function that incorporates the relevant variables.The results of the illustrated example indicate that the proposed method provides credible and accurate estimations of failure probability.As a result,the obtained approximate function can be used as an alternative to the specific analysis process in c-φslope reliability analyses.展开更多
Based on the research of Particle Swarm Optimization (PSO) learning rate, two learning rates are changed linearly with velocity-formula evolving in order to adjust the proportion of social part and cognitional part; t...Based on the research of Particle Swarm Optimization (PSO) learning rate, two learning rates are changed linearly with velocity-formula evolving in order to adjust the proportion of social part and cognitional part; then the methods are applied to BP neural network training, the convergence rate is heavily accelerated and locally optional solution is avoided. According to actual data of two levels compound-box in vibration lab, signals are analyzed and their characteristic values are abstracted. By applying the trained BP neural networks to compound-box fault diagnosis, it is indicated that the methods are sound effective.展开更多
Online gradient methods are widely used for training the weight of neural networks and for other engineering computations. In certain cases, the resulting weight may become very large, causing difficulties in the impl...Online gradient methods are widely used for training the weight of neural networks and for other engineering computations. In certain cases, the resulting weight may become very large, causing difficulties in the implementation of the network by electronic circuits. In this paper we introduce a punishing term into the error function of the training procedure to prevent this situation. The corresponding convergence of the iterative training procedure and the boundedness of the weight sequence are proved. A supporting numerical example is also provided.展开更多
Scientific forecasting water yield of mine is of great significance to the safety production of mine and the colligated using of water resources. The paper established the forecasting model for water yield of mine, co...Scientific forecasting water yield of mine is of great significance to the safety production of mine and the colligated using of water resources. The paper established the forecasting model for water yield of mine, combining neural network with the partial least square method. Dealt with independent variables by the partial least square method, it can not only solve the relationship between independent variables but also reduce the input dimensions in neural network model, and then use the neural network which can solve the non-linear problem better. The result of an example shows that the prediction has higher precision in forecasting and fitting.展开更多
In this paper, a gradient method with momentum for sigma-pi-sigma neural networks (SPSNN) is considered in order to accelerate the convergence of the learning procedure for the network weights. The momentum coefficien...In this paper, a gradient method with momentum for sigma-pi-sigma neural networks (SPSNN) is considered in order to accelerate the convergence of the learning procedure for the network weights. The momentum coefficient is chosen in an adaptive manner, and the corresponding weak convergence and strong convergence results are proved.展开更多
BP neural networks is used to mid-term earthquake prediction in this paper. Some usual prediction parameters of seismology are used as the import units of neural networks. And the export units of neural networks is ca...BP neural networks is used to mid-term earthquake prediction in this paper. Some usual prediction parameters of seismology are used as the import units of neural networks. And the export units of neural networks is called as the character parameter W_0 describing enhancement of seismicity. We applied this method to space scanning of North China. The result shows that the mid-term anomalous zone of W_0-value usually appeared obviously around the future epicenter 1~3 years before earthquake. It is effective to mid-term prediction.展开更多
This study introduces an innovative“Big Model”strategy to enhance Bridge Structural Health Monitoring(SHM)using a Convolutional Neural Network(CNN),time-frequency analysis,and fine element analysis.Leveraging ensemb...This study introduces an innovative“Big Model”strategy to enhance Bridge Structural Health Monitoring(SHM)using a Convolutional Neural Network(CNN),time-frequency analysis,and fine element analysis.Leveraging ensemble methods,collaborative learning,and distributed computing,the approach effectively manages the complexity and scale of large-scale bridge data.The CNN employs transfer learning,fine-tuning,and continuous monitoring to optimize models for adaptive and accurate structural health assessments,focusing on extracting meaningful features through time-frequency analysis.By integrating Finite Element Analysis,time-frequency analysis,and CNNs,the strategy provides a comprehensive understanding of bridge health.Utilizing diverse sensor data,sophisticated feature extraction,and advanced CNN architecture,the model is optimized through rigorous preprocessing and hyperparameter tuning.This approach significantly enhances the ability to make accurate predictions,monitor structural health,and support proactive maintenance practices,thereby ensuring the safety and longevity of critical infrastructure.展开更多
With more and more researches about improving BP algorithm, there are more improvement methods. The paper researches two improvement algorithms based on quasi-Newton method, DFP algorithm and L-BFGS algorithm. After f...With more and more researches about improving BP algorithm, there are more improvement methods. The paper researches two improvement algorithms based on quasi-Newton method, DFP algorithm and L-BFGS algorithm. After fully analyzing the features of quasi- Newton methods, the paper improves BP neural network algorithm. And the adjustment is made for the problems in the improvement process. The paper makes empirical analysis and proves the effectiveness of BP neural network algorithm based on quasi-Newton method. The improved algorithms are compared with the traditional BP algorithm, which indicates that the imoroved BP algorithm is better.展开更多
基金financially supported by the National Natural Science Foundation of China(Grant Nos.52271317 and 52071149)the Fundamental Research Funds for the Central Universities(HUST:2019kfy XJJS007)。
文摘To address the problems of low accuracy by the CONWEP model and poor efficiency by the Coupled Eulerian-Lagrangian(CEL)method in predicting close-range air blast loads of cylindrical charges,a neural network-based simulation(NNS)method with higher accuracy and better efficiency was proposed.The NNS method consisted of three main steps.First,the parameters of blast loads,including the peak pressures and impulses of cylindrical charges with different aspect ratios(L/D)at different stand-off distances and incident angles were obtained by two-dimensional numerical simulations.Subsequently,incident shape factors of cylindrical charges with arbitrary aspect ratios were predicted by a neural network.Finally,reflected shape factors were derived and implemented into the subroutine of the ABAQUS code to modify the CONWEP model,including modifications of impulse and overpressure.The reliability of the proposed NNS method was verified by related experimental results.Remarkable accuracy improvement was acquired by the proposed NNS method compared with the unmodified CONWEP model.Moreover,huge efficiency superiority was obtained by the proposed NNS method compared with the CEL method.The proposed NNS method showed good accuracy when the scaled distance was greater than 0.2 m/kg^(1/3).It should be noted that there is no need to generate a new dataset again since the blast loads satisfy the similarity law,and the proposed NNS method can be directly used to simulate the blast loads generated by different cylindrical charges.The proposed NNS method with high efficiency and accuracy can be used as an effective method to analyze the dynamic response of structures under blast loads,and it has significant application prospects in designing protective structures.
文摘The topology structure of the artificial neural network is an intelligent control model,which is used for the intelligent vehicle control system and household sweeping robot.When setting the intelligent control system,the connection point of each network is regarded as a neuron in the nervous system,and each connection point has input and output functions.Only when the input of nodes reaches a certain threshold can the output function of nodes be stimulated.Using the networking mode of the artificial neural network model,the mobile node can output in multiple directions.If the input direction of a certain path is the same as that of other nodes,it can choose to avoid and choose another path.The weighted value of each path between nodes is different,which means that the influence of the front node on the current node varies.The control method based on the artificial neural network model can be applied to vehicle control,household sweeping robots,and other fields,and a relatively optimized scheme can be obtained from the aspect of time and energy consumption.
基金supported by the National Natural Science Foundation of China(Grant No.U2241263)the fellowship of China Postdoctoral Science Foundation(Grant No.2024M750310).
文摘Numerical simulation plays an important role in the dynamic analysis of multibody system.With the rapid development of computer science,the numerical solution technology has been further developed.Recently,data-driven method has become a very popular computing method.However,due to lack of necessary mechanism information of the traditional pure data-driven methods based on neural network,its numerical accuracy cannot be guaranteed for strong nonlinear system.Therefore,this work proposes a mechanism-data hybrid-driven strategy for solving nonlinear multibody system based on physics-informed neural network to overcome the limitation of traditional data-driven methods.The strategy proposed in this paper introduces scaling coefficients to introduce the dynamic model of multibody system into neural network,ensuring that the training results of neural network conform to the mechanics principle of the system,thereby ensuring the good reliability of the data-driven method.Finally,the stability,generalization ability and numerical accuracy of the proposed method are discussed and analyzed using three typical multibody systems,and the constrained default situations can be controlled within the range of 10^(-2)-10^(-4).
基金supported by the National Natural Science Foundation of China(Grant Nos.12474453,12174085,and 12404530).
文摘Physics-informed neural networks(PINNs)have shown considerable promise for performing numerical simulations in fluid mechanics.They provide mesh-free,end-to-end approaches by embedding physical laws into their loss functions.However,when addressing complex flow problems,PINNs still face some challenges such as activation saturation and vanishing gradients in deep network training,leading to slow convergence and insufficient prediction accuracy.We present physics-informed neural networks incorporating lattice Boltzmann method optimized by tanh robust weight initialization(T-PINN-LBM)to address these challenges.This approach fuses the mesoscopic lattice Boltzmann model with the automatic differentiation framework of PINNs.It also implements a tanh robust weight initialization method derived from fixed point analysis.This model effectively mitigates activation and gradient decay in deep networks,improving convergence speed and data efficiency in multiscale flow simulations.We validate the effectiveness of the model on the classical arithmetic example of lid-driven cavity flow.Compared to the traditional Xavier initialized PINN and PINN-LBM,T-PINNLBM reduces the mean absolute error(MAE)by one order of magnitude at the same network depth and maintains stable convergence in deeper networks.The results demonstrate that this model can accurately capture complex flow structures without prior data,providing a new feasible pathway for data-free driven fluid simulation.
基金Basic Research Program of Ministry of Industry and Information Technology of China(Grant No.B2420133003)National Natural Science Foundation of China(Grant Nos.51779060,51679054)
文摘The trajectory tracking control problem is addressed for autonomous underwater vehicle(AUV) in marine environ?ment, with presence of the influence of the uncertain factors including ocean current disturbance, dynamic modeling uncertainty, and thrust model errors. To improve the trajectory tracking accuracy of AUV, an adaptive backstepping terminal sliding mode control based on recurrent neural networks(RNN) is proposed. Firstly, considering the inaccu?rate of thrust model of thruster, a Taylor’s polynomial is used to obtain the thrust model errors. And then, the dynamic modeling uncertainty and thrust model errors are combined into the system model uncertainty(SMU) of AUV; through the RNN, the SMU and ocean current disturbance are classified, approximated online. Finally, the weights of RNN and other control parameters are adjusted online based on the backstepping terminal sliding mode controller. In addition, a chattering?reduction method is proposed based on sigmoid function. In chattering?reduction method, the sigmoid function is used to realize the continuity of the sliding mode switching function, and the sliding mode switching gain is adjusted online based on the exponential form of the sliding mode function. Based on the Lyapu?nov theory and Barbalat’s lemma, it is theoretically proved that the AUV trajectory tracking error can quickly converge to zero in the finite time. This research proposes a trajectory tracking control method of AUV, which can e ectively achieve high?precision trajectory tracking control of AUV under the influence of the uncertain factors. The feasibility and e ectiveness of the proposed method is demonstrated with trajectory tracking simulations and pool?experi?ments of AUV.
基金Supported by National "Twelfth Five-Year" Plan for Science&Technology Support of China(Grant No.2011BAK06B05)National High-tech Research and Development Program of China(863 Program,Grant No.2013AA040203)Shanxi Scholarship Council of China(Grant No.2015-088)
文摘Static load tests are an important means of supervising and detecting a crane＇s lift capacity. Due to space restrictions, however, there are difficulties and potential danger when testing large bridge cranes. To solve the loading problems of large-tonnage cranes during testing, an equivalency test is proposed based on the similarity theory and BP neural networks. The maximum stress and displacement of a large bridge crane is tested in small loads, combined with the training neural network of a similar structure crane through stress and displacement data which is collected by a physics simulation progressively loaded to a static load test load within the material scope of work. The maximum stress and displacement of a crane under a static load test load can be predicted through the relationship of stress, displacement, and load. By measuring the stress and displacement of small tonnage weights, the stress and displacement of large loads can be predicted, such as the maximum load capacity, which is 1.25 times the rated capacity. Experimental study shows that the load reduction test method can reflect the lift capacity of large bridge cranes. The load shedding predictive analysis for Sanxia 1200 t bridge crane test data indicates that when the load is 1.25 times the rated lifting capacity, the predicted displacement and actual displacement error is zero. The method solves the problem that lifting capacities are difficult to obtain and testing accidents are easily possible when 1.25 times related weight loads are tested for large tonnage cranes.
文摘The flash points of organic compounds were estimated using a hybrid method that includes a simple group contribution method (GCM) implemented in an artificial neural network (ANN) with particle swarm optimization (PSO). Different topologies of a multilayer neural network were studied and the optimum architecture was determined. Property data of 350 compounds were used for training the network. To discriminate different substances the molecular structures defined by the concept of the classical group contribution method were given as input variables. The capabilities of the network were tested with 155 substances not considered in the training step. The study shows that the proposed GCM+ANN+PSO method represent an excellent alternative for the estimation of flash points of organic compounds with acceptable accuracy (AARD = 1.8%; AAE = 6.2 K).
文摘Feedforward multi layer neural networks have very strong mapping capability that is based on the non linearity of the activation function, however, the non linearity of the activation function can cause the multiple local minima on the learning error surfaces, which affect the learning rate and solving optimal weights. This paper proposes a learning method linearizing non linearity of the activation function and discusses its merits and demerits theoretically.
基金the National Natural Science Foundation of China (No. 50778131)the National key Technology R&D Pro-gram, Ministry of Science and Technology (No. 2006BAG04B01), China
文摘Ship collision on bridge is a dynamic process featured by high nonlinearity and instantaneity. Calculating ship-bridge collision force typically involves either the use of design-specification-stipulated equivalent static load, or the use of finite element method (FEM) which is more time-consuming and requires supercomputing resources. In this paper, we proposed an alternative approach that combines FEM with artificial neural network (ANN). The radial basis function neural network (RBFNN) employed for calculating the impact force in consideration of ship-bridge collision mechanics. With ship velocity and mass as the input vectors and ship collision force as the output vector, the neural networks for different network parameters are trained by the learning samples obtained from finite element simulation results. The error analyses of the learning and testing samples show that the proposed RBFNN is accurate enough to calculate ship-bridge collision force. The input-output relationship obtained by the RBFNN is essentially consistent with the typical empirical formulae. Finally, a special toolbox is developed for calculation efficiency in application using MATLAB software.
文摘A new analytical method using Back-Propagation (BP) artificial neural network and kinetic spectrophotometry for simultaneous determination of iron and magnesium in tap water, the Yellow River water and seawater is established. By conditional experiments, the optimum analytical conditions and parameters are obtained. Levenberg-Marquart (L-M) algorithm is used for calculation in BP neural network. The topological structure of three-layer BP ANN network architecture is chosen as 15-16-2 (nodes). The initial value of gradient coefficient μ is fixed at 0.001 and the increase factor and reduction factor of μ take the default values of the system. The data are processed by computers with our own programs written in MATLAB 7.0. The relative standard deviation of the calculated results for iron and manganese is 2.30% and 2.67% respectively. The results of standard addition method show that for the tap water, the recoveries of iron and manganese are in the ranges of 98.0%-104.3% and 96.5%-104.5%, and the RSD is in the range of 0.23%-0.98%; for the Yellow River water (Lijin district of Shandong Province), the recoveries of iron and manganese are in the ranges of 96.0%-101.0% and 98.7%-104.2%, and the RSD is in the range of 0.13%-2.52%; for the seawater in Qingdao offshore, the recoveries of iron and manganese are in the ranges of 95.3%-104.8% and 95.3%-104.7%, and the RSD is in the range of 0.14%-2.66%. It is found that 21 common cations and anions do not interfere with the determination of iron and manganese under the optimum experimental conditions. This method exhibits good reproducibility and high accuracy in the determination of iron and manganese and can be used for the simultaneous determination of iron and manganese in tap water and natural water. By using the established ANN- catalytic spectrophotometric method, the iron and manganese concentrations of the surface seawater at 11 sites in Qingdao offshore are determined and the level distribution maps of iron and manganese are drawn.
文摘In this paper, the method based on uniform design and neural network is proposed to model the complex system. In order to express the system characteristics all round, uniform design method is used to choose the modeling samples and obtain the overall information of the system;for the purpose of modeling the system or its characteristics, the artificial neural network is used to construct the model. Experiment indicates that this method can model the complex system effectively.
基金supported by the National Natural Science Foundation of China(No.51878625)the Collaboratory for the Study of Earthquake Predictability in China Seismic Experimental Site(No.2018YFE0109700)the General Scientific Research Foundation of Shandong Earthquake Agency(No.YB2208).
文摘Topography can strongly affect ground motion,and studies of the quantification of hill surfaces’topographic effect are relatively rare.In this paper,a new quantitative seismic topographic effect prediction method based upon the BP neural network algorithm and three-dimensional finite element method(FEM)was developed.The FEM simulation results were compared with seismic records and the results show that the PGA and response spectra have a tendency to increase with increasing elevation,but the correlation between PGA amplification factors and slope is not obvious for low hills.New BP neural network models were established for the prediction of amplification factors of PGA and response spectra.Two kinds of input variables’combinations which are convenient to achieve are proposed in this paper for the prediction of amplification factors of PGA and response spectra,respectively.The absolute values of prediction errors can be mostly within 0.1 for PGA amplification factors,and they can be mostly within 0.2 for response spectra’s amplification factors.One input variables’combination can achieve better prediction performance while the other one has better expandability of the predictive region.Particularly,the BP models only employ one hidden layer with about a hundred nodes,which makes it efficient for training.
基金financially supported by the National Natural Science Foundation of China(Grant No.51278217)
文摘This paper presents an artificial neural network（ANN）-based response surface method that can be used to predict the failure probability of c-φslopes with spatially variable soil.In this method,the Latin hypercube sampling technique is adopted to generate input datasets for establishing an ANN model;the random finite element method is then utilized to calculate the corresponding output datasets considering the spatial variability of soil properties;and finally,an ANN model is trained to construct the response surface of failure probability and obtain an approximate function that incorporates the relevant variables.The results of the illustrated example indicate that the proposed method provides credible and accurate estimations of failure probability.As a result,the obtained approximate function can be used as an alternative to the specific analysis process in c-φslope reliability analyses.
基金Supported by National Natural Science Foundation (No.50575214)
文摘Based on the research of Particle Swarm Optimization (PSO) learning rate, two learning rates are changed linearly with velocity-formula evolving in order to adjust the proportion of social part and cognitional part; then the methods are applied to BP neural network training, the convergence rate is heavily accelerated and locally optional solution is avoided. According to actual data of two levels compound-box in vibration lab, signals are analyzed and their characteristic values are abstracted. By applying the trained BP neural networks to compound-box fault diagnosis, it is indicated that the methods are sound effective.
文摘Online gradient methods are widely used for training the weight of neural networks and for other engineering computations. In certain cases, the resulting weight may become very large, causing difficulties in the implementation of the network by electronic circuits. In this paper we introduce a punishing term into the error function of the training procedure to prevent this situation. The corresponding convergence of the iterative training procedure and the boundedness of the weight sequence are proved. A supporting numerical example is also provided.
基金Supported by "863" Program of P. R. China(2002AA2Z4291)
文摘Scientific forecasting water yield of mine is of great significance to the safety production of mine and the colligated using of water resources. The paper established the forecasting model for water yield of mine, combining neural network with the partial least square method. Dealt with independent variables by the partial least square method, it can not only solve the relationship between independent variables but also reduce the input dimensions in neural network model, and then use the neural network which can solve the non-linear problem better. The result of an example shows that the prediction has higher precision in forecasting and fitting.
文摘In this paper, a gradient method with momentum for sigma-pi-sigma neural networks (SPSNN) is considered in order to accelerate the convergence of the learning procedure for the network weights. The momentum coefficient is chosen in an adaptive manner, and the corresponding weak convergence and strong convergence results are proved.
文摘BP neural networks is used to mid-term earthquake prediction in this paper. Some usual prediction parameters of seismology are used as the import units of neural networks. And the export units of neural networks is called as the character parameter W_0 describing enhancement of seismicity. We applied this method to space scanning of North China. The result shows that the mid-term anomalous zone of W_0-value usually appeared obviously around the future epicenter 1~3 years before earthquake. It is effective to mid-term prediction.
文摘This study introduces an innovative“Big Model”strategy to enhance Bridge Structural Health Monitoring(SHM)using a Convolutional Neural Network(CNN),time-frequency analysis,and fine element analysis.Leveraging ensemble methods,collaborative learning,and distributed computing,the approach effectively manages the complexity and scale of large-scale bridge data.The CNN employs transfer learning,fine-tuning,and continuous monitoring to optimize models for adaptive and accurate structural health assessments,focusing on extracting meaningful features through time-frequency analysis.By integrating Finite Element Analysis,time-frequency analysis,and CNNs,the strategy provides a comprehensive understanding of bridge health.Utilizing diverse sensor data,sophisticated feature extraction,and advanced CNN architecture,the model is optimized through rigorous preprocessing and hyperparameter tuning.This approach significantly enhances the ability to make accurate predictions,monitor structural health,and support proactive maintenance practices,thereby ensuring the safety and longevity of critical infrastructure.
文摘With more and more researches about improving BP algorithm, there are more improvement methods. The paper researches two improvement algorithms based on quasi-Newton method, DFP algorithm and L-BFGS algorithm. After fully analyzing the features of quasi- Newton methods, the paper improves BP neural network algorithm. And the adjustment is made for the problems in the improvement process. The paper makes empirical analysis and proves the effectiveness of BP neural network algorithm based on quasi-Newton method. The improved algorithms are compared with the traditional BP algorithm, which indicates that the imoroved BP algorithm is better.
