The constitutive models of shape memory alloys(SMAs)play an important role in facilitating the widespread application of such types of alloys in various engineering fields.However,to accurately describe the deformatio...The constitutive models of shape memory alloys(SMAs)play an important role in facilitating the widespread application of such types of alloys in various engineering fields.However,to accurately describe the deformation behaviors of SMAs,the concepts in classical plasticity are employed in the existing constitutive models,and a series of complex mathematical equations are involved.Such complexity brings inconvenience for the construction,implementation,and application of the constitutive models.To overcome these shortcomings,a data-driven constitutive model of SMAs is developed in this work based on the artificial neural network(ANN).In the proposed model,the components of the strain tensor in principal space,ambient temperature,and the maximum equivalent strain in the deformation history from the initial state to the current loading state are chosen as the input features,and the components of the stress tensor in principal space are set as the output.The proposed ANN-based constitutive model is implemented into the finite element program ABAQUS by deriving its consistent tangent modulus and writing a user-defined material subroutine.The stress-strain responses of SMA material under various loading paths and at different ambient temperatures are used to train the ANN model,which is generated from the existing constitutive model(numerical experiments).To validate the capability of the proposed model,the predicted stress-strain responses of SMA material,and the global and local responses of two typical SMA structures are compared with the corresponding numerical experiments.This work demonstrates a good potential to obtain the constitutive model of SMAs by pure data and avoid the need for vast stores of knowledge for the construction of constitutive models.展开更多
Complicated loads encountered by floating offshore wind turbines(FOWTs)in real sea conditions are crucial for future optimization of design,but obtaining data on them directly poses a challenge.To address this issue,w...Complicated loads encountered by floating offshore wind turbines(FOWTs)in real sea conditions are crucial for future optimization of design,but obtaining data on them directly poses a challenge.To address this issue,we applied machine learning techniques to obtain hydrodynamic and aerodynamic loads of FOWTs by measuring platform motion responses and wave-elevation sequences.First,a computational fluid dynamics(CFD)simulation model of the floating platform was established based on the dynamic fluid body interaction technique and overset grid technology.Then,a long short-term memory(LSTM)neural network model was constructed and trained to learn the nonlinear relationship between the waves,platform-motion inputs,and hydrodynamic-load outputs.The optimal model was determined after analyzing the sensitivity of parameters such as sample characteristics,network layers,and neuron numbers.Subsequently,the effectiveness of the hydrodynamic load model was validated under different simulation conditions,and the aerodynamic load calculation was completed based on the D'Alembert principle.Finally,we built a hybrid-scale FOWT model,based on the software in the loop strategy,in which the wind turbine was replaced by an actuation system.Model tests were carried out in a wave basin and the results demonstrated that the root mean square errors of the hydrodynamic and aerodynamic load measurements were 4.20%and 10.68%,respectively.展开更多
Developing efficient neural network(NN)computing systems is crucial in the era of artificial intelligence(AI).Traditional von Neumann architectures have both the issues of"memory wall"and"power wall&quo...Developing efficient neural network(NN)computing systems is crucial in the era of artificial intelligence(AI).Traditional von Neumann architectures have both the issues of"memory wall"and"power wall",limiting the data transfer between memory and processing units[1,2].Compute-in-memory(CIM)technologies,particularly analogue CIM with memristor crossbars,are promising because of their high energy efficiency,computational parallelism,and integration density for NN computations[3].In practical applications,analogue CIM excels in tasks like speech recognition and image classification,revealing its unique advantages.For instance,it efficiently processes vast amounts of audio data in speech recognition,achieving high accuracy with minimal power consumption.In image classification,the high parallelism of analogue CIM significantly speeds up feature extraction and reduces processing time.With the boosting development of AI applications,the demands for computational accuracy and task complexity are rising continually.However,analogue CIM systems are limited in handling complex regression tasks with needs of precise floating-point(FP)calculations.They are primarily suited for the classification tasks with low data precision and a limited dynamic range[4].展开更多
Ionosphere delay is one of the main sources of noise affecting global navigation satellite systems, operation of radio detection and ranging systems and very-long-baseline-interferometry. One of the most important and...Ionosphere delay is one of the main sources of noise affecting global navigation satellite systems, operation of radio detection and ranging systems and very-long-baseline-interferometry. One of the most important and common methods to reduce this phase delay is to establish accurate nowcasting and forecasting ionospheric total electron content models. For forecasting models, compared to mid-to-high latitudes, at low latitudes, an active ionosphere leads to extreme differences between long-term prediction models and the actual state of the ionosphere. To solve the problem of low accuracy for long-term prediction models at low latitudes, this article provides a low-latitude, long-term ionospheric prediction model based on a multi-input-multi-output, long-short-term memory neural network. To verify the feasibility of the model, we first made predictions of the vertical total electron content data 24 and 48 hours in advance for each day of July 2020 and then compared both the predictions corresponding to a given day, for all days. Furthermore, in the model modification part, we selected historical data from June 2020 for the validation set, determined a large offset from the results that were predicted to be active, and used the ratio of the mean absolute error of the detected results to that of the predicted results as a correction coefficient to modify our multi-input-multi-output long short-term memory model. The average root mean square error of the 24-hour-advance predictions of our modified model was 4.4 TECU, which was lower and better than5.1 TECU of the multi-input-multi-output, long short-term memory model and 5.9 TECU of the IRI-2016 model.展开更多
In this paper the globally asymptotic stability of more general two-layer nonlinear feedback associative memory neural networks with time delays is examined. The sufficient conditions of existence, uniqueness and glob...In this paper the globally asymptotic stability of more general two-layer nonlinear feedback associative memory neural networks with time delays is examined. The sufficient conditions of existence, uniqueness and globally asymptotic stability of the equilibrum position are given. Finally, two interesting examples to illustrate the theory are given.展开更多
In this paper, the global exponential stability of an equilibrium position for general bidirectional associative memory neural networks are studied. The sufficient conditions of existence and uniqueness of the equilib...In this paper, the global exponential stability of an equilibrium position for general bidirectional associative memory neural networks are studied. The sufficient conditions of existence and uniqueness of the equilibrium position are given. The method of energy function is examined. Two examples are given to illustrate the theory.展开更多
To explore new operational forecasting methods of waves,a forecasting model for wave heights at three stations in the Bohai Sea has been developed.This model is based on long short-term memory(LSTM)neural network with...To explore new operational forecasting methods of waves,a forecasting model for wave heights at three stations in the Bohai Sea has been developed.This model is based on long short-term memory(LSTM)neural network with sea surface wind and wave heights as training samples.The prediction performance of the model is evaluated,and the error analysis shows that when using the same set of numerically predicted sea surface wind as input,the prediction error produced by the proposed LSTM model at Sta.N01 is 20%,18%and 23%lower than the conventional numerical wave models in terms of the total root mean square error(RMSE),scatter index(SI)and mean absolute error(MAE),respectively.Particularly,for significant wave height in the range of 3–5 m,the prediction accuracy of the LSTM model is improved the most remarkably,with RMSE,SI and MAE all decreasing by 24%.It is also evident that the numbers of hidden neurons,the numbers of buoys used and the time length of training samples all have impact on the prediction accuracy.However,the prediction does not necessary improve with the increase of number of hidden neurons or number of buoys used.The experiment trained by data with the longest time length is found to perform the best overall compared to other experiments with a shorter time length for training.Overall,long short-term memory neural network was proved to be a very promising method for future development and applications in wave forecasting.展开更多
Real-time prediction and precise control of sinter quality are pivotal for energy saving,cost reduction,quality improvement and efficiency enhancement in the ironmaking process.To advance,the accuracy and comprehensiv...Real-time prediction and precise control of sinter quality are pivotal for energy saving,cost reduction,quality improvement and efficiency enhancement in the ironmaking process.To advance,the accuracy and comprehensiveness of sinter quality prediction,an intelligent flare monitoring system for sintering machine tails that combines hybrid neural networks integrating convolutional neural network with long short-term memory(CNN-LSTM)networks was proposed.The system utilized a high-temperature thermal imager for image acquisition at the sintering machine tail and employed a zone-triggered method to accurately capture dynamic feature images under challenging conditions of high-temperature,high dust,and occlusion.The feature images were then segmented through a triple-iteration multi-thresholding approach based on the maximum between-class variance method to minimize detail loss during the segmentation process.Leveraging the advantages of CNN and LSTM networks in capturing temporal and spatial information,a comprehensive model for sinter quality prediction was constructed,with inputs including the proportion of combustion layer,porosity rate,temperature distribution,and image features obtained from the convolutional neural network,and outputs comprising quality indicators such as underburning index,uniformity index,and FeO content of the sinter.The accuracy is notably increased,achieving a 95.8%hit rate within an error margin of±1.0.After the system is applied,the average qualified rate of FeO content increases from 87.24%to 89.99%,representing an improvement of 2.75%.The average monthly solid fuel consumption is reduced from 49.75 to 46.44 kg/t,leading to a 6.65%reduction and underscoring significant energy saving and cost reduction effects.展开更多
Traditional recurrent neural networks are composed of capacitors, inductors, resistors, and operational amplifiers.Memristive neural networks are constructed by replacing resistors with memristors. This paper focuses ...Traditional recurrent neural networks are composed of capacitors, inductors, resistors, and operational amplifiers.Memristive neural networks are constructed by replacing resistors with memristors. This paper focuses on the memory analysis,i.e. the initial value computation, of memristors. Firstly, we present the memory analysis for a single memristor based on memristors’ mathematical models with linear and nonlinear drift.Secondly, we present the memory analysis for two memristors in series and parallel. Thirdly, we point out the difference between traditional neural networks and those that are memristive. Based on the current and voltage relationship of memristors, we use mathematical analysis and SPICE simulations to demonstrate the validity of our methods.展开更多
A novel learning method for multi-valued associative memory network is introduced, which is based on Hebb rule, but utilizes more information. According to the current probe vector, the connection weights matrix could...A novel learning method for multi-valued associative memory network is introduced, which is based on Hebb rule, but utilizes more information. According to the current probe vector, the connection weights matrix could be chosen dynamically. Double-valued and multi-valued associative memory are all realized in our simulation experiment. The experimental results show that the method could enhance the associative success rate.展开更多
In this paper, a novel design procedure is proposed for synthesizing high-capacity auto-associative memories based on complex-valued neural networks with real-imaginary-type activation functions and constant delays. S...In this paper, a novel design procedure is proposed for synthesizing high-capacity auto-associative memories based on complex-valued neural networks with real-imaginary-type activation functions and constant delays. Stability criteria dependent on external inputs of neural networks are derived. The designed networks can retrieve the stored patterns by external inputs rather than initial conditions. The derivation can memorize the desired patterns with lower-dimensional neural networks than real-valued neural networks, and eliminate spurious equilibria of complex-valued neural networks. One numerical example is provided to show the effectiveness and superiority of the presented results.展开更多
Based on current research on applications of chaotic neuron network for information processing, the stability and convergence of chaotic neuron network are proved from the viewpoint of energy function. Moreover, a new...Based on current research on applications of chaotic neuron network for information processing, the stability and convergence of chaotic neuron network are proved from the viewpoint of energy function. Moreover, a new auto-associative matrix is devised for artificial neural network composed of chaotic neurons, thus, an improved chaotic neuron network for associative memory is built up. Finally, the associative recalling process of the network is analyzed in detail and explanations of improvement are given.展开更多
With the rapid development of machine learning,the demand for high-efficient computing becomes more and more urgent.To break the bottleneck of the traditional Von Neumann architecture,computing-in-memory(CIM)has attra...With the rapid development of machine learning,the demand for high-efficient computing becomes more and more urgent.To break the bottleneck of the traditional Von Neumann architecture,computing-in-memory(CIM)has attracted increasing attention in recent years.In this work,to provide a feasible CIM solution for the large-scale neural networks(NN)requiring continuous weight updating in online training,a flash-based computing-in-memory with high endurance(10^(9) cycles)and ultrafast programming speed is investigated.On the one hand,the proposed programming scheme of channel hot electron injection(CHEI)and hot hole injection(HHI)demonstrate high linearity,symmetric potentiation,and a depression process,which help to improve the training speed and accuracy.On the other hand,the low-damage programming scheme and memory window(MW)optimizations can suppress cell degradation effectively with improved computing accuracy.Even after 109 cycles,the leakage current(I_(off))of cells remains sub-10pA,ensuring the large-scale computing ability of memory.Further characterizations are done on read disturb to demonstrate its robust reliabilities.By processing CIFAR-10 tasks,it is evident that~90%accuracy can be achieved after 109 cycles in both ResNet50 and VGG16 NN.Our results suggest that flash-based CIM has great potential to overcome the limitations of traditional Von Neumann architectures and enable high-performance NN online training,which pave the way for further development of artificial intelligence(AI)accelerators.展开更多
A Long Short-Term Memory(LSTM) Recurrent Neural Network(RNN) has driven tremendous improvements on an acoustic model based on Gaussian Mixture Model(GMM). However, these models based on a hybrid method require a force...A Long Short-Term Memory(LSTM) Recurrent Neural Network(RNN) has driven tremendous improvements on an acoustic model based on Gaussian Mixture Model(GMM). However, these models based on a hybrid method require a forced aligned Hidden Markov Model(HMM) state sequence obtained from the GMM-based acoustic model. Therefore, it requires a long computation time for training both the GMM-based acoustic model and a deep learning-based acoustic model. In order to solve this problem, an acoustic model using CTC algorithm is proposed. CTC algorithm does not require the GMM-based acoustic model because it does not use the forced aligned HMM state sequence. However, previous works on a LSTM RNN-based acoustic model using CTC used a small-scale training corpus. In this paper, the LSTM RNN-based acoustic model using CTC is trained on a large-scale training corpus and its performance is evaluated. The implemented acoustic model has a performance of 6.18% and 15.01% in terms of Word Error Rate(WER) for clean speech and noisy speech, respectively. This is similar to a performance of the acoustic model based on the hybrid method.展开更多
Aiming at the problem of insufficient consideration of the correlation between components in the prediction of the remaining life of mechanical equipment,the method of remaining life prediction that combines the self-...Aiming at the problem of insufficient consideration of the correlation between components in the prediction of the remaining life of mechanical equipment,the method of remaining life prediction that combines the self-attention mechanism with the long short-term memory neural network(LSTM-NN)is proposed,called Self-Attention-LSTM.First,the auto-encoder is used to obtain the component-level state information;second,the state information of each component is input into the self-attention mechanism to learn the correlation between components;then,the multi-component correlation matrix is added to the LSTM input gate,and the LSTM-NN is used for life prediction.Finally,combined with the commercial modular aero-propulsion system simulation data set(C-MAPSS),the experiment was carried out and compared with the existing methods.Research results show that the proposed method can achieve better prediction accuracy and verify the feasibility of the method.展开更多
Accurate forecasting of electricity spot prices is crucial for market participants in formulating bidding strategies.However,the extreme volatility of electricity spot prices,influenced by various factors,poses signif...Accurate forecasting of electricity spot prices is crucial for market participants in formulating bidding strategies.However,the extreme volatility of electricity spot prices,influenced by various factors,poses significant challenges for forecasting.To address the data uncertainty of electricity prices and effectively mitigate gradient issues,overfitting,and computational challenges associated with using a single model during forecasting,this paper proposes a framework for forecasting spot market electricity prices by integrating wavelet packet decomposition(WPD)with a hybrid deep neural network.By ensuring accurate data decomposition,the WPD algorithm aids in detecting fluctuating patterns and isolating random noise.The hybrid model integrates temporal convolutional networks(TCN)and long short-term memory(LSTM)networks to enhance feature extraction and improve forecasting performance.Compared to other techniques,it significantly reduces average errors,decreasing mean absolute error(MAE)by 27.3%,root mean square error(RMSE)by 66.9%,and mean absolute percentage error(MAPE)by 22.8%.This framework effectively captures the intricate fluctuations present in the time series,resulting in more accurate and reliable predictions.展开更多
This paper proposes a concurrent neural network model to mitigate non-linear distortion in power amplifiers using a basis function generation approach.The model is designed using polynomial expansion and comprises a f...This paper proposes a concurrent neural network model to mitigate non-linear distortion in power amplifiers using a basis function generation approach.The model is designed using polynomial expansion and comprises a feedforward neural network(FNN)and a convolutional neural network(CNN).The proposed model takes the basic elements that form the bases as input,defined by the generalized memory polynomial(GMP)and dynamic deviation reduction(DDR)models.The FNN generates the basis function and its output represents the basis values,while the CNN generates weights for the corresponding bases.Through the concurrent training of FNN and CNN,the hidden layer coefficients are updated,and the complex multiplication of their outputs yields the trained in-phase/quadrature(I/Q)signals.The proposed model was trained and tested using 300 MHz and 400 MHz broadband data in an orthogonal frequency division multiplexing(OFDM)communication system.The results show that the model achieves an adjacent channel power ratio(ACPR)of less than-48 d B within a 100 MHz integral bandwidth for both the training and test datasets.展开更多
This paper studies the neural networks by means of neural functions.The memoryfunction of neural networks is investigated and its mathematical model is given.The model issynthesized by a piecewise-linear resistive net...This paper studies the neural networks by means of neural functions.The memoryfunction of neural networks is investigated and its mathematical model is given.The model issynthesized by a piecewise-linear resistive network which exhibits many properties of artificialneural network such as parallelism,real-time processing capability,distribution,adaptation.Inaddition,all parameters of the network are expressed analytically by the patterns and featureswhich are memorized in the network.展开更多
Associative memory, one of the major cognitive functions in the hippocampal CA3 region, includes auto-associative memory and hetero-associative memory. Many previous studies have shown that Alzheimer＇s disease （AD）...Associative memory, one of the major cognitive functions in the hippocampal CA3 region, includes auto-associative memory and hetero-associative memory. Many previous studies have shown that Alzheimer＇s disease （AD） can lead to loss of functional synapses in the central nervous system, and associative memory functions in patients with AD are often impaired, but few studies have addressed the effect of AD on hetero-associative memory in the hippocampal CA3 region. In this study, based on a simplified anatomical structure and synaptic connections in the hippocampal CA3 region, a three-layered Hopfield-like neural network model of hippocampal CA3 was proposed and then used to simulate associative memory functions in three circumstances： normal, synaptic deletion and synaptic compensation, according to Ruppin＇s synaptic deletion and compensation theory. The influences of AD on hetero-associative memory were further analyzed. The simulated results showed that the established three-layered Hopfield-like neural network model of hippocampal CA3 has both auto-associative and hetero-associative memory functions. With increasing synaptic deletion level, both associative memory functions were gradually impaired and the mean firing rates of the neurons within the network model were decreased. With gradual increasing synaptic compensation, the associative memory functions of the network were improved and the mean firing rates were increased. The simulated results suggest that the Hopfield-like neural network model can effectively simulate both associative memory functions of the hippocampal CA3 region. Synaptic deletion affects both auto-associative and hetero-associative memory functions in the hippocampal CA3 region, and can also result in memory dysfunction. To some extent, synaptic compensation measures can offset two kinds of associative memory dysfunction caused by synaptic deletion in the hippocampal CA3 area.展开更多
Without assuming the smoothness,monotonicity and boundedness of the activation functions, some novel criteria on the existence and global exponential stability of equilibrium point for delayed bidirectional associativ...Without assuming the smoothness,monotonicity and boundedness of the activation functions, some novel criteria on the existence and global exponential stability of equilibrium point for delayed bidirectional associative memory (BAM) neural networks are established by applying the Liapunov functional methods and matrix_algebraic techniques. It is shown that the new conditions presented in terms of a nonsingular M matrix described by the networks parameters,the connection matrix and the Lipschitz constant of the activation functions,are not only simple and practical,but also easier to check and less conservative than those imposed by similar results in recent literature.展开更多
基金supported by the National Natural Science Foundation of China(NSFC)(Grant No.12322203).
文摘The constitutive models of shape memory alloys(SMAs)play an important role in facilitating the widespread application of such types of alloys in various engineering fields.However,to accurately describe the deformation behaviors of SMAs,the concepts in classical plasticity are employed in the existing constitutive models,and a series of complex mathematical equations are involved.Such complexity brings inconvenience for the construction,implementation,and application of the constitutive models.To overcome these shortcomings,a data-driven constitutive model of SMAs is developed in this work based on the artificial neural network(ANN).In the proposed model,the components of the strain tensor in principal space,ambient temperature,and the maximum equivalent strain in the deformation history from the initial state to the current loading state are chosen as the input features,and the components of the stress tensor in principal space are set as the output.The proposed ANN-based constitutive model is implemented into the finite element program ABAQUS by deriving its consistent tangent modulus and writing a user-defined material subroutine.The stress-strain responses of SMA material under various loading paths and at different ambient temperatures are used to train the ANN model,which is generated from the existing constitutive model(numerical experiments).To validate the capability of the proposed model,the predicted stress-strain responses of SMA material,and the global and local responses of two typical SMA structures are compared with the corresponding numerical experiments.This work demonstrates a good potential to obtain the constitutive model of SMAs by pure data and avoid the need for vast stores of knowledge for the construction of constitutive models.
基金This work is supported by the National Key Research and Development Program of China(No.2023YFB4203000)the National Natural Science Foundation of China(No.U22A20178)
文摘Complicated loads encountered by floating offshore wind turbines(FOWTs)in real sea conditions are crucial for future optimization of design,but obtaining data on them directly poses a challenge.To address this issue,we applied machine learning techniques to obtain hydrodynamic and aerodynamic loads of FOWTs by measuring platform motion responses and wave-elevation sequences.First,a computational fluid dynamics(CFD)simulation model of the floating platform was established based on the dynamic fluid body interaction technique and overset grid technology.Then,a long short-term memory(LSTM)neural network model was constructed and trained to learn the nonlinear relationship between the waves,platform-motion inputs,and hydrodynamic-load outputs.The optimal model was determined after analyzing the sensitivity of parameters such as sample characteristics,network layers,and neuron numbers.Subsequently,the effectiveness of the hydrodynamic load model was validated under different simulation conditions,and the aerodynamic load calculation was completed based on the D'Alembert principle.Finally,we built a hybrid-scale FOWT model,based on the software in the loop strategy,in which the wind turbine was replaced by an actuation system.Model tests were carried out in a wave basin and the results demonstrated that the root mean square errors of the hydrodynamic and aerodynamic load measurements were 4.20%and 10.68%,respectively.
文摘Developing efficient neural network(NN)computing systems is crucial in the era of artificial intelligence(AI).Traditional von Neumann architectures have both the issues of"memory wall"and"power wall",limiting the data transfer between memory and processing units[1,2].Compute-in-memory(CIM)technologies,particularly analogue CIM with memristor crossbars,are promising because of their high energy efficiency,computational parallelism,and integration density for NN computations[3].In practical applications,analogue CIM excels in tasks like speech recognition and image classification,revealing its unique advantages.For instance,it efficiently processes vast amounts of audio data in speech recognition,achieving high accuracy with minimal power consumption.In image classification,the high parallelism of analogue CIM significantly speeds up feature extraction and reduces processing time.With the boosting development of AI applications,the demands for computational accuracy and task complexity are rising continually.However,analogue CIM systems are limited in handling complex regression tasks with needs of precise floating-point(FP)calculations.They are primarily suited for the classification tasks with low data precision and a limited dynamic range[4].
基金Project supported by the National Key Research and Development Program of China(Grant No.2016YFA0302101)the Initiative Program of State Key Laboratory of Precision Measurement Technology and Instrument。
文摘Ionosphere delay is one of the main sources of noise affecting global navigation satellite systems, operation of radio detection and ranging systems and very-long-baseline-interferometry. One of the most important and common methods to reduce this phase delay is to establish accurate nowcasting and forecasting ionospheric total electron content models. For forecasting models, compared to mid-to-high latitudes, at low latitudes, an active ionosphere leads to extreme differences between long-term prediction models and the actual state of the ionosphere. To solve the problem of low accuracy for long-term prediction models at low latitudes, this article provides a low-latitude, long-term ionospheric prediction model based on a multi-input-multi-output, long-short-term memory neural network. To verify the feasibility of the model, we first made predictions of the vertical total electron content data 24 and 48 hours in advance for each day of July 2020 and then compared both the predictions corresponding to a given day, for all days. Furthermore, in the model modification part, we selected historical data from June 2020 for the validation set, determined a large offset from the results that were predicted to be active, and used the ratio of the mean absolute error of the detected results to that of the predicted results as a correction coefficient to modify our multi-input-multi-output long short-term memory model. The average root mean square error of the 24-hour-advance predictions of our modified model was 4.4 TECU, which was lower and better than5.1 TECU of the multi-input-multi-output, long short-term memory model and 5.9 TECU of the IRI-2016 model.
文摘In this paper the globally asymptotic stability of more general two-layer nonlinear feedback associative memory neural networks with time delays is examined. The sufficient conditions of existence, uniqueness and globally asymptotic stability of the equilibrum position are given. Finally, two interesting examples to illustrate the theory are given.
基金Supported by the National Natural Science Foundation of China
文摘In this paper, the global exponential stability of an equilibrium position for general bidirectional associative memory neural networks are studied. The sufficient conditions of existence and uniqueness of the equilibrium position are given. The method of energy function is examined. Two examples are given to illustrate the theory.
基金The National Key R&D Program of China under contract No.2016YFC1402103
文摘To explore new operational forecasting methods of waves,a forecasting model for wave heights at three stations in the Bohai Sea has been developed.This model is based on long short-term memory(LSTM)neural network with sea surface wind and wave heights as training samples.The prediction performance of the model is evaluated,and the error analysis shows that when using the same set of numerically predicted sea surface wind as input,the prediction error produced by the proposed LSTM model at Sta.N01 is 20%,18%and 23%lower than the conventional numerical wave models in terms of the total root mean square error(RMSE),scatter index(SI)and mean absolute error(MAE),respectively.Particularly,for significant wave height in the range of 3–5 m,the prediction accuracy of the LSTM model is improved the most remarkably,with RMSE,SI and MAE all decreasing by 24%.It is also evident that the numbers of hidden neurons,the numbers of buoys used and the time length of training samples all have impact on the prediction accuracy.However,the prediction does not necessary improve with the increase of number of hidden neurons or number of buoys used.The experiment trained by data with the longest time length is found to perform the best overall compared to other experiments with a shorter time length for training.Overall,long short-term memory neural network was proved to be a very promising method for future development and applications in wave forecasting.
基金founded by the Open Project Program of Anhui Province Key Laboratory of Metallurgical Engineering and Resources Recycling(Anhui University of Technology)(No.SKF21-06)Research Fund for Young Teachers of Anhui University of Technology in 2020(No.QZ202001).
文摘Real-time prediction and precise control of sinter quality are pivotal for energy saving,cost reduction,quality improvement and efficiency enhancement in the ironmaking process.To advance,the accuracy and comprehensiveness of sinter quality prediction,an intelligent flare monitoring system for sintering machine tails that combines hybrid neural networks integrating convolutional neural network with long short-term memory(CNN-LSTM)networks was proposed.The system utilized a high-temperature thermal imager for image acquisition at the sintering machine tail and employed a zone-triggered method to accurately capture dynamic feature images under challenging conditions of high-temperature,high dust,and occlusion.The feature images were then segmented through a triple-iteration multi-thresholding approach based on the maximum between-class variance method to minimize detail loss during the segmentation process.Leveraging the advantages of CNN and LSTM networks in capturing temporal and spatial information,a comprehensive model for sinter quality prediction was constructed,with inputs including the proportion of combustion layer,porosity rate,temperature distribution,and image features obtained from the convolutional neural network,and outputs comprising quality indicators such as underburning index,uniformity index,and FeO content of the sinter.The accuracy is notably increased,achieving a 95.8%hit rate within an error margin of±1.0.After the system is applied,the average qualified rate of FeO content increases from 87.24%to 89.99%,representing an improvement of 2.75%.The average monthly solid fuel consumption is reduced from 49.75 to 46.44 kg/t,leading to a 6.65%reduction and underscoring significant energy saving and cost reduction effects.
基金supported by the National Natural Science Foundation of China(61876097,61673188,61761130081)the National Key Research and Development Program of China(2016YFB0800402)+1 种基金the Foundation for Innovative Research Groups of Hubei Province of China(2017CFA005)the Fundamental Research Funds for the Central Universities(2017KFXKJC002)
文摘Traditional recurrent neural networks are composed of capacitors, inductors, resistors, and operational amplifiers.Memristive neural networks are constructed by replacing resistors with memristors. This paper focuses on the memory analysis,i.e. the initial value computation, of memristors. Firstly, we present the memory analysis for a single memristor based on memristors’ mathematical models with linear and nonlinear drift.Secondly, we present the memory analysis for two memristors in series and parallel. Thirdly, we point out the difference between traditional neural networks and those that are memristive. Based on the current and voltage relationship of memristors, we use mathematical analysis and SPICE simulations to demonstrate the validity of our methods.
文摘A novel learning method for multi-valued associative memory network is introduced, which is based on Hebb rule, but utilizes more information. According to the current probe vector, the connection weights matrix could be chosen dynamically. Double-valued and multi-valued associative memory are all realized in our simulation experiment. The experimental results show that the method could enhance the associative success rate.
基金Project supported by the National Natural Science Foundation of China(Grant Nos.61503338,61573316,61374152,and 11302195)the Natural Science Foundation of Zhejiang Province,China(Grant No.LQ15F030005)
文摘In this paper, a novel design procedure is proposed for synthesizing high-capacity auto-associative memories based on complex-valued neural networks with real-imaginary-type activation functions and constant delays. Stability criteria dependent on external inputs of neural networks are derived. The designed networks can retrieve the stored patterns by external inputs rather than initial conditions. The derivation can memorize the desired patterns with lower-dimensional neural networks than real-valued neural networks, and eliminate spurious equilibria of complex-valued neural networks. One numerical example is provided to show the effectiveness and superiority of the presented results.
基金National Natural Science Foundation of P.R.China(No. 69735101)
文摘Based on current research on applications of chaotic neuron network for information processing, the stability and convergence of chaotic neuron network are proved from the viewpoint of energy function. Moreover, a new auto-associative matrix is devised for artificial neural network composed of chaotic neurons, thus, an improved chaotic neuron network for associative memory is built up. Finally, the associative recalling process of the network is analyzed in detail and explanations of improvement are given.
基金This work was supported by the National Natural Science Foundation of China(Nos.62034006,92264201,and 91964105)the Natural Science Foundation of Shandong Province(Nos.ZR2020JQ28 and ZR2020KF016)the Program of Qilu Young Scholars of Shandong University.
文摘With the rapid development of machine learning,the demand for high-efficient computing becomes more and more urgent.To break the bottleneck of the traditional Von Neumann architecture,computing-in-memory(CIM)has attracted increasing attention in recent years.In this work,to provide a feasible CIM solution for the large-scale neural networks(NN)requiring continuous weight updating in online training,a flash-based computing-in-memory with high endurance(10^(9) cycles)and ultrafast programming speed is investigated.On the one hand,the proposed programming scheme of channel hot electron injection(CHEI)and hot hole injection(HHI)demonstrate high linearity,symmetric potentiation,and a depression process,which help to improve the training speed and accuracy.On the other hand,the low-damage programming scheme and memory window(MW)optimizations can suppress cell degradation effectively with improved computing accuracy.Even after 109 cycles,the leakage current(I_(off))of cells remains sub-10pA,ensuring the large-scale computing ability of memory.Further characterizations are done on read disturb to demonstrate its robust reliabilities.By processing CIFAR-10 tasks,it is evident that~90%accuracy can be achieved after 109 cycles in both ResNet50 and VGG16 NN.Our results suggest that flash-based CIM has great potential to overcome the limitations of traditional Von Neumann architectures and enable high-performance NN online training,which pave the way for further development of artificial intelligence(AI)accelerators.
基金supported by the Ministry of Trade,Industry & Energy(MOTIE,Korea) under Industrial Technology Innovation Program (No.10063424,'development of distant speech recognition and multi-task dialog processing technologies for in-door conversational robots')
文摘A Long Short-Term Memory(LSTM) Recurrent Neural Network(RNN) has driven tremendous improvements on an acoustic model based on Gaussian Mixture Model(GMM). However, these models based on a hybrid method require a forced aligned Hidden Markov Model(HMM) state sequence obtained from the GMM-based acoustic model. Therefore, it requires a long computation time for training both the GMM-based acoustic model and a deep learning-based acoustic model. In order to solve this problem, an acoustic model using CTC algorithm is proposed. CTC algorithm does not require the GMM-based acoustic model because it does not use the forced aligned HMM state sequence. However, previous works on a LSTM RNN-based acoustic model using CTC used a small-scale training corpus. In this paper, the LSTM RNN-based acoustic model using CTC is trained on a large-scale training corpus and its performance is evaluated. The implemented acoustic model has a performance of 6.18% and 15.01% in terms of Word Error Rate(WER) for clean speech and noisy speech, respectively. This is similar to a performance of the acoustic model based on the hybrid method.
基金the National Natural Science Foundation of China(Nos.51875451 and 51834006)。
文摘Aiming at the problem of insufficient consideration of the correlation between components in the prediction of the remaining life of mechanical equipment,the method of remaining life prediction that combines the self-attention mechanism with the long short-term memory neural network(LSTM-NN)is proposed,called Self-Attention-LSTM.First,the auto-encoder is used to obtain the component-level state information;second,the state information of each component is input into the self-attention mechanism to learn the correlation between components;then,the multi-component correlation matrix is added to the LSTM input gate,and the LSTM-NN is used for life prediction.Finally,combined with the commercial modular aero-propulsion system simulation data set(C-MAPSS),the experiment was carried out and compared with the existing methods.Research results show that the proposed method can achieve better prediction accuracy and verify the feasibility of the method.
基金partially supported by projects funded by the National Key R&D Program of China(2022YFB2403000)the State Grid Corporation of China Science and Technology Project(522722230034).
文摘Accurate forecasting of electricity spot prices is crucial for market participants in formulating bidding strategies.However,the extreme volatility of electricity spot prices,influenced by various factors,poses significant challenges for forecasting.To address the data uncertainty of electricity prices and effectively mitigate gradient issues,overfitting,and computational challenges associated with using a single model during forecasting,this paper proposes a framework for forecasting spot market electricity prices by integrating wavelet packet decomposition(WPD)with a hybrid deep neural network.By ensuring accurate data decomposition,the WPD algorithm aids in detecting fluctuating patterns and isolating random noise.The hybrid model integrates temporal convolutional networks(TCN)and long short-term memory(LSTM)networks to enhance feature extraction and improve forecasting performance.Compared to other techniques,it significantly reduces average errors,decreasing mean absolute error(MAE)by 27.3%,root mean square error(RMSE)by 66.9%,and mean absolute percentage error(MAPE)by 22.8%.This framework effectively captures the intricate fluctuations present in the time series,resulting in more accurate and reliable predictions.
基金supported by ZTE Industry-University-Institute Cooperation Funds under Grant No.HC-CN-20220722010。
文摘This paper proposes a concurrent neural network model to mitigate non-linear distortion in power amplifiers using a basis function generation approach.The model is designed using polynomial expansion and comprises a feedforward neural network(FNN)and a convolutional neural network(CNN).The proposed model takes the basic elements that form the bases as input,defined by the generalized memory polynomial(GMP)and dynamic deviation reduction(DDR)models.The FNN generates the basis function and its output represents the basis values,while the CNN generates weights for the corresponding bases.Through the concurrent training of FNN and CNN,the hidden layer coefficients are updated,and the complex multiplication of their outputs yields the trained in-phase/quadrature(I/Q)signals.The proposed model was trained and tested using 300 MHz and 400 MHz broadband data in an orthogonal frequency division multiplexing(OFDM)communication system.The results show that the model achieves an adjacent channel power ratio(ACPR)of less than-48 d B within a 100 MHz integral bandwidth for both the training and test datasets.
文摘This paper studies the neural networks by means of neural functions.The memoryfunction of neural networks is investigated and its mathematical model is given.The model issynthesized by a piecewise-linear resistive network which exhibits many properties of artificialneural network such as parallelism,real-time processing capability,distribution,adaptation.Inaddition,all parameters of the network are expressed analytically by the patterns and featureswhich are memorized in the network.
基金the National Natural Science Foundation of China,No.30870649the Natural Science Foundation of Tianjin,No.08JCYBJC03300
文摘Associative memory, one of the major cognitive functions in the hippocampal CA3 region, includes auto-associative memory and hetero-associative memory. Many previous studies have shown that Alzheimer＇s disease （AD） can lead to loss of functional synapses in the central nervous system, and associative memory functions in patients with AD are often impaired, but few studies have addressed the effect of AD on hetero-associative memory in the hippocampal CA3 region. In this study, based on a simplified anatomical structure and synaptic connections in the hippocampal CA3 region, a three-layered Hopfield-like neural network model of hippocampal CA3 was proposed and then used to simulate associative memory functions in three circumstances： normal, synaptic deletion and synaptic compensation, according to Ruppin＇s synaptic deletion and compensation theory. The influences of AD on hetero-associative memory were further analyzed. The simulated results showed that the established three-layered Hopfield-like neural network model of hippocampal CA3 has both auto-associative and hetero-associative memory functions. With increasing synaptic deletion level, both associative memory functions were gradually impaired and the mean firing rates of the neurons within the network model were decreased. With gradual increasing synaptic compensation, the associative memory functions of the network were improved and the mean firing rates were increased. The simulated results suggest that the Hopfield-like neural network model can effectively simulate both associative memory functions of the hippocampal CA3 region. Synaptic deletion affects both auto-associative and hetero-associative memory functions in the hippocampal CA3 region, and can also result in memory dysfunction. To some extent, synaptic compensation measures can offset two kinds of associative memory dysfunction caused by synaptic deletion in the hippocampal CA3 area.
文摘Without assuming the smoothness,monotonicity and boundedness of the activation functions, some novel criteria on the existence and global exponential stability of equilibrium point for delayed bidirectional associative memory (BAM) neural networks are established by applying the Liapunov functional methods and matrix_algebraic techniques. It is shown that the new conditions presented in terms of a nonsingular M matrix described by the networks parameters,the connection matrix and the Lipschitz constant of the activation functions,are not only simple and practical,but also easier to check and less conservative than those imposed by similar results in recent literature.
