In this paper,the recurrent neural network structure of a bidirectional long shortterm memory network(Bi-LSTM)with special memory cells that store information is used to characterize the deep features of the variation...In this paper,the recurrent neural network structure of a bidirectional long shortterm memory network(Bi-LSTM)with special memory cells that store information is used to characterize the deep features of the variation pattern between logging and seismic data.A mapping relationship model between high-frequency logging data and low-frequency seismic data is established via nonlinear mapping.The seismic waveform is infinitely approximated using the logging curve in the low-frequency band to obtain a nonlinear mapping model of this scale,which then stepwise approach the logging curve in the high-frequency band.Finally,a seismic-inversion method of nonlinear mapping multilevel well–seismic matching based on the Bi-LSTM network is developed.The characteristic of this method is that by applying the multilevel well–seismic matching process,the seismic data are stepwise matched to the scale range that is consistent with the logging curve.Further,the matching operator at each level can be stably obtained to effectively overcome the problems that occur in the well–seismic matching process,such as the inconsistency in the scale of two types of data,accuracy in extracting the seismic wavelet of the well-side seismic traces,and multiplicity of solutions.Model test and practical application demonstrate that this method improves the vertical resolution of inversion results,and at the same time,the boundary and the lateral characteristics of the sand body are well maintained to improve the accuracy of thin-layer sand body prediction and achieve an improved practical application effect.展开更多
Nowadays,Internet has become an indispensable part of daily life and is used in many fields.Due to the large amount of Internet traffic,computers are subject to various security threats,which may cause serious economi...Nowadays,Internet has become an indispensable part of daily life and is used in many fields.Due to the large amount of Internet traffic,computers are subject to various security threats,which may cause serious economic losses and even endanger national security.It is hoped that an effective security method can systematically classify intrusion data in order to avoid leakage of important data or misuse of data.As machine learning technology matures,deep learning is widely used in various industries.Combining deep learning with network security and intrusion detection is the current trend.In this paper,the problem of data classification in intrusion detection system is studied.We propose an intrusion detection model based on stack bidirectional long short-term memory(LSTM),introduce stack bidirectional LSTM into the field of intrusion detection and apply it to the intrusion detection.In order to determine the appropriate parameters and structure of stack bidirectional LSTM network,we have carried out experiments on various network structures and parameters and analyzed the experimental results.The classic KDD Cup’1999 dataset was selected for experiments so that we can obtain convincing and comparable results.Experimental results derived from the KDD Cup’1999 dataset show that the network with three hidden layers containing 80 LSTM cells is superior to other algorithms in computational cost and detection performance due to stack bidirectional LSTM model’s ability to review time and correlate with connected records continuously.The experiment shows the effectiveness of stack bidirectional LSTM network in intrusion detection.展开更多
Accurate sag source location and precise sag type recognition are both essential to verifying the responsible party for the sag and taking countermeasures to improve power quality.In this paper,an attention-based inde...Accurate sag source location and precise sag type recognition are both essential to verifying the responsible party for the sag and taking countermeasures to improve power quality.In this paper,an attention-based independently recurrent neural network(IndRNN)for sag source location and sag type recognition in sparsely monitored power system is proposed.Specially,the given inputs are voltage waveforms collected by limited meters in sparsely monitored power system,and the desired outputs simultaneously contain the following information:the located lines where sag occurs;the corresponding sag types,including motor starting,transformer energizing and short circuit;and the fault phase for short circuit.In essence,the responsibility of the proposed method is to automatically establish a nonlinear function that relates the given inputs to the desired outputs with categorization labels as few as possible.A favorable feature of the proposed method is that it can be realized without system parameters or models.The proposed method is validated by IEEE 30-bus system and a real 134-bus system.Experimental results demonstrate that the accuracy of sag source location is higher than 99%for all lines,and the accuracy of sag type recognition is also higher than 99%for various sag sources including motor starting,transformer energizing and 7 different types of short circuits.Furthermore,a comparison among different monitor placements for the proposed method is conducted,which illustrates that the observability of power networks should be ensured to achieve satisfactory performance.展开更多
In the industrial fields, the mechanical equipment will inevitably wear out in the process of operation. With the accumulation of losses, the probability of equipment failure is increasing. Therefore, if the remaining...In the industrial fields, the mechanical equipment will inevitably wear out in the process of operation. With the accumulation of losses, the probability of equipment failure is increasing. Therefore, if the remaining useful life(RUL) of the equipment can be accurately predicted, the equipment can be maintained in time to avoid the downtime caused by equipment failure and greatly improve the production efficiency of enterprises. This paper aims to use independently recurrent neural network(IndRNN) to learn health degradation of turbofan engine and make accurate predictions of its RUL, which not only effectively solves the problem of gradient explosion and vanishing, but also increases the interpretability of neural networks. IndRNN can be used to process longer time series which matches the scene with high frequency sampling sensor in industrial practical applications. The results demonstrate that IndRNN for RUL estimation significantly outperforms traditional approaches, as well as convolutional neural network(CNN) and long short-term memory network(LSTM) for RUL estimation.展开更多
Memristor-based neuromorphic computing shows great potential for high-speed and high-throughput signal processing applications,such as electroencephalogram(EEG)signal processing.Nonetheless,the size of one-transistor ...Memristor-based neuromorphic computing shows great potential for high-speed and high-throughput signal processing applications,such as electroencephalogram(EEG)signal processing.Nonetheless,the size of one-transistor one-resistor(1T1R)memristor arrays is limited by the non-ideality of the devices,which prevents the hardware implementation of large and complex networks.In this work,we propose the depthwise separable convolution and bidirectional gate recurrent unit(DSC-BiGRU)network,a lightweight and highly robust hybrid neural network based on 1T1R arrays that enables efficient processing of EEG signals in the temporal,frequency and spatial domains by hybridizing DSC and BiGRU blocks.The network size is reduced and the network robustness is improved while ensuring the network classification accuracy.In the simulation,the measured non-idealities of the 1T1R array are brought into the network through statistical analysis.Compared with traditional convolutional networks,the network parameters are reduced by 95%and the network classification accuracy is improved by 21%at a 95%array yield rate and 5%tolerable error.This work demonstrates that lightweight and highly robust networks based on memristor arrays hold great promise for applications that rely on low consumption and high efficiency.展开更多
Relation classification is a crucial component in many Natural Language Processing(NLP) systems. In this paper, we propose a novel bidirectional recurrent neural network architecture(using Long Short-Term Memory,LSTM,...Relation classification is a crucial component in many Natural Language Processing(NLP) systems. In this paper, we propose a novel bidirectional recurrent neural network architecture(using Long Short-Term Memory,LSTM, cells) for relation classification, with an attention layer for organizing the context information on the word level and a tensor layer for detecting complex connections between two entities. The above two feature extraction operations are based on the LSTM networks and use their outputs. Our model allows end-to-end learning from the raw sentences in the dataset, without trimming or reconstructing them. Experiments on the SemEval-2010 Task 8dataset show that our model outperforms most state-of-the-art methods.展开更多
基金supported by the National Major Science and Technology Special Project(No.2016ZX05026-002).
文摘In this paper,the recurrent neural network structure of a bidirectional long shortterm memory network(Bi-LSTM)with special memory cells that store information is used to characterize the deep features of the variation pattern between logging and seismic data.A mapping relationship model between high-frequency logging data and low-frequency seismic data is established via nonlinear mapping.The seismic waveform is infinitely approximated using the logging curve in the low-frequency band to obtain a nonlinear mapping model of this scale,which then stepwise approach the logging curve in the high-frequency band.Finally,a seismic-inversion method of nonlinear mapping multilevel well–seismic matching based on the Bi-LSTM network is developed.The characteristic of this method is that by applying the multilevel well–seismic matching process,the seismic data are stepwise matched to the scale range that is consistent with the logging curve.Further,the matching operator at each level can be stably obtained to effectively overcome the problems that occur in the well–seismic matching process,such as the inconsistency in the scale of two types of data,accuracy in extracting the seismic wavelet of the well-side seismic traces,and multiplicity of solutions.Model test and practical application demonstrate that this method improves the vertical resolution of inversion results,and at the same time,the boundary and the lateral characteristics of the sand body are well maintained to improve the accuracy of thin-layer sand body prediction and achieve an improved practical application effect.
基金This work was supported by Scientific Research Starting Project of SWPU[Zheng,D.,No.0202002131604]Major Science and Technology Project of Sichuan Province[Zheng,D.,No.8ZDZX0143]+1 种基金Ministry of Education Collaborative Education Project of China[Zheng,D.,No.952]Fundamental Research Project[Zheng,D.,Nos.549,550].
文摘Nowadays,Internet has become an indispensable part of daily life and is used in many fields.Due to the large amount of Internet traffic,computers are subject to various security threats,which may cause serious economic losses and even endanger national security.It is hoped that an effective security method can systematically classify intrusion data in order to avoid leakage of important data or misuse of data.As machine learning technology matures,deep learning is widely used in various industries.Combining deep learning with network security and intrusion detection is the current trend.In this paper,the problem of data classification in intrusion detection system is studied.We propose an intrusion detection model based on stack bidirectional long short-term memory(LSTM),introduce stack bidirectional LSTM into the field of intrusion detection and apply it to the intrusion detection.In order to determine the appropriate parameters and structure of stack bidirectional LSTM network,we have carried out experiments on various network structures and parameters and analyzed the experimental results.The classic KDD Cup’1999 dataset was selected for experiments so that we can obtain convincing and comparable results.Experimental results derived from the KDD Cup’1999 dataset show that the network with three hidden layers containing 80 LSTM cells is superior to other algorithms in computational cost and detection performance due to stack bidirectional LSTM model’s ability to review time and correlate with connected records continuously.The experiment shows the effectiveness of stack bidirectional LSTM network in intrusion detection.
基金This work was partly supported by National Natural Science Foundation of China(No.61903296)Key Project of Natural Science Basic Research Plan in Shaanxi Province of China(No.2019ZDLGY18-03)+1 种基金Thousand Talents Plan of Shaanxi Province for Young Professionals,Project of Shaanxi Science and Technology(No.2019JQ-329)Doctoral Scientific Research Foundation of Xi’an University of Technology(No.103-451116012).
文摘Accurate sag source location and precise sag type recognition are both essential to verifying the responsible party for the sag and taking countermeasures to improve power quality.In this paper,an attention-based independently recurrent neural network(IndRNN)for sag source location and sag type recognition in sparsely monitored power system is proposed.Specially,the given inputs are voltage waveforms collected by limited meters in sparsely monitored power system,and the desired outputs simultaneously contain the following information:the located lines where sag occurs;the corresponding sag types,including motor starting,transformer energizing and short circuit;and the fault phase for short circuit.In essence,the responsibility of the proposed method is to automatically establish a nonlinear function that relates the given inputs to the desired outputs with categorization labels as few as possible.A favorable feature of the proposed method is that it can be realized without system parameters or models.The proposed method is validated by IEEE 30-bus system and a real 134-bus system.Experimental results demonstrate that the accuracy of sag source location is higher than 99%for all lines,and the accuracy of sag type recognition is also higher than 99%for various sag sources including motor starting,transformer energizing and 7 different types of short circuits.Furthermore,a comparison among different monitor placements for the proposed method is conducted,which illustrates that the observability of power networks should be ensured to achieve satisfactory performance.
基金supported by 2019 Industrial Internet Innovation Development Project of Ministry of Industry and Information Technology of China “Comprehensive Security Defense Platform Project for Industrial/Enterprise Networks”。
文摘In the industrial fields, the mechanical equipment will inevitably wear out in the process of operation. With the accumulation of losses, the probability of equipment failure is increasing. Therefore, if the remaining useful life(RUL) of the equipment can be accurately predicted, the equipment can be maintained in time to avoid the downtime caused by equipment failure and greatly improve the production efficiency of enterprises. This paper aims to use independently recurrent neural network(IndRNN) to learn health degradation of turbofan engine and make accurate predictions of its RUL, which not only effectively solves the problem of gradient explosion and vanishing, but also increases the interpretability of neural networks. IndRNN can be used to process longer time series which matches the scene with high frequency sampling sensor in industrial practical applications. The results demonstrate that IndRNN for RUL estimation significantly outperforms traditional approaches, as well as convolutional neural network(CNN) and long short-term memory network(LSTM) for RUL estimation.
基金Project supported by the National Key Research and Development Program of China(Grant No.2019YFB2205102)the National Natural Science Foundation of China(Grant Nos.61974164,62074166,61804181,62004219,62004220,and 62104256).
文摘Memristor-based neuromorphic computing shows great potential for high-speed and high-throughput signal processing applications,such as electroencephalogram(EEG)signal processing.Nonetheless,the size of one-transistor one-resistor(1T1R)memristor arrays is limited by the non-ideality of the devices,which prevents the hardware implementation of large and complex networks.In this work,we propose the depthwise separable convolution and bidirectional gate recurrent unit(DSC-BiGRU)network,a lightweight and highly robust hybrid neural network based on 1T1R arrays that enables efficient processing of EEG signals in the temporal,frequency and spatial domains by hybridizing DSC and BiGRU blocks.The network size is reduced and the network robustness is improved while ensuring the network classification accuracy.In the simulation,the measured non-idealities of the 1T1R array are brought into the network through statistical analysis.Compared with traditional convolutional networks,the network parameters are reduced by 95%and the network classification accuracy is improved by 21%at a 95%array yield rate and 5%tolerable error.This work demonstrates that lightweight and highly robust networks based on memristor arrays hold great promise for applications that rely on low consumption and high efficiency.
基金supported by the National Natural Science Foundation of China (No. 61572505)ChanXueYan Prospective Project of Jiangsu Province (No. BY201502305)
文摘Relation classification is a crucial component in many Natural Language Processing(NLP) systems. In this paper, we propose a novel bidirectional recurrent neural network architecture(using Long Short-Term Memory,LSTM, cells) for relation classification, with an attention layer for organizing the context information on the word level and a tensor layer for detecting complex connections between two entities. The above two feature extraction operations are based on the LSTM networks and use their outputs. Our model allows end-to-end learning from the raw sentences in the dataset, without trimming or reconstructing them. Experiments on the SemEval-2010 Task 8dataset show that our model outperforms most state-of-the-art methods.
