In recent years, theoretical and practical research on event-based communication strategies has gained considerable research attention due primarily to their irreplaceable superiority in resource-constrained systems（...In recent years, theoretical and practical research on event-based communication strategies has gained considerable research attention due primarily to their irreplaceable superiority in resource-constrained systems（especially networked systems）. For networked systems, event-based transmission scheme is capable of improving the efficiency in resource utilization and prolonging the lifetime of the network components compared with the widely adopted periodic transmission scheme. As such, it would be interesting to 1） examining how the event-triggering mechanisms affect the control or filtering performance for networked systems, and 2） developing some suitable approaches for the controller and filter design problems. In this paper, a bibliographical review is presented on event-based control and filtering problems for various networked systems. First, the event-driven communication scheme is introduced in detail according to its engineering background, characteristic, and representative research frameworks. Then, different event-based control and filtering（or state estimation） problems are categorized and then discussed. Finally, we conclude the paper by outlining future research challenges for event-based networked systems.展开更多
This paper is concerned with the robust H ∞ filter problem for networked environments, which are subject to both transmission delay and packet dropouts randomly. By employing random series which have Bernoulli distri...This paper is concerned with the robust H ∞ filter problem for networked environments, which are subject to both transmission delay and packet dropouts randomly. By employing random series which have Bernoulli distributions taking value of 0 or 1, the data transmission model is obtained. Based on state augmentation and stochastic theory, the sufficient condition for robust stability with H ∞ constraints is derived for the filtering error system. The robust filter is designed in terms of feasibility of one certain linear matrix inequality （LMI）, which is formed by adopting matrix congruence transformations. A numerical example is given to show the effectiveness of the proposed filtering method.展开更多
The networked control system NCS is regarded as a sampled control system withoutput time-variant delay. White noise is considered in the model construction of NCS. By using theKalman filter theory to compute the filte...The networked control system NCS is regarded as a sampled control system withoutput time-variant delay. White noise is considered in the model construction of NCS. By using theKalman filter theory to compute the filter parameters, a Kalman filter is constructed for this NCS.By comparing the output of the filter and the practical system, a residual is generated to diagnoseme sensor faults and the actuator faults. Finally, an example is given to show the feasibility ofthe approach.展开更多
In this paper we propose a Filter-based Uniform Algorithm （FbUA） for optimizing top-κ query in distributed networks, which has been a topic of much recent interest. The basic idea of FhUA is to set a filter at each...In this paper we propose a Filter-based Uniform Algorithm （FbUA） for optimizing top-κ query in distributed networks, which has been a topic of much recent interest. The basic idea of FhUA is to set a filter at each node to pre vent it from sending out the data with little chance to contrib ute to the top-κ result. FbUA can gain exact answers to top-κ query through two phrases of round trip communications between query station and participant nodes. The experiment results show that FbUA reduces network bandwidth consumption dramatically.展开更多
Deep Convolution Neural Networks(DCNNs)can capture discriminative features from large datasets.However,how to incrementally learn new samples without forgetting old ones and recognize novel classes that arise in the d...Deep Convolution Neural Networks(DCNNs)can capture discriminative features from large datasets.However,how to incrementally learn new samples without forgetting old ones and recognize novel classes that arise in the dynamically changing world,e.g.,classifying newly discovered fish species,remains an open problem.We address an even more challenging and realistic setting of this problem where new class samples are insufficient,i.e.,Few-Shot Class-Incremental Learning(FSCIL).Current FSCIL methods augment the training data to alleviate the overfitting of novel classes.By contrast,we propose Filter Bank Networks(FBNs)that augment the learnable filters to capture fine-detailed features for adapting to future new classes.In the forward pass,FBNs augment each convolutional filter to a virtual filter bank containing the canonical one,i.e.,itself,and multiple transformed versions.During back-propagation,FBNs explicitly stimulate fine-detailed features to emerge and collectively align all gradients of each filter bank to learn the canonical one.FBNs capture pattern variants that do not yet exist in the pretraining session,thus making it easy to incorporate new classes in the incremental learning phase.Moreover,FBNs introduce model-level prior knowledge to efficiently utilize the limited few-shot data.Extensive experiments on MNIST,CIFAR100,CUB200,andMini-ImageNet datasets show that FBNs consistently outperformthe baseline by a significantmargin,reporting new state-of-the-art FSCIL results.In addition,we contribute a challenging FSCIL benchmark,Fishshot1K,which contains 8261 underwater images covering 1000 ocean fish species.The code is included in the supplementary materials.展开更多
It is a challenge to track the maneuvering targets with noise disturbance and unknown dynamics. In this paper, an adaptive recurrent neural network tracking filter (ARNNF) for use in maneuvering target tracking was p...It is a challenge to track the maneuvering targets with noise disturbance and unknown dynamics. In this paper, an adaptive recurrent neural network tracking filter (ARNNF) for use in maneuvering target tracking was provided. The scheme is based on recurrent neural networks of which the recurrence provides a potentially unlimited memory depth adjusted by the network adaptively ( i.e. , it finds the best duration to represent the input signals past), and thus can actually capture the dynamics of the system that produced a temporal signal. On the other hand, recurrent neural network can approximate arbitrary nonlinear functions in L 2 space. The theoretical analysis indicates that the ARNNF can track the maneuvering targets with optimal filtering performance. Comparisons with IMM and AIMM algorithm show that ARNNF has better performance, and furthermore the ARNNF does not rely on the assumption with the known maneuvering target models, measurement noise and system noise.展开更多
A new approach for the design of two-dimensional （2-D） linear phase FIR digital filters based on a new neural networks algorithm （NNA） is provided. A compact expression for the transfer function of a 2-D linear ph...A new approach for the design of two-dimensional （2-D） linear phase FIR digital filters based on a new neural networks algorithm （NNA） is provided. A compact expression for the transfer function of a 2-D linear phase FIR filter is derived based on its frequency response characteristic, and the NNA, based on minimizing the square-error in the frequency-domain, is established according to the compact expression. To illustrate the stability of the NNA, the convergence theorem is presented and proved. Design examples are also given, and the results show that the ripple is considerably small in passband and stopband, and the NNA-based method is of powerful stability and requires quite little amount of computations.展开更多
A new nonlinear image filter using fuzzy neural network based on genetic algorithm is proposed. The learning of network parameters is performed by genetic algorithm with the efficient binary encoding scheme. In the fo...A new nonlinear image filter using fuzzy neural network based on genetic algorithm is proposed. The learning of network parameters is performed by genetic algorithm with the efficient binary encoding scheme. In the following, fuzzy reasoning embedded in the network aims at restoring noisy pixels without degrading the quality of fine details. It is shown by experiments that the filter is very effective in removing impulse noise and significantly outperforms conventional filters.展开更多
A novel statistical method based on particle filtering is presented for multiple vehicle acoustic signals separation problem in wireless sensor network. The particle filtering method is able to deal with non-Gaussian ...A novel statistical method based on particle filtering is presented for multiple vehicle acoustic signals separation problem in wireless sensor network. The particle filtering method is able to deal with non-Gaussian and nonlinear models and non-stationary sources. Using some instantaneously mixed observations of several real-world vehicle acoustic signals, the proposed statistical method is compared with a conventional non-stationary Blind Source Separation algorithm and attractive simulation results are achieved. Moreover, considering the natural convenience to transmit particles between sensor nodes, the algorithm based on particle filtering is believed to have potential to enable the task of multiple vehicles recognition collaboratively performed by sensor nodes in distributed wireless sensor network.展开更多
Based on wave digital filter(WDF) principles, this paper presents a digital model of cellular neural networks(CNNs). The model can precisely simulate the dynamic behavior of CNNs.
The algorithm of Hopfield neural network filtering and estimation is studied. The model of vehicular dead reckoning system fitting for the algorithm is constructed, and the design scheme of system filtering and estima...The algorithm of Hopfield neural network filtering and estimation is studied. The model of vehicular dead reckoning system fitting for the algorithm is constructed, and the design scheme of system filtering and estimation based on Hopfield network is proposed. Compared with Kalman filter, the algorithm does not require very precise system model and the prior knowledge of noise statistics and does not diverge easily. The simulation results show that the vehicular dead reckoning system based on Hopfield network filtering and estimation has the good position precision, and needn't require the inertial sensors with high precision. Therefore, the algorithm has the good practicability.展开更多
Battery management systems (BMS) must estimate the state-of-charge (SOC) of the battery accurately to prolong its lifetime and ensure a reliable operation. Since batteries have a wide range of applications, the SOC es...Battery management systems (BMS) must estimate the state-of-charge (SOC) of the battery accurately to prolong its lifetime and ensure a reliable operation. Since batteries have a wide range of applications, the SOC estimation requirements and methods vary from an application to another. This paper compares two SOC estimation methods, namely extended Kalman filters (EKF) and artificial neural networks (ANN). EKF is a nonlinear optimal estimator that is used to estimate the inner state of a nonlinear dynamic system using a state-space model. On the other hand, ANN is a mathematical model that consists of interconnected artificial neurons inspired by biological neural networks and is used to predict the output of a dynamic system based on some historical data of that system. A pulse-discharge test was performed on a commercial lithium-ion (Li-ion) battery cell in order to collect data to evaluate those methods. Results are presented and compared.展开更多
Wireless sensor networks (WSNs) are networked systems that are able to sense various events and report the events to a user to enable appropriate responses. One of security threats to a WSN is false data injection att...Wireless sensor networks (WSNs) are networked systems that are able to sense various events and report the events to a user to enable appropriate responses. One of security threats to a WSN is false data injection attacks in which an attacker steals some sensor nodes in the network and injects forged event messages into the network through the captured nodes. As a result, the intermediate nodes on the forwarding paths of the false event messages waste their limited energy. Additionally, the network cannot provide the user with correct information. There have been many studies on en-route detection of false event messages for WSNs. Yang et al. proposed the commutative cipher-based en-route filtering scheme (CCEF) which establishes a secure session between a sink node and a cluster head (CH) based on the commutative cipher. In CCEF, each intermediate node on the path between the sink node and the CH receives an event message and verifies the authenticity of the session based on a probability. Due to the probabilistic approach, it is hard to adapt to the change of false traffic ratio in the network and energy inefficiency may occur. We propose a filtering scheme which applies a deterministic approach to assign filtering nodes to a given session. The proposed method consumes less energy than that of CCEF without sacrificing security.展开更多
An optimal design approach of high order FIR digital filter is developed based on the algorithm of neural networks with cosine basis function . The main idea is to minimize the sum of the square errors between the amp...An optimal design approach of high order FIR digital filter is developed based on the algorithm of neural networks with cosine basis function . The main idea is to minimize the sum of the square errors between the amplitude response of the desired FIR filter and that of the designed by training the weights of neural networks, then obtains the impulse response of FIR digital filter . The convergence theorem of the neural networks algorithm is presented and proved, and the optimal design method is introduced by designing four kinds of FIR digital filters , i.e., low-pass, high-pass, bandpass , and band-stop FIR digital filter. The results of the amplitude responses show that attenuation in stop-bands is more than 60 dB with no ripple and pulse existing in pass-bands, and cutoff frequency of passband and stop-band is easily controlled precisely .The presented optimal design approach of high order FIR digital filter is significantly effective.展开更多
In this paper, artificial neural networks are used for predicting single fiber efficiency in the process of removing smaller particles from gas stream by fiber filters. For this, numerical simulations are obtained of ...In this paper, artificial neural networks are used for predicting single fiber efficiency in the process of removing smaller particles from gas stream by fiber filters. For this, numerical simulations are obtained of a classic model of literature for fiber efficiency, which is numerically solved along with the convection diffusion equation in polar coordinates for particle concentration, with associated initial and boundary conditions. A sufficient number of examples from two numerical simulations are employed to construct a database, from which parameters of a novel neural model are adjusted. This model is constructed based on the back propagation algorithm in order to map two features, namely Peclet number and packing density, which are extracted from the numerical simulations into the corresponding single fiber efficiency. The results indicate that the developed neural model can be trained in a reasonable computational time and is capable of estimating single fiber efficiency from examples of the test set with a maximum error of 1.7%.展开更多
Localization of the sensor nodes is a key supporting technology in wireless sensor networks （WSNs）. In this paper, a real-time localization estimator of mobile node in WSNs based on extended Kalman filter （KF） is ...Localization of the sensor nodes is a key supporting technology in wireless sensor networks （WSNs）. In this paper, a real-time localization estimator of mobile node in WSNs based on extended Kalman filter （KF） is proposed. Mobile node movement model is analyzed and online sequential iterative method is used to compute location result. The detailed steps of mobile sensor node self-localization adopting extended Kalman filter （EKF） is designed. The simulation results show that the accuracy of the localization estimator scheme designed is better than those of maximum likelihood estimation （MLE） and traditional KF algorithm.展开更多
基金supported by National Natural Science Foundation of China(No.61329301)the Royal Society of the UK+2 种基金the Research Fund for the Taishan Scholar Project of Shandong Province of Chinathe China Postdoctoral Science Foundation(No.2016M600547)the Alexander von Humboldt Foundation of Germany
文摘In recent years, theoretical and practical research on event-based communication strategies has gained considerable research attention due primarily to their irreplaceable superiority in resource-constrained systems（especially networked systems）. For networked systems, event-based transmission scheme is capable of improving the efficiency in resource utilization and prolonging the lifetime of the network components compared with the widely adopted periodic transmission scheme. As such, it would be interesting to 1） examining how the event-triggering mechanisms affect the control or filtering performance for networked systems, and 2） developing some suitable approaches for the controller and filter design problems. In this paper, a bibliographical review is presented on event-based control and filtering problems for various networked systems. First, the event-driven communication scheme is introduced in detail according to its engineering background, characteristic, and representative research frameworks. Then, different event-based control and filtering（or state estimation） problems are categorized and then discussed. Finally, we conclude the paper by outlining future research challenges for event-based networked systems.
基金Supported by National Basic Research Program of China （973 Program） （2010CB731800） and National Natural Science Foundation of China （60974059, 60736026, 61021063）
基金supported by National Natural Science Foundation of China (No. 61004088)the Key Foundation for Basic Research from Science and Technology Commission of Shanghai (No. 09JC1408000)the Aeronautic Science Foundation of China (No. 20100157001)
文摘This paper is concerned with the robust H ∞ filter problem for networked environments, which are subject to both transmission delay and packet dropouts randomly. By employing random series which have Bernoulli distributions taking value of 0 or 1, the data transmission model is obtained. Based on state augmentation and stochastic theory, the sufficient condition for robust stability with H ∞ constraints is derived for the filtering error system. The robust filter is designed in terms of feasibility of one certain linear matrix inequality （LMI）, which is formed by adopting matrix congruence transformations. A numerical example is given to show the effectiveness of the proposed filtering method.
基金Supported by National Natural Science Foundation of China （60574085, 60736026, 60721003）, the National High Technology Research and Development Program of China （863 Program） （2006AA04Z428）, and German Research Foundation （DFG）（DI 773/10）
文摘The networked control system NCS is regarded as a sampled control system withoutput time-variant delay. White noise is considered in the model construction of NCS. By using theKalman filter theory to compute the filter parameters, a Kalman filter is constructed for this NCS.By comparing the output of the filter and the practical system, a residual is generated to diagnoseme sensor faults and the actuator faults. Finally, an example is given to show the feasibility ofthe approach.
基金Supported by the National Natural Science Foun-dation of China (60503036 ,60473073) Fok Ying Tong EducationFoundation (104027)
文摘In this paper we propose a Filter-based Uniform Algorithm （FbUA） for optimizing top-κ query in distributed networks, which has been a topic of much recent interest. The basic idea of FhUA is to set a filter at each node to pre vent it from sending out the data with little chance to contrib ute to the top-κ result. FbUA can gain exact answers to top-κ query through two phrases of round trip communications between query station and participant nodes. The experiment results show that FbUA reduces network bandwidth consumption dramatically.
基金support from the Strategic Priority Research Program of the Chinese Academy of Sciences under Grant No.XDA27000000.
文摘Deep Convolution Neural Networks(DCNNs)can capture discriminative features from large datasets.However,how to incrementally learn new samples without forgetting old ones and recognize novel classes that arise in the dynamically changing world,e.g.,classifying newly discovered fish species,remains an open problem.We address an even more challenging and realistic setting of this problem where new class samples are insufficient,i.e.,Few-Shot Class-Incremental Learning(FSCIL).Current FSCIL methods augment the training data to alleviate the overfitting of novel classes.By contrast,we propose Filter Bank Networks(FBNs)that augment the learnable filters to capture fine-detailed features for adapting to future new classes.In the forward pass,FBNs augment each convolutional filter to a virtual filter bank containing the canonical one,i.e.,itself,and multiple transformed versions.During back-propagation,FBNs explicitly stimulate fine-detailed features to emerge and collectively align all gradients of each filter bank to learn the canonical one.FBNs capture pattern variants that do not yet exist in the pretraining session,thus making it easy to incorporate new classes in the incremental learning phase.Moreover,FBNs introduce model-level prior knowledge to efficiently utilize the limited few-shot data.Extensive experiments on MNIST,CIFAR100,CUB200,andMini-ImageNet datasets show that FBNs consistently outperformthe baseline by a significantmargin,reporting new state-of-the-art FSCIL results.In addition,we contribute a challenging FSCIL benchmark,Fishshot1K,which contains 8261 underwater images covering 1000 ocean fish species.The code is included in the supplementary materials.
文摘It is a challenge to track the maneuvering targets with noise disturbance and unknown dynamics. In this paper, an adaptive recurrent neural network tracking filter (ARNNF) for use in maneuvering target tracking was provided. The scheme is based on recurrent neural networks of which the recurrence provides a potentially unlimited memory depth adjusted by the network adaptively ( i.e. , it finds the best duration to represent the input signals past), and thus can actually capture the dynamics of the system that produced a temporal signal. On the other hand, recurrent neural network can approximate arbitrary nonlinear functions in L 2 space. The theoretical analysis indicates that the ARNNF can track the maneuvering targets with optimal filtering performance. Comparisons with IMM and AIMM algorithm show that ARNNF has better performance, and furthermore the ARNNF does not rely on the assumption with the known maneuvering target models, measurement noise and system noise.
文摘A new approach for the design of two-dimensional （2-D） linear phase FIR digital filters based on a new neural networks algorithm （NNA） is provided. A compact expression for the transfer function of a 2-D linear phase FIR filter is derived based on its frequency response characteristic, and the NNA, based on minimizing the square-error in the frequency-domain, is established according to the compact expression. To illustrate the stability of the NNA, the convergence theorem is presented and proved. Design examples are also given, and the results show that the ripple is considerably small in passband and stopband, and the NNA-based method is of powerful stability and requires quite little amount of computations.
文摘A new nonlinear image filter using fuzzy neural network based on genetic algorithm is proposed. The learning of network parameters is performed by genetic algorithm with the efficient binary encoding scheme. In the following, fuzzy reasoning embedded in the network aims at restoring noisy pixels without degrading the quality of fine details. It is shown by experiments that the filter is very effective in removing impulse noise and significantly outperforms conventional filters.
基金the National "863" High Technology Development Program (2006AA01Z216)the MajorResearch Program of the Science and Technology Commission of Shanghai Municipality of China (054SGA1001).
文摘A novel statistical method based on particle filtering is presented for multiple vehicle acoustic signals separation problem in wireless sensor network. The particle filtering method is able to deal with non-Gaussian and nonlinear models and non-stationary sources. Using some instantaneously mixed observations of several real-world vehicle acoustic signals, the proposed statistical method is compared with a conventional non-stationary Blind Source Separation algorithm and attractive simulation results are achieved. Moreover, considering the natural convenience to transmit particles between sensor nodes, the algorithm based on particle filtering is believed to have potential to enable the task of multiple vehicles recognition collaboratively performed by sensor nodes in distributed wireless sensor network.
文摘Based on wave digital filter(WDF) principles, this paper presents a digital model of cellular neural networks(CNNs). The model can precisely simulate the dynamic behavior of CNNs.
文摘The algorithm of Hopfield neural network filtering and estimation is studied. The model of vehicular dead reckoning system fitting for the algorithm is constructed, and the design scheme of system filtering and estimation based on Hopfield network is proposed. Compared with Kalman filter, the algorithm does not require very precise system model and the prior knowledge of noise statistics and does not diverge easily. The simulation results show that the vehicular dead reckoning system based on Hopfield network filtering and estimation has the good position precision, and needn't require the inertial sensors with high precision. Therefore, the algorithm has the good practicability.
文摘Battery management systems (BMS) must estimate the state-of-charge (SOC) of the battery accurately to prolong its lifetime and ensure a reliable operation. Since batteries have a wide range of applications, the SOC estimation requirements and methods vary from an application to another. This paper compares two SOC estimation methods, namely extended Kalman filters (EKF) and artificial neural networks (ANN). EKF is a nonlinear optimal estimator that is used to estimate the inner state of a nonlinear dynamic system using a state-space model. On the other hand, ANN is a mathematical model that consists of interconnected artificial neurons inspired by biological neural networks and is used to predict the output of a dynamic system based on some historical data of that system. A pulse-discharge test was performed on a commercial lithium-ion (Li-ion) battery cell in order to collect data to evaluate those methods. Results are presented and compared.
文摘Wireless sensor networks (WSNs) are networked systems that are able to sense various events and report the events to a user to enable appropriate responses. One of security threats to a WSN is false data injection attacks in which an attacker steals some sensor nodes in the network and injects forged event messages into the network through the captured nodes. As a result, the intermediate nodes on the forwarding paths of the false event messages waste their limited energy. Additionally, the network cannot provide the user with correct information. There have been many studies on en-route detection of false event messages for WSNs. Yang et al. proposed the commutative cipher-based en-route filtering scheme (CCEF) which establishes a secure session between a sink node and a cluster head (CH) based on the commutative cipher. In CCEF, each intermediate node on the path between the sink node and the CH receives an event message and verifies the authenticity of the session based on a probability. Due to the probabilistic approach, it is hard to adapt to the change of false traffic ratio in the network and energy inefficiency may occur. We propose a filtering scheme which applies a deterministic approach to assign filtering nodes to a given session. The proposed method consumes less energy than that of CCEF without sacrificing security.
基金This project was supported by the National Natural Science Foundation of China (50277010)Doctoral Special Fund of Ministry of Education (20020532016) and Fund of Outstanding Young Scientist of Hunan University.
文摘An optimal design approach of high order FIR digital filter is developed based on the algorithm of neural networks with cosine basis function . The main idea is to minimize the sum of the square errors between the amplitude response of the desired FIR filter and that of the designed by training the weights of neural networks, then obtains the impulse response of FIR digital filter . The convergence theorem of the neural networks algorithm is presented and proved, and the optimal design method is introduced by designing four kinds of FIR digital filters , i.e., low-pass, high-pass, bandpass , and band-stop FIR digital filter. The results of the amplitude responses show that attenuation in stop-bands is more than 60 dB with no ripple and pulse existing in pass-bands, and cutoff frequency of passband and stop-band is easily controlled precisely .The presented optimal design approach of high order FIR digital filter is significantly effective.
文摘In this paper, artificial neural networks are used for predicting single fiber efficiency in the process of removing smaller particles from gas stream by fiber filters. For this, numerical simulations are obtained of a classic model of literature for fiber efficiency, which is numerically solved along with the convection diffusion equation in polar coordinates for particle concentration, with associated initial and boundary conditions. A sufficient number of examples from two numerical simulations are employed to construct a database, from which parameters of a novel neural model are adjusted. This model is constructed based on the back propagation algorithm in order to map two features, namely Peclet number and packing density, which are extracted from the numerical simulations into the corresponding single fiber efficiency. The results indicate that the developed neural model can be trained in a reasonable computational time and is capable of estimating single fiber efficiency from examples of the test set with a maximum error of 1.7%.
基金Project supported by the Shanghai Leading Academic Discipcine Project (Grant No.S30108)the National Natural Science Foundation of China (Grant No.60872021)the Science and Technology Commission of Shanghai Municipality (Grant No.08DZ2231100)
文摘Localization of the sensor nodes is a key supporting technology in wireless sensor networks （WSNs）. In this paper, a real-time localization estimator of mobile node in WSNs based on extended Kalman filter （KF） is proposed. Mobile node movement model is analyzed and online sequential iterative method is used to compute location result. The detailed steps of mobile sensor node self-localization adopting extended Kalman filter （EKF） is designed. The simulation results show that the accuracy of the localization estimator scheme designed is better than those of maximum likelihood estimation （MLE） and traditional KF algorithm.
基金Supported by National Natural Science Foundation of China （60874063）, and Innovation and Scientific Research Foundation of Graduate Student of Heilongjiang Province （YJSCX2012-263HLJ）
