This paper investigates interception missiles’trajectory tracking guidance problem under wind field and external disturbances in the boost phase.Indeed,the velocity control in such trajectory tracking guidance system...This paper investigates interception missiles’trajectory tracking guidance problem under wind field and external disturbances in the boost phase.Indeed,the velocity control in such trajectory tracking guidance systems of missiles is challenging.As our contribution,the velocity control channel is designed to deal with the intractable velocity problem and improve tracking accuracy.The global prescribed performance function,which guarantees the tracking error within the set range and the global convergence of the tracking guidance system,is first proposed based on the traditional PPF.Then,a tracking guidance strategy is derived using the integral sliding mode control techniques to make the sliding manifold and tracking errors converge to zero and avoid singularities.Meanwhile,an improved switching control law is introduced into the designed tracking guidance algorithm to deal with the chattering problem.A back propagation neural network(BPNN)extended state observer(BPNNESO)is employed in the inner loop to identify disturbances.The obtained results indicate that the proposed tracking guidance approach achieves the trajectory tracking guidance objective without and with disturbances and outperforms the existing tracking guidance schemes with the lowest tracking errors,convergence times,and overshoots.展开更多
Aiming at the reliability analysis of small sample data or implicit structural function,a novel structural reliability analysis model based on support vector machine(SVM)and neural network direct integration method(DN...Aiming at the reliability analysis of small sample data or implicit structural function,a novel structural reliability analysis model based on support vector machine(SVM)and neural network direct integration method(DNN)is proposed.Firstly,SVM with good small sample learning ability is used to train small sample data,fit structural performance functions and establish regular integration regions.Secondly,DNN is approximated the integral function to achieve multiple integration in the integration region.Finally,structural reliability was obtained by DNN.Numerical examples are investigated to demonstrate the effectiveness of the present method,which provides a feasible way for the structural reliability analysis.展开更多
In view of the failure of GNSS signals,this paper proposes an INS/GNSS integrated navigation method based on the recurrent neural network(RNN).This proposed method utilizes the calculation principle of INS and the mem...In view of the failure of GNSS signals,this paper proposes an INS/GNSS integrated navigation method based on the recurrent neural network(RNN).This proposed method utilizes the calculation principle of INS and the memory function of the RNN to estimate the errors of the INS,thereby obtaining a continuous,reliable and high-precision navigation solution.The performance of the proposed method is firstly demonstrated using an INS/GNSS simulation environment.Subsequently,an experimental test on boat is also conducted to validate the performance of the method.The results show a promising application prospect for RNN in the field of positioning for INS/GNSS integrated navigation in the absence of GNSS signal,as it outperforms extreme learning machine(ELM)and EKF by approximately 30%and 60%,respectively.展开更多
An adaptive integral dynamic surface control approach based on fully tuned radial basis function neural network (FTRBFNN) is presented for a general class of strict-feedback nonlinear systems,which may possess a wid...An adaptive integral dynamic surface control approach based on fully tuned radial basis function neural network (FTRBFNN) is presented for a general class of strict-feedback nonlinear systems,which may possess a wide class of uncertainties that are not linearly parameterized and do not have any prior knowledge of the bounding functions.FTRBFNN is employed to approximate the uncertainty online,and a systematic framework for adaptive controller design is given by dynamic surface control. The control algorithm has two outstanding features,namely,the neural network regulates the weights,width and center of Gaussian function simultaneously,which ensures the control system has perfect ability of restraining different unknown uncertainties and the integral term of tracking error introduced in the control law can eliminate the static error of the closed loop system effectively. As a result,high control precision can be achieved.All signals in the closed loop system can be guaranteed bounded by Lyapunov approach.Finally,simulation results demonstrate the validity of the control approach.展开更多
Long-term time series forecasting stands as a crucial research domain within the realm of automated machine learning(AutoML).At present,forecasting,whether rooted in machine learning or statistical learning,typically ...Long-term time series forecasting stands as a crucial research domain within the realm of automated machine learning(AutoML).At present,forecasting,whether rooted in machine learning or statistical learning,typically relies on expert input and necessitates substantial manual involvement.This manual effort spans model development,feature engineering,hyper-parameter tuning,and the intricate construction of time series models.The complexity of these tasks renders complete automation unfeasible,as they inherently demand human intervention at multiple junctures.To surmount these challenges,this article proposes leveraging Long Short-Term Memory,which is the variant of Recurrent Neural Networks,harnessing memory cells and gating mechanisms to facilitate long-term time series prediction.However,forecasting accuracy by particular neural network and traditional models can degrade significantly,when addressing long-term time-series tasks.Therefore,our research demonstrates that this innovative approach outperforms the traditional Autoregressive Integrated Moving Average(ARIMA)method in forecasting long-term univariate time series.ARIMA is a high-quality and competitive model in time series prediction,and yet it requires significant preprocessing efforts.Using multiple accuracy metrics,we have evaluated both ARIMA and proposed method on the simulated time-series data and real data in both short and long term.Furthermore,our findings indicate its superiority over alternative network architectures,including Fully Connected Neural Networks,Convolutional Neural Networks,and Nonpooling Convolutional Neural Networks.Our AutoML approach enables non-professional to attain highly accurate and effective time series forecasting,and can be widely applied to various domains,particularly in business and finance.展开更多
Neuromorphic photonic computing has emerged as a competitive computing paradigm to overcome the bottlenecks of the von-Neumann architecture.Linear weighting and nonlinear spike activation are two fundamental functions...Neuromorphic photonic computing has emerged as a competitive computing paradigm to overcome the bottlenecks of the von-Neumann architecture.Linear weighting and nonlinear spike activation are two fundamental functions of a photonic spiking neural network(PSNN).However,they are separately implemented with different photonic materials and devices,hindering the large-scale integration of PSNN.Here,we propose,fabricate and experimentally demonstrate a photonic neuro-synaptic chip enabling the simultaneous implementation of linear weighting and nonlinear spike activation based on a distributed feedback(DFB)laser with a saturable absorber(DFB-SA).A prototypical system is experimentally constructed to demonstrate the parallel weighted function and nonlinear spike activation.Furthermore,a fourchannel DFB-SA laser array is fabricated for realizing matrix convolution of a spiking convolutional neural network,achieving a recognition accuracy of 87%for the MNIST dataset.The fabricated neuro-synaptic chip offers a fundamental building block to construct the large-scale integrated PSNN chip.展开更多
A closed-chain robot has several advantages over an open-chain robot, such as high mechanical rigidity, high payload, high precision. Accurate trajectory control of a robot is essential in practical-use. This paper pr...A closed-chain robot has several advantages over an open-chain robot, such as high mechanical rigidity, high payload, high precision. Accurate trajectory control of a robot is essential in practical-use. This paper presents an adaptive proportional integral differential (PID) control algorithm based on radial basis function (RBF) neural network for trajectory tracking of a two-degree-of-freedom (2-DOF) closed-chain robot. In this scheme, an RBF neural network is used to approximate the unknown nonlinear dynamics of the robot, at the same time, the PID parameters can be adjusted online and the high precision can be obtained. Simulation results show that the control algorithm accurately tracks a 2-DOF closed-chain robot trajectories. The results also indicate that the system robustness and tracking performance are superior to the classic PID method.展开更多
The tight wavelet neural network was constituted by taking the nonlinear Morlet wavelet radices as the excitation function. The idiographic algorithm was presented. It combined the advantages of wavelet analysis and n...The tight wavelet neural network was constituted by taking the nonlinear Morlet wavelet radices as the excitation function. The idiographic algorithm was presented. It combined the advantages of wavelet analysis and neural networks. The integrated wavelet neural network fault diagnosis system was set up based on both the information fusion technology and actual fault diagnosis, which took the sub-wavelet neural network as primary diagnosis from different sides, then came to the conclusions through decision-making fusion. The realizable policy of the diagnosis system and established principle of the sub-wavelet neural networks were given. It can be deduced from the examples that it takes full advantage of diversified characteristic information, and improves the diagnosis rate.展开更多
In the experiment of combined navigation filtering using wavelet neural network, the initial parameters of the network have the influence of randomness on network convergence and navigation accuracy. A combined naviga...In the experiment of combined navigation filtering using wavelet neural network, the initial parameters of the network have the influence of randomness on network convergence and navigation accuracy. A combined navigation filtering method based on wavelet neural network optimized by mind evolution algorithm is proposed. Firstly, the efficient global search ability of the mind evolution algorithm was used to quickly and accurately obtain the initial parameters of the appropriate wavelet neural network, and then the optimized wavelet neural network was applied to directly predict the position and velocity error data. This method is different from the traditional filtering method, while avoiding the drawbacks of the neural network. The simulation experiments with wavelet neural network and GA-wavelet network were carried out. The results show that the proposed method can effectively improve the accuracy of the integrated navigation system and provide a feasible path for combined navigation filtering.展开更多
In this paper, a study of control for an uncertain2-degree of freedom(DOF) helicopter system is given. The2-DOF helicopter is subject to input deadzone and output constraints. In order to cope with system uncertaintie...In this paper, a study of control for an uncertain2-degree of freedom(DOF) helicopter system is given. The2-DOF helicopter is subject to input deadzone and output constraints. In order to cope with system uncertainties and input deadzone, the neural network technique is introduced because of its capability in approximation. In order to update the weights of the neural network, an adaptive control method is utilized to improve the system adaptability. Furthermore, the integral barrier Lyapunov function(IBLF) is adopt in control design to guarantee the condition of output constraints and boundedness of the corresponding tracking errors. The Lyapunov direct method is applied in the control design to analyze system stability and convergence. Finally, numerical simulations are conducted to prove the feasibility and effectiveness of the proposed control based on the model of Quanser's 2-DOF helicopter.展开更多
Parallel multi-thread processing in advanced intelligent processors is the core to realize high-speed and high-capacity signal processing systems.Optical neural network(ONN)has the native advantages of high paralleliz...Parallel multi-thread processing in advanced intelligent processors is the core to realize high-speed and high-capacity signal processing systems.Optical neural network(ONN)has the native advantages of high parallelization,large bandwidth,and low power consumption to meet the demand of big data.Here,we demonstrate the dual-layer ONN with Mach-Zehnder interferometer(MZI)network and nonlinear layer,while the nonlinear activation function is achieved by optical-electronic signal conversion.Two frequency components from the microcomb source carrying digit datasets are simultaneously imposed and intelligently recognized through the ONN.We successfully achieve the digit classification of different frequency components by demultiplexing the output signal and testing power distribution.Efficient parallelization feasibility with wavelength division multiplexing is demonstrated in our high-dimensional ONN.This work provides a high-performance architecture for future parallel high-capacity optical analog computing.展开更多
For vehicle integrated navigation systems, real-time estimating states of the dead reckoning (DR) unit is much more difficult than that of the other measuring sensors under indefinite noises and nonlinear characteri...For vehicle integrated navigation systems, real-time estimating states of the dead reckoning (DR) unit is much more difficult than that of the other measuring sensors under indefinite noises and nonlinear characteristics. Compared with the well known, extended Kalman filter (EKF), a recurrent neural network is proposed for the solution, which not only improves the location precision and the adaptive ability of resisting disturbances, but also avoids calculating the analytic derivation and Jacobian matrices of the nonlinear system model. To test the performances of the recurrent neural network, these two methods are used to estimate the state of the vehicle's DR navigation system. Simulation results show that the recurrent neural network is superior to the EKF and is a more ideal filtering method for vehicle DR navigation.展开更多
A tightly coupled GPS ( global positioning system )/SINS ( strap down inertial navigation system) based on a GMDH ( group method of data handling) neural network was presented to solve the problem of degraded ac...A tightly coupled GPS ( global positioning system )/SINS ( strap down inertial navigation system) based on a GMDH ( group method of data handling) neural network was presented to solve the problem of degraded accuracy for less than four visible GPS satellites with poor signal quality. Positions and velocities of the satellites were predicted by a GMDH neural network, and the pseudo ranges and pseudo range rates received by the GPS receiver were simulated to ensure the regular op eration of the GPS/SINS Kalman filter during outages. In the mathematical simulation a tightly cou pled navigation system with a proposed approach has better navigation accuracy during GPS outages, and the anti jamming ability is strengthened for the tightly coupled navigation system.展开更多
Artificial Neural Network(ANN)models have been extensively applied in the prediction of water resource variables,and Geographical Information System(GIS)includes powerful functions to visualize spatial data.In order t...Artificial Neural Network(ANN)models have been extensively applied in the prediction of water resource variables,and Geographical Information System(GIS)includes powerful functions to visualize spatial data.In order to provide an efficient tool for environmental assessment and management that combines the advantages of these two modules,a GIS-based ANN water quality prediction system was developed in the present study.The ANN module and ArcGIS Engine module,along with a dynamic database,were imbedded in the system,which integrates water quality prediction via the ANN model and spatial presentation of the model results.The structure of the ANN model could be modified through the graphical user interface to optimize the model performance.The developed system was applied to a real case study for the prediction of the total phosphorus concentration in the Lake Champlain area.The prediction results were verified with the monitoring data,and the performance of the developed model was further evaluated through graphical techniques and quantitative statistical methods.Overall,the developed system provided satisfactory prediction results,and spatial distribution maps of the predicted results were obtained,which coincided with the monitored values.The developed GIS-based ANN water quality prediction system could serve as an efficient tool for engineers and decision makers.展开更多
Driven by continuous scaling of nanoscale semiconductor technologies,the past years have witnessed the progressive advancement of machine learning techniques and applications.Recently,dedicated machine learning accele...Driven by continuous scaling of nanoscale semiconductor technologies,the past years have witnessed the progressive advancement of machine learning techniques and applications.Recently,dedicated machine learning accelerators,especially for neural networks,have attracted the research interests of computer architects and VLSI designers.State-of-the-art accelerators increase performance by deploying a huge amount of processing elements,however still face the issue of degraded resource utilization across hybrid and non-standard algorithmic kernels.In this work,we exploit the properties of important neural network kernels for both perception and control to propose a reconfigurable dataflow processor,which adjusts the patterns of data flowing,functionalities of processing elements and on-chip storages according to network kernels.In contrast to stateof-the-art fine-grained data flowing techniques,the proposed coarse-grained dataflow reconfiguration approach enables extensive sharing of computing and storage resources.Three hybrid networks for MobileNet,deep reinforcement learning and sequence classification are constructed and analyzed with customized instruction sets and toolchain.A test chip has been designed and fabricated under UMC 65 nm CMOS technology,with the measured power consumption of 7.51 mW under 100 MHz frequency on a die size of 1.8×1.8 mm^2.展开更多
A current-mode MOS neuron circuit with 4-bit programmable weights is presented by using CMOS technology. The weights of the neurcn have high resolution and also can easily be digitally stored. The resolution can be ex...A current-mode MOS neuron circuit with 4-bit programmable weights is presented by using CMOS technology. The weights of the neurcn have high resolution and also can easily be digitally stored. The resolution can be extended into high levels such as 8-bit, etc. by the design methodology presented in this paper. The operational principle of the neuron is discussed. Circuit simulation has been made by use of SPICE II. The results give a good agreement for the design requirements.展开更多
This paper presents an integration methodology for ma chining and measuring processes using OMM (On-Machine Measurement) technology b ased on CAD/CAM/CAI integration concept. OMM uses a CNC machining center as a me as...This paper presents an integration methodology for ma chining and measuring processes using OMM (On-Machine Measurement) technology b ased on CAD/CAM/CAI integration concept. OMM uses a CNC machining center as a me asuring station by changing the tools into measuring probes such as touch-type, laser and vision. Although the measurement accuracy is not good compared to tha t of the CMM (Coordinate Measuring Machine), there are distinctive advantages us ing OMM in real situation. In this paper, two topics are handled to show the eff ectiveness of the machining and measuring process integration: (1) inspection pl anning strategy for sculptured surface machining and (2) tool path compensation for profile milling process. For the first topic, as a first step, effective mea suring point locations are determined to obtain optimum results for given sampli ng numbers. Two measuring point selection methods are suggested based on the CAD /CAM/CAI integration concept: (1) by the prediction of cutting errors and (2) by considering cutter contact points to avoid the measurement errors caused by cus ps. As a next step, the TSP (Traveling Salesman Problem) algorithm is applied to minimize the probe moving distance. Appropriate simulations and experiments are performed to verify the proposed inspection planning strategy, and the results are analyzed. For the second topic, a methodology for profile milling error comp ensation is presented by using an ANN (Artificial Neural Network) model trained by the inspection database of OMM system. First, geometric and thermal errors of the machining center are compensated using a closed-loop configuration for the improvement of machining and inspection accuracy. The probing errors are also t aken into account. Then, a specimen workpiece is machined and then the machi ning surface error distribution is measured on the machine using touch-type pro be. In order to efficiently analyze the machining errors, two characteristic err or parameters (W err and D err) are defined. Subsequently, these param eters are modeled by applying the RFB (Radial Basis Function) network approach a s an ANN model. Based on the RBF network model, the tool paths are compensated i n order to effectively reduce the errors by employing an iterative algorithm. In order to validate the approaches proposed in this paper, a concrete case of the machining process is taken into account and about 90% of machining error reduction is successfully accomplished through the proposed approaches.展开更多
The global asymptotic stability of cellular neural networks with delays is investigated. Three kinds of time delays have been considered. New delay-dependent stability criteria are proposed and are formulated as the f...The global asymptotic stability of cellular neural networks with delays is investigated. Three kinds of time delays have been considered. New delay-dependent stability criteria are proposed and are formulated as the feasibility of some linear matrix inequalities, which can be checked easily by resorting to the recently developed interior-point algorithms. Based on the Finsler Lemma, it is theoretically proved that the proposed stability criteria are less conservative than some existing results.展开更多
In the past,convolutional neural network(CNN)has become one of the most popular deep learning frameworks,and has been widely used in Hyperspectral image classification tasks.Convolution(Conv)in CNN uses filter weights...In the past,convolutional neural network(CNN)has become one of the most popular deep learning frameworks,and has been widely used in Hyperspectral image classification tasks.Convolution(Conv)in CNN uses filter weights to extract features in local receiving domain,and the weight parameters are shared globally,which more focus on the highfrequency information of the image.Different from Conv,Transformer can obtain the long‐term dependence between long‐distance features through modelling,and adaptively focus on different regions.In addition,Transformer is considered as a low‐pass filter,which more focuses on the low‐frequency information of the image.Considering the complementary characteristics of Conv and Transformer,the two modes can be integrated for full feature extraction.In addition,the most important image features correspond to the discrimination region,while the secondary image features represent important but easily ignored regions,which are also conducive to the classification of HSIs.In this study,a complementary integrated Transformer network(CITNet)for hyperspectral image classification is proposed.Firstly,three‐dimensional convolution(Conv3D)and two‐dimensional convolution(Conv2D)are utilised to extract the shallow semantic information of the image.In order to enhance the secondary features,a channel Gaussian modulation attention module is proposed,which is embedded between Conv3D and Conv2D.This module can not only enhance secondary features,but suppress the most important and least important features.Then,considering the different and complementary characteristics of Conv and Transformer,a complementary integrated Transformer module is designed.Finally,through a large number of experiments,this study evaluates the classification performance of CITNet and several state‐of‐the‐art networks on five common datasets.The experimental results show that compared with these classification networks,CITNet can provide better classification performance.展开更多
Development of energy-resources-poor remote rural areas of the world has been discussed by many in the past. Harnessing locally available renewable energy resources as an environmentally friendly option is gaining mom...Development of energy-resources-poor remote rural areas of the world has been discussed by many in the past. Harnessing locally available renewable energy resources as an environmentally friendly option is gaining momentum. Smart Integrated Renewable Energy Systems (SIRES) offer a resilient and economic path to “energize” the area and reach this goal. This paper discusses its intelligent control using neural networks and fuzzy logic.展开更多
基金the National Natural Science Foundation of China(Grant No.12072090).
文摘This paper investigates interception missiles’trajectory tracking guidance problem under wind field and external disturbances in the boost phase.Indeed,the velocity control in such trajectory tracking guidance systems of missiles is challenging.As our contribution,the velocity control channel is designed to deal with the intractable velocity problem and improve tracking accuracy.The global prescribed performance function,which guarantees the tracking error within the set range and the global convergence of the tracking guidance system,is first proposed based on the traditional PPF.Then,a tracking guidance strategy is derived using the integral sliding mode control techniques to make the sliding manifold and tracking errors converge to zero and avoid singularities.Meanwhile,an improved switching control law is introduced into the designed tracking guidance algorithm to deal with the chattering problem.A back propagation neural network(BPNN)extended state observer(BPNNESO)is employed in the inner loop to identify disturbances.The obtained results indicate that the proposed tracking guidance approach achieves the trajectory tracking guidance objective without and with disturbances and outperforms the existing tracking guidance schemes with the lowest tracking errors,convergence times,and overshoots.
基金National Natural Science Foundation of China(Nos.11262014,11962021 and 51965051)Inner Mongolia Natural Science Foundation,China(No.2019MS05064)+1 种基金Inner Mongolia Earthquake Administration Director Fund Project,China(No.2019YB06)Inner Mongolia University of Technology Foundation,China(No.2020015)。
文摘Aiming at the reliability analysis of small sample data or implicit structural function,a novel structural reliability analysis model based on support vector machine(SVM)and neural network direct integration method(DNN)is proposed.Firstly,SVM with good small sample learning ability is used to train small sample data,fit structural performance functions and establish regular integration regions.Secondly,DNN is approximated the integral function to achieve multiple integration in the integration region.Finally,structural reliability was obtained by DNN.Numerical examples are investigated to demonstrate the effectiveness of the present method,which provides a feasible way for the structural reliability analysis.
基金supported in part by the National Natural Science Foundation of China(No.41876222)。
文摘In view of the failure of GNSS signals,this paper proposes an INS/GNSS integrated navigation method based on the recurrent neural network(RNN).This proposed method utilizes the calculation principle of INS and the memory function of the RNN to estimate the errors of the INS,thereby obtaining a continuous,reliable and high-precision navigation solution.The performance of the proposed method is firstly demonstrated using an INS/GNSS simulation environment.Subsequently,an experimental test on boat is also conducted to validate the performance of the method.The results show a promising application prospect for RNN in the field of positioning for INS/GNSS integrated navigation in the absence of GNSS signal,as it outperforms extreme learning machine(ELM)and EKF by approximately 30%and 60%,respectively.
基金supported by the China Postdoctoral Science Foundation (200904501035 201003548)+3 种基金the National Natural Science Foundation of China (60835001907160289101600460804017)
文摘An adaptive integral dynamic surface control approach based on fully tuned radial basis function neural network (FTRBFNN) is presented for a general class of strict-feedback nonlinear systems,which may possess a wide class of uncertainties that are not linearly parameterized and do not have any prior knowledge of the bounding functions.FTRBFNN is employed to approximate the uncertainty online,and a systematic framework for adaptive controller design is given by dynamic surface control. The control algorithm has two outstanding features,namely,the neural network regulates the weights,width and center of Gaussian function simultaneously,which ensures the control system has perfect ability of restraining different unknown uncertainties and the integral term of tracking error introduced in the control law can eliminate the static error of the closed loop system effectively. As a result,high control precision can be achieved.All signals in the closed loop system can be guaranteed bounded by Lyapunov approach.Finally,simulation results demonstrate the validity of the control approach.
文摘Long-term time series forecasting stands as a crucial research domain within the realm of automated machine learning(AutoML).At present,forecasting,whether rooted in machine learning or statistical learning,typically relies on expert input and necessitates substantial manual involvement.This manual effort spans model development,feature engineering,hyper-parameter tuning,and the intricate construction of time series models.The complexity of these tasks renders complete automation unfeasible,as they inherently demand human intervention at multiple junctures.To surmount these challenges,this article proposes leveraging Long Short-Term Memory,which is the variant of Recurrent Neural Networks,harnessing memory cells and gating mechanisms to facilitate long-term time series prediction.However,forecasting accuracy by particular neural network and traditional models can degrade significantly,when addressing long-term time-series tasks.Therefore,our research demonstrates that this innovative approach outperforms the traditional Autoregressive Integrated Moving Average(ARIMA)method in forecasting long-term univariate time series.ARIMA is a high-quality and competitive model in time series prediction,and yet it requires significant preprocessing efforts.Using multiple accuracy metrics,we have evaluated both ARIMA and proposed method on the simulated time-series data and real data in both short and long term.Furthermore,our findings indicate its superiority over alternative network architectures,including Fully Connected Neural Networks,Convolutional Neural Networks,and Nonpooling Convolutional Neural Networks.Our AutoML approach enables non-professional to attain highly accurate and effective time series forecasting,and can be widely applied to various domains,particularly in business and finance.
基金financial supports from National Key Research and Development Program of China (2021YFB2801900,2021YFB2801901,2021YFB2801902,2021YFB2801904)National Natural Science Foundation of China (No.61974177)+1 种基金National Outstanding Youth Science Fund Project of National Natural Science Foundation of China (62022062)The Fundamental Research Funds for the Central Universities (QTZX23041).
文摘Neuromorphic photonic computing has emerged as a competitive computing paradigm to overcome the bottlenecks of the von-Neumann architecture.Linear weighting and nonlinear spike activation are two fundamental functions of a photonic spiking neural network(PSNN).However,they are separately implemented with different photonic materials and devices,hindering the large-scale integration of PSNN.Here,we propose,fabricate and experimentally demonstrate a photonic neuro-synaptic chip enabling the simultaneous implementation of linear weighting and nonlinear spike activation based on a distributed feedback(DFB)laser with a saturable absorber(DFB-SA).A prototypical system is experimentally constructed to demonstrate the parallel weighted function and nonlinear spike activation.Furthermore,a fourchannel DFB-SA laser array is fabricated for realizing matrix convolution of a spiking convolutional neural network,achieving a recognition accuracy of 87%for the MNIST dataset.The fabricated neuro-synaptic chip offers a fundamental building block to construct the large-scale integrated PSNN chip.
基金Project supported bY the National Natural Science Foundation of China (Grant No.50375085), and the Natural Science Foundation of Shandong Province (Grant No.Y2002F13)
文摘A closed-chain robot has several advantages over an open-chain robot, such as high mechanical rigidity, high payload, high precision. Accurate trajectory control of a robot is essential in practical-use. This paper presents an adaptive proportional integral differential (PID) control algorithm based on radial basis function (RBF) neural network for trajectory tracking of a two-degree-of-freedom (2-DOF) closed-chain robot. In this scheme, an RBF neural network is used to approximate the unknown nonlinear dynamics of the robot, at the same time, the PID parameters can be adjusted online and the high precision can be obtained. Simulation results show that the control algorithm accurately tracks a 2-DOF closed-chain robot trajectories. The results also indicate that the system robustness and tracking performance are superior to the classic PID method.
文摘The tight wavelet neural network was constituted by taking the nonlinear Morlet wavelet radices as the excitation function. The idiographic algorithm was presented. It combined the advantages of wavelet analysis and neural networks. The integrated wavelet neural network fault diagnosis system was set up based on both the information fusion technology and actual fault diagnosis, which took the sub-wavelet neural network as primary diagnosis from different sides, then came to the conclusions through decision-making fusion. The realizable policy of the diagnosis system and established principle of the sub-wavelet neural networks were given. It can be deduced from the examples that it takes full advantage of diversified characteristic information, and improves the diagnosis rate.
文摘In the experiment of combined navigation filtering using wavelet neural network, the initial parameters of the network have the influence of randomness on network convergence and navigation accuracy. A combined navigation filtering method based on wavelet neural network optimized by mind evolution algorithm is proposed. Firstly, the efficient global search ability of the mind evolution algorithm was used to quickly and accurately obtain the initial parameters of the appropriate wavelet neural network, and then the optimized wavelet neural network was applied to directly predict the position and velocity error data. This method is different from the traditional filtering method, while avoiding the drawbacks of the neural network. The simulation experiments with wavelet neural network and GA-wavelet network were carried out. The results show that the proposed method can effectively improve the accuracy of the integrated navigation system and provide a feasible path for combined navigation filtering.
基金supported by the National Natural Science Foundation of China(61803085,61806052,U1713209)the Natural Science Foundation of Jiangsu Province of China(BK20180361)
文摘In this paper, a study of control for an uncertain2-degree of freedom(DOF) helicopter system is given. The2-DOF helicopter is subject to input deadzone and output constraints. In order to cope with system uncertainties and input deadzone, the neural network technique is introduced because of its capability in approximation. In order to update the weights of the neural network, an adaptive control method is utilized to improve the system adaptability. Furthermore, the integral barrier Lyapunov function(IBLF) is adopt in control design to guarantee the condition of output constraints and boundedness of the corresponding tracking errors. The Lyapunov direct method is applied in the control design to analyze system stability and convergence. Finally, numerical simulations are conducted to prove the feasibility and effectiveness of the proposed control based on the model of Quanser's 2-DOF helicopter.
基金Peng Xie acknowledges the support from the China Scholarship Council(Grant no.201804910829).
文摘Parallel multi-thread processing in advanced intelligent processors is the core to realize high-speed and high-capacity signal processing systems.Optical neural network(ONN)has the native advantages of high parallelization,large bandwidth,and low power consumption to meet the demand of big data.Here,we demonstrate the dual-layer ONN with Mach-Zehnder interferometer(MZI)network and nonlinear layer,while the nonlinear activation function is achieved by optical-electronic signal conversion.Two frequency components from the microcomb source carrying digit datasets are simultaneously imposed and intelligently recognized through the ONN.We successfully achieve the digit classification of different frequency components by demultiplexing the output signal and testing power distribution.Efficient parallelization feasibility with wavelength division multiplexing is demonstrated in our high-dimensional ONN.This work provides a high-performance architecture for future parallel high-capacity optical analog computing.
文摘For vehicle integrated navigation systems, real-time estimating states of the dead reckoning (DR) unit is much more difficult than that of the other measuring sensors under indefinite noises and nonlinear characteristics. Compared with the well known, extended Kalman filter (EKF), a recurrent neural network is proposed for the solution, which not only improves the location precision and the adaptive ability of resisting disturbances, but also avoids calculating the analytic derivation and Jacobian matrices of the nonlinear system model. To test the performances of the recurrent neural network, these two methods are used to estimate the state of the vehicle's DR navigation system. Simulation results show that the recurrent neural network is superior to the EKF and is a more ideal filtering method for vehicle DR navigation.
文摘A tightly coupled GPS ( global positioning system )/SINS ( strap down inertial navigation system) based on a GMDH ( group method of data handling) neural network was presented to solve the problem of degraded accuracy for less than four visible GPS satellites with poor signal quality. Positions and velocities of the satellites were predicted by a GMDH neural network, and the pseudo ranges and pseudo range rates received by the GPS receiver were simulated to ensure the regular op eration of the GPS/SINS Kalman filter during outages. In the mathematical simulation a tightly cou pled navigation system with a proposed approach has better navigation accuracy during GPS outages, and the anti jamming ability is strengthened for the tightly coupled navigation system.
基金Supported by the National Natural Science Foundation of China(Nos.41807247,41807229)the Special Fund for Shandong Post-doctoral Innovation Project。
文摘Artificial Neural Network(ANN)models have been extensively applied in the prediction of water resource variables,and Geographical Information System(GIS)includes powerful functions to visualize spatial data.In order to provide an efficient tool for environmental assessment and management that combines the advantages of these two modules,a GIS-based ANN water quality prediction system was developed in the present study.The ANN module and ArcGIS Engine module,along with a dynamic database,were imbedded in the system,which integrates water quality prediction via the ANN model and spatial presentation of the model results.The structure of the ANN model could be modified through the graphical user interface to optimize the model performance.The developed system was applied to a real case study for the prediction of the total phosphorus concentration in the Lake Champlain area.The prediction results were verified with the monitoring data,and the performance of the developed model was further evaluated through graphical techniques and quantitative statistical methods.Overall,the developed system provided satisfactory prediction results,and spatial distribution maps of the predicted results were obtained,which coincided with the monitored values.The developed GIS-based ANN water quality prediction system could serve as an efficient tool for engineers and decision makers.
基金supported by NSFC with Grant No. 61702493, 51707191Science and Technology Planning Project of Guangdong Province with Grant No. 2018B030338001+2 种基金Shenzhen S&T Funding with Grant No. KQJSCX20170731163915914Basic Research Program No. JCYJ20170818164527303, JCYJ20180507182619669SIAT Innovation Program for Excellent Young Researchers with Grant No. 2017001
文摘Driven by continuous scaling of nanoscale semiconductor technologies,the past years have witnessed the progressive advancement of machine learning techniques and applications.Recently,dedicated machine learning accelerators,especially for neural networks,have attracted the research interests of computer architects and VLSI designers.State-of-the-art accelerators increase performance by deploying a huge amount of processing elements,however still face the issue of degraded resource utilization across hybrid and non-standard algorithmic kernels.In this work,we exploit the properties of important neural network kernels for both perception and control to propose a reconfigurable dataflow processor,which adjusts the patterns of data flowing,functionalities of processing elements and on-chip storages according to network kernels.In contrast to stateof-the-art fine-grained data flowing techniques,the proposed coarse-grained dataflow reconfiguration approach enables extensive sharing of computing and storage resources.Three hybrid networks for MobileNet,deep reinforcement learning and sequence classification are constructed and analyzed with customized instruction sets and toolchain.A test chip has been designed and fabricated under UMC 65 nm CMOS technology,with the measured power consumption of 7.51 mW under 100 MHz frequency on a die size of 1.8×1.8 mm^2.
文摘A current-mode MOS neuron circuit with 4-bit programmable weights is presented by using CMOS technology. The weights of the neurcn have high resolution and also can easily be digitally stored. The resolution can be extended into high levels such as 8-bit, etc. by the design methodology presented in this paper. The operational principle of the neuron is discussed. Circuit simulation has been made by use of SPICE II. The results give a good agreement for the design requirements.
文摘This paper presents an integration methodology for ma chining and measuring processes using OMM (On-Machine Measurement) technology b ased on CAD/CAM/CAI integration concept. OMM uses a CNC machining center as a me asuring station by changing the tools into measuring probes such as touch-type, laser and vision. Although the measurement accuracy is not good compared to tha t of the CMM (Coordinate Measuring Machine), there are distinctive advantages us ing OMM in real situation. In this paper, two topics are handled to show the eff ectiveness of the machining and measuring process integration: (1) inspection pl anning strategy for sculptured surface machining and (2) tool path compensation for profile milling process. For the first topic, as a first step, effective mea suring point locations are determined to obtain optimum results for given sampli ng numbers. Two measuring point selection methods are suggested based on the CAD /CAM/CAI integration concept: (1) by the prediction of cutting errors and (2) by considering cutter contact points to avoid the measurement errors caused by cus ps. As a next step, the TSP (Traveling Salesman Problem) algorithm is applied to minimize the probe moving distance. Appropriate simulations and experiments are performed to verify the proposed inspection planning strategy, and the results are analyzed. For the second topic, a methodology for profile milling error comp ensation is presented by using an ANN (Artificial Neural Network) model trained by the inspection database of OMM system. First, geometric and thermal errors of the machining center are compensated using a closed-loop configuration for the improvement of machining and inspection accuracy. The probing errors are also t aken into account. Then, a specimen workpiece is machined and then the machi ning surface error distribution is measured on the machine using touch-type pro be. In order to efficiently analyze the machining errors, two characteristic err or parameters (W err and D err) are defined. Subsequently, these param eters are modeled by applying the RFB (Radial Basis Function) network approach a s an ANN model. Based on the RBF network model, the tool paths are compensated i n order to effectively reduce the errors by employing an iterative algorithm. In order to validate the approaches proposed in this paper, a concrete case of the machining process is taken into account and about 90% of machining error reduction is successfully accomplished through the proposed approaches.
基金supported by the National Natural Science Foundation of China (60604017) the Natural Science Foundation of Zhejiang Province (Y107657).
文摘The global asymptotic stability of cellular neural networks with delays is investigated. Three kinds of time delays have been considered. New delay-dependent stability criteria are proposed and are formulated as the feasibility of some linear matrix inequalities, which can be checked easily by resorting to the recently developed interior-point algorithms. Based on the Finsler Lemma, it is theoretically proved that the proposed stability criteria are less conservative than some existing results.
基金funded in part by the National Natural Science Foundation of China(42271409,62071084)in part by the Heilongjiang Science Foundation Project of China under Grant LH2021D022in part by the Leading Talents Project of the State Ethnic Affairs Commission,and in part by the Fundamental Research Funds in Heilongjiang Provincial Universities of China under Grant 145209149.
文摘In the past,convolutional neural network(CNN)has become one of the most popular deep learning frameworks,and has been widely used in Hyperspectral image classification tasks.Convolution(Conv)in CNN uses filter weights to extract features in local receiving domain,and the weight parameters are shared globally,which more focus on the highfrequency information of the image.Different from Conv,Transformer can obtain the long‐term dependence between long‐distance features through modelling,and adaptively focus on different regions.In addition,Transformer is considered as a low‐pass filter,which more focuses on the low‐frequency information of the image.Considering the complementary characteristics of Conv and Transformer,the two modes can be integrated for full feature extraction.In addition,the most important image features correspond to the discrimination region,while the secondary image features represent important but easily ignored regions,which are also conducive to the classification of HSIs.In this study,a complementary integrated Transformer network(CITNet)for hyperspectral image classification is proposed.Firstly,three‐dimensional convolution(Conv3D)and two‐dimensional convolution(Conv2D)are utilised to extract the shallow semantic information of the image.In order to enhance the secondary features,a channel Gaussian modulation attention module is proposed,which is embedded between Conv3D and Conv2D.This module can not only enhance secondary features,but suppress the most important and least important features.Then,considering the different and complementary characteristics of Conv and Transformer,a complementary integrated Transformer module is designed.Finally,through a large number of experiments,this study evaluates the classification performance of CITNet and several state‐of‐the‐art networks on five common datasets.The experimental results show that compared with these classification networks,CITNet can provide better classification performance.
文摘Development of energy-resources-poor remote rural areas of the world has been discussed by many in the past. Harnessing locally available renewable energy resources as an environmentally friendly option is gaining momentum. Smart Integrated Renewable Energy Systems (SIRES) offer a resilient and economic path to “energize” the area and reach this goal. This paper discusses its intelligent control using neural networks and fuzzy logic.