High-precision and real-time diagnosis of sucker rod pumping system(SRPS)is important for quickly mastering oil well operations.Deep learning-based method for classifying the dynamometer card(DC)of oil wells is an eff...High-precision and real-time diagnosis of sucker rod pumping system(SRPS)is important for quickly mastering oil well operations.Deep learning-based method for classifying the dynamometer card(DC)of oil wells is an efficient diagnosis method.However,the input of the DC as a two-dimensional image into the deep learning framework suffers from low feature utilization and high computational effort.Additionally,different SRPSs in an oil field have various system parameters,and the same SRPS generates different DCs at different moments.Thus,there is heterogeneity in field data,which can dramatically impair the diagnostic accuracy.To solve the above problems,a working condition recognition method based on 4-segment time-frequency signature matrix(4S-TFSM)and deep learning is presented in this paper.First,the 4-segment time-frequency signature(4S-TFS)method that can reduce the computing power requirements is proposed for feature extraction of DC data.Subsequently,the 4S-TFSM is constructed by relative normalization and matrix calculation to synthesize the features of multiple data and solve the problem of data heterogeneity.Finally,a convolutional neural network(CNN),one of the deep learning frameworks,is used to determine the functioning conditions based on the 4S-TFSM.Experiments on field data verify that the proposed diagnostic method based on 4S-TFSM and CNN(4S-TFSM-CNN)can significantly improve the accuracy of working condition recognition with lower computational cost.To the best of our knowledge,this is the first work to discuss the effect of data heterogeneity on the working condition recognition performance of SRPS.展开更多
We propose a novel image segmentation algorithm to tackle the challenge of limited recognition and segmentation performance in identifying welding seam images during robotic intelligent operations.Initially,to enhance...We propose a novel image segmentation algorithm to tackle the challenge of limited recognition and segmentation performance in identifying welding seam images during robotic intelligent operations.Initially,to enhance the capability of deep neural networks in extracting geometric attributes from depth images,we developed a novel deep geometric convolution operator(DGConv).DGConv is utilized to construct a deep local geometric feature extraction module,facilitating a more comprehensive exploration of the intrinsic geometric information within depth images.Secondly,we integrate the newly proposed deep geometric feature module with the Fully Convolutional Network(FCN8)to establish a high-performance deep neural network algorithm tailored for depth image segmentation.Concurrently,we enhance the FCN8 detection head by separating the segmentation and classification processes.This enhancement significantly boosts the network’s overall detection capability.Thirdly,for a comprehensive assessment of our proposed algorithm and its applicability in real-world industrial settings,we curated a line-scan image dataset featuring weld seams.This dataset,named the Standardized Linear Depth Profile(SLDP)dataset,was collected from actual industrial sites where autonomous robots are in operation.Ultimately,we conducted experiments utilizing the SLDP dataset,achieving an average accuracy of 92.7%.Our proposed approach exhibited a remarkable performance improvement over the prior method on the identical dataset.Moreover,we have successfully deployed the proposed algorithm in genuine industrial environments,fulfilling the prerequisites of unmanned robot operations.展开更多
Time-sensitive networks(TSNs)support not only traditional best-effort communications but also deterministic communications,which send each packet at a deterministic time so that the data transmissions of networked con...Time-sensitive networks(TSNs)support not only traditional best-effort communications but also deterministic communications,which send each packet at a deterministic time so that the data transmissions of networked control systems can be precisely scheduled to guarantee hard real-time constraints.No-wait scheduling is suitable for such TSNs and generates the schedules of deterministic communications with the minimal network resources so that all of the remaining resources can be used to improve the throughput of best-effort communications.However,due to inappropriate message fragmentation,the realtime performance of no-wait scheduling algorithms is reduced.Therefore,in this paper,joint algorithms of message fragmentation and no-wait scheduling are proposed.First,a specification for the joint problem based on optimization modulo theories is proposed so that off-the-shelf solvers can be used to find optimal solutions.Second,to improve the scalability of our algorithm,the worst-case delay of messages is analyzed,and then,based on the analysis,a heuristic algorithm is proposed to construct low-delay schedules.Finally,we conduct extensive test cases to evaluate our proposed algorithms.The evaluation results indicate that,compared to existing algorithms,the proposed joint algorithm improves schedulability by up to 50%.展开更多
Inspired by box jellyfish that has distributed and complementary perceptive system,we seek to equip manipulator with a camera and an Inertial Measurement Unit(IMU)to perceive ego motion and surrounding unstructured en...Inspired by box jellyfish that has distributed and complementary perceptive system,we seek to equip manipulator with a camera and an Inertial Measurement Unit(IMU)to perceive ego motion and surrounding unstructured environment.Before robot perception,a reliable and high-precision calibration between camera,IMU and manipulator is a critical prerequisite.This paper introduces a novel calibration system.First,we seek to correlate the spatial relationship between the sensing units and manipulator in a joint framework.Second,the manipulator moving trajectory is elaborately designed in a spiral pattern that enables full excitations on yaw-pitch-roll rotations and x-y-z translations in a repeatable and consistent manner.The calibration has been evaluated on our collected visual inertial-manipulator dataset.The systematic comparisons and analysis indicate the consistency,precision and effectiveness of our proposed calibration method.展开更多
The accurate and intelligent identification of the working conditions of a sucker-rod pumping system is necessary. As onshore oil extraction gradually enters its mid-to late-stage, the cost required to train a deep le...The accurate and intelligent identification of the working conditions of a sucker-rod pumping system is necessary. As onshore oil extraction gradually enters its mid-to late-stage, the cost required to train a deep learning working condition recognition model for pumping wells by obtaining enough new working condition samples is expensive. For the few-shot problem and large calculation issues of new working conditions of oil wells, a working condition recognition method for pumping unit wells based on a 4-dimensional time-frequency signature (4D-TFS) and meta-learning convolutional shrinkage neural network (ML-CSNN) is proposed. First, the measured pumping unit well workup data are converted into 4D-TFS data, and the initial feature extraction task is performed while compressing the data. Subsequently, a convolutional shrinkage neural network (CSNN) with a specific structure that can ablate low-frequency features is designed to extract working conditions features. Finally, a meta-learning fine-tuning framework for learning the network parameters that are susceptible to task changes is merged into the CSNN to solve the few-shot issue. The results of the experiments demonstrate that the trained ML-CSNN has good recognition accuracy and generalization ability for few-shot working condition recognition. More specifically, in the case of lower computational complexity, only few-shot samples are needed to fine-tune the network parameters, and the model can be quickly adapted to new classes of well conditions.展开更多
Large-scale solar sails can provide power to spacecraft for deep space exploration.A new type of telescopic tubular mast(TTM)driven by a bistable carbon fiber-reinforced polymer tube was designed in this study to solv...Large-scale solar sails can provide power to spacecraft for deep space exploration.A new type of telescopic tubular mast(TTM)driven by a bistable carbon fiber-reinforced polymer tube was designed in this study to solve the problem of contact between the sail membrane and the spacecraft under light pressure.Compared with the traditional TTM,it has a small size,light weight,high extension ratio,and simple structure.The anti-blossoming and self-unlocking structure of the proposed TTM was described.We aimed to simplify the TTM with a complex structure into a beam model with equal linear mass density,and the simulation results showed good consistency.The dynamic equation was derived based on the equivalent model,and the effects of different factors on the vibration characteristics of the TTM were analyzed.The performance parameters were optimized based on a multiobjective genetic algorithm,and prototype production and load experiments were conducted.The results show that the advantages of the new TTM can complete the deployment of large-scale solar sails,which is valuable for future deep space exploration.展开更多
As 3D acquisition technology develops and 3D sensors become increasingly affordable,large quantities of 3D point cloud data are emerging.How to effectively learn and extract the geometric features from these point clo...As 3D acquisition technology develops and 3D sensors become increasingly affordable,large quantities of 3D point cloud data are emerging.How to effectively learn and extract the geometric features from these point clouds has become an urgent problem to be solved.The point cloud geometric information is hidden in disordered,unstructured points,making point cloud analysis a very challenging problem.To address this problem,we propose a novel network framework,called Tree Graph Network(TGNet),which can sample,group,and aggregate local geometric features.Specifically,we construct a Tree Graph by explicit rules,which consists of curves extending in all directions in point cloud feature space,and then aggregate the features of the graph through a cross-attention mechanism.In this way,we incorporate more point cloud geometric structure information into the representation of local geometric features,which makes our network perform better.Our model performs well on several basic point clouds processing tasks such as classification,segmentation,and normal estimation,demonstrating the effectiveness and superiority of our network.Furthermore,we provide ablation experiments and visualizations to better understand our network.展开更多
Cell mechanics is essential to cell development and function,and its dynamics evolution reflects the physiological state of cells.Here,we investigate the dynamical mechanical properties of single cells under various d...Cell mechanics is essential to cell development and function,and its dynamics evolution reflects the physiological state of cells.Here,we investigate the dynamical mechanical properties of single cells under various drug conditions,and present two mathematical approaches to quantitatively characterizing the cell physiological state.It is demonstrated that the cellular mechanical properties upon the drug action increase over time and tend to saturate,and can be mathematically characterized by a linear timeinvariant dynamical model.It is shown that the transition matrices of dynamical cell systems significantly improve the classification accuracies of the cells under different drug actions.Furthermore,it is revealed that there exists a positive linear correlation between the cytoskeleton density and the cellular mechanical properties,and the physiological state of a cell in terms of its cytoskeleton density can be predicted from its mechanical properties by a linear regression model.This study builds a relationship between the cellular mechanical properties and the cellular physiological state,adding information for evaluating drug efficacy.展开更多
Rehabilitation using exoskeleton robots can effectively remediate dysfunction and restore post-stroke survivors’ physical ability. However, low kinematic compatibility and poor self-participation of post-stroke patie...Rehabilitation using exoskeleton robots can effectively remediate dysfunction and restore post-stroke survivors’ physical ability. However, low kinematic compatibility and poor self-participation of post-stroke patients in rehabilitation restrict the outcomes of exoskeleton-based therapy. The study presents an Unpowered Shoulder Complex Exoskeleton (USCE), consisting of Shoulder Girdle Mechanism (SGM), Ball-and-Socket Joint Mechanism (BSM), Gravity Compensating Mechanism (GCM) and Adjustable Alignment Design (AAD), to achieve self-rehabilitation of shoulder via energy transfer from the healthy upper limb to the affected counterpart of post-stroke hemiplegic patients. The SGM and AAD are designed to improve the kinematic compatibility by compensating for displacements of the glenohumeral joint with the adaptable size of USCE for different wearers. The BSM and GCM can transfer the body movement and energy from the healthy half of the body to the affected side without external energy input and enhance the self-participation with sick posture correction. The experimental results show that the USCE can provide high kinematic compatibility with 90.9% movement similarity between human and exoskeleton. Meanwhile, the motion ability of a post-stroke patient’s affected limb can be increased through energy transfer. It is expected that USCE can improve outcomes of home-based self-rehabilitation.展开更多
The widespread usage of Cyber Physical Systems(CPSs)generates a vast volume of time series data,and precisely determining anomalies in the data is critical for practical production.Autoencoder is the mainstream method...The widespread usage of Cyber Physical Systems(CPSs)generates a vast volume of time series data,and precisely determining anomalies in the data is critical for practical production.Autoencoder is the mainstream method for time series anomaly detection,and the anomaly is judged by reconstruction error.However,due to the strong generalization ability of neural networks,some abnormal samples close to normal samples may be judged as normal,which fails to detect the abnormality.In addition,the dataset rarely provides sufficient anomaly labels.This research proposes an unsupervised anomaly detection approach based on adversarial memory autoencoders for multivariate time series to solve the above problem.Firstly,an encoder encodes the input data into low-dimensional space to acquire a feature vector.Then,a memory module is used to learn the feature vector’s prototype patterns and update the feature vectors.The updating process allows partial forgetting of information to prevent model overgeneralization.After that,two decoders reconstruct the input data.Finally,this research uses the Peak Over Threshold(POT)method to calculate the threshold to determine anomalous samples from normal samples.This research uses a two-stage adversarial training strategy during model training to enlarge the gap between the reconstruction error of normal and abnormal samples.The proposed method achieves significant anomaly detection results on synthetic and real datasets from power systems,water treatment plants,and computer clusters.The F1 score reached an average of 0.9196 on the five datasets,which is 0.0769 higher than the best baseline method.展开更多
Terahertz(THz)imaging has drawn significant attention because THz wave has a unique capability to transient,ultrawide spectrum and low photon energy.However,the low resolution has always been a problem due to its long...Terahertz(THz)imaging has drawn significant attention because THz wave has a unique capability to transient,ultrawide spectrum and low photon energy.However,the low resolution has always been a problem due to its long wavelength,limiting their application of fields practical use.In this paper,we proposed a complex one-shot super-resolution(COSSR)framework based on a complex convolution neural network to restore superior THz images at 0.35 times wavelength by extracting features directly from a reference measured sample and groundtruth without the measured PSF.Compared with real convolution neural network-based approaches and complex zero-shot super-resolution(CZSSR),COSSR delivers at least 6.67,0.003,and 6.96%superior higher imaging efficacy in terms of peak signal to noise ratio(PSNR),mean square error(MSE),and structural similarity index measure(SSIM),respectively,for the analyzed data.Additionally,the proposed method is experimentally demonstrated to have a good generalization and to perform well on measured data.The COSSR provides a new pathway for THz imaging super-resolution(SR)reconstruction below the diffraction limit.展开更多
Feature extraction plays an important role in constructing artificial intel-ligence(AI)models of industrial control systems(ICSs).Three challenges in this field are learning effective representation from high-dimensio...Feature extraction plays an important role in constructing artificial intel-ligence(AI)models of industrial control systems(ICSs).Three challenges in this field are learning effective representation from high-dimensional features,data heterogeneity,and data noise due to the diversity of data dimensions,formats and noise of sensors,controllers and actuators.Hence,a novel unsupervised learn-ing autoencoder model is proposed for ICS data in this paper.Although traditional methods only capture the linear correlations of ICS features,our deep industrial representation learning model(DIRL)based on a convolutional neural network can mine high-order features,thus solving the problem of high-dimensional and heterogeneous ICS data.In addition,an unsupervised denoising autoencoder is introduced for noisy ICS data in DIRL.Training the denoising autoencoder allows the model to better mitigate the sensor noise problem.In this way,the represen-tative features learned by DIRL could help to evaluate the safety state of ICSs more effectively.We tested our model with absolute and relative accuracy experi-ments on two large-scale ICS datasets.Compared with other popular methods,DIRL showed advantages in four common indicators of AI algorithms:accuracy,precision,recall,and F1-score.This study contributes to the effective analysis of large-scale ICS data,which promotes the stable operation of ICSs.展开更多
Rapid online analysis of liquid slag is essential for optimizing the quality and energy efficiency of steel production. To investigate the key factors that affect the online measurement of refined slag using laser-ind...Rapid online analysis of liquid slag is essential for optimizing the quality and energy efficiency of steel production. To investigate the key factors that affect the online measurement of refined slag using laser-induced breakdown spectroscopy(LIBS), this study examined the effects of slag composition and temperature on the intensity and stability of the LIBS spectra. The experimental temperature was controlled at three levels: 1350℃, 1400℃, and 1450℃. The results showed that slag composition and temperature significantly affected the intensity and stability of the LIBS spectra. Increasing the Fe content and temperature in the slag reduces its viscosity, resulting in an enhanced intensity and stability of the LIBS spectra. Additionally, 42 refined slag samples were quantitatively analyzed for Fe, Si, Ca, Mg, Al, and Mn at 1350℃, 1400℃, and 1450℃.The normalized full spectrum combined with partial least squares(PLS) quantification modeling was used, using the Ca Ⅱ 317.91 nm spectral line as an internal standard. The results show that using the internal standard normalization method can significantly reduce the influence of spectral fluctuations. Meanwhile, a temperature of 1450℃ has been found to yield superior results compared to both 1350℃ and 1400℃, and it is advantageous to conduct a quantitative analysis of the slag when it is in a “water-like” state with low viscosity.展开更多
Electromyography(EMG)has already been broadly used in human-machine interaction(HMI)applications.Determining how to decode the information inside EMG signals robustly and accurately is a key problem for which we urgen...Electromyography(EMG)has already been broadly used in human-machine interaction(HMI)applications.Determining how to decode the information inside EMG signals robustly and accurately is a key problem for which we urgently need a solution.Recently,many EMG pattern recognition tasks have been addressed using deep learning methods.In this paper,we analyze recent papers and present a literature review describing the role that deep learning plays in EMG-based HMI.An overview of typical network structures and processing schemes will be provided.Recent progress in typical tasks such as movement classification,joint angle prediction,and force/torque estimation will be introduced.New issues,including multimodal sensing,inter-subject/inter-session,and robustness toward disturbances will be discussed.We attempt to provide a comprehensive analysis of current research by discussing the advantages,challenges,and opportunities brought by deep learning.We hope that deep learning can aid in eliminating factors that hinder the development of EMG-based HMI systems.Furthermore,possible future directions will be presented to pave the way for future research.展开更多
In the spectral analysis of laser-induced breakdown spectroscopy,abundant characteristic spectral lines and severe interference information exist simultaneously in the original spectral data.Here,a feature selection m...In the spectral analysis of laser-induced breakdown spectroscopy,abundant characteristic spectral lines and severe interference information exist simultaneously in the original spectral data.Here,a feature selection method called recursive feature elimination based on ridge regression(Ridge-RFE)for the original spectral data is recommended to make full use of the valid information of spectra.In the Ridge-RFE method,the absolute value of the ridge regression coefficient was used as a criterion to screen spectral characteristic,the feature with the absolute value of minimum weight in the input subset features was removed by recursive feature elimination(RFE),and the selected features were used as inputs of the partial least squares regression(PLS)model.The Ridge-RFE method based PLS model was used to measure the Fe,Si,Mg,Cu,Zn and Mn for 51 aluminum alloy samples,and the results showed that the root mean square error of prediction decreased greatly compared to the PLS model with full spectrum as input.The overall results demonstrate that the Ridge-RFE method is more efficient to extract the redundant features,make PLS model for better quantitative analysis results and improve model generalization ability.展开更多
Range is an important factor to the design of autonomous underwater vehicles (AUVs), while drag reduction efforts are pursued, the investigation of body-propeller interaction is another vital consideration. We present...Range is an important factor to the design of autonomous underwater vehicles (AUVs), while drag reduction efforts are pursued, the investigation of body-propeller interaction is another vital consideration. We present a numerical and experimental study of the hull-propeller interaction for deeply submerged underwater vehicles, using a proportional-integral- derivative (PID) controller method to estimate self-propulsion point in CFD environment. The hydrodynamic performance of hull and propeller at the balance state when the AUV sails at a fixed depth is investigated, using steady RANS solver of Star-CCM+. The proposed steady RANS solver takes only hours to reach a reasonable solution. It is more time efficient than unsteady simulations which takes days or weeks, as well as huge consumption of computing resources. Explorer 1000, a long range AUV developed by Shenyang Institute of Automation, Chinese Academy of Sciences, was studied as an object, and self-propulsion point, thrust deduction, wake fraction and hull efficiency were analyzed by using the proposed RANS method. Behind-hull performance of the selected propeller MAU4-40, as well as the hull-propeller interaction, was obtained from the computed hydrodynamic forces. The numerical results are in good qualitative and quantitative agreement with the experimental results obtained in the Qiandao Lake of Zhejiang province, China.展开更多
Supernumerary robotic limbs(SRLs) are a new type of wearable human auxiliary equipment, which is currently a hot research topic in the world. SRLs have broad applications in many fields, and will provide a reference a...Supernumerary robotic limbs(SRLs) are a new type of wearable human auxiliary equipment, which is currently a hot research topic in the world. SRLs have broad applications in many fields, and will provide a reference and technical support for the realization of human-robot collaboration and integration,while playing an important role in improving social security and public services. In this paper, representative SRLs are summarized from the aspects of related literature analysis,research status, ontology structure design, control and driving,sensing and perception, and application fields. This paper also analyzes and summarizes the current technical challenges faced by SRLs, and reviews development progress and key technologies,thus giving a prospect of future technical development trends.展开更多
This paper presents autonomous docking of an inhouse built resident Remotely Operated Vehicle(ROV),called Rover ROV,through acoustic guided techniques.A novel cage-type docking station has been developed.The docking s...This paper presents autonomous docking of an inhouse built resident Remotely Operated Vehicle(ROV),called Rover ROV,through acoustic guided techniques.A novel cage-type docking station has been developed.The docking station can be placed on a deep-sea lander,taking the Rover ROV to the seafloor.Instead of using vision-based pose estimation techniques and expensive navigation sensors,the Rover ROV docking adopts an ultra-short baseline(USBL)and low-cost inertial sensors to build an adaptive fault-tolerant integrated navigation system.To solve the problem of sonar-based failure positioning,the measurement residuals are exploited to detect measurement faults.Then,an adaptation scheme for estimating the statistical characteristics of noise in real-time is proposed,which can provide robust and smooth positioning results.It is more suitable for a compact and low-cost deep-sea resident ROV.Field experiments have been conducted successfully in the Qiandao Lake and the South China Sea area with a depth of 3000 m,respectively.The experimental results show that the functionality of autonomous docking has been achieved.Under the guidance of the navigation system,the Rover ROV can autonomously and efficiently return to the docking station within a range of 100 m even when the amounts of outliers exist in the acoustic positioning data.These achievements can be applied to current ROVs by an easy retrofit.展开更多
National navies equip their submarines with Autonomous Underwater Vehicle(AUV)technology.It has become an important component of submarine development in technologically-advanced countries.Employing advanced and relia...National navies equip their submarines with Autonomous Underwater Vehicle(AUV)technology.It has become an important component of submarine development in technologically-advanced countries.Employing advanced and reliable recovery systems directly improves the safety and operational efficiency of submarines equipped with AUVs.In this paper,based on aerial refueling technology,a cone-shaped recovery system with two different guiding covers(closed structure and frame structure)is applied to the submarine.By taking the Suboff model as the research object,STAR-CCM was used to study the influence of the installation position of the recovery system,and the length of the rigid rod,on the Suboff model.It was found that when the recovery system is installed in the middle and rear of the Suboff model at the same velocity and the same length of the rigid rod,the Suboff model has the good stability and less drag.It experiences the largest drag when being installed in the front of the rigid rod.Moreover,when the recovery system is installed in the front and middle of the rigid rod,the drag increases as its length increases,and the lift decreases as its length increases.Compared with the closed structure guiding cover,the Suboff model will have less drag and better stability when the recovery system uses the frame structure guiding cover.Besides,the deflection and vibration of the rigid rod were also analyzed via mathematical theory.展开更多
Variable Stiffness Actuator(VSA)is the core mechanism to achieve physical human–robot interaction,which is an inevitable development trend in robotic.The existing variable stiffness actuators are basically single deg...Variable Stiffness Actuator(VSA)is the core mechanism to achieve physical human–robot interaction,which is an inevitable development trend in robotic.The existing variable stiffness actuators are basically single degree-of-freedom(DOF)rotating joints,which are achieving multi-DOF motion by cascades and resulting in complex robot body structures.In this paper,an integrated 2-DOF actuator with variable stiffness is proposed,which could be used for bionic wrist joints or shoulder joints.The 2-DOF motion is coupling in one universal joint,which is different from the way of single DOF actuators cascade.Based on the 2-DOF orthogonal motion generated by the spherical wrist parallel mechanism,the stiffness could be adjusted by varying the effective length of the springs,which is uniformly distributed in the variable stiffness unit.The variable stiffness principle,the model design,and theoretical analysis of the VSA are discussed in this work.The independence of adjusting the equilibrium position and stiffness of the actuator is validated by experiments.The results show that the measured actuator characteristics are sufficiently matched the theoretical values.In the future,VSA could be used in biped robot or robotic arm,ensuring the safety of human–robot interaction.展开更多
基金We would like to thank the associate editor and the reviewers for their constructive comments.This work was supported in part by the National Natural Science Foundation of China under Grant 62203234in part by the State Key Laboratory of Robotics of China under Grant 2023-Z03+1 种基金in part by the Natural Science Foundation of Liaoning Province under Grant 2023-BS-025in part by the Research Program of Liaoning Liaohe Laboratory under Grant LLL23ZZ-02-02.
文摘High-precision and real-time diagnosis of sucker rod pumping system(SRPS)is important for quickly mastering oil well operations.Deep learning-based method for classifying the dynamometer card(DC)of oil wells is an efficient diagnosis method.However,the input of the DC as a two-dimensional image into the deep learning framework suffers from low feature utilization and high computational effort.Additionally,different SRPSs in an oil field have various system parameters,and the same SRPS generates different DCs at different moments.Thus,there is heterogeneity in field data,which can dramatically impair the diagnostic accuracy.To solve the above problems,a working condition recognition method based on 4-segment time-frequency signature matrix(4S-TFSM)and deep learning is presented in this paper.First,the 4-segment time-frequency signature(4S-TFS)method that can reduce the computing power requirements is proposed for feature extraction of DC data.Subsequently,the 4S-TFSM is constructed by relative normalization and matrix calculation to synthesize the features of multiple data and solve the problem of data heterogeneity.Finally,a convolutional neural network(CNN),one of the deep learning frameworks,is used to determine the functioning conditions based on the 4S-TFSM.Experiments on field data verify that the proposed diagnostic method based on 4S-TFSM and CNN(4S-TFSM-CNN)can significantly improve the accuracy of working condition recognition with lower computational cost.To the best of our knowledge,this is the first work to discuss the effect of data heterogeneity on the working condition recognition performance of SRPS.
基金This work was supported by the National Natural Science Foundation of China(Grant No.U20A20197).
文摘We propose a novel image segmentation algorithm to tackle the challenge of limited recognition and segmentation performance in identifying welding seam images during robotic intelligent operations.Initially,to enhance the capability of deep neural networks in extracting geometric attributes from depth images,we developed a novel deep geometric convolution operator(DGConv).DGConv is utilized to construct a deep local geometric feature extraction module,facilitating a more comprehensive exploration of the intrinsic geometric information within depth images.Secondly,we integrate the newly proposed deep geometric feature module with the Fully Convolutional Network(FCN8)to establish a high-performance deep neural network algorithm tailored for depth image segmentation.Concurrently,we enhance the FCN8 detection head by separating the segmentation and classification processes.This enhancement significantly boosts the network’s overall detection capability.Thirdly,for a comprehensive assessment of our proposed algorithm and its applicability in real-world industrial settings,we curated a line-scan image dataset featuring weld seams.This dataset,named the Standardized Linear Depth Profile(SLDP)dataset,was collected from actual industrial sites where autonomous robots are in operation.Ultimately,we conducted experiments utilizing the SLDP dataset,achieving an average accuracy of 92.7%.Our proposed approach exhibited a remarkable performance improvement over the prior method on the identical dataset.Moreover,we have successfully deployed the proposed algorithm in genuine industrial environments,fulfilling the prerequisites of unmanned robot operations.
基金partially supported by National Key Research and Development Program of China(2018YFB1700200)National Natural Science Foundation of China(61972389,61903356,61803368,U1908212)+2 种基金Youth Innovation Promotion Association of the Chinese Academy of Sciences,National Science and Technology Major Project(2017ZX02101007-004)Liaoning Provincial Natural Science Foundation of China(2020-MS-034,2019-YQ-09)China Postdoctoral Science Foundation(2019M661156)。
文摘Time-sensitive networks(TSNs)support not only traditional best-effort communications but also deterministic communications,which send each packet at a deterministic time so that the data transmissions of networked control systems can be precisely scheduled to guarantee hard real-time constraints.No-wait scheduling is suitable for such TSNs and generates the schedules of deterministic communications with the minimal network resources so that all of the remaining resources can be used to improve the throughput of best-effort communications.However,due to inappropriate message fragmentation,the realtime performance of no-wait scheduling algorithms is reduced.Therefore,in this paper,joint algorithms of message fragmentation and no-wait scheduling are proposed.First,a specification for the joint problem based on optimization modulo theories is proposed so that off-the-shelf solvers can be used to find optimal solutions.Second,to improve the scalability of our algorithm,the worst-case delay of messages is analyzed,and then,based on the analysis,a heuristic algorithm is proposed to construct low-delay schedules.Finally,we conduct extensive test cases to evaluate our proposed algorithms.The evaluation results indicate that,compared to existing algorithms,the proposed joint algorithm improves schedulability by up to 50%.
基金supported by the National Natural Science Foundation of China(61903357,61902299,62022088)the International Partnership Program of Chinese Academy of Sciences(173321KYSB20200002)+2 种基金Liaoning Provincial Natural Science Foundation of China(2020-MS-032,2021JH6/10500114,2020JH2/10500002)Guangzhou Science and Technology Planning Project(202102021300)China Postdoctoral Science Foundation(2019TQ0239,2019M663636).
文摘Inspired by box jellyfish that has distributed and complementary perceptive system,we seek to equip manipulator with a camera and an Inertial Measurement Unit(IMU)to perceive ego motion and surrounding unstructured environment.Before robot perception,a reliable and high-precision calibration between camera,IMU and manipulator is a critical prerequisite.This paper introduces a novel calibration system.First,we seek to correlate the spatial relationship between the sensing units and manipulator in a joint framework.Second,the manipulator moving trajectory is elaborately designed in a spiral pattern that enables full excitations on yaw-pitch-roll rotations and x-y-z translations in a repeatable and consistent manner.The calibration has been evaluated on our collected visual inertial-manipulator dataset.The systematic comparisons and analysis indicate the consistency,precision and effectiveness of our proposed calibration method.
基金supported in part by the National Natural Science Foundation of China under Grant U1908212,62203432 and 92067205in part by the State Key Laboratory of Robotics of China under Grant 2023-Z03 and 2023-Z15in part by the Natural Science Foundation of Liaoning Province under Grant 2020-KF-11-02.
文摘The accurate and intelligent identification of the working conditions of a sucker-rod pumping system is necessary. As onshore oil extraction gradually enters its mid-to late-stage, the cost required to train a deep learning working condition recognition model for pumping wells by obtaining enough new working condition samples is expensive. For the few-shot problem and large calculation issues of new working conditions of oil wells, a working condition recognition method for pumping unit wells based on a 4-dimensional time-frequency signature (4D-TFS) and meta-learning convolutional shrinkage neural network (ML-CSNN) is proposed. First, the measured pumping unit well workup data are converted into 4D-TFS data, and the initial feature extraction task is performed while compressing the data. Subsequently, a convolutional shrinkage neural network (CSNN) with a specific structure that can ablate low-frequency features is designed to extract working conditions features. Finally, a meta-learning fine-tuning framework for learning the network parameters that are susceptible to task changes is merged into the CSNN to solve the few-shot issue. The results of the experiments demonstrate that the trained ML-CSNN has good recognition accuracy and generalization ability for few-shot working condition recognition. More specifically, in the case of lower computational complexity, only few-shot samples are needed to fine-tune the network parameters, and the model can be quickly adapted to new classes of well conditions.
基金Supported by National Key R&D Program of China (Grant No.2018YFB1304600)National Natural Science Foundation of China (Grant No.51905527)+1 种基金CAS Interdisciplinary Innovation Team of China (Grant No.JCTD-2018-11)State Key Laboratory of Robotics Foundation of China (Grant No.Y91Z0303)。
文摘Large-scale solar sails can provide power to spacecraft for deep space exploration.A new type of telescopic tubular mast(TTM)driven by a bistable carbon fiber-reinforced polymer tube was designed in this study to solve the problem of contact between the sail membrane and the spacecraft under light pressure.Compared with the traditional TTM,it has a small size,light weight,high extension ratio,and simple structure.The anti-blossoming and self-unlocking structure of the proposed TTM was described.We aimed to simplify the TTM with a complex structure into a beam model with equal linear mass density,and the simulation results showed good consistency.The dynamic equation was derived based on the equivalent model,and the effects of different factors on the vibration characteristics of the TTM were analyzed.The performance parameters were optimized based on a multiobjective genetic algorithm,and prototype production and load experiments were conducted.The results show that the advantages of the new TTM can complete the deployment of large-scale solar sails,which is valuable for future deep space exploration.
基金supported by the National Natural Science Foundation of China (Grant Nos.91948203,52075532).
文摘As 3D acquisition technology develops and 3D sensors become increasingly affordable,large quantities of 3D point cloud data are emerging.How to effectively learn and extract the geometric features from these point clouds has become an urgent problem to be solved.The point cloud geometric information is hidden in disordered,unstructured points,making point cloud analysis a very challenging problem.To address this problem,we propose a novel network framework,called Tree Graph Network(TGNet),which can sample,group,and aggregate local geometric features.Specifically,we construct a Tree Graph by explicit rules,which consists of curves extending in all directions in point cloud feature space,and then aggregate the features of the graph through a cross-attention mechanism.In this way,we incorporate more point cloud geometric structure information into the representation of local geometric features,which makes our network perform better.Our model performs well on several basic point clouds processing tasks such as classification,segmentation,and normal estimation,demonstrating the effectiveness and superiority of our network.Furthermore,we provide ablation experiments and visualizations to better understand our network.
基金This work was supported by the National Natural Science Foundation of China(Grant Nos:U1908215,61925307,62003338,and 61933008)CAS Project for Young Scientists in Basic Research(Grant No:YSBR-041)+2 种基金Liaoning Revitalization Talents Program(Grant No:XLYC2002014)Natural Science Foundation of Liaoning Province of China(Grant No:2020-ZLLH-47)Joint fund of Science&Technology Department of Liaoning Province and State Key Laboratory of Robotics,China(Grant No:2019-KF-01-01).
文摘Cell mechanics is essential to cell development and function,and its dynamics evolution reflects the physiological state of cells.Here,we investigate the dynamical mechanical properties of single cells under various drug conditions,and present two mathematical approaches to quantitatively characterizing the cell physiological state.It is demonstrated that the cellular mechanical properties upon the drug action increase over time and tend to saturate,and can be mathematically characterized by a linear timeinvariant dynamical model.It is shown that the transition matrices of dynamical cell systems significantly improve the classification accuracies of the cells under different drug actions.Furthermore,it is revealed that there exists a positive linear correlation between the cytoskeleton density and the cellular mechanical properties,and the physiological state of a cell in terms of its cytoskeleton density can be predicted from its mechanical properties by a linear regression model.This study builds a relationship between the cellular mechanical properties and the cellular physiological state,adding information for evaluating drug efficacy.
基金supported by the National Key R&D Program of China(Grant No.2016YFE0206200)the National Natural Science Foundation of China(Grant Nos.U1908215,91848201,61821005 and 61973316)Liaoning Revitalizaiton Talents Program(Grant No.XLYC2002014).
文摘Rehabilitation using exoskeleton robots can effectively remediate dysfunction and restore post-stroke survivors’ physical ability. However, low kinematic compatibility and poor self-participation of post-stroke patients in rehabilitation restrict the outcomes of exoskeleton-based therapy. The study presents an Unpowered Shoulder Complex Exoskeleton (USCE), consisting of Shoulder Girdle Mechanism (SGM), Ball-and-Socket Joint Mechanism (BSM), Gravity Compensating Mechanism (GCM) and Adjustable Alignment Design (AAD), to achieve self-rehabilitation of shoulder via energy transfer from the healthy upper limb to the affected counterpart of post-stroke hemiplegic patients. The SGM and AAD are designed to improve the kinematic compatibility by compensating for displacements of the glenohumeral joint with the adaptable size of USCE for different wearers. The BSM and GCM can transfer the body movement and energy from the healthy half of the body to the affected side without external energy input and enhance the self-participation with sick posture correction. The experimental results show that the USCE can provide high kinematic compatibility with 90.9% movement similarity between human and exoskeleton. Meanwhile, the motion ability of a post-stroke patient’s affected limb can be increased through energy transfer. It is expected that USCE can improve outcomes of home-based self-rehabilitation.
基金supported by the National Natural Science Foundation of China(62203431)。
文摘The widespread usage of Cyber Physical Systems(CPSs)generates a vast volume of time series data,and precisely determining anomalies in the data is critical for practical production.Autoencoder is the mainstream method for time series anomaly detection,and the anomaly is judged by reconstruction error.However,due to the strong generalization ability of neural networks,some abnormal samples close to normal samples may be judged as normal,which fails to detect the abnormality.In addition,the dataset rarely provides sufficient anomaly labels.This research proposes an unsupervised anomaly detection approach based on adversarial memory autoencoders for multivariate time series to solve the above problem.Firstly,an encoder encodes the input data into low-dimensional space to acquire a feature vector.Then,a memory module is used to learn the feature vector’s prototype patterns and update the feature vectors.The updating process allows partial forgetting of information to prevent model overgeneralization.After that,two decoders reconstruct the input data.Finally,this research uses the Peak Over Threshold(POT)method to calculate the threshold to determine anomalous samples from normal samples.This research uses a two-stage adversarial training strategy during model training to enlarge the gap between the reconstruction error of normal and abnormal samples.The proposed method achieves significant anomaly detection results on synthetic and real datasets from power systems,water treatment plants,and computer clusters.The F1 score reached an average of 0.9196 on the five datasets,which is 0.0769 higher than the best baseline method.
基金"XingLiaoYingCai"Talents of Liaoning Province,China(Grant No.XLYC2007074)Shenyang Young and Middle-aged Science and Technology Innovation Talent Support Program(Grant No.RC200512)+1 种基金Project supported by“XingLiaoYingCai"Talents of Liaoning Province,China(Grant No.XLYC2007074)Shenyang Young and Middle-aged Science and Technology Innovation Talent Support Program(Grant No.RC200512),。
文摘Terahertz(THz)imaging has drawn significant attention because THz wave has a unique capability to transient,ultrawide spectrum and low photon energy.However,the low resolution has always been a problem due to its long wavelength,limiting their application of fields practical use.In this paper,we proposed a complex one-shot super-resolution(COSSR)framework based on a complex convolution neural network to restore superior THz images at 0.35 times wavelength by extracting features directly from a reference measured sample and groundtruth without the measured PSF.Compared with real convolution neural network-based approaches and complex zero-shot super-resolution(CZSSR),COSSR delivers at least 6.67,0.003,and 6.96%superior higher imaging efficacy in terms of peak signal to noise ratio(PSNR),mean square error(MSE),and structural similarity index measure(SSIM),respectively,for the analyzed data.Additionally,the proposed method is experimentally demonstrated to have a good generalization and to perform well on measured data.The COSSR provides a new pathway for THz imaging super-resolution(SR)reconstruction below the diffraction limit.
基金This study is supported by The National Key Research and Development Program of China:“Key measurement and control equipment with built-in information security functions”(Grant No.2018YFB2004200)Independent Subject of State Key Laboratory of Robotics“Research on security industry network construction technology for 5G communication”(No.2022-Z13).
文摘Feature extraction plays an important role in constructing artificial intel-ligence(AI)models of industrial control systems(ICSs).Three challenges in this field are learning effective representation from high-dimensional features,data heterogeneity,and data noise due to the diversity of data dimensions,formats and noise of sensors,controllers and actuators.Hence,a novel unsupervised learn-ing autoencoder model is proposed for ICS data in this paper.Although traditional methods only capture the linear correlations of ICS features,our deep industrial representation learning model(DIRL)based on a convolutional neural network can mine high-order features,thus solving the problem of high-dimensional and heterogeneous ICS data.In addition,an unsupervised denoising autoencoder is introduced for noisy ICS data in DIRL.Training the denoising autoencoder allows the model to better mitigate the sensor noise problem.In this way,the represen-tative features learned by DIRL could help to evaluate the safety state of ICSs more effectively.We tested our model with absolute and relative accuracy experi-ments on two large-scale ICS datasets.Compared with other popular methods,DIRL showed advantages in four common indicators of AI algorithms:accuracy,precision,recall,and F1-score.This study contributes to the effective analysis of large-scale ICS data,which promotes the stable operation of ICSs.
基金financially supported by the National Key R&D Program Projects of China (No.2021YFB3202402)National Natural Science Foundation of China (No.62173321)。
文摘Rapid online analysis of liquid slag is essential for optimizing the quality and energy efficiency of steel production. To investigate the key factors that affect the online measurement of refined slag using laser-induced breakdown spectroscopy(LIBS), this study examined the effects of slag composition and temperature on the intensity and stability of the LIBS spectra. The experimental temperature was controlled at three levels: 1350℃, 1400℃, and 1450℃. The results showed that slag composition and temperature significantly affected the intensity and stability of the LIBS spectra. Increasing the Fe content and temperature in the slag reduces its viscosity, resulting in an enhanced intensity and stability of the LIBS spectra. Additionally, 42 refined slag samples were quantitatively analyzed for Fe, Si, Ca, Mg, Al, and Mn at 1350℃, 1400℃, and 1450℃.The normalized full spectrum combined with partial least squares(PLS) quantification modeling was used, using the Ca Ⅱ 317.91 nm spectral line as an internal standard. The results show that using the internal standard normalization method can significantly reduce the influence of spectral fluctuations. Meanwhile, a temperature of 1450℃ has been found to yield superior results compared to both 1350℃ and 1400℃, and it is advantageous to conduct a quantitative analysis of the slag when it is in a “water-like” state with low viscosity.
基金supported in part by the National Natural Science Foundation of China(U181321461773369+2 种基金61903360)the Selfplanned Project of the State Key Laboratory of Robotics(2020-Z12)China Postdoctoral Science Foundation funded project(2019M661155)。
文摘Electromyography(EMG)has already been broadly used in human-machine interaction(HMI)applications.Determining how to decode the information inside EMG signals robustly and accurately is a key problem for which we urgently need a solution.Recently,many EMG pattern recognition tasks have been addressed using deep learning methods.In this paper,we analyze recent papers and present a literature review describing the role that deep learning plays in EMG-based HMI.An overview of typical network structures and processing schemes will be provided.Recent progress in typical tasks such as movement classification,joint angle prediction,and force/torque estimation will be introduced.New issues,including multimodal sensing,inter-subject/inter-session,and robustness toward disturbances will be discussed.We attempt to provide a comprehensive analysis of current research by discussing the advantages,challenges,and opportunities brought by deep learning.We hope that deep learning can aid in eliminating factors that hinder the development of EMG-based HMI systems.Furthermore,possible future directions will be presented to pave the way for future research.
基金supported by National Key Research and Development Program of China(No.2016YFF0102502)the Key Research Program of Frontier Sciences,CAS(No.QYZDJ-SSW-JSC037)the Youth Innovation Promotion Association,CAS,Liao Ning Revitalization Talents Program(No.XLYC1807110)。
文摘In the spectral analysis of laser-induced breakdown spectroscopy,abundant characteristic spectral lines and severe interference information exist simultaneously in the original spectral data.Here,a feature selection method called recursive feature elimination based on ridge regression(Ridge-RFE)for the original spectral data is recommended to make full use of the valid information of spectra.In the Ridge-RFE method,the absolute value of the ridge regression coefficient was used as a criterion to screen spectral characteristic,the feature with the absolute value of minimum weight in the input subset features was removed by recursive feature elimination(RFE),and the selected features were used as inputs of the partial least squares regression(PLS)model.The Ridge-RFE method based PLS model was used to measure the Fe,Si,Mg,Cu,Zn and Mn for 51 aluminum alloy samples,and the results showed that the root mean square error of prediction decreased greatly compared to the PLS model with full spectrum as input.The overall results demonstrate that the Ridge-RFE method is more efficient to extract the redundant features,make PLS model for better quantitative analysis results and improve model generalization ability.
基金financially supported by the National Natural Science Foundation of China(Grant No.41806122)the Strategic Priority Research Program of the Chinese Academy of Sciences(Grant No.XDA11040102)+2 种基金the State Key Laboratory of Robotics of China(Grant No.2017-Z08)Youth Innovation Promotion Association,CASJiang Xinsong Innovation Fund
文摘Range is an important factor to the design of autonomous underwater vehicles (AUVs), while drag reduction efforts are pursued, the investigation of body-propeller interaction is another vital consideration. We present a numerical and experimental study of the hull-propeller interaction for deeply submerged underwater vehicles, using a proportional-integral- derivative (PID) controller method to estimate self-propulsion point in CFD environment. The hydrodynamic performance of hull and propeller at the balance state when the AUV sails at a fixed depth is investigated, using steady RANS solver of Star-CCM+. The proposed steady RANS solver takes only hours to reach a reasonable solution. It is more time efficient than unsteady simulations which takes days or weeks, as well as huge consumption of computing resources. Explorer 1000, a long range AUV developed by Shenyang Institute of Automation, Chinese Academy of Sciences, was studied as an object, and self-propulsion point, thrust deduction, wake fraction and hull efficiency were analyzed by using the proposed RANS method. Behind-hull performance of the selected propeller MAU4-40, as well as the hull-propeller interaction, was obtained from the computed hydrodynamic forces. The numerical results are in good qualitative and quantitative agreement with the experimental results obtained in the Qiandao Lake of Zhejiang province, China.
基金supported in part by theNational Key R&D Program of China (2018YFB1304600)the Natural Science Foundation of China (51775541)CAS Interdisciplinary Innovation Team (JCTD-2018-11)。
文摘Supernumerary robotic limbs(SRLs) are a new type of wearable human auxiliary equipment, which is currently a hot research topic in the world. SRLs have broad applications in many fields, and will provide a reference and technical support for the realization of human-robot collaboration and integration,while playing an important role in improving social security and public services. In this paper, representative SRLs are summarized from the aspects of related literature analysis,research status, ontology structure design, control and driving,sensing and perception, and application fields. This paper also analyzes and summarizes the current technical challenges faced by SRLs, and reviews development progress and key technologies,thus giving a prospect of future technical development trends.
基金financially supported by the National Key R&D Program of China (Grant No. 2017YFC0306402)the Strategic Priority Research Program of the Chinese Academy of Sciences (Grant No. XDA22040102)
文摘This paper presents autonomous docking of an inhouse built resident Remotely Operated Vehicle(ROV),called Rover ROV,through acoustic guided techniques.A novel cage-type docking station has been developed.The docking station can be placed on a deep-sea lander,taking the Rover ROV to the seafloor.Instead of using vision-based pose estimation techniques and expensive navigation sensors,the Rover ROV docking adopts an ultra-short baseline(USBL)and low-cost inertial sensors to build an adaptive fault-tolerant integrated navigation system.To solve the problem of sonar-based failure positioning,the measurement residuals are exploited to detect measurement faults.Then,an adaptation scheme for estimating the statistical characteristics of noise in real-time is proposed,which can provide robust and smooth positioning results.It is more suitable for a compact and low-cost deep-sea resident ROV.Field experiments have been conducted successfully in the Qiandao Lake and the South China Sea area with a depth of 3000 m,respectively.The experimental results show that the functionality of autonomous docking has been achieved.Under the guidance of the navigation system,the Rover ROV can autonomously and efficiently return to the docking station within a range of 100 m even when the amounts of outliers exist in the acoustic positioning data.These achievements can be applied to current ROVs by an easy retrofit.
基金This work was financially supported by the Innovation Fund from Chinese Academy of Sciences(Grant No.CXJJ-17-M130)the Research Fund of the State Key Laboratory of Robotics(Gant No.Y91Z0904).
文摘National navies equip their submarines with Autonomous Underwater Vehicle(AUV)technology.It has become an important component of submarine development in technologically-advanced countries.Employing advanced and reliable recovery systems directly improves the safety and operational efficiency of submarines equipped with AUVs.In this paper,based on aerial refueling technology,a cone-shaped recovery system with two different guiding covers(closed structure and frame structure)is applied to the submarine.By taking the Suboff model as the research object,STAR-CCM was used to study the influence of the installation position of the recovery system,and the length of the rigid rod,on the Suboff model.It was found that when the recovery system is installed in the middle and rear of the Suboff model at the same velocity and the same length of the rigid rod,the Suboff model has the good stability and less drag.It experiences the largest drag when being installed in the front of the rigid rod.Moreover,when the recovery system is installed in the front and middle of the rigid rod,the drag increases as its length increases,and the lift decreases as its length increases.Compared with the closed structure guiding cover,the Suboff model will have less drag and better stability when the recovery system uses the frame structure guiding cover.Besides,the deflection and vibration of the rigid rod were also analyzed via mathematical theory.
基金This work was supported by the National Key R&D Program of China(2018YFB1304600)National Natural Science Foundation of China(51605474,61821005)+1 种基金Key Research Program of Frontier Sciences,CAS,Grantno.ZDBS-LY-JSCollLiaoning RevitalizationTalents Program(XLYC1807090).
文摘Variable Stiffness Actuator(VSA)is the core mechanism to achieve physical human–robot interaction,which is an inevitable development trend in robotic.The existing variable stiffness actuators are basically single degree-of-freedom(DOF)rotating joints,which are achieving multi-DOF motion by cascades and resulting in complex robot body structures.In this paper,an integrated 2-DOF actuator with variable stiffness is proposed,which could be used for bionic wrist joints or shoulder joints.The 2-DOF motion is coupling in one universal joint,which is different from the way of single DOF actuators cascade.Based on the 2-DOF orthogonal motion generated by the spherical wrist parallel mechanism,the stiffness could be adjusted by varying the effective length of the springs,which is uniformly distributed in the variable stiffness unit.The variable stiffness principle,the model design,and theoretical analysis of the VSA are discussed in this work.The independence of adjusting the equilibrium position and stiffness of the actuator is validated by experiments.The results show that the measured actuator characteristics are sufficiently matched the theoretical values.In the future,VSA could be used in biped robot or robotic arm,ensuring the safety of human–robot interaction.