Abstract:Matching clearance is one of the key parameters to determine the accuracy and reliability of spherical inertia element. It is difficult to accurately measure the radius of the spherical inertia element due to the clearance and rough surface, which further lead to hardly control the matching clearance of ball bowl and ball. A laser differential confocal (LDC)based highprecision measurement method for spherical inertial element matching clearance is proposed in this work. Firstly, the antiscattering laser differential confocal radius measurement method is used to measure the radius of the bowl and the spherical crown, respectively. Then, the differential confocal radius measurement systems are utilized to measure the radius difference between the bowl and the crown. Finally, the matching clearance of the spherical inertia sphere is controlled by using the radius difference. Theoretical analysis and experimental results show that the method can measure the relative expansion uncertainty of matching clearance of the spherical crown and the ball bowl with the precision better than 20×10-6. It provides a new technology for the highprecision measurement and control of the spherical inertia element matching clearance.

Abstract:Hybrid inertial navigation system (HINS) is a novel navigation system, which integrates the advantages of platform INS, strapdown INS and rotational INS. Rotational gimbals and highprecision high resolution optical grating angle sensors are employed in the HINS, which can achieve the selfcalibration of the HINS error parameters without removing it from the vehicle, thus simplifies the ground calibration task and system routine maintenance, and is conducive to the effectiveness of weapon platform and combat system. Aiming at the structure characteristics and typical rotation modes of HINS, this paper conducts the error analysis for the HINS. Then the designing principle and method of the selfcalibration scheme for HINS are given, and a concrete selfcalibration scheme is provided. The verification experiment of the HINS principle prototype was carried out. Experiment results show that the designed selfcalibration scheme can achieve the accurate estimations of the gyro drifts, accelerometer biases, scale factor errors and installation angle deviations using its own rotation mechanism without removing the HINS from the vehicle, and the calibration accuracy could satisfy the requirements of actual engineering. The proposed selfcalibration scheme possesses high theoretical significance and engineering application value.

Abstract:During the fast measurement process of measuring the bearing cylindrical roller diameter, the sensor contact deforms due to the applied force produced when the sensor contact approaches the roller, which may reduce the measurement accuracy. Therefore, this workproposesa contact dynamics research methodology considering the microscopic deformation. Based on Hertz contact theory, nonlinear damping theory and Coulomb friction theory, a kinematic differential equation between the contact and the roller isestablished based on the sensor contact deformation during dynamic measurement. Different measurement speed, position angle and deformation amount areobtained by analyzing the relationship between the normal force and the deformation at the contact point. In addition, the test results from the virtual simulation and the physical prototype experiment show that the roller diameter measurement error is no more than 1 μm. By quantifying the relationship between measurement speed and measurement accuracy, this workcontributes to the bearing cylindrical roller diameter measurement studies, and provides a theoretical basis for efficient and accurate contact measurement of the diameter of cylindrical parts in practical engineering applications.

Abstract:In order to improve the measurement accuracy of the embedded time grating angular displacement sensor, starting from the formation mechanism of the sensing signal, the cause of the short period error is analyzed in detail. Through the winding equivalent analysis and excitation signal analysis, it is determined that the main characteristics of the short period error are the first order error and quadric error. The source of the first order error is the zero residual error and direct current component error, and the source of the quadric error is the excitation signal quadrature error. Aiming at the compensation of the short period error, an error compensation method based on extreme learning machine is proposed. The model optimal parameters are obtained through training the measured values and real sample values. According to the model parameters, the short period error model is established, which is used to realize the short period error compensation. The experiment result shows that the analysis result of short period error is consistent with the actual characteristic of the sensor error, and using the proposed error compensation method the short period error of the sensor is reduced by about 96%, which is a greatly reduction. The comparison and repetitive experiments show that the accuracy of the proposed method is improved doubly compared with that of the harmonic compensation method, the error compensation effect is superior. Besides, the proposed method possesses good measurement stability, which has important theoretical and practical significance for improving the measurement accuracy of the embedded time grating angular displacement sensor.

Abstract:With the development of force metrology within superlarge range,force transfer system with a superlarge range has become a hotspot in the field of force metrology recent years. An uncertainty calculation model of the buildup system with a superlarge rangeconsidering the influence of the buildup structure was developed. A 60 MN buildup system was calibrated and its uncertainty was evaluated and verified. The researchresultsshow that the deformation of the buildup structure would induce indication error in the buildup system. The uncertainty component induced by the indication error is larger than that induced by the force transducer,whichis the main component in the combined uncertainty of buildup system with a superlarge range. In the evaluation, the influence of this component should be fully considered; when extrapolating the indication error and its uncertainty to full range, the power function should be used to perform the curve fitting and extrapolation of the buildup system indication error and its uncertainty, which will guarantee thatthe extrapolation results are accurate and reliable.

Abstract:A novel apparatus is developed to calibrate vacuum gauge for its response characteristic, which consists of calibration system, highspeed data acquisition and control system, pumping system, baking system, and gas supply system. A very fast opening ultrahigh vacuum gate valve and flow restriction element of various dimensions are combined to realize controllable standard pressure establishment time ranging from 1812 ms to 65813 ms. Moreover, to establish transient step over three orders of magnitude of the dynamic vacuum pressure, a volume ratio of 1 5739 between downstream chamber and upstream chamber is employed. Through numerical simulation of the rapid expansion process of rarefied gas in the millisecond range, the function between conductance and time under different flow states is obtained. Experiments are conducted on capacitance diaphragm gauges with optimized response time (107 ms). The result show that the calibration range of the apparatus is between 10-1 and 105 Pa, with a combined standard uncertainty of 14%.

Abstract:Satellite ground stations generally use the SGP4/SDP4 model of two line element(TLE) format to calculate and predict the satellite orbit for guidance. However, with the continues expanding applications of Kaband communication satellite antenna, this orbit prediction model can not meet the requirements of high precision measurement and control of Kaband antenna, especially for highorbit satellite with low elevation angle. A satellite tracking and prediction method based on orbit elements for Kaband TT&C antenna is proposed. The influence of satellite perturbation model on orbit prediction under different precision requirements is analyzed. And the error correction of the electric wave elevation is considered and the orbit prediction calculation process is designed. The orbit calculation model is simplified while guaranteeing the accuracy of orbit prediction. The proposed method is tested using two highorbit satellites with low elevation angles, and the tracking guidance accuracy is less than 0005 degrees which can meet the requirements of highprecision, stable and continuous tracking for Kaband TT&C antenna. This work can greatly improve the highspeed data transmission capacity and availability of communication satellite in bad weather, equipment failure and other emergency situationss.

Abstract:In order to improve the contrast and resolution of ultrasound imaging, this paper studies a target scattering imaging method based on a spherical shell focused array. The partial array elements in the spherical shell focused array transmit the focused sound waves. When the sound waves act on the target scattering occurs, and the scattered echoes are received by the remaining array elements. The transmitting and receiving elements in the array are selected randomly to suppress the side lobes. Using the threedimensional automatic scanning mechanism, the scattered echo intensity of the focused beam along the surface of the measured object can be obtained, then the target imaging is realized. Computer simulation results show that the spherical shell focused array can decrease the mainlobe width and effectively reduce the side lobes of the focused beam. The water tank experiment results show that the scattering imaging method based on spherical shell focused array possesses a submillimeter level imaging resolution and could obtain high contrast image.

Abstract:The IEC standard 61000415 recommends that the sinusoidal and rectangular wave modulations are used to test flicker meter. At present, the researches on voltage flicker detection mainly focus on the sinusoidal wave modulation model, while there are few researches on the quick and accurate detection of the flicker under rectangular wave modulation that is more sensitive to human eyes. Aiming at the quick and accurate detection of the flicker under rectangular wave modulation, a fluctuation component extraction algorithm based on interval sampling improved energy operator is established and a new optimized KRV mutual convolution window function is built. An improved fast Fourier transformation (FFT) spectrum analysis method based on the threespectral line interpolation and new KRV deconvolution window is also derived. Based on this, the flicker detection method with rectangular wave modulation based on improved energy operator and new KRV mutual convolution window is proposed in this paper, and the voltage flicker parameter detection platform based on PXI + LabVIEW architecture is developed. The simulation and experiment results show that the proposed algorithm can effectively realize accurate detection and analysis of the flicker, and has higher accuracy compared with the traditional method in the cases of simple rectangular wave modulation, multifrequency modulation, containing all kinds of harmonic interferences, network fundamental frequency deviation and containing noise interference.

Abstract:Combining modified ensemble empirical mode decomposition (MEEMD) and chaotic oscillator, a distribution network fault weak signal detection method was proposed. At first, a neural network prediction model was set up, which was used to conduct the short term prediction of the zero sequence currents of the lines in the distribution network and filter out the background signal in the fault signal. Then, in order to detect the weak 5th harmonic fault signal when the single phase grounding fault happens in the distribution network, the MEEMD method combining multiscale permutation entropy and complete ensemble empirical mode decomposition (CEEMD) was proposed to process the fault signal in which the background signal was filtered out. The first intrinsic mode function was extracted and used as the input of the chaotic oscillator. The chaotic oscillator has high sensitivity to the external driving force signal with the same frequency as that of the internal driving force signal. The distribution network fault line selection was completed with the phase diagram outputted by the chaotic oscillator.

Abstract:Research on the mechanism, evolution rule and condition monitoring of bonding damage of insulated gate bipolar transistor (IGBT) module is one of the important parts in the reliability research of power electronic devices and systems. Firstly, the bonding damage and its evolution trend are analyzed theoretically. It is pointed out that the bonding damage is the result of the synergistic effect of the electrodynamic force on the bonding lines and the shear stress on the bonding points. The positive feedback process of bonding damage evolution is sorted out. Then, the thermalforce field of bonding lines and bonding points of IGBT module is simulated and analyzed to quantitatively explore the bonding damage mechanism and its evolution rules. Experimental results show that both electrodynamic force on the bonding lines and the shear stress on the bonding points are not only the inducement of bonding damage and its evolution, but also obviously affected by them. Furthermore, the equivalent circuit analysis and experiment study are implemented to verify the feasibility of the bonding resistance monitoring method based on the height of miller platform and collector current. Finally, the limitations of the online monitoring method of IGBT bonding resistance in wind power are analyzed and pointed out. The solutions to key problems of online monitoring of IGBT bonding resistance in wind power are put forward, and the online monitoring program is designed. In addition, a fivelevel state assessment method of bonding damage based on percentage increment of bonding resistance is proposed.

Abstract:Unbound sleep monitoring is an urgent requirement. However, the existingmethods of heart rate and cardiaccycle extraction acquire limited physiological information, which lack of clinical value, or need prior knowledge about physiological information. In this study, an improved realtime heart rate and successive cardiac cycle extraction method based on ballistocardiogram(BCG) is proposed.Thismethod uses an acceleration sensor fixed on the bed side to collect the BCG signal, and the cardiac cycle sequence is obtained from the BCGsignal by the improved cardiac cycle function.The concept of cardiac cycle estimation is introduced to solve the problem that the wrong cardiac cycle occurs in the sequence. According to the cardiac cycleestimation value and by combining thecardiac candidate value and its position difference, a successive cardiac cycle is obtained from the matrix composed of the cardiac candidate value and the position.The heart rate and cardiac cycle obtained by this method are compared with electrocardiogram. Experimentalresults show that the accuracy of heart rate extraction is not lower than 9872%, and the successive cardiac cycle obtained by the two methods is within the consistency limit of 9564%.The accuracy and consistency of the proposed method are verified.

Abstract:The traditional rotary electromechanical converter has the problems of low linearity and small work range of static characteristics.To improve its performance, a novelrotating proportional electromechanical converter is designed.The static characteristics of the rotating proportional electromechanical converter arestudied by means of magnetic circuit analysis, Maxwell 3D static magnetic field finite element analysis and experimental tests. Meanwhile,a prototype isdeveloped and a series of experiments are conducted.Experimentalresults indicate that the proportional electromechanical converter has a working range from -5° to 5°, and the maximum torque is about ±252 mN·m. In addition,it has a positive magnetic stiffness.The experiment and simulation of torque values are concentrated within the 96 mN·m.The relative error is less than 6%, which meansthat electromechanical converter has good repeatability.Furthermore, the nonlinear error of the angletorque curve is less than 15%with a hysteresis of less than 35%,which demonstrates that the currenttorque curve has good linearity and low hysteresis. It proves that the test results are basically consistent with the simulation results.

Abstract:The piezoelectric film sensor is affected by temperature and the piezoelectric quartz sensor is expensive. Thus,a fusion design method is achieved by embedding two piezoelectric quartz sensors into fourrow piezoelectric film sensors. The equivalent installation model and temperature compensation algorithm of piezoelectric film sensor are estabilished. The piezoelectric quartz sensors are used to perform accurate and real time temperature compensation for all piezoelectric film sensors. Under the excitation of sinusoidal wave pavement, a quarter vehicle model is selected to conduct weighinmotion simulation experiment in MATLAB. In the experiments, the vehicle speed is ranged from 10 to 120 km/h and the environment temperature shifts from 0 to 40℃. The results show that weighing error mean value of fusion design is less than 097%, and accuracy is over 23 times than fourrow piezoelectric film sensors, and price is 3 times lower than fourrow piezoelectric quartz sensors. This design can realize high precision and low cost detection for vehicle weighinmotion, which shows good prospects for engineering applications.

Abstract:Threedimensional magnetic dipole model is formulated for the stay cable with 95 mm diameter. The magnetic flux leakage field induced by the brokenwire flaw in the 3D space is analyzed. The spatial distribution of the axial component of the leakage field is highlighted and discussed. Scanning system is constructed to measure the axial component of the leakage field. Experimental results verify the accuracy of the magnetic dipole model on leakage field detection. Threshold method is applied to estimate the effective diffusion angle of the axial component of the leakage field along circumferential direction as ±6°, which determines that the minimum number of elements in the sensor array is 30. A circular magnetic flux leakage sensor array is developed for the stay cable. With the liftoff of 8 mm, the sensor array can successfully detect multiple surface brokenwire flaws by providing a scanning image to the cable. The effectiveness of the element number in the array on the quality of the scanning image is discussed. This study provides a reference for designing circular magnetic flux leakage sensor array for surface flaws detection in largediameter stay cable.

Abstract:Aiming at the problems of big volume, cumbersome operation and complex application scenariocommonly existing in current sixdimensional force sensor, a miniaturized wireless sixdimensional force sensor is designedbased on resistance strain principle in this study. Firstly, the elastomer of the sensorwas designed based on cross beam structure, ANSYS software was used to perform structure statics analysis and the optimal structure dimensions of the sensor were determined. Secondly, based on the design requirements of sensor miniaturization and wireless transmission, the internal hardware circuit of the sensor was designed, including fourarmfullbridge circuit, twostage amplification circuit and data acquisition and transmission circuit,which effectively reduces the volume of the sensor and enhances the flexibility.Finally, the static calibration methodwas designed and optimized, and the static decoupling algorithm based on coupling error modeling was adopted to reduce the interdimensional coupling interference of the sensor and improve the measurement accuracy of the sensor.The experiment results show that the indoor wireless transmission distance of the sensor reaches to 8 m, the transmission rate satisfies the low/mediumfrequency force/torque signal measurement requirement, and the type I and II errors satisfy the requirementofhigh precision applications.

Abstract:In fuzzy Cmeans (FCM) algorithm, the clustering result is sensitive to the initial center points and the clustering process does not take into account the influences of different density points. Thus, an improved FCM algorithm based on density peak and spatial neighborhood information is proposed. The improved algorithm selects the points with local density peaks or large local density values as the initial center points, and highlights high density points′ influence in the clustering. The theoretical analysis and experiments on both synthetic and realworld datasets from the UCI machine learning repository demonstrate that, the proposed algorithm has better antinoise, clustering performance and global convergence ability than traditional FCM algorithm.

Abstract:The study of the complex relationship between heat losses and variable stiffness under different material ratios has important significance to improve the vibration isolation performance of a laminated Magnetorheological rubber bearing(MRB). Based on magnetomechanical coupling theory, the macroscopic and microscopic magnetomechanical coupling models are constructed for the bearing. In further, the magnetic field distribution and stiffness of horizontal shear of the bearing are calculated. Considering the influence of heat losses, the relationship between the range of stiffness and the particle volume ratio of the magnetorheological rubber(MRR) is obtained. The bearing is sheared on the MTS tester, and the experimental data are compared with the theoretical results. Results show that the designed bearing can obtain the maximum range of stiffness up to 428% while the maximum heat loss is limited within 387 W, and the optimal particle volume ratio of the MRR is 12%. Besides, the optimal particle volume ratio of the MRR deviates with the changes of heat losses. Therefore, the balance law of heat losses and variable stiffness obtained by this macroscopic and microscopic theory model can provide a novel way to design the optimal structure and properties of MRB.

Abstract:In system stability analysis, a correct expression of the uncertain parameters is a prerequisite for stability evaluation. However, the parameter distribution that affects system reliability often lacks strict regularity in engineering. Even the parameters generally obey a certain distribution, they always drift. Informationloss is another concern when traditional methods are used to deal with such uncertainties. Therefore, a new method to conduct system reliability analysis under uncertain information is proposed by introducing probabilitybox theory. Firstly, the probabilitybox is used to model uncertain parameters. Secondly, the probabilitybox model of system reliability is obtained by discretizing each parameter into equally confidence levels and calculating Cartesian product with the system reliability equation. Finally, the risk zone and the stable zone are divided with zero as boundary, and the system reliability is quantitatively analyzed by integral calculating the area of probabilitybox. The cantilever beam system is analyzed in the experiments. Experimental results demonstrate that the proposed method is effective, and can also improve the accuracy compared with other related approaches.

Abstract:Aiming at the problem of large torso disturbance of hydraulic quadruped robot in locomotion, the torso disturbance inhibition strategy for the hydraulic quadruped robot based on kinematics and virtual model is proposed. In this study, the generation mechanism of the torso disturbance and its influence are analyzed. The kinematic equation of the whole quadruped robot is established. The torso disturbance is suppressed according to the realtime attitude feedback of the robot. The virtual springdamper elements are introduced in the degrees of freedom of rolling and pitching of the torso, and the virtual force is adjusted to control the torso attitude. Facing to the trot gait of the robot, the foot end trajectory planning of the robot in the swing phase and support phase is carried out. Experiment verifications were carried out on the hydraulic quadruped robot test platform. The experiment results show that the proposed disturbance inhibition strategy could adjust the joint angle according to the torso attitude of the robot,the fluctuation of the robot torso is small and the actual locomotion trajectory of the robot is close to its theoretical one, which verifies the effectiveness of the proposed disturbance inhibition strategy.

Abstract:The equivalent model analysis is an effective method to indirectly assess the aging state of oilpaper insulation, and it is essential and difficult to construct the equivalent model and identify model parameters of the oil paper insulation system. In view of the shortcomings of the extended Debye model, the mixed polarization equivalent model of the oilpaper insulation with the interface polarization branch is introduced to describe the actual polarization process. At the same time, the parameter identification method of dielectric spectrum model in frequency domain for oilpaper insulation mixed polarization process is studied. Firstly, the relationship between frequency spectrum and circuit parameters is derived based on a new polarization equivalent circuit. Secondly, the objective function for solving multivariate equations is established, and then the parameters are solved by ACPSO algorithm. Finally, the feasibility and accuracy of the method for identifying model parameters are verified by measured data. Results show that the mixed equivalent model considering the interface polarization can reflect the relaxation process of oilpaper insulation more truly, and it is more consistent with the actual situation of transformer insulation polarization. Compared with the identification method based on timedomain dielectric spectroscopy, the identification method of the equivalent model of oilpaper insulation based on FDS is simple, accurate and reliable.

Abstract:The application scope of traditional spatial cooperative target vision measurement technology is limited by the installation status of the cooperative markers. Taking the edge contour as the cooperative feature, the 3D structural model is used to construct the cooperative relationship and a spatial cooperative target pose vision measurement method with wider application range is proposed. The method generates 2D target feature templates with different observation orientations, which are used to perform searching and matching with the measurement image. The contour orientation features are used to evaluate the matching degree, and the target pose parameters are solved from the optimal matching results. The offline preprocessing strategy and twostage image pyramid search optimization method are used to accelerate the processing. The digital simulation test and semiphysical simulation test were conducted to verify the accuracy and stability of the method. The absolute error of position measurement in vertical optical axis direction is better than 2 mm, and the relative error of position measurement in parallel optical axis direction is better than 07%. The absolute error of attitude measurement is better than 02° and the single pose measurement takes less than 05 s. The proposed method can meet the requirement of spatial target navigation.

Abstract:To improve the autonomy of intelligent wheelchair/bed system (IWBS) docking, a visual measurement based framework for automatic docking is proposed. The docking process of IWBS is divided into remote guidance stage and shortrange visual docking stage. In the longrange guidance stage, a new artificial landmark on the wheelchair is collected and identified by the visual system on the ceiling. Afterwards, the artificial landmark is detected using a tracking method based on Kalman Filter (KF). The relative pose between the artificial landmark and the camera is then determined according to the P3P (perspectivethreepoint, P3P) visual positioning method. Using the transformation between the wheelchair coordinate system and the world coordinate of the remote guidance stage, the position and orientation of wheelchair are then determined. In shortrange docking stage, the guiding landmark of the docking target is collected using an onboard monocular vision system. The feature of the guiding landmark is used to estimate the relative pose between the wheelchair and the bed. Finally, the automatic docking of IWBS is realized according to the visual feedback in realtime. Numerous experiments are carried and experimental results show that the visual docking methods proposed in this work is feasible and effective.

Abstract:The traditional method of moving object detection is based on lowrank and sparse decomposition algorithm, which has the problems of foreground extraction results affected by noise and incomplete detection of moving objects. To solve these problems, one kind of new lowrank and sparse decomposition model is proposed. Not only the structural distribution characteristics of the video foreground object but also the influence of the dynamic background on the foreground extraction result are considered. The foreground is constrained by using the structured sparse norm. The motion area represented by the section is consisted of a dynamic background part and a foreground part. Then, the generalized alternating direction multiplier method is used to solve the proposed model, and the complexity of the algorithm is analyzed. Finally, simulation experiment is carried out and applied to moving object detection. Experimental results show that the proposed method is more stable and effective than other moving object detection methods based on lowrank and sparse decomposition, which indicates its university. It also has certain antinoise ability for different types of noise.

Abstract:When existing damage imaging algorithms are applied to real complex aeronautical structures, due to the influences of anisotropy of structures, complex boundary reflection and the dispersion characteristics of guided waves, the imaging result is often poor, and the application error is large, or the damage even can not be imaged. To solve the above problems, a structural damage imaging method based on dispersion compensation and pathwave velocity mapping is proposed. Firstly, the dispersion compensation of the scattered wave packet of damage under the influence of anisotropy is applied to optimize the characteristics of guided wave amplitude. Secondly, the problem of large errors of the time of flight caused by structural anisotropy is solved based on pathwave velocity mapping method. The imaging method is validated on the composite structures with stiffeners and openings, and the accuracy of damage imaging is obviously improved.

Abstract:The service robot has been developed rapidly in recent years, its application algorithms are constantly alternating, and the item detection algorithm is one of them. Under the premise of ensuring item recognition accuracy, the item detection speed determines the efficiency of robot item capture. Therefore, this paper will take the long distance and small item scene as the test scene, and improve the existing network model. The aim is to improve the detection speed on the premise of ensuring detection accuracy. Mask regionsbased convolution neural network (Mask RCNN) is a widely used algorithm in the field of item detection. Through the studying its network structure, it is found that the mask branch and excessive full connection layers will take up a lot of network detection time. At the same time, the feature map extracted with mask RCNN has a higher dimension, which will take up a lot of computing memory and produce a large number of computing tasks. To tackle these problems, in this paper, the mask RCNN network is improved by removing the mask branch and redundant full link layer, the lighthead RCNN (LHRCNN) is introduced into the mask RCNN network, and the anchor ratio in the region proposal network (RPN) is adjusted. Finally, the improved Mask RCNN network was tested on the home service robot platform with Kinect Ⅱ. The test results demonstrate that compared with the original mask RCNN, the improved Mask RCNN network can greatly improve the running speed of the algorithm, while ensure the detection accuracy at the same time. The detection time is shortened by more than two times, and the proposed method improves the efficiency of the item catch task of service robot.

Abstract:The difficulty of robot in 3D object recognition and optimal grasping lies in the complex background environment and the irregular shape of the target object. It requires the robot to determine the position and pose of the optimal grasping part of the target while recognizing different 3D targets like human. One kind of deep learning method based on the cascaded faster regionbased convolutional neural networks (RCNN) model is proposed to identify the target object and its optimal grasping pose. The improved Faster RCNN model is proposed at the first level, which can recognize small target objects and accurately locate them in the image. Then, a faster RCNN model at the second level is designed to find the optimal grasping pose of the target object recognized by the previous level to realize the optimal grasping of the robot. Experimental results show that the method proposed in this paper can find the object accurately and determine its optimal grasping pose.

Abstract:Eddy current imaging detection of titanium plate is susceptible to noise from industrial field. It is difficult to extract effective features from the detected images containing noise, which affects the classification accuracy. To address this issue, a classification method for eddy current detection images of titanium plate defects based on stacked sparse denoising autoencoder (SSDAE) deep neural network is proposed. Sparsity constraint is introduced into the denoising autoencoders (DAE) and the autoencoders perform unsupervised selflearning layerbylayer. Then, the autoencoders are stacked and a logistic regression (LR) layer is added to construct the deep neural network. The deep neural network can automatically extract features and classify eddy current detection images of titanium plate defects after supervised finetuning. The feature learning ability is improved by sparsity constraint, and the robustness of deep network is improved by stack combination of denoising autoencoders. Experimental results show that, compared with other traditional methods, the proposed method not only has higher classification accuracy in ideal conditions, but also can resist noise and classify defects of titanium plate in complex conditions more effectively.

Abstract:The saggar that contains Liion battery cathode material is prone to abnormal lateral movement behavior during the calcination process of roller kiln. So, in this paper, the influence of roller table flexibility on the motion characteristics of the saggar in the driving process is analyzed based on flexibility dynamics. Firstly, a simplified model of the saggarroller table driving dynamics is established based on flexibility dynamics. Using this model, the elastic deformation of the roller and the motion behavior of the saggar during transmission are analyzed, and experiment was conducted to verify the effectiveness of the model. Secondly, the corresponding rigid dynamics model of saggarroller table is established. Through comparing the kinematic characteristic variation of the saggar under the transmission conditions of flexible and rigid rollers and the lateral movement variation of the saggar under different roller deflection, the influence of the flexibility of the roller table on the saggar kinematic characteristic is revealed. Based on these, the original roller was improved, and corresponding experiment was carried out to verify the effect of the improved scheme. Results show that: Because of the influence of elastic deformation, obvious outward expansion of the gaps among the saggars occurs while the saggars are transmitted forward, however there is almost no expansion of the gaps among the saggars when the roller shows rigidity; The deflection of the roller has an important influence on the lateral movement of the the saggars. When the deflection of the roller is changed from 0 to 2 mm, the expansion gap of the saggars increases with the increasing of the deflection. When the deflection of the roller exceeds 2 mm, the expansion gap of the saggars decreases with the increasing of the deflection; The experiment results for the improved roller show that the properly increasing the rigidity of the roller can significantly reduce the abnormal lateral motion of the saggars.

Abstract:Aiming to accurately identify the fall and draganddrop abnormal gait during assisted walking of the patients with lower limb dysfunction, a noncontact abnormal gait recognition method based on nodeiteration type fuzzy Petri Net is proposed from the universality and convenience of various user groups. Firstly, the structure of the rehabilitation training robot is discussed, and the behavior characteristics of the abnormal gait of fall and draganddrop often occurring in the course of assistant walking are described. A multichannel proximity sensor array is developed to detect gait information in real time. Integrating the intention vector of walking direction, the gait deviation, frequency and body incline angle are taken as the input parameters of the detection system. Based on the fuzzy membership function, the network ignition mechanism is generated, and the nodeiteration type fuzzy Petri Net is established to recognize the abnormal gait. Finally, an abnormal gait recognition method based on nodeiteration fuzzy type Petri net is proposed, and abnormal gait operator reasoning experiment and multimode walking fall detection experiment of walking rehabilitation training robot are carried out. The experiment results show that the algorithm can accurately recognize the abnormal gait of different walking habit groups using walking rehabilitation robot, improve the safety and comfort of the assisted walking of the users, and the recognition rate of abnormal gait reaches to 912%. This proposed method can be used in the daily living and rehabilitation training of the users with lower limb dysfunction using similar walking aids.