Abstract:The traditional micro electro mechanical system (MEMS) vibrating gyroscope cannot survive in the process of high overload and the parameters change greatly before and after the impact. To address these issues, this article proposes the MEMS solid ring wave gyroscope, including its design, processing and testing. Firstly, a gyroscope structure with a fully symmetrical beam and potting technology is proposed, which can effectively reduce the residual stress in the structure during the impact process. The stop mechanism is utilized to enhance the viability of the gyroscope during the impact process. Based on this structure, the dynamic equation of the gyroscope and the impact oscillation motion function of the sensitive axis are derived. And the sensitive axis impact mode is investigated that the higher natural frequency and the smaller quality factor can improve the anti-shock performance of the gyroscope on the sensitive axis. Secondly, the modal analysis and impact characteristic simulation of the gyro structure are implemented by using the finite element analysis software. Results show that the maximum displacement and stress of the gyroscope structure are 9. 46 μm and 99. 6 MPa, respectively, when the shock amplitude is 15 000 g with 10 ms wide pulse. These results ensure that the gyroscope structure has a good anti impact margin. Thirdly, the mature glass silicon bonding and deep silicon etching process are used to realize the processing of the gyroscope structure. And the vacuum packaging of the gyroscope structure is realized by combining ceramic packaging. The test system of the gyroscope is established by the driving closed loop and detection open loop. Finally, in the laboratory environment, the impact test of the gyroscope prototype is realized by using the impact platform. During the impact process (pulse width 0. 6 ms), there are several peaks of more than 5 000 g. The maximum peak value is 16 050 g, the response time of the gyroscope is about 1 s, and the change of the gyroscope zero position before and after the impact is less than 1% . The anti-overload ability of the prototype developed in this article is verified.

Abstract:The wearable sensors have problems of poor experience, high cost and complex construction. To address these issues, an optically wireless tension sensor is developed with strain-sensitive fluorescent material painted on paper. The novel flexible probe uses the mixture of Rhodamine B (RhB) and PDMS as sensing material and the composite is screen printed onto a piece of fine art paper. The fluorescence characteristics of RhB are found linearly depending on the tension applied to the paper slice. Therefore, the sensing functions of the paper-based fluorescence sensor are calibrated on a customized force loading test bench. The constructed sensor is assembled to a wearable device. Then, the finger flexion could be successfully detected by the fluorescence. The finest resolution of the prepared sensor is 0. 04 N over a measuring range of 6 N. This study presents an optical wearable tensiometer employing a non-lightguide principle for the first time. It has potential application in human health evaluation, rehabilitation exercise and joint posture monitoring.

Abstract:An oxygen compensation algorithm and calibration strategy are proposed to address the problem that the zirconia-based NOx sensor signal is affected by the oxygen content of vehicle exhaust, which may result in poor accuracy. A simulation test is conducted according to the working principle of the NOx sensor, the relationship between pump current IP2 and NOx concentration at different oxygen content is analyzed. The oxygen compensation algorithm is derived, and the relationship between oxygen content and compensation coefficient is analyzed. A real-time compensation scheme for IP2 error is designed. Consequently, the error of IP2 in the range of oxygen content from 0% ~ 20% and NOx from 0~ 3 000×10 -6 is reduced to 23 nA, and the Corresponding NOx concentration error is reduced to within 15×10 -6 . To further improve the accuracy of the sensor measurement, a calibration scheme is designed and tested in comparison with foreign NOx sensors. The results show that the sensor has a measurement error of 0~25×10 -6 in an environment with NOx concentration of 0~2 000×10 -6 under transient engine conditions, which verifies the feasibility of the scheme and well meets the requirements for monitoring NOx concentration in the exhaust gas, and provides a reliable solution for improving the accuracy of zirconia-based NOx sensors.

Abstract:According to the requirement of the gas medium dynamic temperature calibration, a set of gas-dynamic temperature calibration devices based on the shock tube principle is developed. In terms of the issue that the temperature sensors cannot be excited by the shock tube due to the short duration of the shock wave, the traditional structure of the shock tube is reformed in this article. As a result, the continuous flow with the steady temperature after the shock wave procedure is realized, and the effective excitation problem of the sensor is solved. To realize the high-quality dynamic calibration for the rapid response temperature sensors, the impact of temperature amplitude instability on the dynamic calibration results is reduced by the research of temperature amplitude compensation technology during the shock wave procedure. In addition, through the analysis of the influence of shock wave unsteady boundary layer on the temperature, the condition of ignoring the change of temperature with wall parameters is obtained, which improves the reliability of excitation signal generation. In the end, the traceability of the calibration device is analyzed and evaluated with the experiments, which shows that the proposed gas dynamic temperature calibration device can generate the temperature excitation signals with the step rise time of 1. 6 ms, a steady step amplitude of 320℃ , and duration of 40 s, which satisfy the dynamic calibration requirements of the temperature sensors.

Abstract:The number of ( ultra-wide band) UWB ranging measurements received by pedestrians to be located in the complex indoor scene is uncertain. To address this issue, the INS / UWB tight integrated localization algorithm based on a factor graph is proposed. It can be used to fuse UWB ranging measurement from random accessing node and exiting node. Firstly, an INS / UWB tightly integrated factor graph model is constructed, which is based on the pedestrian motion model and the UWB measurement model. Due to the simultaneous modeling of pedestrian position and velocity, there are cycles in the factor graph model. Aiming at the factor graph model with cycles, the sum-product algorithm (SPA) is used to derive the message passing algorithm among different nodes in the factor graph model through two iterations, and the posterior probability density of pedestrian position and velocity is calculated. Furthermore, given the rapid enlarging error deduced by a special ranging measurement vector in the INS / UWB tight integrated localization algorithm, an improved factor graph algorithm based on coordinate transformation is proposed. Simulation results show that the proposed INS / UWB tightly integrated localization algorithm can effectively fuse dynamic UWB ranging measurements in complex indoor scenes. On the premise of meeting the demand of computation and memory consumption, the proposed algorithm can improve positioning accuracy and speed estimation accuracy by 14. 94% and 56. 42% , respectively, compared with the extended Kalman filter ( EKF) algorithm with multi-models.

Abstract:As the nuclear magnetic resonance ( NMR) continues to develop towards the direction of portability and low cost, the miniaturization of the main magnet in the NMR has become a major concern. To realize the miniaturization of NMR, a low-field NMR main magnet based on Halbach structure is designed and implemented. Firstly, the ANSYS finite element simulation software is used to simulate the Halbach structure formed by the specific arrangement of multiple magnetic stripes to adjust the parameters, such as the number, size, length, and position of the magnetic stripes. Secondly, according to the ideal model parameters determined by the simulation, the main magnet, which is a cylinder of 108 mm diameter and 170 mm height weighing 4. 5 kg, is realized and its central magnetic induction strength and uniformity are tested. Finally, based on the test results, a uniform field sheet matrix is designed to improve the uniformity in the 5 mm diameter spherical region at the center of the magnet to 451. 68 ppm with a central magnetic induction of 221. 4 mT using passive uniformity. In the design, it is a good trade-off between magnet performance and production cost.

Abstract:In a complex urban environment, when the GNSS signal fails, the existing methods for estimating the vehicle pose using a monocular camera or an inertial navigation system suffer serious cumulative errors. To address these issues, a vehicle pose fusion estimation algorithm based on the cascaded deep neural network ( CDNN) is proposed. First, CDNN is designed to reduce the cumulative error caused by scale blur and scale drift in monocular cameras. Secondly, to reduce the introduced device noise, a simplified inertial sensor system (RISS) is used to obtain the vehicle lateral and longitudinal acceleration and yaw rate. To reduce the influence of uncertain noise in the system, H∞ filtering is used to fuse the outputs of CDNN and RISS to accurately estimate the vehicle pose while keeping high-frequency output. Compared with the method based on the Kalman filter, experimental results on the KITTI dataset show that, the root mean square error (RMSE) of the easting position estimated by the proposed algorithm is reduced by 41. 3% , and the RMSE of the estimated northing position is reduced by 70. 6% , and the RMSE of the estimated heading angle is reduced by 6 K 6 e . y 6 w % o .

Abstract:The quality of flow temperature data in the low-speed wind tunnel is the key to the character of the flow field of the wind tunnel. This article carries out special research on the compensation method of thermal conductivity error and radiation error in thermocouple measurement. Based on the basic principles of heat transfer, research is implemented on the method of the thermal conductivity error compensation with the fixed value of the immersed length of the thermocouple wire and the method of the radiation error compensation with the fixed value of relative change rate of the surrounding temperature. Simulation results of the quantitative relationship between the immersed length of the thermocouple, the relative change rate of the surrounding temperature, and temperature error are obtained. To evaluate the error mechanism analysis and simulation results, a wind tunnel validation test is carried out. Test results show that the measured thermocouple immersion error results are consistent with the theoretical simulation results, and the relative change rate of surrounding temperature ω<1 can significantly reduce the influence of temperature measurement radiation error, which can reduce the influence of temperature measurement radiation error on temperature uniformity by 7. 12 times. The method presented in this article provides useful exploration and technical basis for the engineering application of the thermocouple in gas medium.

Abstract:The formulation of the prediction model to compensate for the thermal error of machine tools is a common method to effectively solve the decline of machine tool accuracy caused by thermal error. This article proposes an adaptive robust modeling method for thermal error of computer numerical control machine tools based on regularization, which can adaptively select temperature-sensitive points (TSPs) in the modeling process, and has high prediction accuracy and robustness. Firstly, the robustness mechanism of thermal error modeling is analyzed, which is based on the principle of structural risk minimization. Secondly, the sparsity of the solution of least absolute shrinkage and selection operator (LASSO) in regularization algorithms is used to realize adaptive TSP selection. Then, based on the thermal error data under different experimental conditions, the prediction effect of the proposed modeling method is analyzed and compared with the commonly used multiple linear regression, back propagation (BP) neural network, and ridge regression algorithms. Results show that the proposed modeling method has the highest prediction accuracy and robustness, which are 5. 22 and 1. 69 μm, respectively. Finally, the thermal error compensation experiment is implemented by using the established prediction model to evaluate the actual compensation effect of the proposed modeling method.

Abstract:A low noise current amplifier based on the cryogenic current comparator (CCC) is designed and fabricated, which can be used for precision measurement of pA level weak current. The amplifier consists of superconducting proportional coils with a maximum turn ratio of 2 000 ∶ 1, a DC-superconducting quantum interference device and a superconducting shield. The equivalent input current noise, current sensitivity and feedback loop model of the amplifier are established. On this basis, the methods to reduce the amplifier noise and improve the amplifier sensitivity are proposed. Experimental results show that the coupling coefficient between CCC superconducting proportional coils and pick-up coil with 12 turns is 0. 61. When the CCC current amplification ratio is 2 000 ∶ 1, the equivalent input current noise of the amplifier is 30 fA/ Hz 0. 5 at 1 Hz. The current sensitivity is 4. 71 nA/ Φ0 and the performance of the amplifier has been evaluated in the test of peak to peak 6 pA square wave current.

Abstract:The shield tail clearance is difficult to be accurately measured in real time during the on-site construction. To address this issue, this article proposes an automatic shield tail clearance measurement method based on the linear structured light. Firstly, a linear laser is used to project a narrow line-shaped indicator laser between the shield shell and the tube. The line laser image projected by the line laser between the shield shell and the tube is collected in real time through an industrial camera. Then, the target features are recognized by image processing and the end coordinates of shield tail clearance are extracted according to monocular vision imaging principle and line structured light coordinate measurement algorithm. In this way, the accurate solution of shield tail clearance value is realized. Experimental results show that the repeatability measurement accuracy of the proposed method is better than 0. 3 mm, and the absolute measurement accuracy is better than 1. 7 mm. This method can realize the real-time accurate measurement of the shield tail clearance within the range of 0. 2~ 1. 9 m in the depth direction of the camera, which can meet the application requirements of the urban subway tunnel construction.

Abstract:Aiming at the influence of upstream elbow pipe flow field changes on the measurement accuracy of ultrasonic flowmeter, CFD is used to numerical simulate the flow field inside the measuring pipe, and a rectifier is designed to improve the obvious secondary flow and eddy current caused by the bend, so as to reduce the measurement error of ultrasonic flowmeter. The research object is DN15 ultrasonic liquid flowmeter based on time difference method. The flow range is in the range of 0. 1~ 1. 5 m 3 / h. and the distance between the upstream elbow and the flowmeter is 2 ~ 20D. Compared with the measurement errors of ultrasonic flowmeter before and after installing the rectifier, The experimental results verify that the flowmeter has a greater measurement error with the shorter the straight pipe segment when the flowmeter is not rectified. The installed rectifier can improve the speed distribution of the flow field in the pipeline, shorten the length of the straight pipe section to 10D, improve the ultrasonic flowmeter measurement error to meet within ± 1. 5% , verify the correctness of the numerical simulation, and have certain guiding significance for the practical application of the project.

Abstract:Aiming at the ill posed and ill conditioned problems in image reconstruction when electrical capacitance tomography technology is applied to the detection of industrial industrial multiphase flow pipeline, an improved regularized half threshold algorithm is proposed. Taking L1 / 2 norm as the penalty function, the threshold operator in the semi threshold iterative algorithm used to solve L1 / 2 norm is improved. Taking the solution vector of Landweber algorithm with acceleration term as the correction vector, the improved semi threshold regularization model is introduced to optimize the accelerated Landweber algorithm. The experimental results show that the average correlation coefficient of the improved regularized half threshold algorithm in the reconstructed image is as high as 0. 91, the average image error is reduced to 0. 21, and the imaging speed is maintained at 0. 04 s. For complex flow pattern identification, the correlation coefficient of the improved algorithm is increased by 21. 67% and the relative error is reduced by 37. 01% compared with the traditional Landweber iterative algorithm; Compared with Tikhonov regularization algorithm, the correlation coefficient is increased by 22. 61% and the relative error is reduced by 37. 08% ; Compared with the half threshold algorithm, the correlation coefficient and relative error are increased by 14. 85% and reduced by 28. 26% respectively. The results show that the improved regularized half threshold algorithm has a good application prospect for ECT research.

Abstract:The image total station cannot achieve the fully automated measurement of the target point in the prism-free cooperative working mode. To address this issue, a fully automated measurement method of image total station based on the improved YOLOv5 algorithm is proposed. The YOLOv5 algorithm fused with the convolutional block attention module is used to realize the wide-angle camera identification and detection of the reflector. And the target automatic aiming algorithm is applied to realize the accurate aiming of the telephoto camera at the center of the reflector, which realizes the fully automated measurement of the position coordinates of the target point. With the help of the self-developed image total station, the identification and detection experiment of the reflector and the fully automated measurement experiment of the target point are carried out. Experimental results show that the accuracy of identifying and detecting reflector targets by the improved YOLOv5 algorithm can reach 98. 65% . Compared with manual photometric measurement method, the fully automated measurement method of target point has comparable measurement accuracy and increases the measurement efficiency by 1. 5 times. The proposed method has high measurement accuracy and measurement efficiency, which can be widely used in the unattended and fully automated measurement work occasions.

Abstract:Quartz pendulum reed is the core device of the quartz flexible accelerometer, and the machining accuracy of its key feature size is determined by the machining accuracy of the pendulum substrate. The feature size of the substrate surface is routinely measured with a tool microscope. This measurement process is complex, with poor repeatability and low efficiency. In this article, a method for measuring the feature size of the pendulum substrate based on the regional gray gradient difference is proposed. Firstly, the hardware structure of the visual measurement system is designed and developed. Secondly, the edge detection algorithm of the pendulum substrate based on the regional grayscale gradient is established. Thirdly, the search and classification method of the feature edge points is proposed around the structural characteristics of the pendulum substrate. Finally, the feature size measurement experiment of the pendulum substrate is carried out. Experimental results show that the proposed edge detection method based on regional gray gradient can replace the Zernike moment method to improve the edge detection efficiency when the fitting deviation of the center of the pendulum substrate is less than 3 μm. The measurement system can complete the measurement of the characteristic size of the pendulum substrate within 2 s, and the repeatability of the measurement of the characteristic size of the pendulum substrate in different postures is better than 8 μm. Compared with the tool microscope, the reliability and repeatability of the measurement results are improved, and the measurement efficiency is significantly improved.

Abstract:Small pedestrian targets at long distance have problems of few pixels and lack of texture information. The deep convolutional neural network is difficult to extract fine-grained features of small objects. This article proposes a long-distance pedestrian small target detection method. Firstly, based on the YOLOv4 algorithm, the shallower features are introduced to improve the feature pyramid to extract fine-grained features of pedestrian small objects. An adaptive feature fusion layer is proposed based on the gravity model to increase dependency between multiple feature layers and prevent the loss of small target feature information. Then, ESRGAN is utilized to increase pedestrian small target features number and improve pedestrian small target detection accuracy. Finally, the small pedestrian targets are selected with a proportion of 0. 004% ~ 0. 026% in the image pixels to establish the self-made data set. Compared with Faster RCNN, ION, and YOLOv4, the mAP0. 5 values are increased by 25. 2% , 26. 3% and 11. 9% . And the FPS reaches 24. The research results have important application value in the field of long-distance security monitoring

Abstract:The dense stereo matching in underwater binocular stereo vision imaging does not meet the epipolar constraint in air. To address this issue, an underwater corresponding point matching and the 3D measurement method are proposed. The stereo image collected by the underwater binocular camera can be corrected to an image pair conforming to the principle of coplanar line alignment. Then, the display image of the underwater left and right cameras can be obtained by applying the mature stereo matching method in the air to realize the 3D reconstruction of the underwater target. Firstly, the sum of all the light entering the camera is regarded as a light field, and each light is modeled by a 4D light field expression. Based on this, the plane refraction imaging model and the binocular stereo vision model of the camera are established, and the direction vector of the light is calculated. According to the light field expression, the light is transformed into the form of a point vector, the pixel coordinates of any image point on the direction image corresponding to the original image are calculated, and the position mapping relationship is determined. Through interpolation, the left and right images in line with the line alignment principle can be quickly obtained. Finally, the display map corresponding to each light is obtained. Simulation results show that the line alignment error of the direction image is less than 0. 8 pixel. In the pool experiment, the ball calibrated in advance is used as the target, and the random scattered point active projection is used to increase the texture information of the target surface. The root mean square error of multiple measurements of the target ball is 2. 8 mm, which has high measurement accuracy.

Abstract:In order to realize the reliable autonomous landing of unmanned aerial vehicle (UAV), a visual target composed of multiple sets of concentric circles with different radius ratio is proposed. Considering the distortion of concentric circles after imaging, a recursive circle center detection method that combines with the harmonic ratio constraint composed of circle centers, vanishing points, and intersection points with concentric circles is firstly designed to get the sub-pixel projection point of the center of concentric circles. Compared with Hough circle detection algorithm and other traditional methods, this method can extract the sub-pixel coordinates of the projection point of the circle center more robustly and accurately. After this step, the initial pose is calculated from the correspondence between the image coordinate and platform coordinate, where the correspondence is constructed based on the cross-ratio invariance. The optimized pose is derived from the nonlinear optimization function that is designed based on the conic re-projection model. To eliminate the influence of motion blur images on the decision making during the landing process, a measurement keyframe selection model based on motion continuity is proposed. Furthermore, a multi-mode switching control structure is designed to realize the motion prediction of landing platform, as well as the generation and updating of the spline trajectory of UAV, thereby achieving the autonomous landing of UAV. In 1 500 measurement experiments, the average re-projection error of this measurement method can reach 0. 578 pixel with a variance of 0. 009 6. In the field landing experiments, the positioning error of UAV to the landing platform at the height of 2. 5 m height is less than 3. 5 cm, which indicates that the proposed method has higher visual measurement accuracy and stability, and is able to achieve stable approach and tracking of the landing target and autonomous landing.

Abstract:When the traditional 2D edge detectors are applied to detect object edges in low-resolution depth images, the detection accuracy is poor and the recall rate is low. At present, the existing edge extraction methods based on the 3D point-cloud data have poor real-time performance and weak anti-interference ability. To address these issues, an edge optimized extraction method based on the gradient clustering is proposed to fast and stably detect the 3D edges of objects from the organized point-cloud data. First, the flying pixel noise is filtered to eliminate false detection on the edge by analyzing the distance between neighborhood points. Secondly, an edge / noedge point separation method based on the gradient clustering is proposed to fast extract the rough edges of objects. Finally, the combination of the fast parallel thinning and the mask filtering is employed to optimize the rough edge. In this way, the precise edges are obtained. Experiments are implemented on the public datasets and a dataset collected by a TOF depth camera to evaluate the proposed method. Results show that the proposed method is superior to the existing methods in the real-time and detection accuracy. With the real data, the edge detection is accuracy 89% , and the FPS achieves 28 fps.

Abstract:The rate-distortion optimization plays a key role in video coding, which aims to achieve a tradeoff between compression efficiency and video quality distortion. The existing rate-distortion optimization algorithms mainly aim to eliminate time and space redundancy, which ignore subjective perception of video content and result in a large amount of perceptual redundancy in video. To address these issues, a rate distortion algorithm based on visual perception is proposed in this article. Firstly, the Lagrangian multiplier factor is obtained based on the data-driven just noticeable distortion prediction mode, which is more in line with the perception of the human eye. Secondly, the Lagrangian multiplier weight coefficient is based on salient model. Finally, the fusion of two models is applied to rate-distortion optimization, and SW-SSIM is used to evaluate video quality and achieve perceptual video coding optimization. Compared with the third generation audio and video coding standard algorithm, experimental results show that the proposed algorithm reduces bitrate by 12. 15% averagely, and the salience weighted-structural similarity index metric increases by 0. 004 3. Furthermore, the proposed algorithm reduces the perceptual redundancy in video content, and improves the video perceptual quality and coding performance.

Abstract:In order to reduce the influence of the change of muscle contraction force on EMG pattern recognition, this paper proposes the feature of DCSP. Firstly, the spatial projection matrix that maximizes inter-class distance is obtained by the CSP. Then, the new signal after projection is differentiated and normalized. Finally, the data are projected into the low-dimensional space with the smallest intraclass distance and the largest inter-class distance by the uncorrelated linear discriminant analysis, the EMG gesture based on DCSP feature is verified on two datasets. The first dataset contains data from 10 complete limb subjects, and the second dataset contains data from 9 upper limb amputees. Among the four schemes of recognition rate testing, the recognition accuracy of the DCSP feature in this paper is higher than that of the CSP feature, and the highest recognition rate is achieved in all force training and all force testing schemes ( dataset1: 95. 83% , dataset2: 86. 93% ). Compared with CSP feature ( dataset1:89. 01% , dataset2: 70. 03% ), the classification accuracy rates are increased by 6% and 16% , respectively. In the two test schemes of feature spatial distribution, the DCSP feature has a smaller intra-class distance and a larger inter-class distance than the CSP feature. In the comparison results of other studies, the DCSP feature improves the recognition accuracy by about 5% (dataset1) and 8% ( dataset2) compared with the existing robust features, and the performance does not depend on the classifier.

Abstract:The sign language is a complex motion pattern with dynamic changes in various gestures. The effect of gesture feature processing is directly related to the accuracy of sign language recognition. In this article, a new dynamic gesture feature processing method based on the estimation of improved S-transform (IST) spectral of surface electromyography (sEMG) is proposed. The collected sEMG signal is transformed by S-transform, the optimization factor is introduced to adjust the time-frequency resolution, and the IST spectrum is generated. The time and frequency components of the IST spectrum are defined as 2-D random variables, and the matrix elements of the IST spectrum are taken as 2-D random variables. The 2-D kernel density function is obtained by Gaussian kernel density estimation. Simulation and experiments show that the estimation method of the IST spectrum effectively suppresses white noise and strengthens the sEMG transient mutation characteristics of dynamic gestures. Compared with empirical mode decomposition, selfpermutation entropy, and singular value permutation entropy, the accuracy of dynamic gesture recognition based on this method is improved by 10. 0% , 6. 67% and 11. 67% , respectively.

Abstract:In the current information age, the traditional paper-based Braille books have many deficiencies in disseminating knowledge, and are gradually unable to meet the Braille reading needs of the visually impaired. Inspired by wearable devices, this article designs a fingertip Braille display system based on touch screen interaction. According to the structural characteristics of standard Braille, the way visually impaired people perceive Braille, and the miniaturization requirements of fingertip devices, this article designs a new type of Braille dot actuator and optimizes its structure and parameters through finite element analysis. The results of performance analysis and user experiments show that the proposed Braille dot actuator has the advantages of high refresh rate, low power consumption and small size, which could achieve locking operations at the highest and lowest positions. The Braille display system combines the portable fingertip device with the touch screen that can refresh data, and employs the vibrotactile feedback of four piezoelectric ceramic actuators to guide the movement direction of the finger on the touch screen, providing a new way for visually impaired people to read Braille conveniently.

Abstract:It is difficult to extract the weak information of rolling bearing fault under complex and changeable operating environments. To address this issue, an optimized resonance sparse decomposition (RSD) based on maximizing the cyclic pulse index (CPI) is proposed, which takes full advantage of pulse characteristics and cycle period characteristics of fault impacts. Firstly, the variation coefficient of the short-time pulse peak moment is used as the CPI to comprehensively characterize the pulse and periodicity of the bearing fault signal. Then, the quality factors of RSD are optimized by the multi-scale simplified particle swarm optimization algorithm with the objective of maximizing the CPI. Finally, the cyclic pulse spectrum of the low-frequency resonance component is established to automatically identify the bearing faults. The results of simulation and applications in the fault diagnosis of EMU axle box bearings show that the proposed method effectively avoids the misjudgment of resonance frequency band caused by strong pulse interferences, and performs well in the synchronous diagnosis of bearing compound faults under complex working conditions, which demonstrates its engineering applicability in the field of bearing fault diagnosis.

Abstract:An extended magnetic charge model considering the magneto-mechanical coupling effect and dislocation pinning effect is constructed to improve the effectiveness of the weak magnetic detection method for early damage diagnosis of buried pipelines, and further clarify the quantitative relationship between detection signals and stress and defects. The variations of weak magnetic detection signals caused by various abnormal conditions of pipelines are studied. The effectiveness of the model is evaluated by engineering detection experiments. Research results show that the gradient modulus GM has significant symmetrical peak distortion at the volume pit defect, area crack defect or weld defect. The peak value of the gradient modulus GM increases with the increase of defect size, especially the equivalent depth. The influence of weld misalignment defect on the peak shape and peak value of GM is significant, while the influence of undercut defect is not obvious. Comparing the engineering measured data and theoretical calculation data, the maximum position error is ≤1 m, and the maximum peak error is ≤ 15% , which are both within the acceptable range of engineering detection.

Abstract:When the circumferential scan is performed along the outer surface of pipeline by the ultrasonic time-of-flight diffraction (TOFD) technique, the layered dead zone is generated in the near-surface area due to the influence of the pipe curvature and the pulse width of direct longitudinal wave (DLW). In this article, the adaptive deconvolution method is applied to perform deconvolution and autoregressive spectrum extrapolation by selecting the sub-band DLW, whose main frequency is close to that of the overlapped signals, as the reference signal, realizing the extension of effective frequency band and the pulse compression of time-domain signal. Meanwhile, the defect depths are determined according to the tip-diffracted waves in circumferential scan images. Experimental results show that the range of layered dead zone is reduced by about 60% under the condition of 5 MHz central frequency and 87 mm probe center separation (PCS) for the carbon steel pipelines with 100. 0 mm outer wall radius and 30. 0 mm wall thickness, and with 148. 0 mm outer wall radius and 27. 0 mm wall thickness by the adaptive deconvolution method. When the distances from defect tips to the ray path of DLW are no less than 4. 0 mm, the measurement error of the depths for the defect is within 10. 6% . Compared with the conventional spectrum analysis method and the autoregressive spectrum extrapolation method, the adaptive deconvolution method has better performance in reducing dead zone and can accurately quantify the defects being difficult to detect with measurement error within 5. 8% .

Abstract:Aiming at the statistical distribution of azimuth detection of ultra-short-range incoming targets, this article studies the influence mechanism of periodic scanning magnetic signal on the detection accuracy of the laser-magnetic composite azimuth measurement method. The mathematical model of the permanent magnet rotating scanning space magnetic field is formulated, and the periodic scanning magnetic signal equation is derived. Combined with the characteristics of the magnetic signal and the constant threshold time measurement method, the analytical formula of the probability and statistical distribution function of the time measurement of the periodic scanning magnetic signal is deduced. The effects of scanning period, magnetic field intensity, threshold voltage, and noise on the probability and statistical distribution of time measurement are studied. The results show that with the increase in scanning period and magnetic field intensity, the symmetry of the probability distribution function is not affected, the half-width of probability distribution decreases, and the peak value of distribution increases in the range of 0. 75~ 1. 88. With the increase of the threshold detection voltage, the probability distribution firstly shows a trend of decreasing half-width and the peak value is increased by 0. 48, then increasing halfwidth and the peak value decreased by 0. 54, and the rising and falling edges of the distribution curve before and after the peak have different trends. With the increase of equivalent noise voltage, the symmetry of the probability distribution function is not affected, but the distribution half-width increases and the distribution peak decreases in the range of 0. 4~ 0. 48.

Abstract:Autonomous learning of robot pushing and grasping skills in the cluttered environment has been widely studied. The cooperation between them is the key to improving grasping efficiency. In this article, a deep reinforcement learning algorithm GARLDQN based on the generative adversarial network and model generalization is proposed. Firstly, the generated adversarial network is embedded into the traditional DQN to train the coevolution between pushing and grasping. Secondly, some parameters in MDP are formulated based on the goal object, and the hindsight experience replay mechanism (HER) is used for reference to improve the sample utilization of the experience pool. Then, according to the image state, a random (convolution) neural network is introduced to improve the generalization ability of the algorithm. Finally, 12 test cases are designed and compared with the other four methods in terms of grasp success rate and average motion times. In the regular block cases, two indicators are 91. 5% and 3. 406, respectively. In the daily tool scene, two indicators are 85. 2% and 8. 6, respectively. These results show the effectiveness of the GARL-DQN algorithm in solving the problems of robot pushing and grasping cooperation and model generalization.

Abstract:To meet the demand for fast conversion of Buck converter from standby or light load to large load state, based on the dynamic performance of the Buck converter with the traditional average current control method, the improved hybrid control strategy with sliding mode controller and PI controller suitable for DCM/ CCM process is proposed. According to the system state and conduction mode, the PI controller or the sliding mode controller is utilized in the voltage outer loop. In addition, the PI controller is used in the steady-state condition, which is then converted to the sliding mode controller according to the conduction mode in the dynamic condition of the load increase. Furthermore, the accurate judgment of the conduction mode is realized by the unified corrected average inductance current. Thus, the proposed hybrid control strategy can effectively combine the steady-state and dynamic performance advantages of PI control and sliding mode control. The simulation and experimental results show that, compared with that of the traditional average current control, the dynamic response time of this strategy is reduced by about 65% and the voltage drop is reduced by more than 35% .

Abstract:To further improve the control accuracy of the traditional variable structure active disturbance rejection control device and enhance the anti-interference ability of the permanent magnet servo drive system, an improved variable structure active disturbance rejection control strategy is proposed. This method introduces the observation errors of position and velocity into the extended state observer designed based on the variable structure principle to realize the error-free estimation of the state variables, adopts the nonlinear state error feedback control law based on the exponential reaching law design to realize the smooth transition between linear control and nonlinear control, introduces position tracking error on this basis to improve the tracking performance of the servo system. Through experimental analysis, two control strategies of improved variable structure active disturbance rejection control and traditional variable structure active disturbance rejection control are compared. Results show that the position tracking error of the improved control strategy is reduced by about 30% compared with the traditional control strategy. When the load changes suddenly, the tracking error of the traditional control strategy is about 3. 4 times the maximum tracking error before the load change, while the improved variable structure ADRC strategy can still track the given signal accurately.