Abstract:Star sensor is the key instrument for high accurate satellite attitude measurement and flight control. Aiming at the special requirements of low star magnitude of threshold and large field of view, a star sensor optical system is designed, whose focal length is 55 mm, relative aperture is 1. 1 and field of view is 17°×17°. Optical passivity athermalization design is achieved through temperature property matching and optimization of optical glass material and mechanical structure material based on athermal glass map method. The stray light suppression design of two-stage shield and baffles are completed. The simulation analysis and calculation of the ghost distribution of imaging ray in view field and the stray light out of view field are completed with no-sequence ray tracing method. The results show that the RMS of the dispersion spot radius for each view field of the star sensor optical system is less than 4. 5 μm, the encircled energy within 2 × 2 pixels is larger or equal than 96% , the optical MTF in the field of view is larger than 0. 6 at cut-off frequency in the large temperature difference from -35℃ to 45℃ and the distortion is better than 0. 05% . The point source transmission ratio of stray light out of 45° avoided angle reaches 10 -8 . The radius of every order ghost spot at image plane is greater than 0. 8 mm. The optical imaging quality and stray light suppression of the star sensor can meet the requirements of high accurate attitude measurement.

Abstract:The large-scale distributed measurement system is based on the principle of intersection of multi-source observation. The relative position and pose relationship among the measurement nodes directly determines the overall performance of the system. The workspace Measurement Positioning System ( wMPS) is a typical representative of distributed measurement systems. Its orientation methods rely on the interchangeability of target reflector of the laser tracker to construct geometric constraints, which results in inevitable centering error existing at the receiving node of the system and affects the orientation accuracy. This paper studies a distributed system orientation method based on mobile cooperative target. Through designing a cooperative target integrating multiple receiving nodes, there is no need to accurately determine the structural relationship among the target receiving nodes, but to use the relative invariance of the spatial positions of the receiving nodes as rigid constraints to establish the relative position and pose relationship model of the base station with redundant optical plane intersection information. Combined with the initial value calculation model of resection orientation, the orientation result that is not affected by the interchangeable centering error of the receiving node is obtained. Finally, wMPS is used as the experiment verification platform. The results show that this orientation method avoids the high-precision interchangeability of target reflector, has good stability and effectively decreases the influence of the interchangeable centering error of the receiving node. Compared with traditional orientation method using standard ruler, the reference length measurement accuracy of this orientation method is improved by about 40% .

Abstract:The preload torque is an important factor that affects the reliability of the ball screw. At present, the calculation of preload torque is mainly based on the empirical formula. There is a large error between the theoretical calculation value and the actual preload torque value. To calculate the preload torque of the ball screw accurately, based on the contact deformation and load distribution theory, the influence of key machining dimensional errors on the preload torque is analyzed. The relationship model among pitch diameter error, lead error and preload torque of ball screw is formulated. The lead error and pitch diameter of different types of ball screws are measured by the ball screw lead error and the profile detection test bench. The theoretical value of preload torque of ball screw is calculated through the lead error and pitch diameter error. The actual value of preload torque is measured by the preload torque test-bed of ball screw. The measurement results show that the relative error between the theoretical calculation value of preload torque and the test value is 0. 65% ~ 12. 18% , which is less than 15. 07% ~ 45. 57% of the existing models. The consistency between the theoretical model and the actual friction torque is accurate. The correctness of the theoretical model is verified.

Abstract:Refractive index is the most widely used optical parameter in the optical system, which has an extremely important influence on the optical performance. The optical length consists of refractive index and thickness, which directly affects the time delay characteristics of birefringent devices in the optical system. In this article, a measurement method of refractive index and thickness based on the fiber laser frequency splitting effect is proposed. By rotating the birefringent device inserted into the laser cavity, the birefringent parameters of the device at different angles are measured by using the frequency splitting effect. Based on the refractive index ellipsoid of birefringent device, the relationship among phase delay, refractive index, thickness and rotation angle is established. The refractive index and thickness parameters of the device are achieved by fitting calculation. Experimental results show that the thickness measurement error of birefringent elements is 210 nm, and the intrinsic refractive index measurement error is 10 -5 . It could be widely used in the measurement of intrinsic refractive index and thickness of birefringent elements in the infrared band.

Abstract:The gear involute artefact (GIA) is the benchmark for the trace of involute gear profile deviation and the value-transmission of involute profile. It is the standard for calibrating the instruments for measuring gear involute profile. However, there is no class-1 GIA that can meet Chinese national standard of GIA GB/ T 6467- 2010 at present. Requirements of the evaluation range and profile form deviation of class-1 GIA are extremely strict. The arc length within the non-evaluation range of the tooth root is only about 0. 1 mm, which is easy to be affected by the machining error. It may result in undercutting or incomplete machining of the tooth root. Then, the tolerance of profile form deviation is exceeded. In this article, the mathematical model of the relationship between the initial rolling angle error and the involute profile deviation at the tooth root is formulated. The reason for the out of tolerance of the profile form deviation at the tooth root is analyzed, and the model is verified by the Siemens NX motion simulation and CAD simulation. Based on the above model, a machining method for tooth root at GIA with gradually modified is proposed, and a new kind of GIA with three base circle parameters is processed by the proposed method. The profile form deviations f fα < 1 μm within the evaluation range of four teeth are processed, which meets the requirements of Chinese national standard of GIA GB/ T 6467- 2010 for the evaluation range and the profile form deviation of class-1 GIA. This research provides a process method support for the precision manufacturing of class-1 GIA.

Abstract:BeiDou navigation satellite system (BDS-3) has begun to provide stable high precision positioning, navigation and timing service for global users. In this paper, based on China′s time reference system of UTC (NTSC), the performance analysis of BDS-3 undifferenced combined carrier phase time comparison is carried out. The performance of BDS-3 undifferenced combined carrier phase time comparison in zero and long baseline, and the BDS-3 and GPS fusion carrier phase time comparison are carried out and studied by using data of BDS-3. Results show that the standard deviation of zero baseline and common clock comparison between two receivers is smaller than 0. 3 ns. For the long baseline time, the time difference between the Asia-Europe two time keeping laboratories by using BDS-3 undifferenced combined carrier phase time comparison, and the fusion carrier phase time comparison have a good consistency with the results released by International Bureau of Weights and Measures (BIPM) with GPS link. The frequency stability and time stability of result are basically the same as BIPM, and the root mean square of the residual is smaller than 0. 25 ns. Experimental results meet the application requirements for sub-nanosecond time comparison.

Abstract:To improve performance of the magnetohydrodynamic angular velocity sensor in detecting signals at low frequency (<1 Hz), a new magnetohydrodynamic angular velocity sensor with multiple magnetohydrodynamic pumps is proposed, which is based on the magnetohydrodynamic angular velocity sensor. In the aspect of improving the mechanical structure, three magnetohydrodynamic pumps are constructed uniformly around the edge of the fluid channel. To enhance the Coriolis force effect at low frequency, the magnetic pole shape of the fluid pump is optimized to improve the magnitude and stability of the radial flow velocity distribution. Compared with the single pump structure, the simulation and experiment results of the designed structure with multiple MHD pumps show that the radial velocity distribution is increased by 20. 28% , and the magnetic field uniformity of fluid pump channel is increased by 38. 36% . Results indicate that the modified sensor with the applied low frequency compensation current can detect the angular velocity signal at low frequency. Through the whole bandwidth, the amplitude fluctuation error is less than 0. 5% , and the phase difference is reduced to the range of 0~10°.

Abstract:The rail crack monitoring is a hot research topic to ensure the safety of railway foundation components. The ultrasonic guided wave has the advantages of small attenuation, long propagation distance and high detection efficiency in the rail, which can realize longdistance rapid detection of the rail. The traditional ultrasonic guided wave sensors are generally made by the piezoelectric ceramics. Due to their brittleness and hardness, the traditional PZT ceramic sensors are not suitable for long-term online monitoring of the rail. In this article, the Flexible 0 ~ 3 type of PZT/ epoxy piezoelectric composite is prepared and used as an ultrasonic guided wave sensor. The mechanical properties of the flexible piezoelectric composite and its sensing characteristics for guided waves are investigated by experiments. Combined with the finite element simulation analysis and experiment test, the mode of the excited guided waves and the delay arrival time of the reflected wave from cracks in the rail are analyzed. The applicability and effectiveness of the piezoelectric composite sensor for rail crack guided wave signal detection are discussed. Results indicate that the PZT/ epoxy composite sensor has well flexibility and linear sensitivity, which is adaptable to detect the guided wave under ambient temperature between 5℃ and 75℃ . The guided wave detection signals obtained from different sensing paths of the rail are consistent with the finite element simulation signals. Their arrival time of the crack reflection waves are same. The research provides a new type of flexible piezoelectric sensing technology for on-line monitoring of the rail crack.

Abstract:The MEMS piezoelectric vibration table can achieve micron-level precision mechanical vibration, which is applied to the onchip calibration of MEMS inertial sensors. However, the thermal expansion coefficients of piezoelectric ceramics and silicon are different, which could result in residual stresses in MEMS piezoelectric beams after manufacturing and forming. The residual stress may lead to the warpage of the MEMS piezoelectric beam and cause the vibration table to have the initial displacement. To analyze the influence of the initial displacement on the vibration characteristics of the MEMS piezoelectric vibration table, the relationship among the residual stress, the deformation of piezoelectric beam and the initial displacement of vibration table is established. Then, the initial displacement is introduced into the vibration model of the MEMS piezoelectric vibration table, and the vibration characteristics of the MEMS piezoelectric vibration table with different initial displacements (10、 20 and 30 μm) are discussed. Experimental results show that under 10 V ( 150 Hz) sinusoidal excitation, the MEMS piezoelectric vibration table with the initial displacement has bistable vibration characteristics and the vibration waveform distortion, and the maximum relative error is 45% . Results are consistent with the theoretical analysis and prove that the residual stress is the main cause of the distortion of the vibration waveform.

Abstract:Aiming at the demand for accurate and reliable positioning of intelligent vehicles in GNSS-denied environments such as indoor parking lots and tunnels, a low-cost fusion positioning method based on factor graph is proposed. Firstly, a UWB/ INS dynamic fusion positioning framework based on factor graph is established; secondly, through making full use of the static characteristics of the UWB base stations, the maximum theoretical distances between the base stations and the intelligent vehicle are calculated via the position and position confidence interval inferred from the factor graph, and compared with the distance measured by UWB to detect the non-line-ofsight signals; lastly, an adaptive fusion rule is formulated to guide the fusion of UWB/ INS under different conditions, the final positioning result is obtained, and the reliable positioning of intelligent vehicles in GNSS-denied environments is achieved. The real vehicle tests show that the proposed fusion positioning method can achieve the positioning accuracy of 0. 622 m, which is more than one time improvement compared with that of the traditional least square method. The proposed method effectively suppresses the influence of non-line-of-sight errors, has the advantages of low cost, high reliability, strong environment adaptability, and overcomes the shortcomings of traditional methods.

Abstract:Adopting integrated optical technology, an integrated optical waveguide current sensor for high current measurement is designed and developed, the size of the device is 78 mm×18 mm×9 mm. The theoretical analysis of the sensor structure with multiple loop antennas and modulation electrodes is carried out. The COMSOL Multiphysics software is used to establish a three-dimensional model of the sensor, the simulation results show that the designed sensor can be used to measure the pulse high current. The experiment system for pulse high current measurement was built to test the performance of the sensor. The results show that when the pulse current time domain waveform measured by the sensor is compared with that measured by the Pearson current transformer, the mean relative errors of the wavefront time Tf and the duration time Td are 3. 977 5% and 5. 437 5% , respectively. The linear correlation coefficient is 0. 998 97 in the measured current range of 100~3 300 A. When the SNR is 6 dB, the minimum measurable current of the sensor is 66 A.

Abstract:Aiming at the difficulties of the harsh environment of the tunneling working face and the frequently happened slippage movement of the roadheader, which results in the fact that the navigation parameters of the roadheader cannot be accurately obtained, an autonomous navigation method of roadheader aided by two-dimensional mileage is proposed. A two-dimensional mileage measuring device is developed and an autonomous navigation system of roadheader aided by two-dimensional mileage is built. The error equations of the two-dimensional mileage measuring device and strapdown inertial navigation system are established, and a two-dimensional mileage-assisted integrated navigation algorithm is derived. Then, to deal with the problem that the time-varying statistical characteristic of the measurement noise leads to the increase of the system estimation error, a data fusion algorithm based on fuzzy theory is proposed to adjust the noise covariance matrix on line and in real time. The simulation experiment results show that compared with the traditional fusion algorithm, the fuzzy fusion algorithm can reduce the heading angle error by 34. 78% , and the positioning errors of three directions by 44. 33% , 41. 82% and 42. 26% , respectively. The navigation and positioning experiment of roadheader was conducted, a positioning accuracy of 0. 052 m is achieved, which meets the autonomous navigation requirements of roadheader in unmanned tunneling working face.

Abstract:There is large error caused by the burst noise of the received signal strength (RSS) in the indoor fingerprint positioning. To address this issue, a subspace matching method combined with the algorithm of clustering by fast search and find of density peaks (DPC) is proposed. In this way, the large positioning error could be effectively avoided. Specifically, the coverage vector of the access points in the online RSS is used to select a subset of the reference points and the subspaces of the radio map. An improved weighted K-nearest neighbors ( WKNN) approach is applied to achieve estimation. Then, the DPC algorithm is used to select S estimated positions with the largest decision values to determine the position of the target. The simple algorithm needs no learning process in the offline stage, which is especially suitable for large indoor areas with a lot of access points. Compared with the WKNN algorithm, experimental results show that the proposed method improves the positioning accuracy by about 25% . The large positioning error of more than 4 m is eliminated when the reference points is 1. 8 m×1. 8 m. The overall positioning performance is improved effectively.

Abstract:The application of radio frequency identification (RFID) technology is becoming wider and wider in the field of industrial detection. The performance of RFID tags in multi-metal environments is also required to be higher and higher. This article proposes a high-efficiency RFID tag antenna that can roll arbitrarily on metal surfaces. The designed tag antenna evolves from the inverted F antenna (IFA). The antenna uses four radiating arms in the columnar layout. One set of opposite arms is connected to the coaxial inner conductor through the feed plane, the other opposite two arms are connected with the outer conductor of the coaxial line, and a slotted rectangular auxiliary plane is placed in the four radiating arms. The overall size of the tag antenna is 11. 3 mm×11. 3 mm×77. 5 mm, the frequency range is 897~ 932 MHz when rolling in various positions on a metal plane, the bandwidth range of | S11 | ≤-10 dB is 3. 5 ~ 8 MHz, and the antenna radiation efficiency reaches 70% . The tag antenna has the advantages of rolling freely in a metal plane, miniaturization, and high efficiency. The proposed tag antenna is connected with the temperature sensor chip and can be used for temperature detection in the complex environment of the industrial internet of things.

Abstract:The reliable operation or not of smart energy metering equipment affects the accuracy of grid edge measurement and electricity metering. For this reason, this paper proposes an equipment degradation trend analysis method based on parameter optimization BP neural network. Combining with the State Grid Xinjiang High Dry Heat Test Base and the basic error data of the real-time operation of the smart energy metering devices, the Spearman correlation analysis method is used to extract the main environmental stress that affects the basic error value of the smart energy metering equipment; the function fitting interpolation (FFI) is used to eliminate the influence of missing values in the original data on degradation analysis. A BP neural network-based smart energy metering equipment degradation research model is established. Finally, an improved genetic algorithm (IGA) is introduced to optimize the BP neural network parameters to realize the backward forecast and update of the degradation trend of smart energy metering equipment. Several smart energy metering devices with different types in the base were selected to conduct various kinds of experiments. The results show that the model proposed in this paper has a high predictive ability. The average root mean square error of the prediction results is 0. 012 3, and the prediction accuracy is up to 90. 2% .

Abstract:The high precision and localization of weapons and their test systems are the serious problems that our country has to face. When the sampling rate of the data acquisition system reaches the GSPS level, current domestic ADC chips are hardly applicable. Therefore, under the prerequisite of not reducing sampling accuracy, a method of array sampling with multiple ADCs is proposed to improve the sampling rate of the system. Aiming at the channel mismatch problem existing in TIADC systems, an adaptive error estimation algorithm based on first-order statistics and correction method of improved Farrow time delay filter is proposed, and a new estimation strategy for the time error estimation of the 4-channel TIADC system is proposed. Simulation and experiment results show that the algorithm can accurately estimate the time mismatch error and effectively suppress spurious components. Compared with the data before correction, the spurious-free dynamic range is increased by 20 dB. The proposed method has important practical significance for solving the localization issue of high-speed and high-precision data acquisition equipment.

Abstract:According to the relevant standards of the international electrotechnical commission ( IEC), the utilization of pulse and sinusoidal voltages to determine the partial discharge inception voltage (PDIV) of the inverter motor insulation system is a key method to evaluate the insulation performance of low-voltage random-wound inverter-fed motors. However, the current IEC standard is difficult to identify the stator insulation weakness. To solve this problem, combines the PDIV off-line test technology under the repetitive pulse and the sinusoidal voltage, a three-step method is proposed to locate the weak point of the inverter-fed motor insulation. Firstly, the repeated pulse voltage is used to measure the PDIV of turn-to-turn insulation in the motor. Secondly, the PDIVs of phase-to-phase insulation and main wall insulation are measured by the sinusoidal voltage, which are compared with the turn-to-turn insulation PDIV. In this way, the weak point of the motor insulation can be determined. Finally, through the PDIV test in different grounding modes of the motor, it is possible to locate whether the weak point of insulation is located at the stator winding head. In addition, by simulation of stator voltage distribution, weak point identification of three commercial motors and corona life test of motors, the feasibility of the above method is comprehensively verified. The research can effectively locate the insulation weakness of the inverter-fed motor. And the efficiency of diagnosing potential risks in the motor insulation is improved.

Abstract:Aiming at the high missed detection rate of casting defects, a casting defect detection method based on deep learning model fusion is proposed. Firstly, the Faster RCNN network is improved, the feature pyramid structure is used to improve the feature extraction network module, multi-scale feature fusion is realized, and the feature extraction of casting defects is completed. Then, the ROI pooling layer in the network is improved based on ROI Align, and the IOU score is introduced into the judgment process of NMS algorithm. And the improved network is integrated with Cascade RCNN and YOLOv3. Finally, an experiment study was carried out to verify that the fusion model can effectively reduce the missed detection rate of casting defects. The experiment results show that the defect recall rates in the Faster RCNN model and the network model proposed in this paper are increased by 1. 73% and 4. 08% , respectively after the pooling improvement of the region of interest. Using the method of model fusion, in the condition without considering the classification accuracy, the defect recognition rate of the entire model reaches 95. 71% . Compared with single model, while guaranteeing the detection accuracy of casting defects, the method also improves the defect detection recall rate and meets the requirements of industrial applications.

Abstract:Electrical impedance tomography technology ( EIT) is one of the methods to solve the corrosion localization problem of grounding grid. In order to improve the ill-posedness of EIT inverse problem of grounding grid, improve the stability of the solution and the clarity of the reconstructed image, a hybrid regularization ( Tikhonov-TV) grounding grid imaging algorithm combining classical Tikhonov regularization and total variation regularization (TV) is proposed. Firstly, on the basis of cyclic measurement principle, this paper innovatively uses COMSOL and MATLAB co-simulation to obtain the voltage data of grounding grid EIT forward problem model. Secondly, on the basis of theoretical analysis, the field resistivity distributions of the two grounding grid inverse problem models based on prior topological information and unknown topological information are solved respectively with Tikhonov-TV regularization EIT algorithm. Finally, the EIT reconstruction images of grounding grid for Tikhonov, TV and Tikhonov-TV three regularization algorithms were compared with simulation and experiment, and the resistivity mean square error (resistivity MSE) and transversal resistivity curve were adopted to measure the image quality, the experiment results show that the resistivity MSE in corrosion condition at locations 1 and 2 places based on prior topological information reaches 1. 27×10 -15 and 1. 59×10 -15 respectively, and the resistivity mean square error is the minimum. The results show that the Tikhonov-TV regularization algorithm proposed can effectively improve the ill-posedness of EIT inverse problem, and achieve the best convergence performance. The reconstructed images are better than those of Tikhonov and TV regularization algorithms.

Abstract:Tool magazine and automatic tool changer are key functional components used for storing and exchanging tools of CNC machine tools. Their performance condition, especially the tool change accuracy, directly affects whether CNC machine tools can accurately and quickly complete tool change. They are key factors that affect the reliability of CNC machine tools. By setting up the tool change accuracy test bed of the tool magazine and automatic tool changer on CNC machine tools, the measuring method for tool changing accuracy is proposed based on two laser displacement sensors. Then, the positioning parameter calibration method of the laser displacement sensor is proposed by means of the position of machining center workbench, which can be precisely controlled by the numerical control system. By developing the tool change accuracy test method, one million tool changing tests are implemented to reveal the change rule of the tool change accuracy of the tool magazine and automatic tool changer with load and test times, and the change law of the precision of the holder before the insert, before the broach and after the broach. In this way, an important reference is provided for how to reduce the influence of the tool holder on the spindle in the process of the insert.

Abstract:Magnetic induction tomography, as a new detection imaging technology, has important application value in clinical diagnosis and continuous monitoring of cerebral hemorrhage and other diseases. MIT system consists of excitation coil and detection coils, among which there is only one excitation coil, while there are dozen or even dozens of detection coils in previous studies. Due to space limitations and interference between coils, it has important significance to study how to minimize the number of coils without losing imaging accuracy. Aiming at this problem, this paper proposes a bidirectional detection imaging way using only five detection coils to detect magnetic field information in each other′s vertical direction. The complex head model simulation experiments and heterogeneous detection experiments on a single heterogeneous body with radius of 15 mm and two heterogeneous bodies with radius of 7 and 15 mm were carried out, respectively, and the results show that the bidirectional detection imaging way can not only counteract the effect of reducing the number of detection coils, but also have high noise resistance performance and imaging accuracy, which lays a technical foundation for the practical application of MIT.

Abstract:The dynamic characteristics of the milling process system respond to the variation tool clamping length, which affect the related milling stability and machined surface quality. In further, they lead to the uncertainty of milling process planning. To address this issue, the tool overhang length and traditional milling parameters are combined as the inputs. Then, two back propagation neural networks (BPNNs) are introduced to predict the limiting axial cutting depth and the surface roughness, respectively. Furthermore, two BPNNs are used to express the constraints of milling stability and machining quality. Then, an optimization model is formulated to decrease the total cutting time of the rough and finish milling processes, where the tool overhang length and the milling parameters for rough and finish processes are designed as the optimization variables. The sparrow search algorithm is used to solve the optimization model. A case study is implemented on a CNC machine tool to milling the fixture cavity. The optimal configuration of tool overhang length and milling parameters are achieved. The total cutting time is 12. 577 min and the measured surface roughness is 3. 01 μm, which verify the feasibility and effectiveness of the proposed optimization model.

Abstract:The existing stereo simultaneous localization and mapping ( SLAM) methods all use standard stereo cameras, and the assumption of the static environment has influence on their accuracy in dynamic environment. A multi-focal dynamic stereo vision SLAM is proposed. It could overcome the insufficiency of standard stereo cameras that cannot perceive the scene at long distance and wide field of view. The impact of dynamic objects is also removed. To be specific, the stereo calibration method is improved and the calibration parameters are utilized to rectify ORB features instead of rectifying stereo images. For multi-focal stereo images, a feature extraction and matching method is also proposed to increase the number of matched features. Multi-view geometry, regional feature flow and relative distance are used to detect dynamic objects. The feature points on the dynamic objects are eliminated. Compared with ORB-SLAM3 and DynaSLAM, the positioning accuracy of the proposed method on the public data set KITTI is increased by 6. 97% , and the positioning accuracy on the self-made data set is increased by 26. 64% and 32. 09% , respectively.

Abstract:As an essential part of the power transmission, the quality of the cable core is important for power safety and transmission efficiency. However, the knife mark texture in the cross-sectional image of the cable core affects the image segmentation effects seriously, which could reduce the measurement accuracy while using visual measurement for cable structural parameter detection. To solve this problem, this article proposes an improved frequency domain filtering to eliminate the knife mark texture of cable core crosssection image. Firstly, the cable core cross-section images are decomposed by the relative total variation (RTV) model to obtain the texture image. Then, the texture image is processed by fast Fourier transform to achieve the main energy direction of the texture. Finally, the Butterworth notch filter with oval-shaped neighborhood narrow stopband is designed to reduce the energy of the corresponding band-shaped area. Experimental results show that the elimination rate of knife mark texture is 92% , which proves the effectiveness of the proposed method.

Abstract:The accurate positioning of the alignment mark is key to wafer bonding accuracy. To improve the accuracy, speed and adaptability of the alignment mark positioning algorithm, a high-precision alignment mark positioning algorithm based on edge detection and fitting positioning is proposed. The method combines the hierarchical pyramid model to locate the alignment mark stepwise, and achieve the pixel-level coarse positioning result of the target area. According to the coarse positioning result of the target area, the method combines the Canny detector and the improved Gaussian fitting method. In this way, the sub-pixel contour of the alignment mark is fitted. On this basis, the center position of alignment mark is quickly fitted by the center symmetrical alignment mark feature and the improved least square method to realize accurate positioning of the alignment mark. The effectiveness of the method is evaluated through experiments. Results show that the proposed alignment mark positioning algorithm can locate the target area quickly while the repeat positioning error is about 0. 06 μm, which satisfies the requirements of wafer bonding alignment in terms of accuracy, speed and adaptability.

Abstract:It is difficult to analyze the shape and position error of internal surface of parts with complex cavity structure. To address this issue, a method of analyzing the shape and position error of complex cavity of parts is proposed, which is based on industrial CT technology. First, the parts are scanned by using industrial CT and reconstructed into the measurement STL model. Then, the design IGES model is discretized into the STL model and the measurement STL model for registration. Secondly, by using the designed IGES model as the guide, the points in the STL model and the belonging of each surface of the IGES model are determined. And all the surfaces in the STL model are automatically segmented and measured. Finally, the surface of IGES model is selected and the point cloud cluster corresponding to the surface is analyzed. The valve body model of the million-dollar measurement point cloud is selected for the experiment. All the surfaces of the point cloud model are completely segmented. The cylindricity, flatness and parallelism of the selected internal surface are measured to be 1. 11 mm, 0. 61 mm, and 1. 92 mm, respectively. In this method, only the input STL model of part measurement and the IGES model of part design are required. The measured surface of IGES model is selected. The internal and external surface of part can be analyzed automatically.

Abstract:The electronic throttle of marine natural gas engines belongs to the category of nonlinear control. And the boundary of highorder sliding mode control is difficult to estimate. To address this issue, a novel adaptive high-order sliding mode control algorithm is proposed. An adaptive strategy based on the phase plane trajectory convergence process is designed. To increase the practicability of the control algorithm, the detection region is designed based on the adaptive algorithm. The control gain is dual-directional regulation by judging the relative position of the phase trajectory and the detection region to prevent the control accuracy from reducing due to excessive gain. Especially, a robust differentiator is used to estimate the derivative of the sliding variable in the controller. Finally, three test schemes are designed to compare the algorithm with the traditional high-order sliding mode algorithm. Experimental results show that the proposed algorithm can improve the system response speed by 35% and reduce the root mean square error by 37. 5% under the step signal. For the sinusoidal signal, the maximum steady-state error and the root mean square of steady-state error are reduced by 30% and 22% , respectively.

Abstract:The traditional A ∗ algorithm for smart cars has many inflection points that are not smooth and prone to collision. To address these issues, a fusion of A ∗ and DWA is proposed to improve the random obstacle avoidance method. In the A ∗ optimization algorithm, by calculating the slope between nodes, key nodes are selected and redundant nodes are eliminated. The evaluation function is improved. The speed and path smoothness of the A ∗ algorithm are enhanced. The A ∗ optimization algorithm cannot avoid random obstacles and the DWA algorithm has the problems of imperfect speed and safety. To solve these problems, an optimized DWA adaptive algorithm is proposed, which considers speed and safety. Between each two optimized key nodes, an adaptive dynamic window method is used to locally avoid random dynamic-static obstacles. Therefore, smart cars can avoid obstacles autonomously, smoothly, safely and efficiently. Compared with the algorithm of A ∗ , Dijkstra, the improved A ∗ and the basic DWA fusion algorithm, experimental results show that the global path length, number of inflection points, and time are reduced by an average of 2. 9% , 36. 2% , and 24. 7% ; 2. 9% , 40. 7% , and 30. 9% ; 0. 31% , 23. 8% , and 13. 6% in three environments. It can achieve random obstacle avoidance and smooth path, which verifies the effectiveness of this algorithm.

Abstract:This paper studies the influence rule of operating temperature and load changes on the resonant frequency of the transducer and a feedback control system is designed to realize the resonant frequency tracking. Firstly, the resonant frequency calculation model with operating temperature T and load F as independent variables is derived based on the electromechanical equivalent circuit of the magnetostrictive transducer, a temperature-controllable and load-adjustable resonant frequency test system is built; then, the methods of segmental linear fitting and surface numerical fitting are performed experiment on the tested data, and a binary function model of resonant frequency with respect to the temperature and load variables is obtained to meet the requirements of linear correlation and surface fitting goodness, and the relevant parameters of the resonant frequency calculation model are determined. Finally, based on the caculation model, a frequency tracking feedback control system reflecting temperature and load changes was designed and the output characteristics of the transducer were tested. The results show that the vibration acceleration amplitude of the transducer is increased by about 31. 11% in average, and the resonant frequency could be automatically adjusted according to the temperature and load changes to achieve high efficiency and stable operation.

Abstract:In order to meet the accuracy requirement of the trajectory tracking motion of snake robot and eliminate the influence of the external disturbance on the tracking error of the robot, an adaptive trajectory tracking controller for snake robot tracking error prediction is proposed. The proposed controller achieves the prediction of the disturbance variables, friction coefficient and control parameters of the robot, and compensates the control input of the system with the predicted value and virtual control function. The sideslip angle of the snake robot in the trajectory tracking is conteracted, and the negative influence of the interference variables on the robot is avoided. The error stability and control accuracy of the trajectory tracking are improved. After the snake robot model is established, the line-of-sight method is improved using the integral sideslip angle compensation term. The adaptive trajectory tracking controller of snake robot is designed. The position error of the robot converges within 10 s, the angle error is less than 0. 03 rad, and the predicted value error converges within 5 s. Through simulation experiments, the effectiveness and superiority of the proposed controller are verified.

Abstract:In the process industry, the change of working condition results in the decrease of the prediction accuracy of traditional softsensing models. Take the impact of industrial data continuity, sequence, multicollinearity, and huge amount of data on the model establishment into account, a multi-conditions soft sensor regression model framework based on the time-nearest neighbor Laplacian regularization is proposed. To solve the multicollinearity of industrial data, the proposed regression framework utilizes the nonlinear iterative partial least squares method. Meanwhile, the domain adaptation regular term is introduced to mitigate the influence of the change of working conditions on the model. On this basis, the time nearest neighbor Laplacian regular term is proposed, which can maintain the sequence structure of the data during the mapping process. And the model training time is greatly reduced to meet the industrial real-time requirement. In the experiment, the multi-conditions data set of the melamine polymerization process is taken as an example. The results show that compared with the traditional method of partial least squares regression, when the target conditions are conditions 1 to 4, the method in this paper reduces the average root mean square error by 30. 3% , 31. 4% , 29. 3% and 24. 1% , respectively. And compared using with the traditional function of total connecting, using the function of time-nearest neighbor connecting to construct the Laplacian regularization could deduce the training time of the four working conditions by 14. 11, 1. 01, 26. 43, 0. 71 s respectively, and indicated that the accuracy and the training time could be improved.

Abstract:This paper takes the vibration displacement control system of continuous casting mold driven by servo motor as the study object, aiming at the problem of the servo motor requirements of single direction and variable angular speed rotation in system processing control and considering that the selection of the system controller parameters mainly depends on experience, a compound tracking control strategy based on feedforward control and PID feedback control with parameter optimization is proposed. Firstly, an approximate mathematical model between the output speed of servo motor and vibration displacement is established according to the characteristics of the vibration displacement system of continuous casting mold driven by servo motor. Secondly, aiming at the processing constraints of servo motor single direction rotation, it is determined that the vibration displacement system of continuous casting mold takes speed compensation as the feedforward controller to ensure that the output of the system controller is greater than zero. Thirdly, aiming at the problem that the controller parameters of vibration displacement system are selected mainly depending on experience, a strategy adopting ameliorated moth-flame optimization (AMFO) algorithm to optimize the PID controller parameters is proposed, which can achieve high accuracy tracking control of the mold vibration displacement. Finally, simulation and experiment verify the effectiveness of the proposed control strategy, and the experiment results show that the adjustment time of the vibration displacement is shortened by 0. 3 s and the relative error of vibration displacement tracking is reduced by 1. 8% after optimization.