Abstract:This paper proposes a GS/T mixed-sensitivity H_∞ controller design method based on spectrum specifications to tackle the multiple-degree-of-freedom control problem with high precision requirements of drag-free satellite in deep space mission. The method is explicit in physical meaning and can reduce the complexity in the design of drag-free control system. First, the high precision dynamic model of drag-free satellite with two test masses was derived, which shows the strong coupling characteristics of the loops. Next, selection matrices and input decoupling strategies were designed to divide the control system into single-input single-output (SISO) drag-free control loops, suspension control loops, and spacecraft attitude control loops, and closed loop feedback control strategies for each loop were further established. Then, combined with various scientific requirements and spectrum models of external disturbances and sensor noises, constrains for sensitivity functions and complementary sensitivity function of each loop were derived. With the help of these design criteria, weighting function matrices could be chosen quickly and accurately for the design of the drag-free and attitude control system in frequency domain. Simulation results proved the stability and capabilities of the proposed control system in resisting disturbances, which can achieve the ultra-quiet-stable requirement for drag-free control. Finally, on the basis of transfer function analysis, a method to further reduce acceleration noises was given.

Abstract:With the increasing complexity of space exploration requirements and the development of high-resolution payloads, satellite-ground downlink data transmission systems have been requested for transmitting increasing volumes of data. In order to meet the requirements of high throughput, low power, and high reliability for variable coding modulation (VCM) transmission systems of low Earth orbit (LEO) satellites, a new fast accumulated parallel recursive low-density parity-check (LDPC) encoding algorithm based on the second generation digital video broadcast (DVB-S2) standard was proposed, and an efficient encoder with low power was designed. The encoder was lower power consumed by simplifying the intermediate variables of parity-check bits based on the randomness of input information bits and the characteristic of binary operation. By analyzing the similarities between different LDPC codes and reusing the computation units and memories, the utilization of hardware resources was improved. Benefitting from the dynamic encoder structure, the encoder was compatible with three VCM modes, and the correctness was guaranteed when VCM modes changed, which increased the flexibility of the encoder. Furthermore, the new parity-check bits storage schemes that match with the modulation mode could output M parallel bits in sequence and increase the encoder throughput with high efficiency. The proposed encoder design was implemented on the Xilinx XC7K325t-3fbg900 FPGA, and experimental results showed that the maximum encoding throughput was up to 1.104 Gb/s when operating at a system clock of 347.5 MHz, the total throughput was improved by 31.9% compared with the constant coding modulation (CCM) transmission system, and the power consumption of the encoder was reduced by 21.7%.

Abstract:To improve the energy density of lithium (Li) batteries and solve the problems of conventional single-salt electrolytes such as poor thermal stability, low conductivity, and Li dendrites, developing new blended-salts electrolytes has become an important research direction of Li battery electrolytes. Blended-salts can change the ion conductivity, Li ion solubility, viscosity, and other properties of the solution, and can improve the electrochemical performance of the battery by affecting the interface of electrode and electrolyte as well as the surface and interface of electrode materials. This paper collects the latest research literature at home and abroad, and reviews the research progress of blended-salts electrolytes for Li batteries, including different research directions such as Li ion batteries, Li metal batteries, Li sulfur batteries, and solid state batteries. The advantages and disadvantages of commonly used single-salts are summarized, the effects and mechanism explanations of hybrid electrolytes on different surface interfaces of batteries are elaborated, and the applications of blended-salts in different battery systems are analyzed. At last, the existing problems of blended-salts electrolytes and interface reactions are discussed, and the development direction and application of characterization of novel blended-salts and interface reactions are given. The current research studies of blended-salts are mostly limited to performance improvement and its mechanism needs to be further investigated. The existing explanation usually unfolds from single-salts, and the mechanism of synergistic effect of blended-salts is an important research direction in the future.

Abstract:Arrhythmia is characterized by irregular heartbeats, and arrhythmia classification plays a key role in early prevention and diagnosis of cardiovascular diseases. In order to improve the accuracy and speed of arrhythmia classification and realize the automatic recognition of arrhythmia types, a seven-layer hybrid model based on convolutional neural network (CNN) was proposed. To maintain the integrity of the beats, the electrocardiogram signal was dynamically segmented according to the R-R interval to obtain different lengths of heartbeats. The local features of the heartbeats were extracted through the sliding of the convolution kernel of convolution layer, and the average pooling layer performed down-sampling to reduce the dimensionality of the features. The spatial pyramid pooling (SPP) layer extracted the beat features with different pooling sizes. Input features of different lengths were fused by SPP layer to obtain output features of the same length. Extreme learning machine (ELM) as a classifier could improve the speed of classification and shorten the training time. The MIT-BIH arrhythmia database (MITDB) and ten-fold cross-validation method were adopted to complete four classification experiments of arrhythmia. The overall accuracy, sensitivity, specificity, and precision of the classification results in the test set reached 99.16%, 99.85%, 98.89%, and 99.85%, respectively. In the same software environment, the accuracy and training time of hybrid model and single model were verified, and results show that the hybrid model achieved higher accuracy with less training time, which provides a feasible scheme for quickly and accurately identifying the types of arrhythmia.

Abstract:Among the conventional training methods of pattern recognition, the supervised learning methods with abundant labels have achieved significant performance in recognition accuracy. However, in real life, there are problems that samples often lack labels, or existing labeled samples cannot be used directly due to the distributional divergences of the target samples. To resolve these issues, unsupervised domain adaptation is often applied to recognize samples of unlabeled target domain by taking advantage of the data of the source domain with sufficient labels but different distributions. Considering the situation that the distributions of the target recognition samples and the source training samples are different, an optimal representation learning method for unsupervised domain adaption was proposed. Two representation matrices were introduced to the common subspace of the domain samples to better reduce the distributional divergence of the domains. Then, optimization constraints were implemented on the two representation matrices, so as to make the source domain and the target domain optimally represent each other, thereby reducing the distributional divergence between the domains. In this way, the unlabeled target domain samples could be recognized by the fully labeled source domain samples (i.e., transfer learning). Experiments on three common unsupervised domain adaption datasets show that the proposed method outperformed the conventional transfer learning methods and deep learning methods in recognition accuracy, which verifies the validity and robustness of the proposed algorithm.

Abstract:To solve the problems that pipelined coordinate rotation digital computer (CORDIC) algorithm has low output precision, long output time, and large hardware resource consumption, this paper proposes a bi-directional prediction scaling-free CORDIC algorithm. First, the algorithm decomposes the input angle in [0, π/4) into the smaller angle of 2^-i according to the bit value i after binary encoding of the angle. Then it uses the set-up lookup table to perform two-way scaling-free factor rotation based on the initial angle. It is not necessary to judge the next rotation direction according to the intermediate iteration result, which avoids the uncertainty of the iteration direction, reduces the number of iteration units and iterations, and improves the calculation accuracy. Finally, the algorithm transforms [π/4,2π) to [0, π/4) through the angle interval folding technique, so that the calculation interval is extended to the entire circumference [0,2π), which guarantees the calculation range. In addition, the algorithm can be achieved by only using shift and addition and subtraction operations, avoiding multiplication operations. Simulation and verification were carried out on MATLAB and Vivado. Results show that compared with the pipelined and unidirectional scaling-free CORDIC algorithms, when the output bit width was 14 bits, the output accuracy was increased by 47.5% and 18.8% respectively, the maximum output delay was reduced by 53.8% and 40.0% respectively, and the hardware resource consumption was improved to some extent. The algorithm has the characteristics of high output accuracy and short output delay, and its comprehensive performance has been greatly improved.

Abstract:To solve the problem that the pedestrian interaction model is difficult to be effectively constructed in the pedestrian trajectory prediction task, a trajectory prediction algorithm based on graph convolutional network (TP-GCN) was proposed to establish pedestrian interaction and predict future trajectories of pedestrians. First, the long short-term memory was used to extract the trajectory motion features of the trajectory sequences of pedestrians. Then, the pedestrians were considered as the nodes on the graph, and adjacency matrix was built to represent the created interactions. Next, the connection weights between unrelated nodes were screened out according to the blind zone. For trajectory motion features, the graph convolutional network was applied to extract the interactions between the trajectories and increase the extraction of the interaction in each trajectory, and the interaction was then encoded as trajectory interaction features by using long short-term memory. Furthermore, the weights of the graph convolutional network were optimized by the Deep Graph Info method to ensure that the motion pattern of individual accords with those of all the pedestrians in the scene. Finally, the trajectory motion features and trajectory interaction features were decoded using long short-term memory to complete the trajectory prediction. According to the experiment on the public datasets ETH and UCY, the proposed algorithm could make the predictions of pedestrian habits close to the real trajectories based on the interaction model between pedestrians, and the overall prediction accuracy was high. In addition, the ablation experiment and the visualization of the predicted trajectory also verified the effectiveness and interpretability of the algorithm.

Abstract:The identification of key functional parts can improve the retrieval efficiency of assembly model and the reuse level of retrieval, and also provide critical reference information for autonomous design. In order to reduce the subjectivity of the expert system, the rough set theory was introduced into the automatic identification of key subassembly functional parts, and the ranking process of functional parts was driven by the data of the assembly model itself. The characteristics and connection relationships of parts in assembly were analyzed, and assembly model was established based on complex network. The topological layer, part attribute layer data, and part types were extracted as condition attributes and decision attributes. The algorithm based on dynamic hierarchical clustering was used to discretize the decision information table of the subassembly parts. The heuristic reduced algorithm based on attribute importance was adopted to dig knowledge, eliminate redundant condition attributes, and obtain attributes reduction set as well as corresponding attribute weight. Finally, the order of the importance of the subassembly parts with key functions was obtained through comprehensive evaluation. The worm gear reducer model was taken as an example to verify the performance of the proposed method. Experimental results show that the final ranking results of the proposed model were consistent with those of previous research results, and since the model is driven by the assembly model data itself, it is more objective.

Abstract:To improve the frequency and accuracy for measuring ocean hurricane intensity, a joint use of cyclone global navigation satellite system (CYGNSS) and soil moisture active and passive (SMAP) satellites data to measure the hurricane intensity was investigated. First, the satellite data characteristics were introduced. Then, taking Hurricane Florence as an example, a method was proposed by fusing the satellite data to measure the ocean hurricane intensity, including the preprocessing of satellite data, the extraction of high wind speed area using a pixel-level data fusion method, and the measurement of hurricane intensity. Finally, the observation results of ten hurricanes were compared with the highest wind speed measured by the American National Hurricane Center (NHC). The root-mean-square difference, mean absolute error, and correlation coefficient were adopted to analyze the differences among the wind speed measurement results. Results demonstrate that compared with the case of the SMAP satellite data, the proposed method could measure the hurricane intensity at more frequent intervals, and more complete high wind speed area was obtained. The comparison between the hurricane intensity obtained using the proposed method and measured by NHC shows that the mean absolute error ranged between 3.9 and 10.2 m/s, the root-mean-square difference varied between 4.6 and 12.5 m/s, and the correlation coefficient ranged between 0.570 7 and 0.915 2, which confirms the effectiveness of the proposed method in measuring hurricane intensity.

Abstract:The stretch bending process of the variable curvature L-section aluminum alloy door column of an electric multiple unit (EMU) was simulated. Based on the relationship between the stress and strain distribution of the cross-section and the spring-back radius at different stages, the factors affecting the precision of stretch bending process were systematically explored for optimizing processing parameters and controlling methods, which were then verified by experiment. Results show that the main forming defects during bending were corss-section distortions and spring-back, where the cross-section distortion defects were web collapse in straight section and concave deformation of vertical edge in arc segment. The defects of cross-section distortions and spring-back could be reduced to some extent, by adjusting the processing parameters such as the stretching amount. The cross-section distortions were mainly caused by different forces on the vertical side during the stretch bending process. By setting back plate and increasing its load, the distortion value of the web plate in straight section could be reduced. The optimization of the profile of the bending die could decrease the distortion value of the vertical edge section in arc segment. Moreover, the spring-back degree in the small arc side was more serious than that in the large arc side, which was mainly due to the asymmetry of both ends of the variable curvature aluminum alloy door column, affecting the part-mold contact gap of the door column. The spring-back could be alleviated via modifying the die surface by spring-back compensation method. The forming accuracy could be greatly enhanced to meet the design requirements when the amount of pre-elongation, wrap-elongation, and post-elongation was 1%, 4.5%, and 0.5%, respectively, and the back pressure applied to the straight section was 40 kN.

Abstract:HJ1.1mm] Aiming to enable high power impulse magnetron sputtering (HiPIMS) technology to achieve high-speed deposition process while maintaining high ionization rate, the electric and magnetic fields synergistic effect was applied to improve the HiPIMS discharge and deposition characteristics. The substrate ion current density, optical emission spectrum, surface morphology and surface roughness, cross-section morphology, and deposition rate of vanadium films prepared by electro-magnetic fields synergistically enhancing HiPIMS ((E-MF) HiPIMS) were investigated, and the intrinsic physical mechanism of (E-MF) HiPIMS was analyzed. Results show that compared with HiPIMS, the peak of substrate ion current density of the plate workpiece in the discharge of vanadium target (E-MF) HiPIMS increased seven times, and the peak of substrate ion current density of the cylinder-like workpiece in the discharge of copper target (E-MF) HiPIMS increased 14 times. The intensity of Ar⁰ spectral lines, Ar+ spectral lines, V⁰ spectral lines, and V+ spectral lines of (E-MF) HiPIMS all enhanced significantly, and the ionization rate of Ar and V particles increased. The vanadium films prepared by (E-MF) HiPIMS consists of V (111) and V (211). The surface of the vanadium films prepared by (E-MF) HiPIMS was much smoother, and the surface roughness was reduced from 15.0 nm to 9.6 nm. The growing structure of the vanadium films was much denser, and the deposition rate was increased by about 30%. The deposition rate of copper films prepared by (E-MF) HiPIMS increased by about 50% under the cylinder-like workpiece rotation conditions. (E-MF) HiPIMS is a novel discharge mode with high ionization rate and high deposition rate, which can effectively avoid the technical defect of low deposition rate of conventional HiPIMS.

Abstract:To effectively improve weld forming and inhibit forming defects, taking 12 mm-thick 7N01 aluminum alloy as research object, the keyhole drilling process as well as the heat and fluid transport phenomena of electron beam spot welding (EBSW) were analyzed by Ansys Fluent and validated by experiments. In order to reflect the beam energy density distribution, a self-adaptive heat source model considering the characteristics of the active zone of beam was established, and VOF algorithm was used to track the gas-liquid interface in real time. Numerical analysis results show that the beam energy density distribution and the coupling between the beam and the transient pool/keyhole were the key factors to determine weld forming. When the beam was in lower focus mode, the unique energy distribution and the induced plasma insulation, metal vapor recoil pressure, Marangoni flow, and upward transportations of thermal buoyancy (maximum flow rate at about 15 m/s) led to the weld reinforcement and the extension of nailhead area. The beam energy density in the depth direction increased first and then decreased significantly, which resulted in similar evolutions for the welding in depth and width directions and might induce the spiking defects adjacent to the keyhole bottom. In addition, it was found that with the increase in keyhole depth, the energy fluctuation as well as the competition between recoil pressure and surface tension gradually increased, which promoted the periodicity of keyhole drilling process.

Abstract:In order to reduce the fracture accident of disc suspension insulators, the stress level of porcelain and the optimization of its structure were studied. The finite element model of the disc suspension insulator was established to investigate the stress distribution characteristics of the insulator under tension, and the stress on the porcelain under different tension directions was analyzed. The insulator structure was optimized based on finite element method and machine learning method. Results show that under the action of tension, the contact surface between porcelain and cement was partially separated, resulting in higher stress level on the middle wall of porcelain. The greater the angle between the tension direction and the axis of insulator was, the higher the stress level on the porcelain became. On the basis of finite element calculation, the machine learning method was adopted to optimize the structural parameters of the insulator, and the optimal structural parameters were obtained. The stress level of the optimal structure was reduced by 30% compared with that of the original structure. The finite element calculation of the optimal structure shows that the error rate was only 0.644%, indicating that the result is reliable and the optimization effect is significant. Therefore, in the process of cementing of insulator, the cementing strength should be enhanced. In the installation of insulator, the angle between the tension direction of the insulator and the axis should be minimized so as to effectively increase the working life of the insulator. The optimal parameters obtained by structural optimization can provide theoretical guidance and technical support for structural design.

Abstract:Combined with practical cases, this paper explores the modeling method of building environment information based on multi-view images, aiming to improve the efficiency of building environment information modeling and the accuracy of local information modeling of building environment such as eave bottom. The technical route of multi-view image fusion was investigated, low altitude photogrammetry and ground photography were carried out based on practical cases, and the architectural and environmental image data of the practical cases were collected. By constructing connection points and fusing multi-view image data, the ground image was fused with aerial image, and the image of some dead corner areas which are difficult to capture in the air was obtained to realize multi-angle and omni-directional image acquisition. Through the air triangulation processing, dense matching, and texture mapping, the three-dimensional digital model of the building environment information of the practical case was generated. Results show that the integration of low altitude photogrammetry and ground photography image data could significantly improve the modeling efficiency of building environment information and the modeling accuracy of building detail information, and could solve the problem of incomplete information collected by a single image source, which has the advantages of low cost, high efficiency, and easy operation. In the process of data collection and protection of preserved buildings, the proposed method can achieve the rapid and efficient preservation and recording of three-dimensional data as well as realize the preservation and inheritance of multi-dimensional data of historical buildings.

Abstract:Forced layout design in residential area is beneficial for increasing plot ratio, which is an important approach to achieve intensive construction. Existing residential forced layouts are mostly made by designers subjectively based on the results of sunshine simulation analysis, with low design efficiency. In the context of deep learning technology, a generative design method for residential forced layout based on conditional generative adversarial network (CGAN) was proposed, applying pix2pix algorithm to construct residential forced layout generative model. By learning the corresponding relationship between the outline and the general forced layout of low-rise, multi-story, and high-rise residential areas, the model could generate residential forced layout under any outline conditions, which improves the precision and efficiency of residential forced layout design and promotes the efficient use of urban land. Three residential areas in mid-latitude region were taken as examples to verify the application effect of the proposed method and evaluate the sunshine performance of the generated scheme. Results show that the generated low-rise scheme could meet the sunshine requirement of 2 h in Great Cold day (around January 20); 96% of rooms in multi-story scheme and 84% of rooms in high-rise scheme could meet the sunshine requirement. The plot ratio of the high-rise scheme was more than 3.0, that of the multi-story scheme was more than 1.5, and that of the low-rise scheme was more than 0.5, indicating that the generated schemes make effective use of urban land. The constructed model could generate residential forced layout within 3 s, which significantly reduces the design time of forced layout and improves the design efficiency.

Abstract:Energy consumption in cities and towns is related to the individual building and also the group morphology. In order to explore the impact of the spatial forms of small town residential blocks on energy consumption, by taking Changxing County of Zhejiang Province as an example, the morphological characteristics of small town residential blocks were analyzed according to the sample study of the residential blocks in central area, and five typical floor area ratios were extracted on the basis of spatial form types of typical residential blocks. Then, from the perspective of typo-morphology, based on the morphological characteristics of residential blocks in small towns, five “basic forms” and 17 “derived forms” were generated by the five typical floor area ratios. Finally, two-step energy simulation was carried out to analyze the influence of physical forms and immaterial forms on the energy consumption of residential blocks. Results show that the energy consumption intensity of multi-story slab building type was the lowest, and that of low-rise single building type was the highest. The mixed residential block was the most suitable residential block mode for small towns. Under the same floor area ratio, the energy consumption intensity of the block could be reduced by mixing the two types of residential buildings and placing the higher story residential buildings on the north side of the lower floor residential buildings. It is also more conducive to the energy saving of residential blocks by adopting the positive south and north determinant layout. In the process of planning and construction, it is necessary to adjust the factors of immaterial forms through the control of physical forms, thereby saving energy.

Abstract:In view of the low robustness and poor accuracy of traditional visual odometry in dynamic environment, a dense visual odometry based on edge information fusion was proposed. First, the spatial coordinates of pixels based on depth information were calculated, and the K-means algorithm was adopted for scene clustering. Based on the clustering of photometric information and edge information, the photometric consistency error and edge alignment error were constructed respectively, and the residual model was obtained after fusion and regularization of the two errors. Next, the average background depth was introduced to the residual model so as to expand the residual difference between the dynamic and static parts, ensuring correct motion segmentation. Then, a non-parametric statistical model was constructed based on the general characteristics of the cluster residual distribution, and motion segmentation was performed through dynamic thresholds to eliminate dynamic objects and obtain clustering weights. Finally, the weighted cluster residuals were added to the nonlinear optimization function of pose estimation to reduce the effect of dynamic objects and improve the accuracy of pose estimation. Experiments on TUM dataset show that the proposed algorithm could achieve better results in both static and high dynamic environments, and it had higher accuracy and robustness than the existing algorithm in dynamic environment.

Abstract:To solve the problem of wide-band direction-of-arrival (DOA) estimation in the case of colored noise, this paper proposes an effective solution combining covariance matrix difference theory and eigenvector space focusing algorithm. First, based on the covariance matrix difference theory, the difference covariance matrix was decomposed, and the observation model was reconstructed by taking the corresponding eigenvector of the positive eigenvalue part to eliminate the influence of colored noise and “pseudo” peak. Then, for the obtained observation model, a new signal autocorrelation matrix was constructed, and a wide-band DOA estimation method that requires no angle prediction was derived, which is to solve the focus matrix based on the eigenvector signal subspace at different frequency points. In addition, in order to avoid the influence of the eigenvector corresponding to zero eigenvalue on the resolution threshold, the eigenvector matrix was rearranged according to the eigenvalue decreasing sequence, and the focus matrix was reconstructed from the orthogonality relationship between the nonsalient part and the flow pattern matrix of the signal array. Finally, the performances of direction finding accuracy, resolution, robustness, and complexity of the proposed algorithm were analyzed in noise background. Theoretical analysis and simulation results indicate that the method has high accuracy and robustness in the background of colored noise with low complexity and strong practicability, and there is no need to conduct angle prediction.

Abstract:Since the accuracy of random forest algorithm under differential privacy is undesirable when classifying high-dimensional data, the out-of-bag estimate was introduced to calculate the weights of decision trees and features, and the random forest algorithm under differential privacy based on the out-of-bag estimate (RFDP_OOB) was proposed. First, the algorithm generates a part of random forest under differential privacy, and the importance of decision trees and features is evaluated by utilizing the out-of-bag estimate under differential privacy, so as to calculate the weights of the decision trees and features. Then, the features are re-divided into non-essential features through feature weights. Next, in the process of generating the remaining part of the random forest, the pre-pruning operation is performed on the nodes whose best features are non-important features to make them leaf nodes, so as to reduce noise and improve the classification accuracy of the decision tree with better efficiency. Finally, in predicting the classification results, the classification result with the maximum weight of the corresponding decision tree is taken as the classification result of the random forest algorithm, thereby improving the classification accuracy of the random forest algorithm. The privacy and effectiveness of the algorithm were analyzed theoretically, and the experimental results verified the effectiveness of the algorithm. The proposed algorithm can improve the classification accuracy and protect the privacy of data.

Abstract:The finite element numerical simulation of the coupled structure-electromagnetic fields was carried out based on LS-Dyna platform for the electromagnetic constriction process of aluminum/steel bimetallic tube, aiming to guide actual precision forming process of bimetallic tube. The influences of the inner and outer tube wall thickness ratio, the inner and outer tube gap, discharge voltage, and elastic mandrel on forming quality were studied. Results show that small wall thickness ratio caused the weak connection of the tubes, while large wall thickness ratio led to the depression of the inner tube and the crack of the outer tube. As the discharge voltage increased or the distance between two tubes decreased, the constriction impact force of the outer tube was increased, which resulted in a drastic increase in the circumferential stress of inner tube and the occurrence of instability and wrinkling, as well as the crack of the outer tube. On the contrary, with the discharge voltage decreased or the distance between the two tubes increased, the constriction impact force of the outer tube was decreased, and the inner tube could not bear an effective deformation and springback process, leading to smaller effective connection area between the two tubes. The elastic mandrel could restrain macro-defects such as cracking at a certain extent, and appropriately increase the springback of inner tube. Lastly, the electromagnetic forming rules and defect control theory were obtained based on theoretical calculation and finite element simulation. With the purpose of increasing the length of effective connection area and reducing macro-defects, the research parameters were optimized, and it was found that the forming quality is the best when the inner tube wall thickness remains 1 mm, outer tube wall thickness is 1.5 mm, the gap between the tubes is 0.7 mm, and discharge voltage is 50 kV.

Abstract:In order to efficiently predict the rheological viscosity of low viscosity epoxy resin TQ-50 and determine suitable processing window in resin transfer molding (RTM) process, the properties of TQ-50 resin system including curing reaction characteristics, static viscosity, dynamic viscosity, and gel time were systematically investigated by using differential scanning calorimetry (DSC), viscometer, rheometer, and gel time tester. Based on the experimental results, the static rheological model of TQ-50 was established by using Arrhenius equation. It was found that the rheological model was well consistent with the experimental results and could accurately describe the chemical rheological behavior of TQ-50 in low viscosity range. On the basis of the rheological model, three-dimensional prediction maps of temperature, time, and viscosity as well as viscosity curves of temperature and time were established, and the processing window was predicted, which provided guidance for the optimization of process parameters. Results show that when the viscosity was lower than 500 mPa·s, the optional processing window temperature was 50~70 ℃, and the holding time was about 150~180 min. When the viscosity was lower than 200 mPa·s, the optional range window temperature was 55~70 ℃, and the holding time was about 56~96 min. The resin system has a wider range of suitable injection temperature and longer injection time, which can meet the basic requirements of RTM processing.

Abstract:To realize the continuous monitoring of foundation displacement and the prediction of final settlement of long-span bridges based on Beidou navigation satellite system (BDS), the influences of various errors should be eliminated in the collection process of satellite signals. This paper proposes a method to eliminate the multipath errors in BDS signals using stationary wavelet transform. The method simplifies random noises to Gaussian white noises, and combines the determinant criterion of noises and empirical mode decomposition (EMD) method to reduce the random noises in the signal based on autocorrelation function. In order to meet the observation accuracy for foundation displacement of long-span bridges, the measured BDS data and the settlement data obtained by precise levelling were defined by hyperbolic model. Based on the measured data during the construction process and the modelled settlement data, the non-linear relationship of the data was established by BP neural network to realize real-time correction of measured BDS data. The foundation pile of a rail-cum-road bridge under construction was taken as an example, which is over 1 km in length. The BDS data was collected, the bad points were eliminated, and the lost data was compensated by linear interpolation. The real-time measured data of the long-span bridge at foundation pile was denoised and corrected by the proposed method. Results show that compared with the precise levelling results, the proposed method had high precision, which could meet the requirements of continuous monitoring and final settlement prediction of bridge towers of long-span bridges.

Abstract:Vision-based simultaneous localization and mapping (SLAM) is a research hotspot in the field of intelligent driving. However, for the scenes that contain moving targets or inconspicuous close-range features, it is easy to cause ineffective or inaccurate pose estimation between frames. To solve this problem, this paper proposes a SLAM algorithm based on road structured features and scene semantic information. First, for the problem of target moving, a semantic segmentation neural network with an improved pyramid pooling module was designed to obtain the target category corresponding to each pixel in the image. The segmentation results were taken as basis for the elimination of moving points, which avoids the problem of low accuracy of pose calculation caused by moving points participating in feature matching. Then, in view of the lack of effective feature points, the road area in the image was determined based on v-disparity algorithm, and disparity function was obtained to calculate the accurate disparity value of the pixels on the road. Furthermore, a pose calculation method based on road structured features (e.g., lane lines, road boundaries, pavement traffic markings) was proposed. Finally, scene experiments were carried out and results show that the absolute trajectory error of the improved algorithm proposed in this paper was smaller than that of the original algorithm, which proves that the proposed method has higher pose estimation accuracy in scenes with moving targets or inconspicuous close-range features. In addition, an effective dense point cloud map containing semantic information was established, which has good environmental adaptability.

Abstract:In biological and medical research, two-vessel occlusion (2-VO) is often used to establish cerebral ischemia models, while due to the non-quantitative ligation of the artery, the model usually has poor reproducibility, large individual differences within the group, and high mortality. In this paper, a vascular blocking tension meter is proposed by taking metal strain gauge as the core component to realize the quantitative detection of ligation force. Biological experimental verification was carried out in vitroand in vivo, as well as in such aspects as cells and molecules level, tissues and organs level, and overall level. Results show that the linearity and sensitivity of the tension meter were affected by factors such as base parameters, bonding process, circuit parameters, and appearance structure of the metal strain gauge. Compared with the non-instrumental ligation group, the animal mortality of the cerebral ischemia model was reduced by 42.22%. The decrease in cerebral blood flow of the experimental animals corresponded to the increase in ligation force. After hematoxylin-eosin (HE) staining of the hippocampus, light microscopy results show that the neurons were loosely arranged, the number of neurons decreased and even disappeared, the cytoplasmic staining of the neurons became weak, and the dentate gyrus exhibited neuronal pyknosis. A comparison between different ligation force groups shows that the individual differences in the hippocampus morphology of the rats in the 1.5 N ligation force group were the smallest. Morris water maze results show that compared with the sham operation group, the escape latencies of the rats in the 1.5 N ligation group and the 2.5 N ligation group were significantly increased (P<0.05), and the number of platform crossing times in the 1.5 N ligation group was significantly decreased (P<0.05). There was an obvious learning and memory disorder. It was falsified that the greater the ligation force of the 2-VO rats was, the better the ligation effect was, which was obtained in our previous laboratory experiments. The conclusion was drawn that the smaller the standard deviation of the ligation force was, the better the ligation effect was, and the ligation force could be quantitatively controlled. The application of the vascular blocking tension meter to the basic experimental research of medicine is of significance.

Abstract:For the precise and fast position control of an ammunition coordinator, a continuous time-varying feedback control method based on implicit Lyapunov function was proposed. The method is PD-like in the form of control law, but its proportional and differential coefficients depend on the system Lyapunov function, which are differentiable functions of system error variables. First, the dynamic model of the system was established through the second Lagrange equation. Two main nonlinear disturbance terms of the ammunition coordinator system (i.e., friction torque and balance torque) were modeled in detail, where the friction term was modeled via LuGre model. Then, experiments were designed according to the structure of the dynamic model, and key parameters in the friction torque and balance torque terms were identified through genetic algorithm using experimental data. To further shorten the positioning time of ammunition coordinator and improve its coordination efficiency and performance, the model-based feedforward compensation of the friction torque and balance torque were introduced into the control strategy. Experimental results show that the proposed control method was robust to system payload uncertainty. With varying payload, the positioning time and accuracy of the system were guaranteed. In addition, the nonlinear disturbance compensation of the friction torque and balance torque shortened the positioning time by 25.1%, from 2.07 s to 1.55 s. Meanwhile, the positioning accuracy was guaranteed, which verified the effectiveness of the proposed method.