Gait recognition is a hotspot in the field of pattern recognition, information security and clinical medicine in recent years. This paper mainly reviewed the four aspects of the gait data acquisition instruments, common gait databases, gait feature extraction and the methods of gait recognition. Firstly, the advantages and disadvantages, reliability, and application scenarios of the commonly used gait data acquisition instruments were introduced. Secondly, the common gait data sets were compared and analyzed from six aspects of the establishment of institutions, the sample size, the sampling rate, the environments, the instruments and the variables. Third, the existing gait parameter extraction methods were divided into model-based methods and non-model-based methods to elaborate in detail. Fourth, the gait recognition was introduced from support vector machine, auto-encoder and convolutional neural network, respectively, and the above methods from the two application directions of personal identification and abnormal gait identification were compared respectively. Finally, the shortcomings of current research and the future development directions were pointed out based on practical application.

In order to properly express the dynamic mechanical properties of tire under high-frequency excitation condition, a tire dynamic model is established, which discretizes the continuous physical structure of tire. The sidewall portion of the tire and the compressed air are discretized into a series of spring and damping units. The belt portion of the tire is discreted into a series of mass points which are uniformly distributed along the circumferential direction of the belt. The mass points are connected by elastic units that represents the tensile and bending properties of the belt. A number of massless contact elements between adjacent mass points are introduced for the calculation of contact forces in tire-road contact. The data of Tire Model Procedure Test (TMPT) is used for parameter identification and accuracy verification of the established tire dynamic model. The comparison results show that the tire dynamic model established in this paper can not only accurately express the static mechanical properties of the tire, but also express the high-frequency dynamic mechanical properties of the tire on typical uneven road surface well.

To design vehicle collision energy-absorbing structures with excellent crashworthiness, the concave triangle cell structure with negative Poisson's ratio was studied. The axial impact finite element model of the structure was established by explicit dynamic finite element software LS-DYNA. Combined with the optimal Latin hypercube design method, the polynomial surrogate model of the peak crush force (PCF) and the specific energy absorption (SEA) for horizontal size and thickness of the cell wall were constructed, respectively. The Non-dominated Sorting Genetic Algorithm-II (NSGA-II) was adopted to perform the multi-objective optimization design, and a compromise solution was obtained based on the fitness function C. The results show that the wall thickness has more obvious effect on crashworthiness of the energy-absorbing structure than horizontal cell wall size. The PCF can be effectively reduced and SEA can be improved by reasonable matching wall thickness and horizontal cell wall size．

An energy management strategy for fuel cell bus is designed based on the Pontryagin’s minimum principle, in which the fuel cell power is set as the control variable, minimal fuel consumption is taken as the objective function, and the durability of fuel cell is considered as the constraint. An on-line energy management strategy is designed with improving the Pontryagin’s minimum principle in order to adapting to different operating conditions. A fuel cell bus simulation model is built. The simulation results show that the energy management strategy can adapt to different operating conditions well, and it is better than the traditional on/off energy management strategy in terms of fuel economy and fuel cell durability.

The torsional characteristics modeling and piecewise point smooth fit of multi-stage stiffness Dual Mass Flywheel (DMF) are studied. Firstly, based on the analysis of the working principle of multi-stage stiffness DMF, a mathematical model of the torsional characteristics of the DMF with piecewise stiffness and damping nonlinearity is established, and the accuracy of the model is verified through bench test. Secondly, to solve the problems of jump discontinuities in the model, and the errors that may occur during the simulation process, the inverse tangent function is used to smoothly fit the model, and the nonlinear model of the DMF based on the arctangent fitting function is obtained. Finally, the models before and after smooth fitting are compared and analyzed. The results show that the smooth fitting of the arctangent function has little effect on the accuracy of the model, which can make the model piecewise point and the critical point smoothly transition, and improve the model simulation stability.

The issue of Telescoping Path Optimization (TPO) of Single-cylinder Pin-type Multi-section Boom (SPMB) was raised and its model is proposed by using Hopfield Neural Network (HNN). In an energy equation, it is difficult to balance the parameter λ of constrained term and the parameter γ of objective term. When λ dominates, network converges toward constraint satisfaction and a valid solution is often of poor quality. Whereas when γ dominates, network might converge to an invalid solution. To solve this problem, this paper proposes that the λ is gradient rising and the γ is gradient falling. A PID hybridized method based on Constraint Boundary Gap (CBG) control is presented to adaptively regulate the increments of λ and γ. Experimental results show that the telescoping efficiency improves 10%~30% after path optimization. TPO model achieves good optimization effect, and enables network converge almost 100% to a valid solution. The generation proportion of the optimal solution remains high at about 50% seeing that PID control often imposes a solution to the constraint boundary.

In order to analyze the kinematic characteristics of rigid origami mechanism and study the folding process of different patterns, the kinematic model of rigid origami mechanisms is established and the folding state of typical pattern is analyzed. Firstly, the relation equation between the rotation angle of creases at a single vertex is established, the explicit expression of the unknown angle of creases is derived, and the interference solution is eliminated through the mutual test of equation solutions to obtain the actual rotation angle of the arbitrary folding state. Secondly, using the rotation transformation matrix, the real time coordinates of the vertex are calculated. Thirdly, the intersection test algorithm of triangular facets is applied to judge interference of rigid origami mechanism, and the interference in the folding process is detected. Finally, the folding process simulation platform is established, and the folding process of typical rigid origami mechanism is analyzed. The simulation results coincide with the actual folding process, which verifies the correctness of the kinematic analysis method of rigid origami mechanisms and the applicability of the simulation platform.

In the process of FDM 3D printing, it is necessary to heat the printing nozzle and the heating bed of the printing platform until the required temperature of the printing material is reached. The heating process is time-consuming and energy-wasting due to the time lag and poor stability of the heating system. In order to solve the above problems, the paper adopts the fuzzy self-adaptive PID control method to control the heating process of the printing nozzle and the heating bed of the printing platform, and establishes the Matlab/Simulink simulation model of the control system. The simulation results show that the control effect of the fuzzy self-adaptive PID control method on the heating system of FDM 3D printer is better than that of the traditional PID control method. It has the advantages of small overshoot, fast response and more stable control effect.

A near net forming process scheme for railway knuckle forgings based on closed die forging is proposed, and the key technical problems in the design of pre-forged parts in near net forming are studied. The pre-forming process was designed by analyzing the metal flow characteristics in the final forging stage. Combining the multiple response surface method with the numerical simulation, the shape of the pre-forged formed part was optimized by the muli-objective function, in which the objectives include the complete filling and the best deformation uniformity of the forging part and the minimum damage of the material. The optimization was verified by the forming test. The measurement and analysis results show that the proposed method can obtain the shape of the optimized pre-forging. Based on the designed pre-forging, the final forged part without forming defects and high precision can be obtained, which can realize the near net forming of the knuckle forging.

Hot compression tests on Cr8 alloy were carried out on Gleeble-1500D thermal simulation test machine in the deformation temperature range from 900 ℃ to 1200℃ and strain rate range from 0.005 s^-1 to 5 s^-1．X-ray diffraction tests of the samples after hot deformation were carried out, and the hot deformation behavior and dislocation density evolution of Cr8 alloy were investigated．Based on the data of experiment, a two-stage constitutive model containing multiple parameters was established in consideration of dislocation density evolution．The results show that the true stress-true strain curves of Cr8 alloy show the typical characteristics of dynamic recrystallization under low strain rate．The hot deformation activation energy of Cr8 alloy is 423.41 kJ/mol, and the calculated values of the constitutive model agree well with the experimental values．Under the experimental conditions，the total dislocation density of Cr8 alloy reached above 10¹⁴ cm^-2, and the total dislocation density increased with increasing strain rate and decreasing deformation temperature．

The clinch-adhesive hybrid joining technology has many advantages such as good abradability and good energy absorption. This technology is very promising and can be widely used in the aspect of car body. To study the influence of the thickness of adhesive in the clinch-adhesive hybrid joint on the forming quality and the joint performance, a finite element model (FEM) of clinch-adhesive hybrid joining was established, and was validated by the experiment. The FEM is applied to analyze the adhesive layer distribution in the joint and the variation tendency of adhesive layer during the forming process, and the influence of the adhesive thickness on the forming quality was investigated. The results indicate that the adhesive thickness firstly decreases and then increases from the centre of joint bottom to the interlock region, and the thickest part is locate in the bottom corner. During the clinching process, the thickness distribution of adhesive varies with the increase in the punch displacement, the adhesive thickness influences the interlock value and the neck thickness of the joint.

Automatic driving vehicles need lidar to detect the object while driving. Because of the moving of vehicles, the point cloud becomes inaccurate while the traditional Euclidean clustering algorithm can not detect the obstacles in both near and remote areas at the same time, which leads to an inaccurate result of the number of those obstacles. In order to solve this problem, a algorithm is proposed to correct the 3D lidar point cloud, which is inaccurate, meanwhile, improve the Euclidean cluster algorithm, so it be able to change the distance limit according to the distance between the obstacle and lidar. Experimental results illustrate that the proposed algorithm can detect the obstacle in both near and remote areas, and the detect distance is increased compared with the traditional method.

The driver's eye movement experiments during driving are conducted on a typical desert grassland highway section, The coordinates of the point of view of the driver under normal driving conditions, the driver's continuous gazing time and the vehicle speed are collected, which are then converted into an area of visual interest per unit time. Through the correlation analysis and model establishment of the area of interest area and the length of the straight line section per unit time, the shortest straight line length between the curves of the desert grassland road must be guaranteed to drive for 6 s, the ideal length of long straight line segment is driving 60 s, and the length of the longest straight segment is driving 140 s, The values is in the reasonable rang compared with the domestic and foreign relative index, which can provide a theory for the selection of the index of the straight line between the curves in the desert grassland area.

In order to make effective graded warning for road rage, an optimal threshold determination method for discriminating driving anger intensity based on driving behavior features is proposed in this study. Different driving behavior data were acquired in driving states with different anger intensity by conducting timed experiments in busy traffic sections. The steering wheel rotation angel rate (SWRAR), speed of gas pedaling (SGP), and lateral acceleration (LA) were selected as three effective driving behavior features for discriminating different anger intensity by analysis of variance. A method of receiver operating characteristic (ROC) curve analysis was introduced to determine the optimal discrimination thresholds of SWRAR, SGP and LA. For normal state the ranges of the thresholds are 0≤SWRAR<1.8863, 0≤SGP<0.397 4, 0≤LA<0.136 2; for low anger state the ranges are 1.886 3≤SWRAR<2.324 1, 0.397 4≤SGP<0.505 4, 0.136 2≤LA<0.155 9; for moderate anger state the ranges are 2.324 1≤SWRAR<3.083 5, 0.505 4≤SGP<0.614 2, 0.155 9≤LA<0.175 4; and for high anger state the ranges are SWRAR \text{≥} 3.083 5, SGP \text{≥} 0.614 2, LA \text{≥} 0.175 4. The verification performances show that the accuracy of the optimal thresholds for discriminating different anger intensity can reach 79.42%. The results can provide theoretical foundation for developing multilevel warning systems for driving anger based on the optimal discrimination thresholds.

To make Cell Transmission Model (CTM) suitable for urban network traffic simulation with multi nodes and complex traffic conditions, and improve the accuracy of the model in estimation of road actual impedance, a improved road model is constructed by modifying the parameters of existing model. The influence of multi nodes on the accuracy of model is taken into consideration, and a node cell partition method is proposed. Then the Variable CTM (VCTM) for urban multi nodes network is constructed and the actual impedance calculation method of road is improved based on this. Simulation and analysis of case network are carried out by the improved model and VISSIM. The results show that the improve model can reflect the characteristics of urban traffic flow, and the node model is more consistent with the reality. The actual impedance calculated by the model is in good agreement with VISSIM output data. The improved model and the actual impedance calculation method can be well applied to the dynamic traffic assignment.

In this paper, a Multiple Input and Multiple Output Brain-like Emotional Learning Loop model (MIMO-BEL) was introduced, and a driver model was proposed with comprehensive transverse and lengthways control built on cognitive process. By using the adaptive control ability of the brain-like emotional learning loop model to the uncertain system, the driver model can decouple the transverse and lengthways control of the vehicle. Finally, based on the MIMO-BEL, the driver model of the transverse and lengthways integrated control is established. The simulation results show that the proposed driver model possesses good following accuracy of road and speed of vehicle.

The longitudinal berth is designed with a central aisle and double-deck berth units on both sides. With relatively independent longitudinal berths, the passengers lie parallel with the train’s running direction. In collision accidents, compared with the transversal berth, and the kinematic responses of dummies reflect the passive safety protection performance of structures. Analyzing the characteristics of the longitudinal berth in the structural design of a passenger interface, the finite element models of typical collision scenes with dummies were established according to a passive safety evaluation method for a passenger interface based on the anthropomorphic simulation test. The simulation was carried out with the standard load as the boundary condition. The passive safety protection of the structure to occupants was evaluated by analyzing kinematic responses and injury parameters, and compared with 25T transversal berth. The results indicate that, in the longitudinal berth, under forward load, the head and neck injury indexes of the dummies exceed the standard criterion values (HIC ₁₅ is 626 and N_ij is 2.5). Under the backward load, the peak femur compressive force reaches 6.4 kN, which exceeds the criterion value. Due to fundamental changes in the structural design, dummies in the longitudinal and transversal berths can’t avoid secondary impacts and have different injuries. The longitudinal berth restrains occupants more effectively and reduces the risks of injury in secondary impacts.

In order to solve the problem of interface calculation between unsaturated soil and structure coupled force and water, the direct shear tests of remolded loess and concrete interface with different moisture content were carried out. Firstly, the relationships between the initial shear stiffness coefficient, the ultimate shear strength and the normal stress in the hyperbolic model are simplified based on the test rule. Then the water content is introduced to establish the modified hyperbolic model suitable for unsaturated soil, which has six parameters and can be obtained by direct shear test of interface with different moisture content. The feasibility of the model is demonstrated by comparing the calculated results with the test results. The modified hyperbola model can consider both normal stress and moisture content, and can be used to calculate the interface mechanics between unsaturated soil and structure.

In order to meet the requirement of high temperature stability and low temperature crack resistance of asphalt pavement in seasonal frozen soil region of Jilin Province, in this research the diatomite and basalt fiber are simultaneously added into asphalt mixture. Then through the analyses of high temperature stability, low temperature stability, water stability and fatigue properties of compound modified asphalt mixture, the pavement performance of diatomite-basalt fiber composite modified asphalt mixture is evaluated. The results show that the diatomite-basalt fiber composite modified asphalt mixture has good road performance, can significantly improve the high temperature stability of the asphalt mixture, and effectively improve the low temperature crack resistance of the asphalt mixture. The fatigue life of asphalt mixture is greatly improved and the water stability is improved to some extent.

The dynamic water scouring effect caused by the vehicle load was investigated with a self-designed apparatus in this paper. The void ratio, water absorption and split tensile strength of asphalt mixture of AC-13 were monitored during the freeze-thaw cycle test by two different immersion modes (static immersion mode and dynamic scouring immersion mode). The microstructure changes were also tested with the aid of X-ray computed tomography (CT) to explore the mechanism of water damage of asphalt mixture both from the macroscopic and mesoscopic points of view. The experimental results show that with the increase of freezing-thawing cycles, the void rate under dynamic scouring mode and static immersion mode increases by 47.62% and 22.5%, respectively, and the water absorption rate increases by 263.2% and 86.4%, respectively. The freeze-thaw splitting tensile strength ratio (TSR) declines by 34% and 24.9%, respectively. The CT images analysis reveals that new frost-heave voids are produced in the inner part of mixtures, and some of the aggregates are also spalled at the outer edge. Dynamic water scouring will accelerate the shedding of fine aggregates and expand the volume of voids correspondingly. The deterioration of the anti-water property of the mixture under dynamic water scouring was more obvious.

In order to explore the influence of salts on the low-temperature rheological properties and adhesion properties of asphalt mortars, preparation method of asphalt mortars with salt particles replacing mineral powder by volume ratio 1:1 and the bending beam rheological test (BBR) were used to study the effects of different salt-doped asphalt mortars on the low-temperature rheological properties at 30 d and 120 d test age. The corresponding Burgers model was built to analyze the effect of salinity on low temperature viscoelasticity of asphalt. Through the contact angle test, the surface energy parameters of asphalt were obtained, and the adhesion characteristics of different asphalt mortar were evaluated. Scanning electron microscopy (SEM) and infrared spectroscopy (FTIR) were used to explore the microscopic mechanism of salt-doped asphalt mortars. The results showed that compared with control asphalt, the rheological property of salt-doped asphalt mortars at low temperature decreases with the increase of age. At the age of 120d, the contact angles of asphalt increase obviously, the surface energy of salt-doped asphalt mortars decreases and its adhesion property degrades. The dissolution and crystallization process of salt particle and its salt aging effect are the main inducing factors of performance deterioration of salt-doped asphalt mortars.

The concrete life prediction model does not accord with actual service conditions if only chloride ion, sulfate ion or carbonation are considered. To solve this problem, the corrosion current density, which reflects the steel corrosion, and the relative dynamic elastic modulus evaluation parameter, which characterizes the concrete damage, are taken as the marginal distribution function parameters. Then a life prediction model of Magnesium Oxychloride-Coated Reinforced Concrete (MOCRC) is established based on the Copula function. The results show that the Gumble-Copula function can predict the durability of MOCRC under the influence of two coupling factors, and the damage of MOCRC is about 20 000 d. The establishment of the Copula function life prediction model under the influence of two coupling factors provides a new research direction for the life prediction model under the multiple coupling factors.

In order to consider the correlation between each floor′s seismic demand in the vulnerability analysis of the frame structure, a frame structure vulnerability analysis method, which can consider the correlation of each floor seismic demand was proposed. The multiple Copula function method was used to deal with the correlation and edge distribution function of each floor’s seismic demand respectively. The incremental dynamic analysis method was used to get the vulnerability curve of each floor. The edge distribution function of each floor’s seismic demand was obtained by using the kernel density estimation method, which is a nonparametric method, and the corresponding parameters of the alternative Copula function were obtained. Then the optimal Copula function was selected by the minimum distance method, and the joint distribution function of each floor′s seismic demand was obtained, and the vulnerability curves of the frame structure were obtained. The results of the first order boundary method and the Bootstrap resampling method were compared. It is shown that the results of the frame structure vulnerability analysis method proposed in this paper are both located between the two boundaries of the first order boundary method, and are in good agreement with the results of the Bootstrap resampling method, which proves the accuracy of the method to a certain extent.

There exist some problems in traditional analysis method of concrete stress test, such as low fitting degree and high complexity. To solve these problems, an experimental analysis method of spatial stress performance of concrete reinforced stone arch bridge based on enlarged section method is proposed. First, the concrete grade is determined as C30, and the concrete is poured with 150 mm×150 mm×150 mm cube. Ribbed steel bars, 14 mm HRB335 longitudinal tension steel bars and 8 mm HPB stirrups are used as design materials for steel members. Then, the shape and size of concrete reinforced stone arch bridge specimens are set up, the specimens are grouped, and the specimens are made according to the manufacturing process. Third, according to the design and fabrication of specimens, the reinforcement section is divided into meshes and strips, the number of meshes in the transverse and longitudinal directions and the number of strips in the reinforcement section are input. Fourth, the spatial mechanical properties of the reinforced stone arch bridge are tested by low-cycle repeated loading, and the concrete strain at the center of any element is calculated. Fifth, the concrete strain is increased on one side of the section near the axial force, and the next step related calculation is carried out until the concrete compressive strain value near the axial force is larger than the ultimate compressive strain value. Finally, the parameters that meet the preset conditions are outputted, such as the ultimate value of the space stress of the concrete reinforced stone arch bridge, and the test is completed. The experimental results show that the method has high fitting degree, about 98%, low running complexity and feasibility.

To solve the problem of bridge vibration signal with high degree of non-stationary feature, an optimization filtering method of adaptive decomposition and reconstruction is proposed for the bridge health monitoring. On the basis of complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN), the original signal is decomposed into multiple intrinsic mode functions (IMFs) with different feature time-scales. The endpoint mirroring method is used to suppress endpoint effect. In order to find the IMFs with high randomness, the multi-scale permutation entropy (MPE) is introduced to analyze the mean entropy value of each IMF at different scales. Then the noise component and abnormal component are respectively eliminated to complete the first filtering. In order to achieve better similarity and smoothness for the final filter, the optimized reconstruction model is established to complete the second filtering. The research shows that adaptive decomposition stage of the proposed method has better completeness, orthogonality and computational efficiency than traditional ensemble empirical mode decomposition (EEMD) and complementary ensemble empirical mode decomposition (CEEMD) methods. Meanwhile, the proposed method suppresses the mode mixing phenomenon to a certain extent, and the endpoint effect is improved. Through the optimized reconstruction, the final filter signal has high signal-to-noise ratio. Finally, the measured signal analysis once again proves the advantages of the proposed method, the filter result can be used as the basis of bridge health monitoring technology.

In order to improve the quality of multi-focus image fusion, a multi-focus image fusion method based on support vector machines (SVM) and window gradient is proposed in this paper. First, the multi-focus images are decomposed by window empirical mode decomposition (WEMD), and a set of intrinsic mode function components (high frequency part) and residual components (low frequency part) are obtained. WEMD can effectively solve the signal aliasing problem in image decomposition. Then, the fusion rule of low-frequency components is determined by the output of the support vector machine, and the clearer focus area is selected. The window gradient contrast algorithm proposed in this paper is used to guide the fusion of high-frequency components, and the consistency of the image is ensured while maintaining the contrast of the fused image. Finally, the WEMD inverse transform is performed to obtain the fused image. Experiments were carried out on 9 sets of multi-focus images. Results show that the proposed method can obtain better fusion quality than the other five methods in terms of the subjective evaluation and five objective evaluation indicators.

With the constant increase in the number and the scale of cloud data centers, the energy consumption control in cloud computing is becoming increasingly apparent. By introducing a periodic sleep mode with double control of wake up threshold and sleep timer, we propose a clustered virtual machine (VM) allocation strategy. All the VMs in a cloud data center are divided into two modules: The VMs in Module I are always awake, while the VMs in Module II will switch between sleep state and awake state according to the workload of the cloud data center. By establishing a queueing model with double service rates and (N,T) policy asynchronous multiple vacations of partial servers, and using the method of a matrix geometric solution, we evaluate the performance of the clustered VM allocation strategy in terms of average latency of cloud requests and energy saving rate of the system. Theoretical analysis results and simulation results verify the effectiveness of the proposed clustered VM allocation strategy. By constructing a system cost function from the perspective of economics, introducing an An chaotic mapping mechanism and a nonlinear decreasing inertia weight strategy to the Particle Swarm Optimization (PSO) algorithm, we optimize the strategy parameters to achieve a reasonable balance between the response performance and the energy efficiency of the system.

In the medical big data service system, there exists the problem of the unbalanced dynamic load of physiological data flow. The processing power of the traditional method is limited to the window range that can be processed by the node where the operator is located. In the state where the data is gradually increased, the processing capacity is insufficient, the data flow congestion is easy to occur, and the load distribution of the whole system is neglected. To solve these problems, a new dynamic load balancing method for physiological data flow based on network parallel computing environment is proposed in this paper. Firstly, the Hash value of the tuple key is used to obtain the corresponding data block of the node, and the corresponding target node is obtained by using the data block record, and the data tuple is output. At the same time, the entropy of parallel computing is extended to define the heterogeneous cluster and solve it. Then, the parallel computing entropy in the network parallel computing environment is regarded as the measurement index of the dynamic load balancing of physiological data flow in the medical big data service system. Finally, by judging whether load migration is necessary by parallel computing entropy, the way and amount of migration tasks are determined by parallel computing entropy, so we can make migration decision and realize dynamic load balancing of physiological data flow in parallel environment of large data network. The experimental results show that the proposed method is highly feasible, and the calculation performance and dynamic load balance are good.

The tradeoff between performance and security has always been a key issue in ad-hoc network research, and many optimization schemes have been proposed by the researchers. However, there has been no a unified quantitative criterion to compare the performance and benefits of these schemes. This paper proposes a widely used method to evaluate performance and security tradeoff in ad-hoc networks. Firstly, GSPN was used to construct the data transmission model within and between nodes. Then the model is abstracted and refined by Petri net. Finally, combining with Continuous-time Markov Chain (CTMC) the method calculates the performance, security and benefits. In this way, the different strategies can be incorporated into the unified evaluation framework, the optimal equilibrium solution can be selected according to the different environments and requirements. Simulations are carried out on data encryption algorithms, key size and strategy selections, which provides theoretical basis and data support to this model for the effective utilization of network resources.

The validation method based on partial evaluation is a novel and effect method to validate compilers. However, in this method the static inputs and their values are specified by hand in the partial evaluation stage, which possesses blindly and limits the framework to run automatically. This paper solves this limitation by proposing an automatic method, which is based on variable dependency, to specify the static inputs and their values. The validation framework can work automatically shipped with this method, therefore improve the efficiency of testing. Results of the experiment show that this method is applicable and effect.

To solve the problem of heavy decrease in object tracking performance induced by illumination change, a visual tracking method via jointly optimizing the illumination compensation and low-rank sparse representation is proposed. The template illumination is firstly compensated by the developed algorithm, which is based on the average brightness difference between templates and candidates. Second, the candidate set is exploited to sparsely represent templates after illumination compensation, afterwards the similarity between the candidates is considered to implement low rank constraint on the sparse coding matrix, and the sparse error term is included to improve the robustness of the algorithm to local occlusion, thereby an illumination compensation and low rank sparse representation joint optimization model can be constructed. Finally, the sparse coding matrix obtained by solving the model is used to quickly eliminate the irrelevant candidates, and then the local structured evaluation method is employed to achieve the object tracking with high accuracy. As compared to the existing state-of-the-art algorithms, simulation results show that the proposed algorithm can improve the accuracy and robustness of the object tracking significantly in the presence of heavy illumination change.

In the denoising of 3D point cloud data, it is difficult to keep the features of sharp areas while making smooth areas highly smooth. To solve this problem, a denoising method based on normal vector distance classification is proposed. Firstly, the differential geometry information of point cloud data was calculated. A robust method was used to estimate the normal vectors, and the normal vectors were adjusted to the same direction. The curvature was estimated by fitting the local quadratic surface of the sampling point. Then, the point cloud data were divided into smooth areas and sharp areas by calculating the normal vector distance from the sampling point to its tangent plane. Finally the smooth areas and the sharp areas were denoised respectively by weighted local optimal projection algorithm and bilateral filtering algorithm. The Bunny and Fandisk models were tested using weighted local projection algorithm, bilateral filtering algorithm and the proposed method respectively. The test results show that the proposed method can eliminate the isolated points in the noise model, improve the uniformity of point cloud distribution, and enhance the smoothness of the smooth areas. At the same time, it can also keep the geometric features of sharp areas, avoids excessive smoothing and detail feature distortion. Compared with the test data, the error and deviation of the point cloud model after noise reduction are smaller, the average error of the Bunny model is 0.001 1 mm, and the average error of the Fandisk model is 0.000 7 mm.

The existing image automatic annotation algorithm has very limited ability of automatic annotation and semantic description of Thangka image. This paper proposes a semantic annotation method of Thangka image based on domain knowledge ontology. First, the system of Thangka image annotation framework was constructed. Then in the early study of the Thangka image target recognition and classification, on the basis of combining Thangka domain knowledge ontology, two levels of Thangka image annotation were realized, which were local region automatic annotation and ontology?based global annotation. Experimental results prove that the proposed method is effective for Thangka image semantic annotation.

In order to achieve that the fusion results can maintain the global structure, preserve the local details and be robust to the unregistered source images at the same time, this paper proposes a multi-focus image fusion algorithm based on latent low-rank representation combining low-rank representation. First, the algorithm decomposes images into low-rank part (global structure) and saliency part (local structure) by latent low-rank representation. Then it extracts low-rank coefficients from each part by low-rank representation. Third, the algorithm fuses the low-rank coefficients of each part. Fourth, the fusion images of two parts are obtained by inverse transformation. Finally the two parts are added together as the final fusion image. Besides, guided filtering is used to enhance spatial continuity. The dictionaries of low-rank representation are consist of several sub-dictionaries, respectively. The source images are divided into blocks with sliding window technology, then four kinds of Sobel operators are used to classify the blocks into four categories. Learning sub-dictionaries from each category can enhance the ability of detail retention. We select three groups of data, including two groups of perfectly registered images and one group of unregistered images for verifying the validity of the proposed algorithm. Experimental results show that this algorithm outperforms the current mainstream multi-focus image fusion algorithms from both subjective visual effect analysis and objective quality assessment analysis.

In cellular networks with D2D systems, high energy cost and interference between cellular users and D2D users become critical and need to be properly handled. In this paper, we present a hybrid-duplex system model where D2D users can switch the duplex mode between half-duplex and full-duplex. Then a resource allocation scheme based on multi-element joint optimization is proposed. The algorithm first guarantees the quality of service (QoS) of cellular users and then maximizes the system capacity through the joint optimal allocation of the user's access control, power allocation, duplex mode selection and channel allocation. Simulation results show that the proposed scheme can improve the overall performance of the networks greatly.

A dynamic game algorithm for cognitive spectrum sharing is proposed, to solve the spectrum sharing problem in the spectrum market, in which there exist multiple primary users and secondary users that belong to primary service and secondary service respectively. The Priority strategy is introduced to meet the spectrum requirements of mission-critical users and high real-time service users of secondary service. The spectrum price satisfying allocation attribute of scarce resources is obtained by considering the revenue of primary service and the satisfaction of secondary service. The non-cooperative game is used to model the competition process between secondary users to maximize the utilities of secondary users. The simulation results of static game, dynamic game and dynamic game stability show that the algorithm is reasonable and effective in spectrum sharing.

Coordinate system generation and transformation play an important role in many fields. Taking the left-hand coordinate system as an example, a clockwise transformation model is deduced based on any rotation angle for plane rectangular coordinate systems in detail. The model is also suitable for the counterclockwise transformation for the right-hand coordinate system. The study shows that the form of rotation matrix has nothing to do with the size of rotation angle, but relates to the sequences and the rotation directions of coordinate axes, which changes when one of the two sequences changes, and stays the same when both sequences change. Based on the above derivation, a universal three-point model is proposed for the generation a new spatial rectangular coordinate system. By adding the fine criterions to the rotational parameter calculation in the gradual coordinate transformation, the universal model can arbitrarily generate a new coordinate system based on non-collinear three points and realize coordinate transformation. This model breaks the distribution limitations for three points used to generate a new coordinate system in the conventional axis alignment method. Furthermore, the universal model can be applied for generating the coordinate system of control field for the calibration of a non-metric camera which has been widely used in the fields of low-altitude remote sensing and close-range photogrammetry as a photoelectric sensor. The reliability of the proposed universal model is verified.

In order to study the performance of downlink in 5G wireless cellular system, Cooperative Distributed Antenna Transmission (CDAT) is a feasible scheme. When using Frequency Division Multiple Access (FDMA), the CDAT technology inevitably causes interference overload, resulting in the increase in Peak-to-Average Signal Power Ratio (PAPR) of transmitted signals, which will lead to a certain decline in channel quality. To solve this problem, this paper theoretically proposes a joint downlink Minimum Mean Square Error (MMSE) and spatial diversity receiving method in multi-carrier systems. In the case of adjacent macro-cell co-channel interference, Space-Time Block Coded Transmit Diversity (STBC-TD) for spatial diversity, MMSE filtering and MMSE-SVD for multiuser spatial multiplexing are described. In order to restrain PAPR growth under CDAT, Blind-Selected Mapping (SLM) is introduced based on co-channel interference distribution. The detailed mathematical expressions of STBC, Co-Channel Interference (CCI), Signal-to-Interference Noise Ratio (SINR) and Bit Error Ratio (BER) are given. Finally, Monte Carlo computer simulation results show that, under the same BER condition, the SNR performance of the system can be improved by about 1dB. Therefore, the CDAT way with full consideration of CCI can improve the system capacity more effectively. It also provides a new way for studying the downlink performance of wireless cellular multi-access system.

In this paper, an energy guaranteed cost optimal control method based on adaptive dynamic programming (ADP) is presented for decentralized control problem of coupling crosslinking uncertainty reconfigurable robots. Based on joint torque measurement information, a reconfigurable robot dynamics model is established. According to the control precision and energy consumption of the performance index function, the Hamilton Jacobi Bellman (HJB) equation is built, then the policy iteration (PI) method is employed to solve the HJB equation and approximate the optimal control strategy. The asymptotic stability of closed-loop system is proved by Lyapunov theory. The simulation results demonstrate the effectiveness of the algorithm.

Screen blocking during cleaning operation of combine by corn cob seriously affects the yield of corn grain. Amplitude, vibration frequency and vibration angle of the screen are three key motion parameters. In order to explore the influence of three key motion parameters on screen blocking, single-factor test, orthogonal test and verification test were carried out by self-made vibration test bench and modified combine prototype respectively. The change rules of acceleration and mass of blocked corn cob with the increase of motion parameters were revealed through single factor test. The regression model of the influence of three motion parameters on screen blocking was established through orthogonal test. Then the optimal combination of three motion parameters was amplitude of 50 mm, vibration frequency of 7 Hz, direction angle of 55°. The validation test showed that the average mass of blocked corn cob was 10.39 g, the grain impurity rate was 0.9%, and the cleaning loss rate was 1.09%. The optimal combination of motion parameters has good screening performance and meets the production requirements of grain cleaning in combine.

This paper develops a real-time interpretation system of variable spray prescription map based on ARM architecture. The system can extract the parameters such as the position, speed and height of the drone in real time, and address it in the prescription map to read the corresponding grid prescription value, and send to the variable spray system. The addressing uses a delay algorithm to optimize the design so that the interpretation results meet the operational requirements. The interpretation system and the variable system were built on the DJI MG-1 plant protection drone, and the implementability and accuracy of the interpretation system were tested through experiments. The results show that the system can successfully interpret the variable application prescription map in real time. Within the effective spray width, the grid center and grid boundary variable results are significant, consistent with the preset prescription value. The collection with the droplet deposition amount is 0.005~0.7658 μL/cm², the droplet coverage density is 12~72/cm², the variable conversion time is about 0.5 s, and the variable accuracy is 1 m.

The mechanism that how hydraulic pressure is generated in the spider body and then transferred to the leg was thoroughly reviewed. The hydraulic system in the spider leg is investigated and the relationship between biological hydraulic and engineering hydraulic problems is tentatively established. Hainan fishing spider is taken as the research object, and its inner channel model is obtained by electron microscope scanning, tissue section, Micro-CT and other observation and scanning experiments. Based on this model, the flow mechanism is studied by using CFD software. According to the hemolymph flow trajectory obtained from CFD post-processing, a bionic hydraulic system is designed. The functional bionic exploration of spider walking hydraulic system is carried out, and a kind of integrated two-way motion joint of spider is designed, which solves the problems that the existing spider-like joint cannot realize the integration of drive and the driving accuracy cannot be guaranteed. It highly reproduces the biological driving process of spiders. This study may provide a new idea for the design of engineering hydraulic system.

A small bionic electronic nose system for oil and gas detection was designed. By analyzing the structural characteristics and flow field distribution inside the nasal passage of the dog's nasal chamber, an optimized bionic electronic nasal chamber simulating the nasal chamber of the dog was proposed, which effectively solved the problems of bloated and inflexible operation of the oil and gas detection device. The 3-D printed chamber sample was used to evaluate the detection effect of hydrocarbon mixed gas. The 10-fold cross-validation method was used to compare the recognition effects of random forest, naive Bayes, extreme learning machine and support vector machine algorithms. The results show that the sensitivity of the electronic nose system after optimizing the detection chamber is obviously higher than that of the ordinary chamber electronic nose system, and the recognition rate is increased by 5.5%. The CFD analysis results of the 3D simulation chamber also verify the validity of the bionic nasal chamber.