In this paper, the static pressure drop and heat transfer coefficient of the cold side of the benchmark air-cooled Charge Air Cooler (CAC) at different speeds are obtained through experimental methods. Then the experimental values are verified by computational fluid dynamics methods. Results show that for the static pressure drop, the absolute error between experimental values and simulation values is less than 7%, and for heat transfer coefficient the error is less than 15%, which indicates the accuracy of the current simulation. After the validation, computational fluid dynamics method was applied to investigate the influence of the leading edge radius R₁ of the cooling pipe and the airflow direction β on the cold side thermal and hydraulic performance of the CAC under the airflow speed of 6 m/s. Research results show that, when R₁=3.6 mm and β=90°, the Fanning friction factor f achieves the minimum value in each working condition; when R₁=0 mm and β=90°, Nusselt number Nu reaches the maximum value in all working conditions; and when R₁=3.6 mm and β=90°, the Performance Evaluation Criterion (PEC) for the thermal and hydraulic performance of the CAC obtaines the maximum value in each working condition.

A heat pump air conditioning system performance simulation platform for electric vehicle is designed and built, which can be used to analyze the system performance in different working modes. The experimental results are used to verify the simulation model accuracy. The results show that the maximum error between experiment and simulation data of compressor power, heat exchange rate and system Coefficient of Performance (COP) is 4%~10.09% in different working modes. The compressor power gradually increases, the heat exchange rate gradually increases and the system COP gradually decreases with the increase of compressor speed in the cooling mode and heating mode. The influence of inlet air flow rate, inlet air temperature, inlet air mode and other factors of condenser and evaporator on the system performance in the cooling or heating mode are also researched in the paper. The results indicate that high air flow rate will improve the cooling capacity, and the effect of inlet air temperature is dependent on working condition, while the partial inlet air mode can save energy in heating mode.

To increase the generation and efficiency of piezoelectric materials, a cantilever piezoelectric vibrator broadband power generation device is designed. Firstly, the output and phase frequency characteristics of a single piezoelectric vibrator were analyzed, the parameter factors that affect the output were obtained. Then, the series/parallel output performance of the double piezoelectric vibrators connected with rectifiers were analyzed, the output voltage was the larger value of the two piezoelectric vibrators and the frequency bandwidth was increased when the series-parallel piezoelectric vibrators were rectified. Finally, the power generation tests of four piezoelectric vibrators were implemented. The test results indicate that the effective power generation bandwidth of a single piezoelectric vibrator is about 10 Hz, the effective power generation frequency range of the series/parallel four piezoelectric vibrators connected with rectifiers is 44.5 Hz~121.3 Hz and the width is about 77 Hz. The effective power generation bandwidth is improved when the output voltage is 30 V. The device can be used to adaptive power supply in narrow space.

Based on the finite element method， the influences of the interference fit and knurling connection types on the two major press fitting failure problems including plastic deformation and sliding during knurling connection of assembled camshaft were investigated. The results indicate that a large press fitting force and frictional shear stress could be obtained for a large interference fit to inhibit the sliding between the fit surfaces. On the other hand， the cam profile’s radial displacement magnitude and gradient change increase with the e interference fit， resulting in an obvious stress concentration around the transition region of cam profile. To keep the fit surfaces from sliding， the axial knurling connection is suggested when the hardness of shaft is larger than that of cam. Instead， the radial knurling connection is more beneficial to inhibit the occurrence of sliding between the fit surfaces for a assembled camshaft with a softer shaft. The press fitting force obtained from the radial knurling connection test of assembled camshaft validates the rationality of both the 1/36 finite element model and the complete finite element model.

In this paper, the average clearance leakage flow of the whole piston pump is derived by analyzing the clearance leakage flow of each friction pair of the piston pump, and the total leakage flow and volumetric efficiency model are established. The flow of the piston pump is random due to the random uncertainty of the size tolerance and the operating conditions during the manufacturing and operation of the piston pump. The volumetric efficiency which is closely related to the random flow is selected as the reliability criterion, and a practical reliability and sensitivity analysis method for the flow characteristics of the piston pump was proposed. The accuracy of the proposed method was verified by Monte-Carlo stochastic numerical simulation. The results show that the volumetric efficiency reliability of piston pump increases with the tilt angle of swash plate and rotation speed, and decreases with the increase of oil discharge pressure. The sensitivity of the of the basic random variables to the influence of the reliability is obviously different, and the slipper pair clearance is the most sensitive to the influence of reliability in many clearances of friction pairs.

In structural reliability analysis, choosing an appropriate adaptive sampling strategy is the key to constructing a high-precision and high-efficiency surrogate model. An adaptive sampling strategy for structural reliability analysis based on General Learning Function (GLF) for multiple surrogate models is proposed. The adaptive sampling strategy is regarded as a multi-objective optimization process, so the average and the minimum distance between the sample points, whether they are distributed near the limit state function, and the probability density function are all considered by the GLF to ensure that the new sample points can robustly and efficiently improve the surrogate model estimation accuracy of failure probability. Numerical cases and engineering case show that for different surrogate models, the GLF can use a small number of sample points to estimate the structural failure probability with high accuracy and efficiency.

This paper studies the trajectory tracking control of ball and plate system with actuator dynamic being considered. To realize precise trajectory tracking control, three steps are included as following. Firstly, according to the target trajectory, the feedforward control and target state are obtained using differential flatness technology. Secondly, tracking error system is established according to the target state and then a moving horizon Linear-Quadratic Regulator (LQR) controller is designed based on one of three fundamental principles of model predictive control. Thirdly, considering the practical constraints of control inputs, the target controller is achieved by solving Linear Matrix Inequalities（LMIs） under the condition of variable elliptic domain limits. Simulation results show that the closed-loop system has good trajectory tracking performance, steady state performance and robust performance under input constraints.

Aiming at the problem of the nonlinear and time-varying characteristics of ultrasonic motor, a fuzzy self-tuning PID controller is designed, and the parameters of the fuzzy self-tuning PID controller are optimized by using an improved quantum genetic algorithm, which can improve the dynamic performance and adaptability of the system. In order to overcome the shortcomings of traditional quantum genetic algorithm, five improving measures are taken, including the coding mode, population initialization, quantum rotation gate, quantum mutation and adding quantum catastrophe. The simulation results show that the improved quantum genetic algorithm can improve the convergence performance and the premature population problem of traditional quantum genetic algorithm. At the same time, the fuzzy self-tuning PID controller based on the improved quantum genetic algorithm significantly improves the dynamic and stable state performance of the ultrasonic motor system compared with the classical fuzzy self-tuning PID controller.

This paper takes the new energy pumping unit as the research object. In order to improve the ability of new energy pumping unit，first, the maximum wind energy tracking control strategy of the wind turbine is studied, and a wind speed estimation method-support vector regression is presented. Then, the key parameters in the estimation model are optimized by particle swarm optimization. Finally, a feedback linearized sliding mode controller is designed for speed feedback control. By comparing with PID controller, it is verified that the new controller has the characteristics of rapidity and anti-disturbance. Therefore, the maximum wind power tracking control strategy proposed in this paper meets the energy-saving requirements of new energy pumping units.

High-pure MnBi alloy was successfully synthesized by closed cryogenic sintering and then rapid quenching method. The phase transformation of the Mn-Bi samples in the sintering process were studied by differential thermal analyzer (DTA). The phases and magnetic properties of the samples were characterized by X-ray diffractometer (XRD) and vibration sample magnetometer (VSM). The results show that the low-temperature phase MnBi alloy block material with high purity can be obtained at 1∶1.3 (n(Bi)∶n(Mn))，275 ℃, 20 h. At 50~350 K, the magnetization decreases with the increasing of temperature, and the coercivity Hc and residual magnetization Mr decrease firstly and then increase. When the temperature is lower than 200 K, the sample is more likely to approach saturation with the decrease of temperature. At 50 K, the saturation magnetization of the sample is 78.0 Am²/kg. The preparation process used in this research is simple and the quantity production of alloy can be realized , which has an important reference value for the preparation of the new type alloy material.

It is inefficient to test only one specimen using traditional bending fatigue experiment machine. A new bending fatigue experiment machine is proposed, which can test six specimens simultenously. The machine is used to test the fatigue life of H13 steel. Firstly, pressure load is applied to six specimens. Then, the force loaded on the specimens is measured by force sensors. The single load is cancelled immediately by the control system if one of the six specimens’ load is bigger than allowable load. Moreover, the fatigue experiment of other specimens is still proceeding. Finally, the bending fatigue experiment is finished as all specimens engender cracks and break due to fatigue. The fatigue life of the specimens is calculated by analyzing the experiment data. The efficiency of bending fatigue experiment is significantly increased via the test of six specimens simultenously.

The stability of temperature field control directly affects the stability of production and quality. Based on the practical production experience, the solidification behavior of molten copper in the continuous casting system of SCR continuous casting and rolling production line and its influencing factors were studied. The influence mechanisms （temperature field, flow field, and solidification crystallization behavior of copper liquid）of casting process parameters such as casting temperature, casting speed, acetylene flow rate and cooling water control were discussed. The formation reasons (casting temperature, casting speed and cooling water flow rate fluctuate greatly) of microstructure defects were analyzed, and the optimization method of slab quality was proposed. The research work has important reference significance for analyzing SCR production line to improve the quality of copper rod.

In order to accurately analyze the characteristics of bus dwelling time and forecast the dwelling time, this paper analyzes the time structure of bus dwelling process. The concept of station service time is introduced to quantify the length of time that a vehicle is used for passengers to get on and off the train in the process of station service. According to the analysis of manual survey data, it is concluded that the boarding time and alighting time are the influencing factors of the station service time and dwelling time. Compared with the alighting time, the correlations between boarding time and station service time, and between boarding time and dwelling time are stronger, and the maximum value of both has the strongest correlation with the station service time and dwelling time. This paper analyzes the impact of congestion situation inside the bus on the boarding time and alighting time of passengers, as well as the influence of the payment method, age and load of passengers on the boarding time. This paper introduces the concept of vehicle stop deviation distance, and analyzes the influence of the degree of vehicle stop deviation and the number of passengers on the boarding delay loss time, and the influence of vehicle waiting time on the departure delay loss time. Based on the above analysis results, a BP neural network model with eight input layers is constructed. The goodness of fit between the training results and the prediction results is 0.915 and 0.955 respectively, which has a good forecasting effect.

To study the coupling relationship between multi-standard rail transit sub-network systems, the topological index value and the coupling coordination degree are calculated based on the complex network theory, and then the multi-standard composite network is constructed. The change of the topological index value of the composite network is studied. The calculation method for the coupling coordination degree of the multi-standard rail transit network is proposed combining factors such as network scale, topology, robustness, transportation capacity, transfer ratio and functional cost. The Chongqing rail transit network (in October 2019) is finally used to do case analysis. The results show that the composite network has a greater average degree and a smaller shortest path length. The coupling degree of the composite network is 0.988, the coordination degree is 0.678, the degree of coupling is high, and the degree of coordination is average. The values of the coupling degree and coordination degree of each index of the four subsystems are quite different, but as the number of systems increases, the overall coupling and coordination degree tends to decrease.

In order to reduce the adverse impact of vehicle stop and go, a single-lane speed coordination control strategy based on the trajectory is proposed. Firstly, the information collected by road-side units is used to offset and modify the trajectories of Connected Autonomous Vehicles (CAVs) in downstream vehicle clusters to predict the trajectory of Human-driving Vehicle (HV). Then, the velocity distribution of upstream CAVs are planned, so that they can merge into the downstream traffic flow at a more reasonable and smooth speed.; The numerical simulation results show that, within the designed range of 0%~25% CAV permeability and 575 m~975 m bottleneck start position, the higher the permeability, the farther the bottleneck start position, and the smoother the CAV speed guidance curve, the stronger the inhibition of speed oscillation. At the same time, it proves that the speed coordinated control strategy can effectively reduce the fluctuation of traffic flow, improve driver's comfort and overall efficiency of vehicle queue.

The type-1 balance model for the single U-shaped assembly line under the assumptions is established. First, according to the characteristics of the adjacency matrix that expresses the priority relationship of the operations, the set of candidate tasks for the entry and exit are determined. Then the assignment tasks are selected in turn according to the comprehensive rank value. Finally, the single U-shaped assembly line balancing problem is solved. Combined with one example, the solution process of the improved heuristic algorithm is explained and programmed. 32 classic examples are calculated using the improved heuristic algorithm program. The calculation results and evaluation indicators are compared with the heuristic algorithm and genetic algorithm respectively. It is verified that the improved heuristic algorithm is effective, fast and stable.

In order to solve the contradiction between the safety and economy of the main saddle of suspension bridge, the ultimate bearing capacity and influencing factors of the main saddle were studied. By summarizing the design parameters of the main saddle structure of the suspension bridge, the parametric geometric model is established using Autodesk Inventor software, and the large-scale general finite element software ABAQUS is imported for large number of numerical simulation studies. The influencing factors of the cable saddle bearing capacity and their changing law are investigated, and the reasonable optimization of the structure is performed. The analysis results show that the ultimate bearing capacity of the original design cable saddle is 3.00 times the design load, which has sufficient safety reserves. The strength of cast steel material has great influence on the bearing capacity of the main cable saddle, while the material strength of the container steel has little influence. The thickness and number of the transverse ribs has a great influence on the load-bearing capacity of the saddle. The thickness of the sidewall of the saddle groove, whether the longitudinal stiffener is provided with transverse ribs, and the thickness of the web have less influence on the load-bearing capacity of the saddle. The transverse rigidity and ultimate bearing capacity of the saddle will decrease with reduction of the transverse size of the saddle. The ultimate bearing capacity of the optimized saddle is reduced by 10%, while the steel consumption is reduced by 20%.

With the help of the dynamic triaxial test system, the changes of coarse-grain soil rebound modulus under different moisture content, stress loading paths and dry and wet cycles (N) are studied. A several estimation models are used to determine the Ni model as the estimation model of coarse-grain soil modulus, and the correction coefficient and fitting equation are put forward. The test results show that the rebound modulus decreases with the increase of water content, and the rebound modulus corresponding to 4% water content is the largest. When the partial stress is fixed, the surrounding pressure and the rebound modulus are negatively correlated with the rebound modulus. With the increase of N the rebound modulus has the trend of decay, which reaches the maximum at N=1~2, and the decay is reduced at N=3~4, when N>4, the rebound modulus becomes stable. The research results provide a reference basis for the design of coarse-grain soil subgrade.

Truss structures are prone to damages of components and joints after long-term service, so it is necessary to detect the damages of the structure accurately and timely. Considering the stiffness reduction of joints caused by damage, linear springs in six-axis directions are used to simulate the damaged joints, and the spectral element method based on Laplace transform is used to establish the dynamic model of truss structure with damaged joints. A narrow-band pulse excitation is applied to the truss structure to excite the propagation of high-frequency elastic wave in the structure. The time-domain response of the structure is obtained by using the numerical inverse Laplace transform method. The results show that, when the elastic wave passes through the component with damaged joints at both ends, the propagation will be hindered to a certain extent, resulting in the change of displacement amplitude of truss nodes and the delay of the arrival time of the first wave. By analyzing the change of elastic wave propagation in truss caused by joint damage, the damage detection of truss structure can be realized.

A new type of Bamboo-plywood and Steel-tube dual-confined Dust-powder Concrete Column (BSDCC) was proposed for integrated utilization of industrial solid waste and bamboo material. 12 BSDCC specimens were executed to eccentric compression tests. The compressive destruction mechanism was analyzed based on the whole process of surface damage of specimens, the failure modes of thin-walled steel tubes, the load-displacement and load-strain curves. A formula for calculating the bearing capacity of BSDCC subjected to eccentric compression was proposed through nonlinear regression analysis. The test results show that the eccentric compression failure of BSDCC is mainly caused by the cracking failure of bamboo-plywood between binding bars in the middle of the column body and the local buckling failure of bamboo-plywood materials. The ultimate load of specimens is not only related to eccentricity, slender-length ratio, sectional size and wall thickness of steel-tube, but is also affected by the spacing ratio of binding bars. Steel-tube filled with dust-powder concrete can effectively change the ultimate failure mode and significantly improve the ultimate bearing capacity of BSDCC specimens. Compared with the ultimate compressive stress of SBCCB, the ultimately compressive stress of BSDCC is increased by 25% averagely.

In order to study repair effect of concrete column wrapped Carbon Fiber Reinforced Plastics （CFRP） after extreme freeze-thaw environment， according to the requirements of code of China's current concrete durability （GB/T 50476—2008） and technical specifications for strengthening concrete structures with carbon fiber reinforced polymer laminate， axial pressure test was designed by concrete members wrapped by different spacing and dosage CFRP after freeze-thaw cycle 0 times， 50 times， 50 times， 100 times， 150 times and 200 times. The compression properties of concrete repaired with CFRP were studied by sample microstructure and considering three factors： compression load， vertical overall displacement and the ratio of expansion and compression. The results show the freeze-thaw has a more significant impact on compressive strength， but the compressive strength of freezing-thawing damaged concrete can be improved effectively after CFRP repair， up to 630.6%， and the improvement of the bearing capacity of specimens repaired with CFRP is not related to the repair spacing， but to the fiber amount of CFRP. The overall vertical displacement increases with the increase of freezing-thawing times， and decreases with the increase of the repaired CFRP. With the freezing-thawing cycle， the ratio of expansion and compression first decreases， then increases and then decreases， and the first peak value first increases and then decreases， and the second peak gradually decreases from the "convex" peak gradually to the peak of the collapse of the concave， and the distance between the two peaks gradually decreases. Through microstructure analysis， thechange of the ratio of expansion and compression is mainly caused by freezing-thawing increasing compression and hoop effect of CFRP.

To study the effects of different lithium slag contents and recycled coarse aggregate percentages on the double-K fracture toughness of recycled concrete with lithium slag, 36 standard three-point bending beams of recycled concrete with lithium slag, which include 9 mix ratios about lithium slag content of 0%,10%,20% and recycled coarse aggregate percentage of 0%,30%,50%, were prepared for fracture test. Based on the analysis of crack propagation process, a prediction model for fracture toughness of concrete was proposed. The research results show that the double K fracture toughness of concrete all increases first and then decreases with the increase of the recycled coarse aggregate percentage. The addition of lithium slag can improve the double K fracture toughness of the concrete and offset the deficiency of the concrete fracture performance with high recycled coarse aggregate percentage. The double K properties are optimal when the lithium slag content and recycled coarse aggregate account for 20 % and 30 % respectively. The initial fracture toughness and unstable fracture toughness reach 0.54 MPa·m^1/2 and1.07 MPa·m^1/2, increasing by 42.1% and 23 % respectively compared with the reference group.

In order to analyze the dynamic features of the special transition section of different types of filling adjacent to fill, considering the background of Haoji (Haole Baoji-Jian) heavy-haul railway engineering, the method, combining the field excitation test and numerical simulation, was used to reveal the difference of dynamic features between transverse and longitudinal of the transition section subgrade between improved soil and group A filler. The results show that under the same load condition, the dynamic response of the cross section of the A-filled subgrade is greater than that of the modified expansive soil, which is specifically manifested by the increase of dynamic stress by 5~8% and acceleration by 40~50%. When the trains with a speed of 120 km/h and an axle load of 25~30 t, the dynamic response of the longitudinal road surface at the junction of the transition section increases by 0.8~13%, and to certain extent it is affected by the direction of the vehicle. The research results can provide a theoretical basis for the fine construction and maintenance of the filling transition section of heavy-haul railway.

A method applying extension analytic hierarchy process （EAHP） was proposed to evaluate asphalt mixture road performance whose comprehensive evaluation index has the characteristics of multi-hierarchic and multi-factors. This method used interval number instead of point number to construct judgment matrix， and used correlation function to simplify extension interval number and calculate weight vector. Through the construction of comprehensive evaluation system of asphalt mixture pavement performance， the determination of extension interval number judgment matrix， the calculation of feature vector， and the consistency test， the single weight and total sorting weight were obtained. Based on the classification and grading method of asphalt mixture pavement performance index and considering the impact of weights， the asphalt mixture road performance in three sections of Jicao Expressway was comprehensively evaluated. The results show that the comprehensive score of 23 section is the highest and the road performance is the best.

Aiming at the problem of redundant fail data in the existing N-cover fail data reduction method, a fail data reduction method combining pattern detection and local search is proposed. A local search method combining configuration checking strategy was proposed. By analyzing the logical relationship between fail data and the coverage time of ports, the concepts of fail frequency and defect coverage degree were proposed. Meanwhile, the configuration checking strategy is combined to avoid visiting the same solution space, and the search breadth is increased by reducing redundant iterations. The experimental results for benchmark circuits demonstrate that, compared with existing methods, the proposed method effectively reduces the amount of fail data and improves efficiency while ensuring the quality of diagnosis.

As the existing camera planning schemes are unable to deal with multiple targets and achieve real-time effect, in this paper, focusing on automatically optimizing the camera visual effect, a real-time camera planning algorithm is proposed for dynamic multiple targets. The method first adopts the toric camera space among the multiple targets, then takes the visual properties of the targets as the input of the non-linear optimization function. With the model predictive control, we can optimize the targets visual effect in real time. By analyzing the operation time, efficiency and performance of the experiments, it is verified that this method can effectively optimize the visual effect of the targets in the scene and it promotes the research of the multi-target camera planning algorithm.

This paper proposed a COVID-19 chest CT image segmentation method based on Federated Learning (FL) and blockchain to automatically segment the area in lung affected by COVID-19. Firstly, in the situation where the sample data of patients is limited and it is distributed in different institutions, which cannot be easily collected, we applied FL method. Then, we used blockchain network to replace the central server in FL to solve the “single point of failure” problem. Finally, we designed a lightweight separable convolution U-NET to reduce the cost of computation and time. Experimental results show that the method has good performance after training, and its dice metric can achieve 63.26%, which is helpful for diagnosis of COVID-19.

Due to the diverse species of pests and the large differences between classes, it is difficult to achieve accurate classification. Based on the two processes of target positioning and target recognition when the human visual system recognizes objects, a pest classification model PestNet is designed. The model is mainly composed of Object Positioning Module (OPM) and Multi-Feature Fusion Module (MFFM). OPM integrates the shallow detail information and deep spatial information of the image through a U-shaped network structure, preliminarily delineating significant areas, and outputting spatial semantic features. MFFM performs bilinear pooling operations on spatial semantic features and abstracts semantic features to weaken background information and increase detailed features. In addition, training is assisted by target area clipping and masking to improve model classification accuracy. We conducted experiments on the disease and insect pest data set IP102, and the model classification accuracy rate reached 77.40%, which can realize the classification and recognition of large-scale pest images under complex background.

This article proposes a service recommendation approach for a type of Internet of Things (IoT) automation-driven applications. The approach combines text analysis, heterogeneous information networks and deep learning to find related objects in the IoT by analyzing text, and then recommends the services. First, we construct a text fusion heterogeneous information network, which integrates text and text relationships. Then we design meta-paths and relationship weights. Finally, we train a neural network model for recommending the services. This paper conducts an experimental evaluation on a real data set. The results show that the proposed approach can be used for recommendation task and outperforms some classical methods.

In order to improve the ability of multi-verse optimization algorithm to solve practical problems, a modified multi-verse optimization algorithm with slime mould foraging is proposed. The algorithm uses slime foraging behavior to further seek optimal solutions between local optimum and global optimum. By comparing with 10 other similar algorithms tested on 12 benchmark-functions, the results show that the convergence speed and solution quality of the proposed algorithm in this paper are better than other algorithms, with better solving ability and optimization performance, and can be used as an effective tool for problem optimization.

Aiming to attest the real integrity state of cloud system and improve scalability and efficiency of existing schemes, a Scalable Collective Remote Attestation (SCRA) scheme towards cloud system is presented in this paper. First, this paper proposes an attestation relationship tree structure to determine the cooperative relationship among cloud nodes, so that each node could implement collective attestation to promote efficiency and scalability. Second, to improve the accuracy, this paper introduces the identity-based aggregation signature technology to quickly review the validity of attestation report. Finally, we conduct security analysis and SCRA is simulated at large scale to assess scalability. Compared to current cloud system attestation methods, experimental results show that SCRA is not only able to effectively prove large-scale cloud system, but could also perform well in terms of scalability.

In this paper, the Non-line-of-sight (NLOS) error is identified by virtual anchor high-dimensional coordinates and its influence on positioning accuracy is weakened. In N-dimensional space, the N+1-dimensional coordinates of the anchor point are introduced, and the ranging value is modified based on the N+1-dimensional coordinates of the anchor nodes. Then the N+1-dimensional coordinates of the anchor nodes are solved by the modified value, and the positioning result of the node to be solved is optimized by iteration. The NLOS identification and localization optimization method of virtual anchor high-dimensional coordinates are tested. The results show that multiple NLOS errors can be identified by introducing virtual anchor high-dimensional components, and the localization accuracy can be effectively improved by iterative correction.

In order to facilitate the transformation of libraries to new forms of smart libraries， face recognition plays a very important role in it. Image segmentation， on the other hand， is one of the most fundamental studies in face recognition. In this study， we first propose a hybrid successive ant colony optimization algorithm （BACO）， and further propose a BACO-based multi-threshold image segmentation model for facial image segmentation. In order to verify the performance of the BACO-based multi-threshold image segmentation model， it is compared with nine other similar methods for experiments. For the obtained experimental results， the segmentation results were first evaluated using PSNR and FSIM， and the evaluation was further analyzed using Wilcoxon signed-rank test. The obtained results clearly confirm that the BACO-based multi-threshold image segmentation model obtains the optimal performance. Thus， the proposed multi-threshold image segmentation model lays a solid foundation for the next step of facial recognition and even smart library construction.

Massive Multiple-input Multiple-output (MIMO) systems may cause serious problems named pilot contamination due to the limitation of the number of pilot sequences and interference of users. which will result in the reduction of the system performance. In order to suppress pilot contamination, a Divided Pilot Reuse Strategy (DPRS) is proposed to redistribute the pilot sequences. The pilot sequences are divided into central user groups and edge user groups according to the degree of user interference in the cell, and the fully orthogonal pilot sequences are allocated for edge users. In addition, a pilot overhead optimization strategy is proposed for edge users, analyzing the pilot overhead of the system, and seeking the optimal solution of both pilot contamination suppression and pilot overhead. Theoretical analysis and simulation results show that the proposed method can suppress pilot contamination and improve system performance.

In order to solve the problem that the Direction of Arrival (DOA) estimation of one-dimensional wideband signal needs angle estimation, a DOA estimation algorithm without angle estimation is designed on the basis of bilateral correlation transform algorithm. Firstly, the wideband signal is decomposed into several narrow-band data models at different frequency points by using Discrete Fourier Transform (DFT). Secondly, the narrow-band data at different frequency points are focused on the same reference frequency point by the improved focusing matrix, thus, the narrow-band signal model at a single frequency point is obtained. Finally, the improved array covariance matrix sparse iterative estimation algorithm is used to solve the problem. Theoretical research shows that the proposed method has higher estimation accuracy and resolution than the traditional algorithm under the condition of low SNR and multiple snapshots, and the computational complexity of the algorithm is effectively reduced by combining with compressed sensing theory. The simulation results verify the reliability of the above conclusions.

In order to overcome the shortcomings of traditional compressed sensing reconstruction algorithms, such as heavy dependence on sparseness and low reconstruction accuracy, an Optimized Discrete Differential Evolution (ODDE) algorithm is proposed to analyze the evolutionary population. ODDE can improve the learning and evolution of the population while realizing effective clustering. The differential evolution particle coding method and evolution mechanism are redefined, and the ODDE is applied to the compressed sensing reconstruction method. The sparse unknown signal is equivalent to the particle coding, and the sparse signal is reconstructed accurately through population iterative evolution. The simulation results show that compared with traditional reconstruction algorithms such as StOMP, the reconstruction accuracy of this method is significantly improved and the reconstruction time is reduced.

In order to effectively mine network protocol vulnerabilities， prevent malicious attackers from divulging protocol confidential information， and maintain the security of protocol running environment， a vulnerability detection method of instant messaging network protocol based on passive clustering algorithm is proposed. In passive clustering algorithm， the first declarator priority mechanism is used to select the cluster head， and the gateway node is defined according to the balance principle between network robustness and energy efficiency； the protocol to be detected is formally defined to obtain the detailed workflow of the protocol； the positive and negative samples of the protocol are oversampled by AFL fuzzy detection tool to obtain the complete sample set， and the forward feedback network and support vector are obtained The machine is regarded as the generation model and the discrimination model in the generative adversary network， and the detection case data is obtained by using Lagrange algorithm， which is substituted into the protocol system to complete the vulnerability detection. Simulation results show that the proposed method has high accuracy and efficiency of vulnerability detection， and can effectively ensure the security of network protocol operation.

This paper presents an on-line trajectory planning method based on the reinforcement learning for driving the quadrotor to its destination accurately and suppressing the swing motion of the slung-payload effectively. To deal with the unknown external disturbances, the desired trajectory of the Unmanned Aerial Vehicle (UAV) is divided into two parts: the positioning trajectory planning and the disturbance rejection trajectory planning. The positioning trajectory planning can be designed in advance to guide the UAV to reach the desired position, and the disturbance rejection trajectory planning can compensate the unknown external disturbances based on the reinforcement learning strategy and suppress the swing motion of the slung-payload simultenously. The Lyapunov based stability analysis is employed to prove the stability of the closed-loop system, the convergence of the UAV′s position and the swing motion of the slung-payload. Finally, real-time comparing experiments are performed to verify the effectiveness of the proposed trajectory generation method and its robustness to external disturbances and variation of the mass of the slung-payload.

To solve the problem that the feature extracted by the manifold learning algorithm in Euclidean space is not significant enough, this paper proposes a sparse graph preserving projection algorithm based on Riemannian manifold and applies it to bearing fault diagnosis. Firstly, the symmetric positive definite matrix of the original data is calculated to construct the Symmetric Positive Definite manifold (SPD manifold). Secondly, the sparse structure of the SPD manifold is explored by using the regularization technique. On this basis, the intrinsic graph of within-class and the penalty graph of the between-class of the sample are established respectively. Finally, the feature extraction of the data is realized by the method of graph embedding. Experimental results show that the sparse graph preserving projection algorithm based on Riemannian manifold can extract significant features.

In order to improve the stability of Brushless DC Motor (BLDCM), a method of Q-learning algorithm optimized BP neural network PID controller (QBP-PID) is proposed. QBP-PID uses BP Neural Network (BPNN) to adjust the PID gain, and then optimizes the key weights in BPNN by modifying the weight momentum factor through Q-learning. Therefore the controller has better learning and online correction abilities, and the BLDCM can achieve better control effect. The simulation results show that QBP-PID has better adaptive ability, anti-interference ability and stronger robustness than the traditional PID, Fuzzy PID (Fuzzy-PID) and BP neural network PID (BP-PID) controllers.

In order to solve the intelligent vehicle path stability tracking control problem on high-speed road with low surface adhesion coefficient, a lateral and longitudinal coupling integrated controller based on nonlinear model predictive control is proposed in this paper, and the real-time control is realized. First of all, the desired vehicle lateral velocity is calculated according to the desired path information. Second, the objective function is designed as the errors of the differences between actual lateral and longitudinal vehicle speeds and the desired speeds. Third, considering the actuator constraints, a nonlinear model predictive controller is designed and the optimal solutions of the front wheel angle and the driving force are obtained to realize the vehicle path stability tracking control. Fourth, in order to improve the real-time performance of the controller, the nonlinear programming problem is transformed into algebraic equations, and the parallel Newton method is used to solve the equations. Finally, through decoupling the coupling relationship of the equations between time domains, the computational complexity of the optimization problem is reduced.The simulation experiment of the controller and the hardware-in-the-loop experiment are carried out under the double line shifting condition. The experimental results verify the effectiveness of the controller designed in this paper, and realize the real-time control of vehicle path tracking.

In order to solve the problem that the uncertainty in the servo system of Permanent Magnet Linear Synchronous Motor (PMLSM) affects the servo accuracy, an Adaptive Fuzzy Fractional Order Sliding Mode Control (AFFOSMC) method was proposed to ensure that the actuator position of the PMLSM follows the given position precisely. Firstly, the mathematical model of PMLSM with the uncertainty of parameter change and external disturbance was established. Then, Fractional Order Sliding Mode Control (FOSMC) was designed to ensure the robustness of the system. Because it was difficult to determine the switching control gain of FOSMC in application, AFFOSMC was used to estimate the uncertainty of the system. At the same time, an Adaptive Fuzzy Reaching Regulator (AFRR) was designed to compensate the estimation error of AFFOSMC. Therefore, the chattering can be further weakened and the performance can be improved without the uncertain boundary information. Finally, experiments show that the proposed AFFOSMC method can effectively reduce the impact of parameter changes and external disturbances on the system, improve the robust control performance and accurate tracking response of the system.

At present, weed identification mostly uses weeds images taken in laboratory, and it is difficult to quickly and accurately identify weeds in a complex field environment. In order to improve low weed identification accuracy and slow detection speed in natural field environments, according to the characteristics of the weed image data, this study proposed a weed recognition model based on Xception lightweight convolutional network by improving the activation function and pooling layer. Firstly, the Xception was improved using Exponential Linear Unit (ELU) as the activation function of the model, and the global max pooling layer was used to downsample the convolution of the last layer. Then, the weed image was processed by background segmentation and data enhancement, and the best model was obtained by fine tuning on the pre-trained model. Compared with the four standard deep convolutional network models of VGG16, VGG19, ResNet50 and Inception-V3 under the same experimental conditions, the experimental results show that the overall performance of the model in this paper was optimal with the test accuracy rate as high as 98.63%. The scale of the model was 83.5MB, and the average time for detecting a single weed image was only 63.8ms. This study can provide theoretical basis and technical support for the implementation of precision spraying in natural field environment.

The effect of light culture technology on the explants growth of Taxus cuspidate was studied. The young stems of Taxus cuspidate were used as the explants. The optimal dosage of sucrose in the culture medium were researched. The orthogonal test was applied to study the optimal light culture conditions (CO₂ concentration, light intensity, light cycle) for explants of Taxus cuspidate. The results show that the optimal concentration of sucrose was 10 g/L, under the 3∶1 of red to blue light ratio, 1500 mg/L of CO₂ concentration, 12.5 mol·m^-2·s^-1 of light intensity, and 10 h/d of photoperiod. Under this condition, the explants have the highest bud germination rate and healing rate, short bud germination time and initial healing time, and the lowest browning rate. The optimal combination of factors in the light culture on explants of Taxus cuspidate was as follows: CO₂ concentration was 3000 mg/L, the light intensity was 37.5 μmol·m^-2·s^-1, and the photoperiod was 14 h/d. The verification test showed that compared with the control group (without light), the bacteria rate, bud germination time and initial healing time of explants under light culture are reduced by 4.33%, 2.33 d, and 3.75 d, respectively. Bud germination rate, healing rate, fresh weight growth rate of callus and paclitaxel content are increased by 16%, 6.33%, 0.35 times, 0.09 mg/L, respectively, which meant that the optimization is obvious. This study suggests that the light culture provides theoretical basis and technical support for tissue culture of Taxus cuspidate under light environment.

Based on the structure of the human medial meniscus, we designed a free bionic medial meniscus prosthesis to replace the irreparable meniscus in clinic treatment. The feasibility of the prosthesis was verified by a three-dimensional finite element analysis method. The data results show higher stresses in the meniscus prosthesis compared to the wholesome meniscus, with the region of high stress distribution in the medial middle and posterior in both models. The results of the study illustrate that the designed meniscus prosthesis has similar biomechanical properties to the healthy meniscus. The main advantage of the data results validating the prosthesis is that no supplementary fixation pattern is required, avoiding failure due to prosthetic fixation failure.