To reduce the influence of the lateral motion on the vehicle mass and the estimation accuracy of the road gradient， an estimation algorithm considering the lateral motion is proposed， the vehicle dynamics model is corrected by the acceleration， and the forgetting factor is used to enhance the new data to adapt to the minimum value of the time-varying characteristics of the vehicle system. The vehicle mass is estimated by the quadratic algorithm， and the mass estimation result is input into the road gradient estimation in real time； in addition， two gradient estimation models of vehicle kinematics and dynamics are established， and the acceleration correction term is added to the model， and the strong tracking filtering algorithm is designed respectively. A time-varying interactive multi-model fusion algorithm is proposed to estimate the road slope for the two models. The estimated road slope is obtained according to the weight coefficients of the two slope estimation models and the transition probability between the models. The proposed algorithm was tested and evaluated on a real vehicle in the Nong'an Automobile Proving Ground of the Technology Center of China FAW Co.， Ltd. Compared with the fusion estimation algorithm that did not consider the lateral direction， it improves the estimation accuracy of road slope when the vehicle moves laterally.

A DIP-LSTM model with dynamic interactive poling layer is proposed， which enables neighboring vehicles to share hidden states of LSTM network by pooling to get the characteristic of historical trajectory， and then realizes interactive modeling of time-space relationship between target vehicle and surrounding vehicles. NGSIM from USA and High-D from Germany are used to train and test the model， and the accuracy， robustness and transferability of the model are verified. The results show that compared with the traditional model prediction method， the DIP-LSTM network show advantages in prediction accuracy and long-time prediction considering multi-vehicle interactive information， and the model has good transferability and robustness， which significantly improves the practicability and universality of intelligent vehicle trajectory prediction model.

Aiming at the problem that the load potential energy is wasted in the load falling stage of the existing electric servo actuator， this paper proposes a new electric servo actuator implementation scheme based on energy recovery. The project uses an accumulator parallel to the transmission system to recover the potential energy dissipated during the load falling stage and provide extra thrust during the load lifting stage. Combined with the theory of power bond graph， a dynamic energy consumption analysis method of mechanical system in complex energy domain is introduced. The dynamic energy consumption model is established， and the comparison simulation test is carried out. The simulation results show that the response time can be reduced by 36.9%， and the motor starting torque can be reduced by 43.3%. When lifting the load， the energy consumption is reduced by about 31%， and 36% of potential energy can be recovered. In the working process， the bearing capacity of the ball screw pair is significantly reduced， which is helpful to reduce friction heat and prolong the service life of the screw.

The paint defects that exist during the automotive painting process affect the overall appearance quality of the car. In response to the problems of missed inspection，low efficiency，and high implementation cost of traditional inspection schemes in manual inspection，a paint defect detection method based on the improved YOLOv7 algorithm is proposed. A dataset of automotive paint defects was constructed，consisting of 4023 images，including 5 types of automotive paint defects； In response to the problem of insufficient detection accuracy of YOLOv7 algorithm on small defects，GAM attention mechanism and SPPFCSPC module were introduced into the original network to improve the algorithm's ability to extract small defect features. At the same time，an improved ELAN module was used to improve the network structure to reduce the problem of small target information loss caused by deep network，ensuring that the network model is reduced while improving the recognition accuracy of small features； Based on the constructed dataset，the defect detection performance of different algorithms was tested and the effectiveness of the module was verified. The experimental results show that this method significantly improves the detection ability of small defects on paint surfaces，with an average detection accuracy of 88.9%，which is the highest detection accuracy compared to various algorithms.

Auto mechanical automatic transmission system can not meet the operation requirements of low energy consumption and high power at the same time. In order to improve the operation performance of the transmission system， a multi-objective optimization of auto mechanical automatic transmission system parameters based on particle swarm optimization algorithm is proposed. This method first learns and analyzes the overall dynamic model of the vehicle and the motor efficiency， then establishes the objective function according to the maximum speed， acceleration performance and climbing performance of the vehicle during driving， and finally uses particle swarm optimization algorithm to solve the optimal solution of the objective function， uses the inertia weight of the vehicle during driving to make the global search ability reach a balanced state， and realizes automatic particle optimization based on the learning factor， To realize the multi-objective optimization of the parameters of the automotive mechanical automatic transmission system. The experimental results show that the proposed method has low energy consumption， high operating efficiency and good dynamic performance. It can effectively improve the operation performance of the mechanical automatic transmission system of automobiles， and has a certain role in promoting the non delayed transmission of automobiles.

To alleviate the radial power deviation caused by flow inertia in automotive shock absorbers， a fault detection algorithm based on dynamic analysis of automotive shock absorber indicator diagram is proposed. Calculate the flow inertia of the shock absorber fluid in the left and right enclosed areas， substitute the radial force in the compression stroke and the radial force in the rebound stroke into the energy and momentum equations， establish the dynamic equation of the car shock absorber， draw the measured indicator diagram， and input it as a sample into the support vector machine for classification. Set the judgment threshold for threshold judgment， and complete the fault detection of the indicator diagram. The experimental results show that the proposed method's fault detection accuracy is good， as the indicator diagram of the automotive shock absorber obtained during insufficient liquid supply， double valve leakage， piston jamming， and sand resistance is basically consistent with the measured indicator diagram.

To reduce the difficulty in autonomous loading operations， a mining haul truck dump body optimization method for payload balancing is constructed. Firstly， a differential iterative algorithm for reconstruction of material pile in dump body was built for calculation of center of gravity. Secondly，based on such calculation algorithm， the discrete element method （DEM） simulation for loading operation was constructed and validated. Thirdly， the influence of dump body structure on position of piled material gravity center was studied using DEM simulation experiments. At last， an optimization on dump body structure was accomplished with the object of payload balancing in cross section and the restraints of longitudinal payload distribution， safety and capacity. The comparison results indicate that the optimized dump body can generate more balanced payload distribution in extreme loading positions. Besides， it has larger loading area and higher operation efficiency.

At the temperature of 653-713 K and the strain rate of 0.01-10 s^-1， Hot compression test of w（GNP/7075Al）= 0.5% composite was applied， and strain compensated Arrhenius and BP neural network model were established. At the same time， the hot processing map and dynamic recrystallization volume fraction prediction model of the composite were established. The hot deformation behavior of the composite was studied， and the hot processing parameters of the composite were determined. The results show that the predicted values of flow stress obtained by BP neural network model are in good agreement with the experimental results. The highest correlation coefficient is 99.998 3%， and the minimum absolute value of average relative error is 0.5%. It shows that neural network has high prediction accuracy for the hot deformation behavior of w（GNP/7075Al）= 0.5% composites. The optimum deformation temperature and strain rate of w（GNP/7075Al）= 0.5% composite are 685-705 K and 0.01-0.1 s^-1， respectively. Dynamic recrystallization （DRX） tends to occur at low strain rates and high deformation temperatures. Numerical simulation and hot extrusion test show that the profile with good surface quality can be extruded under the temperature of 693K and extrusion speed 1mm/min.

The problem of vehicle location estimation in the vehicle-infrastructure cooperative systems was formulated into the construction and optimization of a spatial-temporal graph model， and a spatial-temporal graph optimization cooperative localization（STGO-CL） method was proposed. In the graph model， the location of the vehicle at different times in the perception area constituted the nodes， and the absolute and relative location of the vehicle calculated by the vehicle end and the road end fused with high-precision（HD） map constituted the edges. And time delay compensation constraints are added. In the solution process， the LM method was used to solve the objective function to realize the optimal state estimation of the vehicle location in the perception area and realize the vehicle-infrastructure-map cooperative localization. Used CARLA to establish straight and curve simulation experimental scenes to verify the proposed method. The experimental results demonstrate that the average localization error of the proposed method is 0.29 m. The localization performance is improved by 97.1% and 55.4% respectively compared with GPS or RSU localization alone. Compared with the proposed method without HD map， the proposed method is improved by 42.0%. In terms of time delay compensation， the localization performance under 200 ms time delay can be improved by 67.0%. The use of spatial-temporal graph model to realize the vehicle-infrastructure-map cooperative localization can effectively improve the environmental perception performance of the vehicle-infrastructure cooperative systems.

The purpose is to investigate the mesoscopic cracking behavior of heterogeneous cement-stabilized macadam （CSM） materials using the discrete element method （DEM）. According to Weibull distribution material random field， the heterogeneous fracture models with different homogenization levels were established. The virtual semicircular bending （SCB） tests were carried out to simulate the mesoscopic cracking process. The influence of mortar heterogeneity on mesoscopic cracking behavior of CSM was further studied. The results show that the CSM cracks tend to propagate along the weak mortar and the interfacial transition zone， and the fracture of mortar matrix was dominant. The peak strength and failure deformation of CSM material increase with the increase of mortar homogeneity. A small amount of weak mortar matrix is beneficial to improve the structure deformation characteristics and enhance its overall strength. The optimum crack resistance of CSM materials can be obtained by maintaining reasonable homogenization levels of mortar matrix.

In order to solve the difficult problem of not being able to scientifically classify and effectively evaluate the business flow due to the huge construction content and complicated business during the evaluation of the degree of intelligence of airports， this study establishes the evaluation system of airport core business flow based on object-attribute double dimensions from the perspective of task object-oriented and functional attribute-oriented， and proposes 90 basic indicators. On this basis， the scores of the indicator layer of the two dimensions of Lukou Airport were calculated by the superior and inferior solution distance method. The research results show that： The establishment of the two-dimensional evaluation index system helps to clarify the problem of multiple types of business flows and complicated flows in smart airports， and facilitates the evaluation of the degree of smartness in subsequent airports； The index layer score based on the task object dimension shows that the degree of intelligent construction outside the segregated area for passenger flow is the highest at 73.76. In the evaluation of each indicator divided by functional attribute dimension， the quantity/weight indicator of the perception category has the highest rating of 79.68， 21.4% higher than the average value； the time indicator of the prediction category has the highest rating of 80.26， 9.1% higher than the average value； and the event indicator of the decision-making category has the highest rating of 76.30， 10.2% higher than the average value. The degree of intelligent construction of the related indicators in the prediction and decision-making categories is relatively balanced， while the lowest score of the perception category indicator is 54.36 and the highest score is 79.68， and there is a big difference in the degree of intelligent construction of each business.

A risk assessment method that combines risk probability and risk loss has been established to address the problem of insufficient consideration of driver status and risk loss in intersection risk assessment. The entropy weight method is utilized to determine the risk probability based on drivers' physiological information， while direct risk loss is calculated using the energy conversion theorem， and indirect risk loss is identified through the introduction of environmental vulnerability metrics. To prioritize intersection risk， a clustering algorithm is applied to rank both risk probability and risk loss. Subsequently， the intersection risk rank is determined through a risk matrix analysis. Empirical using data from 19 intersections in Xi'an city demonstrates that intersection risk levels can be categorized into four tiers： relatively safe， generally safe， relatively risky， and very risky， with a predominant concentration in the relatively risky category. The proposed intersection risk assessment method based on drivers' physiological information matches the perceptions of test drivers with 90% accuracy， providing a new method for driver risk management.

To study the impact of connected autonomous vehicle （CAV） on traffic flow， based on the intelligent driver model （IDM）， a car-following model for CAV is constructed by considering the rear vehicle and velocity difference of multiple front vehicles simultaneously. Then critical stability condition is deduced by applying the linear stability analysis theory. Taking a rear vehicle and five-head vehicles into consideration， the numerical simulation is performed. The results show that under the backward looking effect context only when the backward weight ratio belongs to an appropriate range then the traffic flow stability can be enhanced. Furthermore， accounting for both the rear vehicle and the velocity difference of multiple preceding vehicles can also reduce the instability of traffic flow caused by time delay. The acceleration of the vehicle under the new model is gentler and more conducive to improve the stability and safety of traffic flow.

To solve the problem of supply-demand mismatch caused by the imbalance of time-space passenger flow， this paper proposes an electric bus scheduling method with multiple vehicle types based on the characteristics of passenger flow variety. In view of the operating characteristics difference of multiple vehicle types， considering the time-of-use electricity price and coordinated charging strategy， an electric bus scheduling model is established to jointly optimize the vehicle driving plan， charging scheme and fleet configuration. A genetic algorithm model is designed to solve the problem， and a network in Nanjing are taken as examples to verify the effect of the model. compared with the small and large type scheduling scheme， the total system cost of the scheduling scheme with multiple vehicle types is reduced by 8.26% and 11.25%， respectively. The results show that the proposed method can not only improve the public transport service level， but also effectively reduce the operation cost of the public transport system.

Truck is easy to rollover due to improper operation because of its high center of gravity （c.g）. This paper proposes a vehicle rollover prediction method which could be put into use quickly based on existing roadside measurement equipment. First， a regression prediction model of vehicle c.g height based on improved KNN was proposed to realize the roadside dynamic estimation of vehicle c.g. Then， a roll estimation model based on lateral load transfer rate （LTR） was established to realize the roadside estimation of roll degree. Finally， an online ARIMA model was proposed to predict vehicle roll state online. The results of MATLAB/Simulink and TruckSim co-simulation showed that the proposed rollover prediction method could accurately predict the LTR trend and the LTR limit value in the whole process， thus providing a basis for roadside equipment to transmit specific warnings in advance on the curve.

Considering the vehicle type， driving style， and the most important objects （MIO） affecting the vehicle lane change at different stages， cooperative lane-change obstacle avoidance model （CLAM） was constructed by describing the "vehicle-vehicle interaction" mechanism in the vehicle obstacle avoidance process as a force relationship； the vehicle lane change avoidance execution events under emergencies were extracted according to the lane change execution segment extraction criterion to establish a vehicle obstacle avoidance micro-trajectory dataset to unexpected events. The cooperative vehicle lane change obstacle avoidance was transformed into a multi-constraint optimal control problem. The cooperative lane-change obstacle avoidance model-optimistic algorithm strategy （CLAM-OA strategy） was designed with the optimization algorithm as a bridge. The results show that compared with the data-driven LSTM model， the outputs of the CLAM-OA strategy have significantly lower errors and more stable results in different time domains of vehicle speed and displacement.

Taking the typical mountainous two-lane highway bend and straight road as the research object， based on traffic trajectory data extracted by UAV aerial video， the risk prediction model of passenger car following under the movement interruption of truck was constructed by the light gradient boosting machine algorithm （LGBM）. The support vector machine （SVM） and random forest machine （RF） were used to verify the validity of the model， and the risk mechanism of the key characteristic parameters of the model was analyzed. The experimental results show that the accuracy of the risk prediction model based on the LGBM algorithm is 96.9%， which is superior. The speed difference and the following distance are the key characteristic parameters of the model， and the single factor importance on the straight road is greater. Compared with the curve， the dangerous driving behavior of straight road section is prominent， and the unstable following characteristics such as large lateral offset are obvious； the results of the model interpreter show that when the speed difference is less than 0.5 m/s and the car-following distance is greater than 40 m， it is a safe car-following state.

In order to solve the efficiency and safety problems of vehicle passage during ramp confluence in an intelligent network environment， a cooperative control method for intelligent vehicles based on game theory was proposed. In this method， a three-way game with vehicles on the outside of the ramp as the main body was constructed， and the vehicle revenue function was established from three aspects of safety， lane change and vehicle efficiency. Based on Shapley's value， the optimal control strategy of the vehicle in a three-way game was obtained. In the experiment， the merging area of Chongqing Yuma Road and Baotou direction of Chongqing Inner Ring Expressway was selected as the prototype of the road network， and the vehicle running scene in the merging area was co-simulated by SUMO/TraCI and Python. The results show that compared with the evolutionary game control method， the overall average travel time ratio is reduced by 3.22%， and the average delay time is reduced by 14.47%，the time-to-collision （TTC） less than 3 s decreased by 26.56%. It effectively reduces the traffic conflict time and improves the traffic operation efficiency in the confluence area.

Aiming at the problem of balance and efficiency of yard crane operation， a time phased balance strategy （TPBS） is designed. Considering the constraints of safe distance between yard cranes and yard capacity， a mixed integer programming model is established to minimize the variance of workload and total operation time of double yard cranes. In order to solve the established model and analyze the calculation example， Simulated annealing genetic algorithm （SAGA） is developed. Through the comparative experiments of different algorithms and the comparative experiments and analysis of three balancing strategies when the number of containers is 80~500， the experimental results show that the target values of SAGA are reduced by 7.32% and 20.66% respectively compared with GA and SA. The time division balancing strategy can promote the balance of yard crane operation， shorten the completion time of container loading and unloading tasks， effectively improve the utilization of yard space， and help improve the overall operation efficiency of yard crane.

To improve the longitudinal seismic performance of long-span soft-lighten arch bridges in high-intensity near-fault regions， the applicability of viscous dampers was discussed for an upper-deck RC arch bridge through the nonlinear time history analysis. The structural system and dynamic characteristics of the novel arch bridge were compared with those of the conventional RC arch bridge. The seismic damage path was explored and the layout schemes of the viscous damper were optimized. The structural response and the mitigation due to dampers were investigated under the near-field and far-field ground motions. The influence of high-order mode was analyzed. The rationality of the mitigation design was investigated based on fragility analysis. The results show that the medium height column is vulnerable to near-field longitudinal and vertical design earthquake， while the arch rib remains elastic. The overall mitigation effect is the best when the dampers are located on the abutments and high columns. The S-shaped distribution of shear force and bending moment envelope of high column with damper is more obvious than the others due to the influence of high-order mode. The structural responses are larger with significant energy dissipation and efficient mitigation under near-field impulse and far-field long-period earthquakes than the other conditions， and the hysteresis loop of the damper would suddenly change under the influence of displacement pulse. The damage probability of mitigation design with excellent application is effectively decreased under four types of near-field and far-field earthquakes. However， the damper should meet the seismic requirements of large force and stroke.

A double-sided shear test was carried out on the sintered clay brick block bonded with carbon fiber reinforced composite material （CFRP） to investigate the failure morphology and bonding properties of the CFRP-clay brick block interface under different sulfate dry-wet cycles， and the sulfate corrosion was discussed. Under the degradation mechanism of CFRP-clay brick interface bond performance， a bond strength degradation model is proposed. The results show that the failure morphology of the CFRP-clay brick block interface is greatly affected by the sulfate dry-wet cycle； the interface bonding properties （ultimate load-bearing capacity， peak shear stress） first slightly increase with the cycle time and then decrease rapidly. On this basis， according to the existing interface theory， a CFRP-clay brick block interface bond-slip model affected by the cycle time is proposed. The predicted value of the model is compared and analyzed with the experimental data. The model can accurately express the impact the law of degradation of interfacial bonding properties caused by sulphate dry-wet cycles.

It is not modeling analysis characteristics lead to poor analysis effect in the analysis of interstory deformation characteristics of building structure. Therefore， the distribution characteristics of interstory deformation of building structure after earthquake are studied. It built structure 3D model is established by building information model technology. The generalized interstory displacement spectrum method is introduced to construct the interstory displacement formula and it realize the analysis of interstory deformation characteristic distribution of building structure. The results show that the interlayer displacement increases with the acceleration under different accelerations. The bending moment， shear force and interstory displacement angle of a one-story building reach 237 kN·m， 210 kN and 1/231 rad， respectively， showing the biggest changes. The analysis results are in agreement with the practice and it can be used to analyze the interlayer deformation characteristics effectively.

To improve the prediction accuracy of joint points of the heatmap， this paper proposes a postprocessing method of human pose heatmap based on sub-pixel localization. The method includes two strategies： the first is the sub-pixel shift processing of the flipped image heatmap， which can eliminate the unaligned deviation from the original image heatmap； the second is the heatmap decoding for local region surface fitting to achieve sub-pixel localization of the joint points. The heatmap postprocessing method in this paper is independent of the network model and can be applied to the current heatmap-based human pose estimation models without any modification. To verify the effectiveness of the proposed method， experiments have been carried out by using two publicly available datasets named COCO2017 and MPII. The average precision can be improved by 0.9 and 1.1 on COCO2017， respectively， by adopting two deep learning models， i.e.， HRNet-W32-256×192 model and Simple Baseline-W32-256×192 model.

Aiming at the problem of large amount of calculation caused by exhaustive sample points in global K-means clustering algorithm， this paper proposes an efficient global K-means clustering algorithm based on weighted space partition. Firstly， the sample space is divided into grids， and then the density criterion and distance criterion are proposed to filter the grids， and the grids with large density and far distance from each other are retained as candidate center grids. In order to avoid the limitation that the global K-means algorithm only selects candidate centers in the sample set， the weight criterion and the center iteration strategy are proposed to expand the candidate centers and increase the diversity of the candidate centers. Finally， the candidate centers were traversed by incremental clustering to obtain the final clustering result. The experimental results on UCI data sets show that compared with the global K-means algorithm， the computational efficiency of the new algorithm is improved by 89.39%~95.79% on average under the premise of ensuring the clustering accuracy. Compared with K-means ++， IK-+ and the recently proposed CD algorithm， the new algorithm has higher accuracy and overcomes the problem of unstable clustering results caused by random initialization.

In order to cluster massive cloud data accurately， a fuzzy clustering algorithm for massive cloud data based on peak density is proposed. The cloud data with noise is separated by BP neural network， and the output noise is reconstructed by singular value decomposition to obtain the noise output by the joint algorithm. The cloud data with noise is subtracted from the output noise to obtain the cloud data after noise removal. The density peak is combined with the optimized fuzzy clustering algorithm to adaptively form the initial clustering center， determine the number of clusters， and finally realize the fuzzy clustering of massive cloud data. Experimental results show that the clustering effect and efficiency of the proposed algorithm are significantly better than other algorithms.

In order to improve the detection capability of detectors for multiscale ships in SAR images and ensure the real-time performance of the detection networks， an improved anchor-free model based on attention mechanism for ship detection is proposed. On the basic framework of the off-the-shelf YOLOX， a lightweight dilated convolutional attention module （DCAM） is embedded in front of feature pyramid network （FPN） to adjust the relationship between receptive field and multiscale fusion， and strengthen the representation ability of features. The detection head is redesigned by introducing the center-ness prediction branch， which can weight the classification scores of the anchor points， in the meantime， the loss function of the proposed model is also revised to optimize the final detection performance. Through the comparative experiments on dataset SSDD， the proposed model in this paper is superior to the mainstream deep learning detection models， with an accuracy of 94.73%， and achieves the best trade-off between detection accuracy and detection speed.

In the face of manifold data， the local density of density peaks clustering （DPC） algorithm is easy to find the wrong cluster center and the allocation strategy is easy to cause the residual samples far from the cluster center to be misallocation. In view of the above problems， this paper proposes density peaks clustering with second-order K-nearest neighbors and multi-cluster merging. Firstly， the minimum second-order K-nearest neighbor is used to define the local density， highlighting the density difference between the cluster center and the non-cluster center， so as to find the correct cluster center； Secondly， the K-nearest neighbor is used to find the local representative points of the sample and determine the core points， and the core points are used to guide the micro-cluster division； Finally， the inter-cluster attraction defined by the minimum second-order K-nearest neighbor and shared nearest neighbor is used to merge the micro-clusters， which avoids the misallocation of samples away from the cluster center， and the micro-cluster merging process does not require iteration. In this paper， DPC-SKMM algorithm is compared with IDPC-FA， DPCSA， FNDPC， FKNN-DPC， DPC algorithm. Experimental results show that DPC-SKMM algorithm can cluster manifolds and UCI data sets effectively.

Unmanned Aerial Vehicle（UAV） target tracking tasks are often suffer from small target size， large scale variance and frequent viewpoint change. To address these issues， in this paper， an UAV target tracking algorithm based on high-resolution siamese network is proposed. Firstly， a high-resolution network is improved as the feature extraction backbone network （Lite-high resolution network， L-HRNet）， and a dynamic multi-template strategy is used to mine the inter-frame information of the video. Secondly， a multi-frame feature fusion module is constructed to obtain fusion features that are beneficial to target localization. Finally， an anchor-free strategy is selected to locate the target position and obtain accurate tracking results. The experimental results show that the success rate and accuracy of the proposed algorithm are 66.0% and 84.7% on DTB70 dataset， 65.7% and 84.3% on UAV123 dataset respectively， which improves the target tracking performance effectively.

A monocular-vision-based method to reproduce the forward-looking visual field of the vehicle closed cockpit was proposed. Accurate off-line calibration and on-line monocular image acquisition， forward-looking image processing， and perspective image display are performed in a multi-threaded parallel framework. So forward-looking images can be reproduced with low distortion and high fluency on the screen. This paper also built a hardware system of vehicle screen-displayed closed cockpit. Through laboratory and field tests， the effectiveness and efficiency of our method are verified from the aspects of system delay， forward-looking angle of view， image distortion， high-speed straight driving， turning， obstacle avoidance and driving experience. This method provides a good theoretical and practical basis for the subsequent improvement of the vehicle closed cockpit.

When the multi-motors driven robotic arm system operates in a complex environment， its joint position is easily disturbed by external interference such as load， resulting in large tracking error and synchronization error， which causes system performance degradation. To solve this problem， a new multi-motors ring coupling control （NRCC） strategy is proposed. The synchronous proportional coefficient is set in NRCC to ensure the coordinated operation of multiple motors. The active disturbance rejection compensation controller （ADRCC） and the adjacent mean error processor are designed. The ADRCC compensates the position control signal of the motor twice through the adjacent mean error signal， which reduces the synchronization error between multi-motors. Meanwhile， a novel adaptive neuro-fuzzy inference system （ANFIS） optimizes exponential reaching rate sliding mode tracking controller （ANFIS-SMC） and disturbance observer are proposed to ensure the position tracking performance of motors. The simulation results show that the proposed control strategy effectively reduces the synchronization error between multiple motors and ensures the high-precision tracking performance of the motors.

The total fuel consumption will only decrease when the increased fuel consumption during formation is less than the reduced fuel consumption during platoon operation. In order to form a commercial vehicle platoon with fuel saving potential， this paper proposes a density clustering based commercial vehicle formation strategy. First， the commercial vehicle model in Trucksim was used to collect fuel consumption data， and an equivalent fuel consumption model was established through fitting and simplified in the distance domain； Second， utilizing the optimization problem of two vehicle formation to obtain formation standards with fuel saving potential； Then， in density clustering， the dispersed commercial vehicles are clustered into sub platoon clusters using this standard， and the optimal equivalent fuel consumption function is defined to determine the final platoon form. Finally， the effectiveness and superiority of the formation strategy proposed in this paper were verified through co-simulation.

The binocular vision sensor suffers from the weakness of long-distance measurement and unstable perception in various outdoor environment， the application of binocular vision is widely but limitedly. Inspired by hawk eyes， by simulating the physiological structure of dual fovea and the search experience which trained by predation， the modified reference-based super resolution is employed for enhancing the binocular vision perception ability and the accuracy of target candidates. The super resolution part aims to improve the global perception ability in vision field， the reference-based super resolution part aims to enhance the specified target candidates by manual references addition. For the orchard 3D point cloud navigation map， the maximum decrease of error ratio and standard deviation is 12.2% and 2.305 under various lighting conditions. For the 3D point cloud reconstruction of fruit tree operation， the quality of point cloud is greatly improved， which can provide accurate 3D spatial information， and restore the 3D spatial information of each branch of fruit tree more completely， which meet the needs of orchard operation.

In response to the problem of heavy adhesion and high resistance of the rice direct seeder skateboard in the disturbed and saturated paddy field， the non-smooth surface of the loach body was observed microscopically； Using Fluent software for simulation and analysis， the principle of reducing viscosity and resistance on the non-smooth surface has been revealed； The Box Behnken response surface method was used to design the experiment and obtain the regression equations of the total resistance F_B of the non-smooth surface of the groove with the groove width w， groove depth h， and inflow velocity v. The optimization analysis results showed that the maximum drag reduction rate reached 10.052% when w was 4.5 mm， h was 4 mm， and v was 1.25 m/s； Based on the optimal parameters， a biomimetic sliding plate for rice direct seeding machine was designed， and indoor paddy soil tank experiments were conducted， with a drag reduction rate of 10.23% ultimately. This study can provide new ideas for the development of viscosity reduction and resistance reduction technology for paddy field soil contacting components.

Existing reliability allocation methods are less applied in the field of agricultural machinery， and in order to better meet the reliability allocation needs of agricultural machinery， this paper proposes a reliability allocation method for agricultural machinery based on improved integrated factors method（IFM）. The method integrates six influencing factors including subsystem severity A₁， improvement potential A₂， complexity A₃， maintainability A₄， environmental conditions A₅ and technology level A₆ for the characteristics of agricultural machinery， and forms the weight factor of subsystem failure rate to realize the allocation of the specified overall system reliability target to the subsystem level. For newly developed products， the evaluation reference standard for the value of each influence factor is given， which can be quantitatively calculated by FMECA combined with expert scores. Taking 2BMQF-6/12 no-till planter as an example， the improved IFM is applied to allocate its reliability， and the feasibility of the proposed method is verified by comparing it with the IFM， FOO， Kim's and Yadav's methods. The allocation results of the proposed method are reasonable， the process is complete， and the factors considered are relatively comprehensive， which can provide a reference for the reliability allocation of agricultural machinery and a basis for the reliability design of agricultural machinery products.