For the problem that it is difficult to adapt to various complex working conditions to obtain the best shift performance of two-gear pure electric vehicles under the traditional control strategy， this paper proposes an intelligent shift control method. According to the motor efficiency and vehicle driving acceleration in different gears， combined with the battery state， the traditional two-parameter optimal economic shifting law and the optimal power shifting law are obtained respectively， and the optimal shifting interval is analyzed to obtain the optimal shifting interval of the vehicle. According to the accelerator pedal opening degree and accelerator pedal change rate， the driver's driving intention is recognized through fuzzy control， and combined with the driver's intention and the vehicle speed within a reasonable range， a two-stage fuzzy control intelligent shift controller is designed， and the validity of the designed adaptive shift strategy is verified through simulation. The simulation results show that in the face of different working conditions， compared with the traditional three-parameter fuzzy gearshift control， the adaptive gearshift strategy designed in this paper can accurately combine the vehicle performance and driver's intention， reduce the frequency of gearshift， and improve the economy.

To explore the underlying causes of uneven path utilization in urban mixed road networks， revealed preference （RP） and stated preference （SP） joint surveys were adopted to systematically collect data on drivers' attributes and path choices in urban mixed road networks. Path choice models based on random regret theory were developed， incorporating multidimensional path attributes and individual travel characteristics， with scale effects introduced to enhance predictive accuracy. The results show that path attributes such as distance， number of traffic lights， tolls， historical average travel time， additional time during congestion， and the likelihood of congestion significantly influence path selection. Personal travel frequency and historical path preferences also play critical roles in decision-making. The improved model considering path attributes achieves a hit rate of 0.703， and the improved model integrated drivers' attributes has a hit rate of 0.872， demonstrating superior fitting performance and predictive capability compared to traditional models. The study results provides a quantitative basis for refined urban traffic management strategies and practical insights for optimizing personalized navigation services.

In order to investigate the factors influencing the severity of injury in self-driving car accidents， a structural equation model considering mediating effects was constructd to analyze the correlation between the takeover mode of self-driving cars and the severity of accidents. Using the self-driving car accident data in AVOID dataset， 17 influencing factors were selected from four aspects： vehicle characteristics， weather characteristics， road characteristics and collision conditions， and the mediating effect and factor structure analysis were used to explore the influence of each factor on the severity of self-driving accidents. The results show that road type can significantly reduce the severity of self-driving car accidents. Road condition， weather-cloudy， weather-rain， collision-left rear and lighting-daylight indirectly affect the severity of self-driving car accidents through the mediating effect. Among them， road condition can significantly reduce the severity of traffic accidents. Weather-rain significantly increases the severity of traffic accidents of self-driving cars. The takeover mode of the self-driving car when it is involved in a traffic accident has a significant negative effect on accident severity and the two have a complex correlation.

To address the issues of low efficiency and limited solution space caused by manual hyperparameter tuning when using traditional neural network models for vehicle lane-changing intention recognition， a lane-changing intention recognition model based on PSO-GRU was proposed. First， a multi-step trajectory smoothing method was developed considering the characteristics of horizontal and vertical trajectories in the NGSIM dataset. Next， the trajectory data were mined for parameters and reconstructed into the dataset to establish a vehicle lane-changing trajectory database. The importance of features in the database was evaluated using the random forest algorithm. Finally， a combined model for lane-changing intention recognition based on PSO-GRU was built， with the PSO algorithm used to optimize the hyperparameters of the GRU network to find the optimal configuration. The results demonstrate that， compared to traditional deep learning and machine learning algorithms， the proposed PSO-GRU model exhibits higher recognition accuracy and more stable performance.

To explore the differences in the impact of built environment on residents' car dependence in different periods，a light gradient boosting machine（LightGBM）model was constructed based on the residents' travel survey data of Hangzhou in 2010 and 2023. Nonlinear analysis was carried out using partial dependence plots and shapley additive explanations.The results show that the LightGBM model accurately captured the nonlinear relationship between built environment and residents' car dependence.The built environment has a significant impact on residents' car dependence，and there are significant differences in different periods.Compared with 2010，the impact of built environment on residents' car dependence in 2023 has significantly weakened.The influence of population and road density shows a threshold effect.When the population density of the residential area exceeds 20 000 people/km2 or the road density exceeds 0.04 km/buffer zone，the proportion of car travel among young people increases instead.This indicates that it is necessary to accurately grasp the impact of threshold effect to avoid the waste of construction resources and the negative effects brought by over-construction.

Based on the global positioning system （GPS） trajectory data of taxis in Beijing and the map matching function of ArcGIS， appropriate threshold values were selected， a taxi travel information database was constructed， and the travel characteristics of taxis were analyzed. Firstly， a taxi travel network was built， and spatial and temporal distribution features of travel trajectories were analyzed using kernel density and hot spot analysis methods， revealing the changing trend of taxi travel density during peak hours. Then， considering the impact of land use on taxi travel distribution， a dual-constraint population weight opportunity model based on land use was established to predict taxi travel distribution. The results show that the model improves the prediction accuracy of taxi inter-district travel， with the root mean square error reduced from 4.54 trips to 4.41 trips， and the similarity between the predicted results and actual data improved from 0.961 to 0.970. The average relative error in the number of trips decreased from 0.48% to 0.19%.

In order to analyze the impact of transit transfer preferential （TTP） policy on residents' transfer behavior， a questionnaire of combining Revealed Preference （RP） and Stated Preference （SP） was designed to collect data on respondents' socio-economic attributes， travel characteristics， psychological perceptions， and preferences in different transfer scenarios. Using an integrated choice and latent variable （ICLV） model， a transfer behavior model was established to explore the decision-making mechanism of residents under the TTP policy. The analysis results show that the fitting effect of the ICLV model is better that that of the SEM-Logit model. In the ICLV model， individuals' socio-economic attributes and travel characteristics can directly influence transfer behavior and indirectly influence it through psychological latent variables. The perceived ease of use and travel habits among the psychological latent variables have positive and negative impacts on residents' transfer behavior， respectively. Attitude acts as a moderating variable， where a better attitude towards the TTP policy enhances the positive impact of preferential magnitude on transfer behavior. When perceived transfer convenience and public transportation travel habits under the TTP policy increase 1%， the probability of choosing to transfer increase by 4.67% and decrease by 4.62%， respectively. By introducing variables such as individuals' socio-economic attributes and travel characteristics to estimate latent variables， the method overcomes the measurement error problem inherent in using questionnaire items to estimate latent variables thereby improving the explanatory power of the model estimation. The conclusions can provide theoretical support and reference value for government departments in planning， formulating， and implementing the TTP policy.

In order to enhance the friendliness of driver-automation cooperative lateral control system， a novel shared lateral control algorithm considering driver group characteristics was proposed. Firstly， a driver imitated steering model was introduced， and then the linear-varying closed-loop driver-vehicle-road model was established. Secondly， for the purpose of handling disturbance factors such as time-varying model parameters， large road curvature， and the insufficient adaptation of linearization of nonlinear vehicle dynamics， a predictive controller based on linear time-varying model predictive control （LTV-MPC） framework was designed. Thirdly， a driver-automation cooperative factor model was constructed according to the aggregate lateral error as well as lateral offset sensitivity of different driver groups， aiming to achieve differentiation adjustment of the intervention or assistance degree from the intelligent system during the co-driving. Finally， the comparative experimental results demonstrate that the shared lateral control algorithm exhibits excellent group adaptability and enhances driver-automation friendliness while ensuring path-tracking accuracy.

Based on the vehicle specific power（VSP） model， with particular emphasis on the characteristics of plug-in hybrid electric vehicles（PHEVs）， this study proposes NO _x emission factor calculation methods for both charge-depleting and charge-sustaining modes. With the objective of minimizing average per-vehicle NO _x emissions within the intersection area， and subject to constraints including average per-vehicle delay and average number of stops， a signal timing optimization model is established for intersections under mixed traffic environments comprising conventional fuel vehicles and hybrid electric vehicles. Analysis results indicate that the proposed model achieves a 31.29% reduction in average per-vehicle NO _x emissions. Meanwhile， average per-vehicle delay and average number of stops exhibit consistent variation trends with average per-vehicle NO _x emissions， with reduction rates of 43.99% and 28.88%， respectively， outperforming the Webster model which considers only traffic flow conditions. Finally， parameter sensitivity analysis reveals that traffic volume levels， left-turn vehicle proportions， and hybrid vehicle penetration rates exert significant influences on average per-vehicle NO _x emissions. The intersection signal timing optimization method proposed in this study， which balances environmental benefits and traffic efficiency， provides an effective basis for urban traffic management and policy formulation.

To address the issues of insufficient resolution， signal overlap， and reduced signal-to-noise ratio that existing Ground Penetrating Radar（GPR） technology still faces in multi-layer dense rebar scenarios， a multi-layer rebar detection method based on spread spectrum and polarization gain compensation is proposed. The proposed approach optimizes the spread spectrum technique to effectively mitigate the issues of frequency discontinuities and information loss during data fusion. Additionally， it leverages the electromagnetic reflective properties of metal and the amplification mechanism of polarization reversal to enhance the intensity of rebar reflection signals， thereby significantly improving detection performance. The numerical simulation results obtained using CST demonstrate that the polarization deflection characteristics of metal reflection signals can significantly enhance the SNR of rebar reflections in echo signals. The polarization matching transmission efficiency can be improved by up to approximately 20 dB. Additionally， simulations and post-processing analyses were performed on GPR data before and after spectrum expansion using gprMax. The results demonstrate that the GPR frequency expansion technique proposed in this study effectively enhances distance resolution， resolves the issue of overlapping reflection signals from closely spaced double-layered rebars， and successfully achieves their independent localization， with a maximum error of only 5.25%. In addition， the robustness analysis under multiple SNR conditions is also carried out in this paper to further validate the applicability and reliability of the method in complex practical environments.

In order to accurately predict the service life of the hob in the rock-breaking process of Tunnel Boring Machine（TBM） to improve the boring efficiency and determine the time of changing the hob， this study analyzed the rock-breaking mechanism of the disk hob and established the calculation model of the normal force of rock-breaking hob of TBM based on the shear and tensile failure mechanism. The Rabinowicz abrasive wear model was improved based on the dense core effect to establish the wear rate and life prediction model of TBM disc cutters considering dense core effect and compression-shear failure mode. The engineering applicability of the prediction model proposed in this study was verified by the comparing the prediction results with the existing theoretical model calculation results， numerical simulation results and on-site monitoring results. The results show that the predicted values of this model are closer to the measured values in the engineering field. Compared with the existing theoretical model， the average prediction accuracies of cutter wear rate and cutter life are improved by 9.1% and 11.7% respectively， which effectively demonstrates the necessity of comprehensively considering the dense core effect and compression-shear failure mode in the calculation of TBM cutters breaking rock. The cutter wear rate and life index can be used as reliable indexes for evaluating the cutter wear characteristics during the rock breaking process of TBM， which can provide theoretical guidance for the design and timing of cutters change program in TBM tunnel and other similar projects.

To tackle the issues that vehicle driving risk influencing factors are complex and unknown in the cooperative vehicle-infrastructure environment， and that traditional data-driven and physics-based model-driven prediction models lack sufficient interpretability and adaptability， a method is proposed in which an ego vehicle driving potential field is constructed using surrounding vehicles' motion state information acquired by onboard sensors， and a generalized force model is established by integrating ego vehicle driving state information to construct a vehicle-to-vehicle interaction information network. A generalized force model integrates the vehicle's driving state to create a vehicle-to-vehicle interaction network， while a vehicle-to-road network is formed using road sign and marking data. These networks are then fused into a dynamic interaction information network. To handle the dynamic nature of these forces， a routing mechanism is introduced. A switching Transformer model， combining the Transformer framework with a hybrid expert model， forms the foundation of the proposed vehicle driving risk prediction model， which is enhanced by the interaction information fusion network. Results show that the switching Transformer model outperforms the standard Transformer， with a 3.48% reduction in Mean Absolute Error（MAE）， a 2.19% reduction in Root Mean Square Error（RMSE）， and a 1.39% increase in the coefficient of determination （R2） in the comprehensive test set. In the lane-changing test set， the model shows improved adaptability， reducing MAE by 11.1%， RMSE by 9.0%， and increasing R2 by 5.4%. Additionally， in risk prediction scenarios， the model demonstrates significant improvements in Inverse Time to Collision （ITTC）， with increases of 0.288 7， 0.200 5， 0.714 2， and 0.288 8 for car-following， parking， exiting， and mixed lane-changing and exiting scenarios.

Reliability prediction of GPGPU programs under soft error perturbations presents significant challenges， particularly when fault injection samples are limited. This paper proposes a novel method for constructing a high-accuracy reliability prediction model using a small number of fault injection samples. The approach introduces guiding rules to optimize fault injection experiments by reducing redundancy and enhancing the quality of collected samples. Furthermore， data augmentation techniques are employed to mitigate the challenges of insufficient training data， enabling effective model learning. Experimental validation on various benchmark programs demonstrates that the proposed method achieves an average prediction accuracy of 91.56% with only 300 fault injection samples. These results highlight the method's capability to reduce experimental overhead while maintaining high prediction performance.

To address weak generalization and low accuracy in existing detection methods， a WSNs traffic anomaly detection model based on an improved convolutional neural network （CNN） and density peak clustering （DPC） is proposed. This model consists of a data processing layer， a feature extraction layer， and an anomaly detection output layer. In the data processing Layer， the gramian angular fields （GAF） is used to transform one-dimensional WSNs traffic time series into two-dimensional images. In the feature extraction layer， CNN is used to extract high-dimensional spatial features of WSNs data. An improved slime mold optimization algorithm is designed and the least absolute shrinkage and selection operator （Lasso） is introduced to enhance the generalization ability of CNN network. In the traffic anomaly detection output layer， DPC is used to cluster the extracted high-dimensional spatial features， and anomaly detection of WSNs data is ultimately achieved by defining anomaly detection rules. The experimental results show that compared to other WSNs traffic anomaly detection schemes， the detection accuracy of the proposed detection model has increased by 15.33%.

In order to solve the problems of unequal distribution and convergence of the equivalent solutions obtained by the multi-modal multi-objective optimization algorithm， a multi-modal multi-objective optimization algorithm based on density clustering （MMO_DBSCAN_BWO） is proposed. Firstly， the chaotic mapping initializes the population， makes the individuals evenly distributed in the decision space， and enhances its global search ability. Secondly， density clustering algorithm is used to generate niches based on non-dominated solution sets of different levels， identify stray individuals， and propose hierarchical evolutionary strategies and hierarchical ring topological evolutionary strategies to enhance search ability. Then， the archival mechanism is introduced to preserve the optimal solution of individual history. Finally， the optimal Pareto solution with uniform distribution is selected based on special congestion distance non-dominated sorting and Euclidean distance truncation mechanism. The overall performance of MMO_DBSCAN_BWO is better than that of the comparison algorithm by testing on the 2020 CEC multimodal multiobjective test function and comparing with the three algorithms. At the same time， the algorithm is applied to the hyperparameter optimization problem of JavaScript malicious code detection graph model， and the accuracy of the model is further improved， reaching 99.3%， which proves the effectiveness of the algorithm.

Density Peak Clustering algorithm faces challenges in detecting density peaks in manifold data， and its assignment strategy often misallocates samples far from density peaks. To address these issues， this paper proposes a Natural Neighbor Graph Optimization and Micro-Cluster Merging Density Peak Clustering algorithm for manifold data. First， based on the natural neighbor graph， a novel local density measurement method is designed using geodesic distances between vertices， which accurately characterizes the internal structure and distribution properties of manifold data. Second， by analyzing the connection relationships between vertices in the natural neighbor graph， the algorithm automatically identifies representative samples and determines local cores to guide micro-cluster partitioning. Finally， a new similarity measurement criterion for micro-clusters is defined based on geodesic distances between samples， which optimizes clustering performance throughmicro-cluster merging. The proposed algorithm is compared with four improved density peaks clustering algorithms and the original density peaks clustering algorithm. Experimental results show that the proposed algorithm can be effectively applied to clustering analysis of manifold data and real-world datasets.

To address the challenges of internal parameter perturbations and external disturbances in improving trajectory tracking accuracy and suppressing link vibrations for the two-degree-of-freedom serial flexible manipulator in practical implementations， a singular perturbation theory-based composite control strategy is proposed. Firstly， the dynamic model of the two-degree-of-freedom serial flexible manipulator was established using the assumed modes method and Hamilton's principle. Subsequently， based on singular perturbation theory， the system was decomposed into slow and fast subsystems. Then， for the slow subsystem， an incremental nonlinear dynamic inversion control strategy was designed using feedback linearization techniques， aimed at trajectory tracking control of the manipulator. For the fast subsystem， an error sign robust integral control strategy was developed through the integration of error signs and nonlinear feedback， intended for suppressing link chattering of the manipulator. The stability of the closed-loop system was proven using Lyapunov stability theory. Simulation results demonstrate that the proposed composite control strategy exhibits significant advantages in trajectory tracking control and disturbance rejection performance.

To enhance the covertness of the Orthogonal Chirp Division Multiplexing with Index Modulation （OCDM-IM）， a variable time-width and index mapping scheme is proposed to construct the corresponding VT-OCDM-IM system. By selecting time-width and index mapping methods using the same pseudorandom sequences， the system disrupts signal periodicity and enhances covert performance in scenarios where detecting signal periodicity features is critical. Simulation and experimental results show that altering the number of subcarriers alongside variable time-width improves reliability under certain conditions while enhancing covert performance without compromising the system's transmission rate advantage.

To maintain a safe distance between adjacent vehicles in a vehicle platoon， a prescribed performance output feedback model-free decentralized control algorithm based solely on local information is proposed. This algorithm ensures that distance control errors are constrained within a preset range， speed disturbances are attenuated， and string stability is guaranteed. Additionally， a prescribed performance model-free speed control algorithm was designed to address the autonomous speed control problem of the leading vehicle during the separation or merging of vehicle platoon. This algorithm can achieve the integrated control of autonomous cruise of sub-fleets after vehicle platoon separation and stable merging of multiple vehicle platoons. The simulation results have verified the effectiveness， flexibility， and ability to assure string stability of the proposed model-free decentralized control algorithm for nonlinear vehicle platoon systems.

A dual belt residual film bundling device has been designed to address the problems of complex structure， low service life of bundling belts， adjustable tightness of film bundling， uncertain unloading position， and poor compatibility with existing residual film recycling machines in the existing residual film bundling process. This article introduces the composition and working process of the residual film bundling structure， with a focus on optimizing the design of the bundling device， tensioning device， and film carrying box. By conducting a dynamic analysis of the bundling process， the structural and operational parameters of each functional component were determined. To verify the operational performance and efficiency of the dual belt residual film bundling device， three factor three-level quadratic regression experiments were conducted with bundling chamber angle， bundling belt friction coefficient， and bundling belt linear speed as experimental factors， and residual film bundling rate， residual film bundling density， and bundling efficiency as response values. A regression model was established to analyze the effects of each factor on residual film bundling rate， residual film bundling density， and bundling efficiency， and parameter optimization and experimental verification were carried out.The experimental results show that the factors affecting the bundling rate of residual film are in descending order： bundling chamber angle>bundling belt linear velocity>bundling belt friction coefficient； The factors that affect the density of residual film bundling are in the following order： bundling belt line speed>bundling chamber angle>bundling belt frictioncoefficient； The factors that affect packaging efficiency are in the following order： the friction coefficient of the strapping belt>the angle of the strapping chamber>the linear velocity of the strapping belt. Theoretical optimal design parameters： bundling chamber angle is 33°， packing belt friction coefficient is 0.83， and packing belt linear velocity is 1.03 m/s. The research results provide reference for the design and high-quality operation of residual film bundling devices.