To improve the stability and accuracy of the sound quality evaluation model, the vehicle interior sound quality evaluation model(ALSTM) based on long short-term memory (LSTM) network and attention mechanism was proposed. The steady noise samples of different brands of vehicles at the right ear of drivers under different working conditions were collected, and the subjective evaluation test of noise samples was carried out with annoyance as evaluation index to establish the evaluation data set of interior sound quality. On the basis of the data set, the sound quality evaluation model based on LSTM network was constructed with Mel-scale frequency cepstral coefficient (MFCC) of noise samples as feature input, and the attention mechanism was introduced to optimize the model. The experimental results show that the proposed evaluation model of sound quality can effectively evaluate the vehicle interior noise, and the accuracy in the test set is as high as 97.07%. Compared with other methods, the stability, convergence speed and classification accuracy of the ALSTM model are improved.

To solve the problem of vehicle dynamics state variations in autonomous vehicles due to the uncertainty of road conditions on seacrossing bridges, the vehicle roll control strategy was proposed based on the deep deterministic policy gradient (DDPG) algorithm, and the generalization capability under different speeds was discussed. The vertical model of the seacrossing bridge was constructed to provide dynamic road environment. The vehicle dynamics and vehicle tracking error models were established for incorporating the dynamic characteristics of vehicle roll, sideslip and yaw and for establishing the criteria for roll stability. The state space and action space for the DDPG algorithm were designed, and the reward function was formulated based on the vehicle roll state. The numerical simulation results show that by the DDPG algorithm, good performance is achieved in each episode with robust learning and problemsolving capabilities in complex and uncertain environments. The vehicle roll angle and lateral distance error are ensured within acceptable and minor fluctuation ranges to achieve safe vehicle control.

 To solve the severe miss detection problem in pedestrian detection caused by insufficient pixel information of distant targets and occlusion induced loss of human pattern information, the pedestrian detection method was proposed based on dual key points combinations. The discriminative semantic features of pedestrians were effectively extracted and fused by utilizing key points of the head and center regions for significantly reducing the pedestrian miss detection rate. The deformable convolution was introduced into the deep aggregation backbone feature network to enlarge the receptive field and enhance the semantic information of human pattern. The dual branch joint detection module based on key points combinations was designed, and the positive samples for different branches were redefined to strengthen the semantic information of small scale and occluded targets. The results of the dual branch detection were fused using the non maximum suppression (NMS) algorithm. The results show that on the CityPerson validation dataset, the average miss detection rates of the normal, small scale and heavily occluded subsets reach 8.24%, 11.81% and 30.59%, respectively. Especially, for the heavily occluded subset, the miss detection rate is reduced by 15.71% compared to the traditional method ACSP. By the proposed method, the detection speed reaches 16 frames per second. On the CrowdHuman dataset, the average precision and average miss detection rate are 86.30% and 45.52%, respectively. Compared with other state of the art methods, the proposed method exhibits superior performance in average precision, miss detection rate and detection speed, which demonstrates significant application value in complex scenarios with dense pedestrian crowds.

To explore the characteristics of traffic volume data and improve the accuracy of parameter estimation for traffic volume data fitting, the parameter estimation method for traffic volume data fitting was proposed based on the maximum correlation entropy criterion. The traffic volume data was preprocessed, and the probability distribution of the data was specified. The maximum correlation entropy criterion was used to estimate the model parameters, and the gradient ascending method was used to output the model parameters. The estimated values of the probability distribution parameters were obtained, and the results were compared with those of the traditional methods. Based on the measured data and the four assumed distribution models of Normal distribution，Logarithmic normal distribution，Webull distribution and Rayleigh distribution, the performance evaluation was conducted.The results show that the proposed method has the best fitting performance, and the comprehensive evaluation indexes are respective 0.979 00, 0.726 08, 1.397 69 and 1.494 50, with strong accuracy and reliability.

Under nitrogen atmosphere, the thermogravimetric tests were carried out on rape straw, polypropylene, acrylonitrile butadiene styrene resin (ABS resin) ， rape straw/polypropylene  with mass ratio of 1∶1 and rape straw/ABS resin with mass ratio of 1∶1 at temperature range of 30-800 ℃ with heating rate of 20 ℃/min. The apparent activation energy was measured by Friedman method， and the kinetic analysis was also carried out. The catalytic pyrolysis of rape straw and the cocatalytic pyrolysis of rape straw/polypropylene and rape straw/ABS resin were tested, and the threephase yield was calculated and analyzed. The results show that there is interaction among the copyrolysis reactants, and blending is conducive to reducing the difficulty of pyrolysis reaction and improving the spontaneity of reaction. According to the GC/MS analysis, the highest yield of monocyclic aromatic hydrocarbons is about 22.11% when the blending ratio of straw and polypropylene is 1∶1, and the highest yield of monocyclic aromatic hydrocarbons reaches 32.87% when the blending ratio of straw and ABS resin is the same.

Numerical and experimental studies were performed on the flow and heat transfer characteristics of impingement-film composite cooling structures on center cone. The results show that the gas recirculation occurs on the surface of center cone, and the recirculation vortex is generated. The vortex scale is increased with the increasing of blowing ratio. The hot wall temperature of center cone is increased along the flow direction, and the increasing of blowing ratio can improve the overall cooling effectiveness and increase the relative pressure drop. Increasing film-hole diameter can improve the overall cooling effectiveness and reduce the relative pressure drop. With the increasing of impinging-hole diameter, the overall cooling effectiveness is decreased slightly， and the relative pressure drop is decreased to some degree. The variation of impinging distance has little influence on the flow and heat transfer law.

To investigate the photo-thermal conversion performance of TiN nanofluid in the direct absorption solar collectors（DASC）, the ethylene glycol (EG) based TiN nanofluid with polyacrylic acid (PAA) stabilizing ligand was prepared. The long-term stability of the TiN-PAA nanofluid was characterized by conducting the static storage test at ambient conditions to show that the average transmittance of the TiN-PAA nanofluid with 100 mg/L was increased by 1.908% after ambient storage for 310 days. The photo-thermal conversion performance of the TiN-PAA nanofluid in DASC was evaluated. The effects of nanoparticles concentration, mass flow rate and irradiation intensity on photo-thermal conversion performance of the collector with TiN-PAA nanofluid were discussed. The results indicate that the stagnation temperature and efficiency of the TiN-PAA nanofluid are much higher than those of the base fluid EG after 2.0 h irradiation, and the stagnation temperature of TiN-PAA nanofluid with 100 mg/L is 80.12 ℃. The efficiency of TiN-PAA nanofluid is improved with the increasing of mass flow rate, and the efficiency of 75 mg/L TiN-PAA nanofluid is increased by 32.69% as the mass flow rate is increased from 1 mL/min to 3 mL/min. However, the efficiency of TiN-PAA nanofluid is decreased as the irradiation intensity is increased. The efficiency of the 75 mg/L TiN-PAA nanofluid reaches 69.10% when the irradiation intensity is 800 W/m2.

To solve the problems that the current segment-based video abnormal event detection model with only considering the normal or abnormal of segments could lead to the abnormal fragmentation, the abnormal start/stop location spreading and the difficulty of abnormal event classification, the video abnormal event detection and classification method was proposed based on salient features and space-time graph network. The video anomaly temporal fragment integration and refinement method based on space-time fusion graph network was proposed to integrate the continuous anomaly regions. Considering the feature transferability of the space-time fusion graph network, the anomaly regions were refined to effectively solve the problems of uncertainty in anomaly discrimination and fragmentation of anomaly fragments. To solve the difficult classification problem of weakly supervised anomalous events caused by the intrinsic features of difficultly effective expressing, the abnormal event feature learning and classification method was proposed. The feature similarity graphs and abnormal similarity graphs were established in abnormal regions, and the graph convolutional networks were used for the abnormal event feature fusion learning. The abnormal-imbalance loss function was designed to improve the classification performance of abnormal events. The results show that by the proposed method, the AUC reaches 85.37% on the UCF-Crime dataset, which is 9.83% higher than the baseline of SULTANI method and 0.89% higher than that by the state-of-the-art method of THAKARE method. For the abnormal event classification, the average accuracy of 74.06% is achieved, which is increased by 4.39% compared with the existing ZHOU method. The proposed method can detect and locate video anomalous events more effectively with better anomalous event classification performance.

To generate orbital angular momentum vortex waves with different modes, the microstrip antenna was designed using graphene materials, and the antenna dimensions and materials were provided. The antenna was light in weight, simple in structure and easy to manufacture, which could generate orbital angular momentum vortex waves with mode numbers l of -2, 0 and +2 only by adjusting the feed phase difference. Through parameterized scanning, the antenna properties were evaluated in terms of impedance matching, return loss, operating bandwidth and gain, and the influence of graphene on antenna properties was analyzed. The simulation results show that when the chemical potential of graphene is 0.9 eV, the return loss of microstrip antenna at 14.90 GHz can reach -47.00 dB with the gain of 9.25 dB, and the performance is improved significantly compared with the traditional antenna. Adjusting the chemical potential of graphene can also improve the return loss of antenna.

To identify the overvoltage types of high-speed electric multiple units trains and provide data support for fault tracing and train improvement and optimization, the fuzzy identification method was proposed. The wavelet transform was used to decompose the overvoltage time domain signal in 17 layers under the db4 wavelet basis, and the energy values under each decomposition scale were extracted and normalized to obtain the overvoltage signal characteristic quantities. Based on fuzzy theory, the standard models of train lift bow, incoming split phase, outgoing split phase, high frequency resonance, ferromagnetic resonance and VCB operation overvoltage were established. The identification of overvoltage was completed by the closeness calculation and the proximity selection principle. The results show that by the proposed method, the identification accuracy of overvoltage is more than 90%.

To better characterize the damage process of cement stabilized crushed stone, based on the gray level co-occurrence matrix theory, the damage characterization indexes of moveable grey level pair(Psgl) and damage degree(D)  were proposed. The damage processes of cement stabilized crushed stone materials with two typical structures of skeleton void type and skeleton compact type were analyzed. The damage characterization indexes of Psgl and loss degree parameter of D of cement stabilized crushed stone sections were calculated under different damage degrees. The results show that the damage process of cement stabilized crushed stone can be characterized by the moveable gray scale on Psgl. With the increasing of load, the damage degree parameter D of cement stabilized gravel is gradually increased in the form of quadratic parabola, and the cumulative damage degree of the damage development accounts for about 80% of the whole loading process. The damage development rate of the skeleton void structure specimen is higher than that of the skeleton chamber structure specimen.

To investigate the secondary lining support timing for expanding the existing tunnel in grade V surrounding rock, the finite element numerical simulation method of Midas GTS was used to establish finite element numerical simulation model, and the concepts of load release rate and displacement release rate were introduced. Combining the example of the Muyucao tunnel expansion project, the numerical simulation data was compared with the monitoring measured data. The results show that compared with new tunnel construction, the deformation volume in the expansion process of existing tunnel is smaller, and the rate of change is slower. As the cutting face is continuously advanced, the load release rate and displacement release rate are both increased continuously and eventually stabilized. As the proportion of secondary lining load is increased, the crown subsidence and horizontal displacement are both decreased to some extent, which illuminates that the early construction of secondary lining in the tunnel expansion process is beneficial for controlling rock mass deformation. The deformation will continue to occur even after the lining is constructed, and the subsequent monitoring data should be fed back in time to ensure tunnel safety. The reasonable support timing for the secondary lining in the expansion process of existing highway tunnel with load release rate and displacement release rate as control indicators is obtained.

To investigate the influence of solvents on asphalt viscosity, the single and dual solvent diluted asphalts were prepared using hydrocarbon solvent (HS) and ethyl acetate (EA). The Brookfield rotational viscosity tests were conducted to evaluate the effects on viscosity reduction. The viscosity prediction model was established based on fuzzy inference system (FIS), and the predicted values were compared with the interpolated experimental results for error analysis. The results show that the viscosity reduction curve of the asphalt with single HS follows inverse exponential function. When the HS content ranges from 10% to 30%, the viscosity is sharply decreased from 88 000 to 5 800 mPa·s. For dual solvent systems with HS and EA contents of respective 10% and 25%, the asphalt viscosities are dropped below respective 3 500 mPa·s and 150 mPa·s, which demonstrates significant viscosity reduction, and EA shows superior viscosity reduction efficiency. The error metrics between predicted value and test value of MAE, MAPE, RMSE, R2 and paired t-test p-value are respective 12.216 7, 0.068 5, 18.159 4, 0.999 3 and 0.667, which indicates that the prediction model can reliably reflect the experimental outcomes. The proposed dual solvent viscosity prediction model exhibits high accuracy, and the developed interactive viscosity calculation tool offers practical convenience.

 To reveal the nonlinear dynamic behaviors of the braking system for bridge erection machines, the dynamic model of braking system with non-constant friction coefficient and viscoelastic damping was established. The viscoelastic damping was characterized by the Caputo fractional differential model, and the equivalent damping coefficient and equivalent stiffness coefficient of the viscoelastic damping were deduced. Based on the Lyapunov stability theory, the stability conditions of the steady-state solution were obtained, and the existence conditions for the periodic motion of system were established. The bifurcation diagram, phase diagram and Poincaré diagram of the system were obtained by numerical solutions. The effects of viscoelastic damping and braking force on the nonlinear dynamics properties of system were discussed. The results show that there are complex nonlinear dynamic behaviors in the bridge girder erection machine. The vibration amplitude, stiffness characteristics, damping effects, range of generated periodic motion and bifurcation characteristics of the braking system can be simultaneously affected by changing the braking force and viscoelastic damping parameters, which has complex influence on the working stability and working life of system.

To solve the problem of the lack of specific theoretical support for the current anchored sealing gasket section design method with relying on engineering experience and analogy, the ABAQUS finite element software was used to establish the two-dimensional plane strain models of various anchored sealing gasket sections under the most unfavorable condition with 15 mm staggered joint and the normal condition with 0 mm staggered joint. The effects of the gasket cross-section parameters of bottom thickness, foot size, bottom hole height and hole shape and the material parameter of rubber hardness on the waterproof performance of anchored sealing gaskets were investigated. The results show that the bottom thickness and foot size have slight impact on the waterproof performance of the anchored sealing gasket. When the central hole is set in arch shape, the gasket demonstrates superior waterproof performance with low closing compression force. As the height of bottom hole is increased, the waterproof performance of gasket is decreased, while the maximum contact stress and the closing compression force is also reduced.

The seismic response mechanism and characteristics of layered sedimentary valleys on inclined slopes were investigated. By setting viscoelastic artificial boundary and equivalent node force, the seismic input was realized. Considering the change of inclined bedrock and sedimentary sequence, the propagation process of seismic wave in the valley was simulated and analyzed based on the finite element method. The results show that there are significant differences in ground motion response under different sedimentary sequences. For normal sedimentation case, the peak time history of surface horizontal acceleration in the valley is the highest, and the corresponding peak period of the response spectrum is the smallest. For the surface displacement response, the abnormal depositions of soft interlayer deposition and reverse deposition in the low frequency band show more obvious response, while the normal deposition shows significant response in the middle and high frequency bands. Soft interlayer sedimentary valleys can filter out some high frequency seismic waves and show a certain degree of "seismic isolation" effect. When the seismic waves with different frequency components are input, the reverse sedimentation situation shows that the area near the edge of the valley is relatively safe. The effects of inclined bedrock surfaces, sedimentary sequences and seismic characteristics on the seismic response of valleys should be comprehensively considered in the actual engineering sites.