According to the transmission system characteristics of pure electric loader, the ramp anti-slip control strategy was developed based on the operating parameters of loader and the torque and speed uploaded by the walking motor controller to the vehicle controller. By the vehicle controller, the conditions for the pure electric loader to enter the anti-slip mode was determined, and the proportional-integral（PI） algorithm was used to calculate the target torque delivered to the walking motor controller during the anti-slip mode for making the loader stable on the slope without pedal action. Based on Simulink, the ramp anti-slip control algorithm model of pure electric loader was established. By the Simulink automatic code generation technology, the vehicle controller programs were generated to conduct test verification and parameter optimization. The sliding distance of loader in slope parking and slope starting stage was used to measure the ramp anti-slip effect. The results show that the proposed ramp anti-slip strategy can effectively realize the standing slope on each slope. The maximum sliding distance is 8.0 cm when the loader stops on the slope at 30% slope, and the maximum sliding distance is 4.0 cm when the loader starts on the slope, which can meet the requirements for safe parking on slope.

To solve the problems of severe centralization and vulnerability to data tampering in traditional seafood traceability systems, the blockchain technology was employed to design the seafood traceability system based on hazard analysis and critical control points (HACCP). The multiple linear regression was used to process the sample data indicators and confirm the critical control points in the farming process. Based on the determination of key information, the smart contracts were designed to verify the uploaded data at each stage of seafood traceability. From the perspective of system feasibility, the detailed designs were conducted for system functionality, consensus topology and traceability coding. The seafood traceability system was developed and subjected to function and performance testing. The results show that the system transaction throughput is consistently around 320 transactions per second. The average transaction latency is approximate 0.5 s, and the transaction success rate is about 99.8%. The performance is fully capable of meeting the needs of production practices.

The three-dimensional dislocation cultivation system of factory hydroponic lettuce can be developed on the basis of increasing the number of cultivated plants per unit area without affecting the light of lettuce.To improve the mechanization rate of the system and reduce labor intensity, according to the characteristics of the three-dimensional dislocation cultivation, the planting cup was proposed to raise to a certain height before pushing the planting tray, and the colonization cup could cross the cultivation trough plug. The colonization tray was pushed up the cultivation trough, and a set of planting tray handling equipment was developed. Taking the lettuce colonization tray as test object, the Box-Behnken experimental design method was used to explore the factors affecting the success rate of the handling equipment on the plate, and the optimal parameter combination was discussed. The test results show that when the horizontal installation error of the cultivation tank is 0-5 mm with the guide plate angle of 60° and the advancing speed of 0.22 m/s, the largest success rate of the upper plate of 99.33% can be achieved， which can meet the three-dimensional dislocation production requirements of factory hydroponic lettuce.

To realize the on-line detection of pH value of heterogeneous systems and provide stable reference potential, the polydopamine modified Ag/AgCl solid film reference electrode was developed by magnetron sputtering deposition of silver film and chlorination of NaClO solution to obtain AgCl layer and dopamine deposition modification film. The stability, reversibility and cyclic voltammetry characteristics of polydopamine modified electrodes with deposition times of 4, 8, 12 and 16 h were compared in phosphate buffered saline（PBS） solution. The results show that the standard deviation of response potential of reference electrode in PBS solution is about 1.00 mV. It can recover to relatively stable potential in different pH buffer solutions, and the deviation of the reference electrode is less than 5.00 mV with good reversibility. The response potential drift is small in 50 cyclic voltammetry tests. According to the sensitivity of the reference electrode ranging from 2.00 to 2.50 mV/pCl, the polydopamine modified electrode is not sensitive to chloride ion concentration, which confirms that the polydopamine modified reference electrode is feasible. The reference electrode with deposition time of 12 h has the best comprehensive performance in various tests. The application tests illuminate that the reference electrode is more suitable for the detection of pH of cultivation substrate.

 VOF model and user-defined function were applied to simulate the flow boiling process of water in microchannel with V-cavities. The effects of V-cavity asymmetry (ε) and distribution on the bubble growth and detachment and the transition of gas-liquid two-phase flow in microchannel were analyzed. The results show that the magnitude of the V-cavity asymmetry can change the bubble detachment diameter and detachment time. The bubble detachment diameter of the V-cavity with asymmetry ε of 1.0 is 0.063 mm and 0.025 mm smaller than that of the V-cavity with ε of 0 and ε of 0.5, and the bubble detachment time is shortened by 5.20 ms and 2.40 ms, respectively. The transition of gas-liquid two-phase flow from bubbly flow to confined bubble flow is 2.7 mm backward along the flow direction, and the occupied area of the segment annular flow in microchannel is reduced slightly. When the V-cavity with ε of 1.0 has different distribution along the flow direction, the gas-liquid two-phase flow in microchannel presents different characteristics. The V-cavity with increasingly dense distribution can better inhibit the coalescence of bubbles in microchannel without segment annular flow, which can ensure the existence of liquid film near the wall to avoid the local drying for improving the stability and reliability of flow boiling heat transfer in microchannel.

 To solve the problem that the pedestrian dead reckoning (PDR) indoor signals were susceptible to interference from environment and multipath effects, the optimal indoor PDR modelling method based on multi-model fusion was proposed. The system model of the multi-model fusion indoor PDR modelling approach was given with four key stages of step detection, step length projection, direction projection and position projection. In the step detection stage, the peak detection algorithm, local maximum algorithm and advance over zero detection algorithm were integrated, and in the step projection stage, the Weinberg method and Kim method were integrated. The Kalman filter algorithm was used to correct the errors of step detection and step projection. The comparison with traditional algorithms in terms of step number, step length, direction and position errors in different scenarios was completed. The results show that the fused model combines the feature recognition results of traditional step detection and step length estimation algorithms, which can realize the optimization of signal characteristics in the process of step detection and step length estimation. In the handheld scene, the step detection is accurate, and the step length estimation median error range is 0.060 m or less with the minimum direction estimation average absolute error of 3.06° and the position estimation average error of 0.235 3 m, which achieves good indoor walking status recognition and position estimation performance.

 To improve the data detection accuracy of low-cost sensors in the original navigation system and the control effect of ship trajectory tracking, the data fusion algorithm was introduced to propose a new trajectory tracking algorithm. The multiple coordinate transformations of ship hull were conducted, and two Kalman filters were used to fuse more accurate ship heading angle, coordinates and velocity. The control algorithm based on eliminating trajectory velocity deviation and heading angle deviation was proposed. The traditional PID controller was designed as cascade control system with PI controller and  differential controller. The differential variables in the inner loop were precisely detected and transmitted by the existing accelerometers and gyroscopes to eliminate the adjusting differential parameters for achieving good control results of systems with low detection accuracy. The each required output of the two vessels on the ship was calculated to improve the effect of trajectory tracking. The multiple cruise tests of unmanned working ship were conducted on the experimental platform with real-time monitoring the ship status data and operating trajectory through upper computer，and the overall optimized trajectory tracking effect was analyzed. The results show that the heading angle, positioning and velocity fused by the Kalman filter have smaller data variance and are closer to ideal values. For the control algorithm， the overshoot of the system does not exceed 3%, and the response speed is fast enough with the steady-state error of about zero. After optimization，the success rate of the straight section for trajectory tracking is increased from 80% to 95%, and the success rate of the turning section is increased from 60% to 90%. The maximum average yaw distance of the straight section after optimization is decreased from 0.83 m to 0.12 m, and the maximum average yaw distance of the turning section is decreased from 1.25 m to 0.22 m. After turning, the average adjustment distance of 0.95 meters is required to enter the straight state, and the tracking effects of straight and turning sections are significantly improved.

To provide drivers with real-time and accurate pedestrian information and reduce traffic accidents, the detection of enhanced YOLOv3-tiny (DOEYT) pedestrian detection algorithm was proposed. The robust feature extraction network was established, and the asymmetric max-pooling was used for down sampling to prevent the loss of lateral pedestrian features due to the increased receptive field. Hardswish was employed as activation function for the convolutional layers to optimize network performance, and the global context (GC) self-attention mechanism was used to capture holistic feature information. In the classification and regression network, the three-scale detection strategy was adopted to improve the accuracy of small-scale pedestrian target detection. The k-means++ algorithm was used to regenerate dataset anchor boxes for enhancing network convergence speed. The pedestrian detection dataset was constructed and divided into training and testing sets to evaluate DOEYT performance. The results show that by the asymmetric max-pooling, Hardswish function and GC self-attention mechanism, AP values are increased by 14.4%, 7.9% and 10.8%, respectively. On the testing set, DOEYT achieves average precision of 91.2% and detection speed of 103 frames per second, which demonstrates that the proposed algorithm can quickly and accurately detect pedestrians for reducing the risk of traffic accidents.

The dynamic graph structure was proposed to address the non-Euclidean property of vehicle interactions for realizing the exchange of vehicle information among highway vehicles. The multi-modal driving behavior trajectory prediction model was designed with incorporating graph convolutional neural network (GCN), convolutional social pooling (CS) and long short-term memory network (LSTM). The LSTM encoder-decoder was used as foundation framework, and the vehicle interaction relationships were effectively extracted by convolution and graph convolution. The maximum pooling and average pooling techniques were introduced to achieve feature extraction and background information retention. The results show that by the proposed model, the root mean square error (RMSE) is 4.03 m in long-term horizon (5 s) and is significantly improved by 10.8% compared to the baseline model. Compared to other deep learning models, the proposed model exhibits higher accuracy. The prediction performance of the model is 8% to 11% better than that of the baseline model in different traffic scenarios. The ablation experiments can confirm the effectiveness of each module in the model. The proposed model can predict the probability distribution and corresponding trajectories of vehicles in various modalities over long-term horizon in the future.

The performance of Beidou satellite-based augmentation system (BDSBAS) single frequency augmentation service was evaluated and analyzed according to the shortcomings of the research on the corresponding operational performance support of the BDSBAS with the actual data. The calculation process of BDSBAS accuracy enhancement and integrity enhancement was comprehensively sorted out based on the relevant international standards. The process of accuracy enhancement and integrity enhancement was achieved through the publicly broadcast BDSBAS-B1C messages, and the enhancement results were used to evaluate the service performance of BDSBAS in China in the second quarter of 2022 in terms of precision, integrity and availability. The results show that during the test period, the BDSBAS single frequency augmentation service can support the accuracy and availability required for approach procedure with vertical guidance (APV)-I operation, and the horizontal and vertical accuracies respectively range from 1.5 to 5.5 m and from 2.5 to 10.3 m with the availability of APV-I operation of 99.871% (better than 99.000%). Although the transient integrity does not support APV-I, the positioning error is completely enveloped by the protection level without misleading information and dangerous misleading information events.

 Kitchen waste is composed of complex and diverse ingredients with a variety of organic matter and inorganic salts, which is perishable and foul-smelling under high temperature conditions in summer. To reduce the deterioration rate of kitchen waste in the collection and storage process and cool down the kitchen waste in community residences for short time before treatment，the intelligent temperature-controlled dustbin was designed based on the fan-pad cooling device. Several temperature sensors were distributed inside the dustbin and controlled by Siemens PLC.The execution conditions of the fan-pad cooling device were programmed, and the PID algorithm was used for software optimization, while the data was visualized by touch screen. The intelligent control of the fan-pad cooling system was realized to maintain the internal temperature of the dustbin in the set range，and the system debugging and test verification were conducted. The results show that the detection error of the intelligent temperature control system is less than 0.5 ℃, which meets the accuracy requirements. When the outdoor temperature is as high as 35 ℃, the average temperature in the dustbin can be maintained at about 25 ℃, which meets the constant temperature storage requirements of kitchen waste.

To solve the frost failure problem on super-hydrophobic surfaces in low temperature and high humidity environment, the slippery surface was fabricated by coating with lubricant oil on the rough super-hydrophobic surface. The wettability and microstructures of slippery surfaces were determined. The experimental platforms of frost prevention system and vertical defrosting test system were designed and constructed to investigate the dynamic behavior of frosting and defrosting on slippery surface, super-hydrophobic surface and pure copper surface. The results show that on the slippery surface, the dynamic lubricant film can be generated in the porous and rough structures of the substrate, which can effectively reduce the frost crystal nucleation and prevent the formation of "ice bridge". The frost mass and propagation speed wave on slippery surface are significantly lower than those on super-hydrophobic surface and pure copper surface, which shows better frost inhibition performance. During the process of vertical defrosting and drainage, the defrosting efficiency on each surface is proportional to the frosting mass. The mass and surface coverage of residual water upon defrosting on slippery surfaces are slightly higher than that on super-hydrophobic surface and lower than that on pure copper surface, which exhibits good defrosting and drainage performance.

To solve the problem that conventional linear Kalman filtering was increasingly unable to meet the demand of multi-motorized target tracking accuracy, a cooperative tracking method based on adaptive multi-model particle filtering was proposed. The host vehicle and the cooperative vehicle respectively executed the adaptive interactive multi model particle filter (AIMM-PF) algorithm to obtain the motion states of the target vehicles in the environment. By the cooperative vehicle, the tracked target state was sent to the host vehicle through vehicle-to-vehicle communication. The data association and data fusion techniques based on the Hungarian algorithm and the fast covariance crossover algorithm were utilized to achieve cooperative tracking of multiple maneuvering targets. The V2V communication, radar and localization simulation system were built to sense and track seven target vehicles within 200 meters range with two intelligent vehicles as the host vehicle and the cooperative vehicle, and the simulation experiments were completed. The results show that compared with the traditional single-vehicle tracking, by the cooperative tracking, the sensing range is expanded, and the tracking error is reduced by 31.1% without affecting the tracking efficiency.

To realize fast and accurate measurement of grain moisture with low cost, the miniaturized channel state information (CSI) acquisition equipment was used for grain moisture detection. Two feature selection algorithms of random forest and principal component analysis were adopted to extract the feature subcarriers of the CSI amplitude index, and the ten kinds of grain moisture were classified based on the selected feature subcarriers. Considering that the application in the mobility scene was limited by power consumption and arithmetic power, the breadth learning system with simple structure, fast operation speed and low arithmetic power requirement was selected for processing CSI data and was compared with the traditional convolutional neural network(CNN) in terms of accuracy and training time. The enhancement nodes of the broad learning system (BLS) were dynamically increased. The experimental results show that the principal component analysis(PCA) algorithm maximally eliminates the redundant information in the CSI data. Compared with the CNN, the BLS can achieve faster speed  and better accuracy. Therefore, the PCA-BLS combination achieves the best classification results. Increasing the number of enhancement nodes can increase the training time, but the recognition accuracy is improved to some extent.

To investigate the effects of volume replacement rate of rubber coarse and fine aggregates on the stress-strain curve, the concept of volume replacement rate of rubber coarse and fine aggregates was defined. By collecting 110 existing experimental stress-strain curves from published literatures, the effects of rubber coarse and fine aggregate volume replacement rate on the key parameters of stress-strain relationship of fiber reinforced polymer(FRP) confined rubber concrete with different sections were analyzed. Based on the analysis results, the stress-strain relationship model and compressive strength model of FRP confined rubber concrete were established. The factor effects of the volume replacement rate of rubber coarse and fine aggregates and the cross-sectional shape of rubber concrete were considered in the proposed models. The model evaluation results show that the compressive strength ratio of FRP confined rubber concrete is increased with the increasing of the volume replacement rate of rubber coarse and fine aggregates, the FRP confinement stress ratio and the cross-sectional corner radius ratio. The proposed compressive strength model can accurately predict the compressive strength of FRP confined rubber concrete with rubber particles replacing part of fine aggregate or coarse and fine aggregate, and the average error value is 1.03. The predicted curves of the proposed stress-strain model are consistent with the experimental stress-strain curves of FRP confined rubber concrete.

For the air conditioning with secondary pump variable flow system, considering the regional cooling situation, the multi-objective regression method was used to solve the load forecasting problem for improving the accuracy of load forecasting. For the central air conditioning, two multi-objective regression load forecasting models of multi-objective support vector regression（SVR） and multi-objective long short-term memory（LSTM）neural network were proposed. The two models were used to train and predict on the data of the secondary pump variable flow system of the hospital in Shanghai, and the results were compared with those of the single objective regression prediction model. The results show that the prediction accuracies of the two multi-objective prediction models are higher than that of the single objective regression prediction model, and the multi-objective SVR load forecasting model has higher prediction accuracy than the multi-objective LSTM load forecasting model.

To solve the problems that the existing detection algorithm of HRNetV2p could not balance the detection accuracy of defects at various scales, the feature fusion module combined with channel attention mechanism was introduced into the detection algorithm of HRNetV2p, which could adaptively adjust the ratio of spatial-semantic information in the fusion features and could improve the network ability to preserve semantic information in shallow features. The surface defect detection dataset of IIDD for malleable iron was established for data labelling and data statistics. The CG dense skip transmission unit and the CG adaptive fusion module were introduced into the HRNetV2p network for adaptively adjusting the spatial-semantic information ratio of the front-layer features through three operations of integration, recalibration and reintegration. The experimental setup and evaluation index were given, and the performance experiments of the improved HRNetV2p algorithm on the malleable iron surface defect dataset of IIDD were completed. The results show that the average detection accuracy AP50 of the improved HRNetV2p algorithm on IIDD is 91.3%, which is 2.6% higher than the average detection accuracy of the original HRNetV2p. The detection accuracies of large, medium and small scale defects are improved by 2.7%, 2.7% and 5.6%, respectively.

To eliminate the influence of unknown disturbance and uncertainty on vehicle longitudinal control, the layered vehicle longitudinal control system based on disturbance observer and terminal sliding mode control was designed. The disturbance observer was used to observe the system disturbance in real time and realize the feedback compensation of system for reducing the switching gain of sliding mode control and suppressing the chattering problem of sliding mode control. The stability of the control system was proved by Lyapunov function, and the joint simulation experiments were conducted by Carsim/Simulink. The results show that under the two simulation conditions, the controller has good tracking effect on the reference speed, and the maximum speed error is only 0.22 m/s. The disturbance observer can accurately and quickly estimate the system disturbance with the estimation error of only 0.003 14, and the switching logic actuator performs well. Compared with the traditional sliding mode control, the designed controller has better stability with decreased chattering problem of sliding mode control, and the stable tracking of vehicle longitudinal speed is realized.

To investigate the chaotic characteristics of mixed traffic flow and discern the factors influencing the degree of chaos in mixed traffic platoons, the Cao method and the improved Cao method were employed based on the traditional traffic flow theory to determine the delay time and embedding dimension of the mixed traffic flow. The phase space of mixed traffic flow sequences was reconstructed to determine the chaotic characteristics by calculating the maximum Lyapunov exponent. The influential parameters of the proportion of intelligent connected vehicle (ICV) using cooperative adaptive cruise control (CACC) and the delay time in mixed traffic flow were analyzed. The results show that when the maximum Lyapunov exponent of the headway sequence during the car-following process is less than zero, chaos exists in the mixed traffic flow. Increasing the proportion of CACC vehicles can mitigate chaos in certain time intervals. The car-following system tends to be stable when the proportion of CACC vehicles reaches 0.6. The delay time of CACC vehicles significantly affects chaos, and maintaining low communication delays is essential for CACC vehicles to effectively suppress chaos.