Simulation analysis and experimental research were conducted to investigate the problem of operating consistency deviation of each cylinder in a high specific power V8 engine. Through the simulation analysis， the causes of the differences in the operation of each cylinder were clarified， the cam profile was optimized， and the effects of the profile optimization proposal for engine test on the operation consistency， power and economy performance of the V8 engine were analyzed. The test results show that the optimized intake cam profile can significantly improve the load deviation among cylinders under high speed and high load conditions， with a maximum improvement of 3.1% in the load consistency of each cylinder， while the engine full load， economy and emissions performance are the same as the base profile. The engine performance under partial load conditions are comparable to the basic profile， which confirms the effect of profile optimization.

To overcome the shortcomings of traditional in-wheel motor applied in electric vehicles， an integrated dual-rotor in-wheel motor （DRIWM） was proposed， which can realize single inner motor drive， single outer motor drive and dual-motor coupling drive. How to choose the most suitable drive mode for current driving condition and dynamically distribute the power of inner and outer motors is the key to solve the vehicle energy consumption problem. On the basis of the driving characteristics analysis of the motor， the drive mode switching rules of the DRIWM driven electric vehicle were formulated based on the optimal system efficiency， and the specific working range of each drive mode was determined. In the coupling drive mode， a torque distribution strategy based on the minimal energy consumption of the system was proposed by the efficiency characteristics of inner and outer motors. The simulation results show that， when the vehicle is running on the gradients of 5%， 10% and 15% respectively， the system power consumption is reduced by 4.1%， 2.7% and 1.6% using the proposed torque distribution strategy compared with no optimization strategy. In the NEDC cycle， the three drive modes of the DRIWM can freely switch with the changes of vehicle speed and demand torque. Most of the working points of inner and outer motors are distributed in the higher efficiency range， which indicate that the two motors can work with the best efficiency while bearing the best demand torque to reduce the vehicle energy consumption. The experimental results of dynamic characteristics of the DRIWM show that the inner and outer motors have fast speed response and the ability of cruise control， which verifies the rationality of the motor structure scheme.

The combustion of monomethylhydrazine （MMH）gel droplet under a nitrogen tetroxide （NTO） environment are simulated considering chemical kinetics. At first， the one-dimensional counterflow and zero-dimensional ignition of the MMH/NTO mixture are conducted. The results show that MMH will decompose into CH₃NNH and H₂ immediately under the current condition， and MMH/NTO mixture has a two-stage ignition. Then， single MMH/NTO droplet combustion under the pressure of 0.5 MPa and temperature of 1 000 K is carried out. The gel formation， expansion and fragmentation， as well as the mutual diffusion of the MMH vapor with NTO and forming a non-premix flame surface are found. Due to the physical process of gel formation， expansion and fragmentation， the temporal evolution of droplet radius is oscillating. Between the sequential gel fragmentation， the temperature at the flame surface will gradually decrease due to the consumption of MMH. With the advance of time， the frequency of gel droplet expansion and break-up increases， leading to a rising in the gas-liquid interface temperature. In addition， it is also found that there are two ignition phenomena for the single MMH gel droplet， and the temperature in the surroundings decreases due to the decomposition of NTO. Finally， the effects of initial temperature and pressure on the combustion process are compared. The higher initial temperature in the surroundings results in the faster the droplet expansion break-up frequency， and the shorter lifetime of the gel droplet. As the pressure increases， the flame surface is closer to the droplet， which is similar to the conventional droplet evaporation combustion process.

In this study， the influence of aspect ratio on the pressure drop intersection， internal flow field， and particle deposition of ACT and SCT carriers were investigated. 1D and 3D 1/4 orifice discrete phase models of DPF were established. The results showed that with the increase of the ash content and the decrease of the inlet flow rate， the pressure drop intersection point of ACT and SCT carriers decreased. High aspect ratios （1.6） resulted in better pressure drop characteristics for SCT carriers， while low aspect ratios were better for ACT carriers. The airflow velocity in the inlet/outlet orifice of ACT carriers was higher compared to that of SCT carriers. High aspect ratio carriers had significantly higher airflow velocity in the inlet orifice， while the velocity in the outlet was lower. The particle deposition on the wall of the pore channel was uneven. Increasing the aspect ratio of the carrier improved the particle trapping efficiency. The results of this paper can provide scientific and theoretical guidance for the selection of DPF for agricultural and have certain engineering application value.

Two internal side reactions that have the greatest impact on the battery aging mode are introduced. The negative region equation of the traditional pseudo two-dimensional model is improved， and the electrochemical degradation model of lithium-ion batteries is proposed. The response surface analysis method is applied to establish the aging characteristic parameters that can comprehensively describe the degradation of battery performance. A long short-term memory neural network is established to predict the future capacity. The aging characteristic parameters obtained based on the mechanism model and historical capacity retention rate are as the input of the network. Verification results of capacity forecast show that the prediction error is within 2%.

This paper mainly focuses on the simulation and experiments of spinning process of copper alloy charge liner. Based on the Simufact/Ansys platform， a three-dimensional finite element model of biconical copper alloy charge liner spinning is established. The stress and strain distribution of the charge liner， during the process of power spinning， is analyzed by numerical simulation method， and the conclusion of analysis is that higher stresses lie in the front area and biconical transition arc area of the charge liner. Aiming at stress-strain distribution of charge liner after spinning， the influences of spinning wheel installation angle， radius of the spinning wheel and the spinning wheel feed ratio on the stress and strain distribution are respectively obtained. The spinning experiments are carried out with the optimized process parameters， the experimental results show that the size of the charge liner meets the accuracy requirements， which indicates that the simulation method can play a guiding role in the actual spinning processing.

In order to improve the running performance of variable-gauge trains when working on different gauge lines， a PCA assigned signal-to-noise ratio-based method is used to optimize the suspension parameters. It established the dynamic simulation model of 1 435/1 520 mm high-speed variable-gauge train based on SIMPACK. The key suspension parameters were selected by parameter test method Pareto diagram. Based on the optimal Latin hypercube design method， 200 groups of key suspension parameters were selected for simulation test. The signal-to-noise ratio is used to analyze the dynamic performance. Assigning the weights of various dynamic indicators through principal component analysis， and obtain the optimized values of key suspension parameters for variable-gauge trains under two different gauge combinations. The results of parameter optimization are tested and verified， which show that the dynamic performance of the optimized suspension parameters on the two gauges meets the standard requirements， and the vehicle has better dynamic performance.

In order to improve the fault management system of complex system， a fault diagnosis strategy based on interpretive structural modeling （ISM） is proposed. According to the fault mechanism analysis， the fault correlation matrix of system elements is established， and the interpretive structural model is applied to transform the complex system fault correlation relationship into an intuitive hierarchical model through matrix transformation， so as to realize the structure and hierarchical system fault propagation； The PageRank algorithm is introduced to evaluate the impact and influence of system element failures； The main cause of fault transmission is clarified through the magnitude of the impact related to component failure and the fault transmission logic， so as to provide a basis for fault diagnosis. Finally， a device is used as an example to verify the effectiveness of the method.

When measuring the position of vehicle mass center by Tilting Method（TM）， there would be some problems such as inaccurate calculation of tire force point and large tilting of the vehicle， so Improved Tilting Method（ITM） is proposed in this paper. In the process of measuring the position of vehicle mass center by ITM， the force direction of the tire is always straight up， which does not need to design additional wheel load weighing device and can reduce the influence of adverse factors such as friction. In order to keep the supporting plates horizontal， a parallelogram mechanism is used to constrain them. Based on the balance of static force and moment， the calculation model of ITM is established. By means of numerical simulation， the key components of the system were verified and the measurement process was simulated. The results show that the stiffness and strength of parallelogram mechanism and test bench frame meet the safety requirements. To determine the vehicle mass center， the vehicle titling angle by ITM is smaller than that by TM， which means that ITM can supply higher precision and better experiment safety.

Bobbin tool friction stir welding （BT-FSW） was used to weld 2024 aluminum alloy with thickness of 4 mm， and the effect of thermal cycle temperature on the microstructure and mechanical properties of the joint was analyzed. The results show that the peak temperature on the Retreating Side （RS） is higher than that on the advancing side （AS）， with a temperature difference of about 20 °C. The average grain size in the stir zone （SZ） is 10.1 μm， the average grain size in each region on the RS is larger than that of the AS， and the proportions of small-angle grain boundaries is smaller than that on the AS. The fine S-phase precipitates are formed in the SZ， reduced number and increased size of precipitates in the heat affected zone （HAZ） resulting in the mechanical properties are reduced. The hardness of the joint is distributed in a "W" shape， with the smallest hardness value of approximately 72.3 HV at the junction of the thermal-mechanically affected zone （TMAZ） and the HAZ on the RS.

Due to the extremely complex internal structure of asphalt concrete， it exhibits different changing characteristics at different fatigue stages. In order to obtain the fatigue damage evolution process of asphalt concrete， research was conducted based on CT three-dimensional reconstruction technology. Conducting fatigue tests on asphalt concrete， obtaining CT scan images and 3D reconstruction images of different fatigue stages； The difference box dimension method is used to calculate the Fractal dimension of 3D CT reconstruction image， and the relationship between fatigue damage times and Fractal dimension is obtained； Using Miner's linear fatigue damage model to redefine the damage variables of asphalt concrete， it is concluded that the fatigue damage evolution process exhibits linear changes. By constructing epoxy asphalt concrete specimens for experimental testing， the results show that the proposed method can accurately calculate the relationship between compressive displacement and crack volume of asphalt concrete and fatigue damage， and infer the fatigue damage evolution process of asphalt concrete.

There are some problems in travel feature research on the scale of urban agglomeration， such as insufficient basic data and difficult horizontal comparison， thus web crawler is used to obtain some fundamental data such as migration OD flow matrix among the cities in China， and then the cumulative proportion curve features of migration intensity are analyzed. Further， a gravity model is built to fit and analyze the change characteristics of inter-city migration intensity along with distance within different urban agglomerations. This model uses Baidu migration data to represent the inter-city travel intensity， take the shortest road network distance as the distance parameter， and combine with urban permanent population. What's more， the concept of “accessibility” is defined as the average travel time weighted by central city “quality”， then the accessibility under road transport， railway transport and comprehensive transport is calculated respectively and the relationship between accessibility and distance-decay coefficient is analyzed.^{It is found that the travel distance-decay effects are obvious for most of the city clusters and the migration interactions have reached the city-scale level.}

To study the day-to-day route choice behavior of travelers under the mixed traffic flows of cars， conventional buses and customized buses and its impact on the traffic network， the characteristics of the three travel modes were analyzed first. At the same time， considering the exclusive bus lane as the public transit priority mode， the links were divided into links without exclusive bus lanes and links with exclusive bus lanes， so as to establish the link impedance function under mixed traffic flows. Based on the link impedance function， the traffic network theory was further used to establish the mixed day-to-day equilibrium model. A traffic network with multiple OD travel demand was selected for numerical simulation. The simulation results showed that the influence of experience dependence on customized bus travelers is greater than that on car travelers. Setting exclusive bus lanes can improve the route flows of customized buses， but the significance of the effect is different under different circumstances. When the proportion of conventional bus travelers is too large， the total travel time of all travelers may increase， while the larger the proportion of car travelers attracted by customized buses， the lower the total travel time， and whether the setting of exclusive bus lanes will reduce the total travel time is closely related to the size of these two proportions.

In order to study the car-following characteristics of Internet connected mixed traffic flow， the establishment of Internet connected mixed traffic flow car-following model can help to understand its car-following characteristics and improve the stability of mixed traffic flow. Considering the optimal velocity and optimal velocity changes with memory based on front and rear headway space， the velocity difference and acceleration difference of multiple front vehicles， a car-following model named multiple front and rear optimal velocity changes with memory （MFROVCM ） which is suitable for the interactive penetration of mixed traffic flow with connected and autonomous vehicles （CAV） and human-driven vehicles （HV） was constructed. The stability analysis of the model shows： Compared with OVCM model， the unstable area is reduced by 53.17%； compared with BL-OVCM model， the unstable area is reduced by 15.44%， and the stability of MFROVCM model is better than other comparison models. The simulation results show that under the same disturbance conditions， MFROVCM model has better traffic flow stabilization performance. With the increase of CAV permeability， the fluctuation amplitude of overall traffic flow velocity decreases， and the time to restore stability gradually decreases. The model can be applied to the car-following simulation of CAV and HV mixed traffic flow， and provides a theoretical basis and model basis for the traffic controlstrategy of networked mixed traffic flow.

The prediction model for shear capacity of corroded reinforced concrete （RC） beams based on the mechanism method usually introduces a series of assumptions and correction coefficients， resulting in low accuracy of calculation results and limited applicability. In the present study， based on the data drive， considering the reliability of the black box model and the rationality of the input features， the key basic characteristics of the corroded RC beams， such as geometric dimensions， longitudinal reinforcement ratio， stirrup yield strength， stirrup corrosion loss， and concrete strength， were selected. A practical model of shear capacity based on an interpretable machine learning algorithm was established. The results show that the corrosion loss， effective height， shear-to-span ratio and beam width are sensitive to the structural shear capacity. The practical model of shear capacity of corroded RC beams based on the data-driven can reveal the underlying mapping relationship between the basic features and the shear capacity. The proposed model have good applicability and prediction accuracy compared with the empirical model and black box model.

In order to evaluate the effect of filler particles on the properties of asphalt mastic， its high-temperature rheological performance of asphalt mastic is investigated by using a modified multiple stress creep recovery （MSCR） test. The difference in J_nr values between an asphalt binder and its corresponding asphalt mastic is proposed as an index to evaluate the filler stiffening effect in asphalt mastic quantitatively. The results showed that the filler stiffening effect was highly related with the type of asphalt binder. For the SBS-modified asphalt mastic， the addition of mineral filler tends to stiffen the binder until a critical filler content where any further addition of filler would not yield any stiffening effect in the resulting asphalt mastic. It was found that the critical filler contents can be regarded as appropriate filler content for the SBS-modified mastic. It was also found that the critical filler contents can be regarded as appropriate filler content for the SBS-modified mastic and they are found to be 120% and 100% of asphalt binder by mass for the asphalt mastics made by modified binders with 3.5% and 4.0% SBS respectively. Multiple linear regression modelsfor quantifying the filler stiffening effect are derived and the effect of creep stress， SBS content and filler content on filler stiffening is discussed.

A new method is proposed to measure real-time full-field soil deformation of an unconfined specimen by taking advantage of 3D-DIC technology and multi-camera photogrammetry. This measurement is achieved based on texture on specimen surface and a multi-camera system. Generation of 2D point clouds is made first using 3D-DIC， photogrammetric analysis is then applied to extract the 3D coordinates of each point in the 2D point clouds and follows by point cloud assembly， triangular mesh generation， end cut， surface reconstruction， and volume and deformation calculation by homemade software GeoTri3D. In order to verify the feasibility of the proposed method， unconfined tests were performed on a standard-steel cylinder， sand， and silt specimen. The unconfined compression test results on standard-steel cylinder verified that the proposed method is accurate in volume measurement and different-time and full-field deformation of sand and silt specimens could be captured based on the natural or man-made texture on the specimen surface. In addition， for any area of interest on specimen surface， the measurement resolution can be further increased. The proposed method enables total volume， full-field deformation， and localized deformation measurement at a high resolution， overcomes the limitations associated with the other methods such as low accuracy， resolution， and limited area of interest.

To study the influence of hybrid fibers on the flexural properties of UHPC， steel fibers were mixed with three new types of micro-nonmetallic fiber. Through the center point loading test of seven groups of four-sided simply supported two-way slabs， the loading process， failure modes and load-deflection curves of each group were studied， and the initial crack load， peak load and bending toughness of hybrid fiber boards were analyzed. The results show that hybrid fibers can improve the initial crack load， peak load， energy absorption value and deformation performance of UHPC slabs. The energy absorption value of hybrid fiber board is higher than that of steel fiber （SF） doped 1.3%. The post-peak toughness indexes T9（5.5） and T14（8）， T19（10.5） of hybrid fiber UHPC slabs were higher than those of single-doped with 1.3%SF and 1.8%SF specimens， and hybrid fiber improved the residual bearing capacity and deformation capacity. When SF is mixed with polyvinyl alcohol fiber （PVA）， glass fiber （GF） and basalt fiber （BF）， the hybrid toughening effect of SF/PVA and SF/BF is better than SF/GF. Through energy absorption method， hardening index and post-peak toughness index， the strengthening and toughening effects of hybrid fiber UHPC slabs at different loading stages can be evaluated comprehensively.

In current augmented reality systems， the most popular touch input method suffers from inefficiency and occlusion. To address these challenges， this paper proposes a novel text entry method that combines gesture keyboard with an imaginary keyboard， where users draw gestures in a transparent area. First， we verified that it is feasible and practical to use an imaginary keyboard in an augmented reality environment. Second， we conducted a user study to evaluate the performance of three entry methods： touch， gesture with visual feedback and gesture without visual feedback （the method of this paper）. The results show that the average speed of the proposed method can reach 16.95 WPM （Words Per Minute）， which is 6.92 WPM higher than touch input.

In order to accurately achieve steam turbine fault warning， a steam turbine fault warning algorithm based on association rules and multivariate state estimation is proposed. Mining different types of turbine fault features from historical data using association rules to construct original decision tables； Conduct attribute reduction on the results of association rule mining and extract the optimal set of fault feature reduction； Using the optimal attribute set combination reduction decision table to classify different types of turbine fault information. Based on historical data of steam turbine faults and real-time monitoring data， a dynamic memory matrix is constructed， and the fault state of the steam turbine is calculated using multivariate state estimation method. The fault feature vector is calculated using nonlinear Euclidean distance to estimate the deviation distance from the observed data. Use similarity function to estimate the state of the steam turbine， set fault alarm threshold， and achieve steam turbine fault warning. The test results show that the proposed algorithm can accurately classify the types of steam turbine faults， and the estimated values are basically consistent with the actual values， which can accurately achieve steam turbine fault warning.

In order to promote the development of computer vision technology and improve the utilization of remote sensing image information， the proposed method proposes fine-grained intelligent recognition of sensitive small targets in complex background remote sensing images. The proposed method first utilizes a median filtering algorithm to remove interference noise from the original remote sensing image. After graying out the image， a reasonable selection of background segmentation threshold is used to achieve the division of complex backgrounds and recognition targets in the remote sensing image， avoiding the impact of background information in the remote sensing image on the accuracy of subsequent target recognition. Input the processed remote sensing image into the CNN network， and use spatial selection method to refine and extract the main information features in the remote sensing image based on the output of the network convolutional layer feature map， obtaining fine-grained features of the remote sensing image for sensitive small target recognition； Construct an ELM small target recognition model based on the ELM classification mechanism to achieve fine-grained recognition of sensitive small targets in remote sensing images. Experiments have shown that the proposed method can achieve precise recognition of specific small targets in remote sensing images containing multiple categories of subjects， effectively improving the utilization efficiency of information in remote sensing images， and has significant significance for obtaining important intelligence.

In order to further improve the compression efficiency of Efficient Video Coding （HEVC） and make it more suitable for high-definition video compression. By utilizing the powerful mining ability of deep learning for video features， this paper proposes a multi input multi-scale residual convolutional neural network and network iterative training method， which significantly improves the performance of HEVC loop filtering. And a novel pixel based interpolation filtering method was proposed to further improve the compression efficiency of the encoding. The experimental results show that the algorithm proposed in this paper can reduce BD rate by an average of 7.47% in RA encoding mode. Compared with the two existing encoding optimization algorithms， the optimization algorithm proposed in this paper effectively improves compression efficiency while enhancing video quality.

Aiming at the randomness and correlation of passenger flow in urban public transportation hubs， a real-time data feature and XGBoost algorithm based passenger flow prediction for urban public transportation hubs is proposed. Firstly， the automatic encoder is optimized using the AdamOptimizer algorithm， and the real-time passenger flow data feature matrix of urban public transportation hubs is inputted into the optimized automatic encoder to extract data features， Then， an XGBoost model is constructed as the passenger flow prediction model for urban public transportation hubs. Differential evolution algorithm is used to iteratively optimize the model parameters， and the data features are input into the trained XGBoost model to achieve passenger flow prediction for urban public transportation hubs. The experimental results show that the proposed methods have lower RMSE and MAPE， and require less prediction time.

The purpose of person re-identification （re-id） is to identify images of the same person on different cameras. Although unsupervised models have better generalization than supervised models， unsupervised clustering will be more susceptible to noise interference. To address this problem， this paper proposes a model reverse backbone net （RBNet） that can reduce noise interference， using RBNet to learn the keypoints of the human body output from the pose detection model， adjusting the local spatial information and generating masks， and augmenting the specified location attention with the generated masks. The experimental results show that comparing baseline's cross-domain experimental results from Market-1501 to DukeMTMC-reID， mAP is enhanced by 7.0% and Rank-1 by 6.4%. Strengthening the attention to different local information can effectively improve the model accuracy.

In view of the problems that the classification loss function of existing models in the field of voiceprint recognition cannot effectively distinguish the separability between categories and lack of attention to the quality of voiceprint data， a new classification loss function DV-Softmax is proposed in this paper. Firstly， the working principle of the existing boundary loss function in voiceprint field is introduced. Secondly， the mining loss function in the field of object detection is introduced， and the concept of fuzzy sample is proposed based on it. Then， the MV-Softmax loss function is introduced in the field of face recognition， and fuzzy samples are added to make it adaptive to emphasize the difference between different samples and guide the feature learning. Finally， the voicing recognition was studied on Voxceleb1 and SITW data respectively. The experimental results show that compared with the existing boundary loss function， the equal error rate of DV-Softmax is reduced by 8% and 5.4%， respectively， which verifies that the DV-Softmax loss function effectively solves the separability between categories and concerns the quality of sample voice print data， and has a good performance in the field of voice print recognition。

Aiming at the problem that the existing siamese network only use spatial information， and face the challenges of object obstruction， disappearance， apparent severe deformation and so on， which leads to the decrease of tracking accuracy， a temporal salient attention siamese tracking network is proposed. Through the information exchange “bridge”， the network on the one hand adds salient attention to the current frame， and guides the network to focus on learning the object characteristics； on the other hand， the features of historical object in the memory network are screened， and they are used as additional templates to provide the external appearance information of object， at the same time， the changing rules of the external information and spatial position of object are studied to guide the subsequent detection and classification process. In order to further improve the ability of temporal attention， a multi-scale feature extraction unit is proposed to make up for the insufficient feature extraction of backbone network. The model is tested on Got-10k data set， and compared with the object tracking algorithm STMTrack， the AO value is improved by 2.4%. According to the visualization results， this network has higher accuracy in the challenges of object obstruction and disappearance.

In the application scenario of the Internet of Things， it is difficult to meet the processing needs of emergency tasks by prioritizing task offloading based on scalar information such as maximum tolerance delay. The most critical task is called an emergency task. To ensure that emergency tasks are prioritized， this paper proposes a method of prioritizing tasks based on their criticality， and conducts research on the decision-making problem of priority task offloading， taking into account the caching of edge server task handlers， with the optimization objectives of minimizing comprehensive delays， social loss rate， and load imbalance degree. A multi-objective optimization task offloading decision problem model was established， and an improved multi-objective grey wolf optimizer was proposed to solve the problem. This algorithm introduces the best effort evolution strategy of grey wolf individuals， an external archive generation strategy based on improved differential evolution operator， and a weighted maximum method optimal solution preservation strategy to improve algorithm performance. Simulation experiments show that the algorithm proposed in this paper can effectively reduce the comprehensive delay and social loss rate， optimize load balancing between edge servers， ensure priority processing of emergency tasks， and its algorithm performance is superior to other algorithm schemes.

A building structure stability monitoring algorithm based on the improved ELM-Markov Model is proposed to address the direct impact of structural stability on building safety. Firstly， the time-frequency map of the building structure acceleration signal is obtained through the S-transform， and the texture features of the acceleration signal time-frequency map are obtained using the gray level co-occurrence matrix. Sensitive feature vectors are extracted by combining the intra class and inter class scatter matrices， Then， ELM Markov Model is constructed by combining Extreme learning machine （ELM） and Markov Model， and the fitting error of ELM is divided into Markov state and predicted by error， and the predicted value of ELM is revised. Then， the improved gray wolf algorithm is introduced to optimize the state number of ELM Markov Model. Finally， the sensitive feature vector is input into the optimized ELM Markov Model to realize the stability monitoring of building structures. The experimental results show that the proposed method has small monitoring error， strong robustness， and high efficiency.

A real-time control system for a six-degree-of-freedom robotic arm based on a hydraulic drive is designed to meet the needs of heavy-duty operations. A single IPC is used to construct the real-time control system architecture of the hydraulic robotic arm， i.e.， the upper-level task scheduling and the lower-level electro-hydraulic servo drive are completed by a single controller. An improved parabolic-based trapezoidal velocity profile optimization strategy is designed； an inverse kinematic iterative solution method incorporating levenberg-marquardt （LM） and Quasi-Newton methods is proposed； an anti-integration saturation PI controller is introduced to eliminate the effect of integration saturation. The continuous trajectory tracking performance of the robot arm in Cartesian space and the overall performance of the control system are tested by simulation and physical prototype. The experimental results show that the control system is stable and reliable， with strong real-time performance， and meets the engineering control requirements.

Aiming at the different target requirements of each user in the Underlay cognitive radio network， a power allocation scheme that flexibly adjusts the target according to its actual needs is proposed. A dual-objective adjustment factor is proposed to configure the performance indicators of user power consumption and quality of service （QoS）， and a performance selection objective function is constructed to effectively solve the problem of only considering unilateral performance improvement under the traditional power allocation algorithm. In order to verify the effectiveness of the scheme， the user power and signal to noise ratio under different adjustment factors are simulated and analyzed. Experiments show that by adjusting the target factor， low power consumption and high quality of service can be selected to meet the different needs of users in the same scenario.

The 3-degrees of freedom （DOF） helicopter has three typical features： nonlinearity， underactuation， and suffering from uncertainties and external disturbances. It has received much attention from well-known research groups， owing to its convivence for the verification of motion planning and robust control. However， the lack of accurate models for motion and actuator dynamics leading to the limitation for the motion planning and control algorithms. Motivated by this fact， we study the model of a 3-DOF helicopter platform as a multi-rigid-body system， which is equipped with low-precision encoders for attitude measurements. The detailed work includes： ① modelling the dynamics for three bodies respectively； ② identifying the parameters of the multi-channel motion as well as the parameters of the motor-propeller lifting component； ③ linearizing and analyzing the nonlinear model； ④ verifying both the nonlinear and linearized models. Finally， a benchmark model for the platform is obtained， by which， the model-based controller can be designed. Some experiments are designed to show the completeness and high-accuracy of the benchmark model. Two “feedforward + linear quadratic regulator （LQR）-based feedback” controllers for the angular trajectory tracking are designed based on the identified model and implemented on the experimental setup， where the tracking performance of these controllers can be benchmarked against other advanced algorithms.

In order to better obtain shortwave digital signals due to noise interference in the channel， a modulation recognition algorithm for shortwave digital signals under low signal-to-noise ratio is proposed. The short wave digital signal model to be identified is constructed， and the transmitted signals are collected to obtain pulse noise， Gaussian noise， and intersymbol interference signals. The roll off and the amplitude frequency response of the Matched filter are calculated using instantaneous values， and the noise effect is suppressed by combining phase correction. The instantaneous information is optimized by wavelet filtering technology， and the average of instantaneous amplitude， instantaneous phase and Instantaneous phase and frequency absolute value is calculated to complete modulation recognition of short wave digital signal. Through experiments， it has been proven that the proposed algorithm can effectively achieve digital signal modulation recognition with minimal noise impact and high accuracy.

In order to optimize the quality of deep space communication， a multi-scale temporal suppression algorithm for inter code interference in deep space communication networks is proposed. The proposed method first clarifies the transmission mode of communication signals in the communication channel during the deep space communication process based on the deep space communication structure， and then establishes a deep space communication network channel model. Utilize FFT algorithm to conduct multi-scale analysis of communication signals in deep space communication network channels， and transform the transmission signals from time domain to frequency domain； Introducing a threshold algorithm to determine the frequency of inter symbol interference by traversing the amplitude of the signal in the frequency domain. Combining blind equalization algorithm with minimum mean square algorithm， a blind equalizer is constructed to achieve multi-scale temporal suppression of inter symbol interference in deep space communication networks by balancing the communication signal's own sequence. The experiment shows that the proposed method can ensure the quality of signal reception at the receiving end while reducing inter symbol interference in deep space communication networks， providing an important guarantee for the effective transmission of deep space communication data.

A new security-oriented cooperative adaptive predictive cruise control （CAPCC） algorithm is presented for connected and automated vehicle （CAV） platoons subject to constraints and Denial of Service （DoS） attacks of predecessors' acceleration. Firstly， using the kinematic characteristics， the acceleration estimator is designed to alleviate the effect of DoS attacks on cruise controllers. Then the synthetical trajectory signals are designed to flexibly represent various traffic situations of CAV platoons. Secondly， by adopting the receding horizon optimization principle， the secure CAPCC scheme of the vehicle platoon with satisfactions of the safety constraints is formulated in the framework of distributed model predictive control. Moreover， some sufficient conditions in the form of linear matrix inequalities are obtained to ensure stability and string stability of the platoon in the presence of safety constraints and the acceleration DoS attacks. Finally， some simulation experiments in representative traffic scenarios verify the effectiveness of the presented algorithm.

Aiming at the problems of single sensor target localization technology being unable to adapt to complex environments and susceptible to external interference， a distributed multi-sensor target joint localization method based on state data covariance cross fusion algorithm is proposed. Observing the position information of target objects through distributed multiple sensors and using the position information as target state data； On this basis， input the target state observation values of each sensor into the Kalman filter for error compensation， in order to improve the accuracy of subsequent target joint positioning； Therefore， the covariance cross fusion algorithm is used to fuse the observation values of each sensor after error compensation， and the fusion value is the final target position， thereby completing the joint target localization. The experimental results show that the intersection to union ratio of the positioning results for 10 target objects is close to 1， and the obtained target object position coordinates are closest to the actual position coordinates of the object. The proposed method has strong anti-interference performance and high accuracy in target localization.