The objective of this study is to analyze the drying shrinkage behavior of cement-stabilized macadam（CSM） and improve its anti-cracking performance. In this paper， previous studies conducted on the mechanism of shrinkage and cracking in CSM and shrinkage mitigation strategies are reviewed， and summarized. The drying shrinkage performance of CSM materials is mainly studied through shrinkage mechanism and shrinkage prevention and improvement measures. As a cement-based material， the existing research on the drying shrinkage mechanism of CSM mainly refers to the available dry shrinkage mechanism of cement concrete. Four models are always used to characterize the shrinkage at different stages according to the change of relative humidity（RH） inside the mixture： capillary tension， surface free energy， disjoining pressure and interlayer water movement. The drying shrinkage mitigation strategies of CSM have achieved a series of results in recent years. According to the different mechanisms CSM shrinkage performance improvement methods can be classified into gradation， cement content， CSM molding method， and additive modification. The results show that reducing the amount of cement， adopting the skeleton of dense gradation， employing vibration stirring or vibration compaction， and adding rubber， steel slag， fiber or other modifiers can effectively reduce the drying shrinkage of CSM and improve its shrinkage performance and road serviceability. Additionally， the shrinkage mechanism， comprehensive optimization method， drying shrinkage measurement and evaluation method and shrinkage prediction model of CSM are assessed. The development trend and prospect of the shrinkage performance of CSM are generated.

Topology identification of distribution network is an important work to ensure the safe and stable operation of distribution network. It could provide structure data for system power flow calculation， load capacity distribution， fault diagnosis， power network state estimation， which is the foundation of distribution network system analysis.The existing research of distribution network topology identification into two categories could be divided in this paper ： the first type of method is based on historical topology information， including matrix method， innovation graph method and optimal matching method. The second type of method is based on real-time measurement information， including correlation judgment method， signal injection method， linear programming method and machine learning method. Finally，the application range， main used data and characteristics of the existing methods was analyzed，the future research direction of distribution network topology identification was proposed.

This paper introduces the research status of controllable reactor by consulting literatures， analyzes the regulation principle of controllable reactor and points out the realization mode of controllable compensation capacity of reactor. Based on this， Aiming at the problems of various structures and different names of the existing reactors， more than 20 kinds of controllable reactors are classified from the perspective of the working principle of controllable reactors. The characteristics of each type of controllable reactors are summarized. And a comparative analysis of controllable reactors is made from the control accuracy， response speed， harmonic content and other aspects. Finally， the development of controllable reactor is prospected. The results show that， according to the regulation principle of controllable reactor， controllable reactor can be divided into two types： mechanical controllable reactor and magnetic adjustable controllable reactor. Magnetic controllable reactor has obvious advantages and good practical value； besides， the DC magnetic controllable reactor has the characteristics of small capacity control， which is widely used. The future development of controllable reactor can focus on six aspects： magnetization characteristics， intelligent protection and control， controllable reactor-transformer magnetic integrated equipment， et al.

In order to study the failure of single and double lap joints after high temperature aging， a batch of Al-CFRP single lap joints and Al-CFRP-Al double lap joints were fabricated. The quasi-static tensile tests were carried out at －40， 20 and 80°C after aging at 80°C for 0， 10， 20 and 30 days. The single regression and binary regression models were established to analyze summarize the failure strength and failure forms of lap joints and reveal the aging mechanism. The results show that： with the increase of aging time and service temperature， the failure strength of the joint decreases significantly， and the service temperature has a greater influence on the failure strength than the aging time， and the correlation is higher. The failure strength of double lap joint is significantly higher than that of single lap joint. In engineering application， double lap joint can be used to replace single lap joint to obtain greater bearing capacity under certain conditions. With the increase of aging time， the failure mode of lap joint changes from cohesive failure to mixed failure， accompanied by fiber tearing. The longer the aging time is， the more obvious the fiber tearing is. The SEM observation shows that the aging CFRP resin decreases significantly， and the bonding strength between CFRP matrix and resin decreases， which leads to the fiber tearing phenomenon more easily.

Aiming at the high sensitivity and low accuracy of threshold selection in time domain extrapolation， a multi-criteria threshold selection method based on principal component analysis is proposed. First， a pressure wireless test system to collect load data under continuous operating conditions of hydraulic excavators was set up； and the fitting test indexes of Generalized Pareto distribution (GPD) under various candidate thresholds are obtained through a variety of test criteria， and the optimal threshold is selected based on each test index and principal component analysis. Secondly， the over-threshold data is extracted and the GPD distribution is used to fit the over-threshold data. The correlation coefficient between the fitting result and the over-threshold data is greater than 0.995， and the GPD distribution generates random data consistent with the number of over-threshold data and replaces the original over-threshold data to achieve extrapolation of the load spectrum in time domain. Finally， the frequency characteristics of the one-fold extrapolated data and the original data are analyzed， and the amplitude correlation coefficient between them is 0.987， and the mean correlation coefficient is 0.993. The results show that the threshold selection method in this paper is suitable for time-domain extrapolation of the load spectrum.

The mechanical model of a two-degree-of-freedom amplitude-limited vibration system with multiple rigid constraints is established. Combined with the periodic and impact Poincaré mapping， the pattern types of periodic impact motions and their occurrence domains are identified in a two-parameter system. With the irreversible mutual transition of adjacent one-periodic impact motions， there are a series of singularities， hysteresis and tongue-shaped transition domains on the critical line of the distribution domains. Based on the sampling range of dynamic parameter， the dynamic response of the vibration system under extreme mass distribution conditions is analyzed. The simulation results show that the non-smooth bifurcation behaviors such as Grazing bifurcation and Sliding bifurcation leads to the one-periodic impact motion sequence transits to incomplete and complete chattering-impact motions. Excessive energy loss in instantaneous impact leads to sticking phenomenon， which changes the original structure of the vibration system and reduces the degree of freedom of the system. Due to the simultaneous action of complete chattering-impact at multiple rigid constraints， the transient static state appears in extreme control parameter domain.

When high-capacity transition metal oxides want to replace current low-capacity commercial carbon as anode materials for lithium-ion batteries， they must be designed to solve the pulverization and conductivity issues. In this paper，N-doped carbon-based Co/Co₃O₄@C core-shell structure composites by pyrolysis and hydrothermal oxidation were innovatively synthesized by us. By adjusting the hydrothermal time， complete structure， regular morphology and uniform size product can be achieved. The core-shell structure composed of nitrogen-doped carbon can shorten the transmission path of lithium-ion， which leads to the high conductivity of these composites. In addition， there are more pores in the amorphous carbon， which can not only make full contact with the electrolyte during charging and discharging， but also reduce the volume expansion generated in the cycling process and delay the rate of capacity decay caused by the structural collapse. The results of the electrochemical show that the discharge capacity was stable at 620 mA·h/g （carbon content 56.8 wt%） after 50 cycles of constant current of 0.1 A/g， which is higher than its theoretical specific capacity. After 250 cycles， the reversible capacity can reach 572 mA·h/g and the coulomb efficiency can be maintained at about 99.8% under the constant current of 2A/g. The results show that the N-doped carbon-based Co/Co₃O₄ nanoparticle core-shell structure composites with good dispersion can improve the conductivity and electrochemical performance of the electrode of the high capacity of lithium ion batteries with a long cycle life.

In order to evaluate the efficiency of road transportation more comprehensively， objectively and flexibly， the road transportation process was disassembled into three parts： infrastructure construction， passenger and cargo transportation and transportation performance evaluation， and fully considered the impact of undesired output and indicator uncertainty on efficiency. As a result， a road transportation efficiency evaluation method based on fuzzy double frontiers network data envelopment model was proposed. The main advantage of the model was that it can monitor the variation of internal sub-process efficiency in the imprecise system and highlight the role of inefficiency decision unit in efficiency evaluation. Finally， the model was used to evaluate the road transportation efficiency in Jilin Province， and the reliability of the model was proved by the rank-sum test.

In view of the deficiency of existing research on analyzing the resilience of urban rail transit network based on topological network efficiency， a resilience assessment method based on network performance response function is proposed， which is the weighted sum of OD passenger flow loss ratio and network service efficiency loss ratio. The Delphi-entropy weigh method is used to determine the comprehensive weight of two indicators， and a recovery optimization model with the maximum network resilience index is established， and the adaptive genetic algorithm is adopted to solve the developed model. Taking Xi'an rail transit network as an example， four hypothetical perturbance scenarios are proposed considering random attack and intentional attack. The differences of network resilience repair effects of target recovery strategy， random recovery strategy and preference recovery strategy under four hypothetical perturbance scenarios are compared and analyzed. The results show that the target recovery strategy has the best repair effect on rail transit network， followed by preference recovery strategy. Compared with random attack strategy， different recovery strategies have different network resilience under intentional attack strategy. When selecting the sequence of repairing damaged stations， we should not only consider the importance of damaged stations in network topology， but also consider the impact of passenger flow on network performance. Increasing the input of repair resources can shorten the recovery time and improve the repair efficiency， but the increase of repair resources is not proportional to the improvement of network resilience. The research conclusion can provide decision-making basis for the resilience assessment and emergency repair recovery of urban rail transit network.

With the development and popularization of connected and automated vehicles（CAVs）， the traffic flow on the road will transition to a mixed traffic flow composed of CAVs and human-driven vehicles （HDVs） for a long period of time in the future. To reveal the relationship among the flow， density， and speed of mixed traffic flow， a fundamental diagram model of mixed traffic flow in connected automated vehicle environment is established， taking into account CAVs degradations and platooning intensity as the starting point. First， different types of vehicles after CAVs degradations are simulated by using specific car-following models. The proportion of different types of vehicles under the influence of platooning intensity is determined. Based on this， the fundamental diagram model is derived and established which considers both vehicles degradations and platooning intensity. This makes up for the shortcomings of existing studies that do not fully consider the two aspects. What's more， it makes the model more suitable for the actual situation of mixed traffic flow. Finally， the SUMO simulation experiment is designed to verify the theoretical conclusions. The results show that the flow-density scatter points obtained by simulation in different scenarios are consistent with the corresponding theoretical curves， which verifies the correctness of the theoretical model established in this paper.

To explore cargo electric tricycles' risk behavioral characteristics of driving behaviors and its factors， a questionnaire scale of "personality trait-attitude/perception-behaviors" factors of risk behavior was designed and implemented. The data of 906 electric tricycle riders and 280 control group were analyzed， and four risk behavior models of road lapse， errors， speeding and red-light violations were given by structural equation. Based on the information entropy of the sample data， the risk factors were given weight to analyze the influencing mechanism of each factor on the risky riding behavior of cargo electric tricycles. Studies have shown that people who are prone to anger （β=0.332，0.309） and hardly to altruism （β=-0.215，-0.156） are more likely to make risk behaviors of errors and speeding， and people who are prone to anger（β=0.275）， have a strong normlessness（β=0.164） and have a poor altruism（β=-0.209） are more likely to make risk behaviors of lapses. People with a high anxiety（β=0.144） and a strong normlessness（β=0.231） are more likely to make risk behavior of red-light running. People with a poor safety attitude are prone to all four risk behaviors. The comprehensive risk scores of the four types of risk behaviors are 0.367， 0.176， 0.321 and 0.136 respetively.

Based on the requirements of bidirectional green wave coordination control for the head of the platoon on the arterial street， aiming at the incoming traffic flow of different movements at the upstream intersection， considering the influence of traffic flows and saturation flows， the setting of intersection phase sequences， the duration of arterial coordination phases and other factors， the computing method of the total clearance time of queuing vehicles was given. The calculation of bias-split in different directions were analyzed， the calculation processes of intersection offset and arterial green wave bandwidth were presented， and the algebraic method of bidirectional green wave coordination control for the head of the platoon considering the clearance time was established. The findings show that， compared with traffic signal optimization software SYNCHRO and TRANSYT， better results in reducing the average delay and stops of the coordinated platoon can be achieved by this method.

In order to better predict the short-term passenger flow of urban rail transit， a prediction method based on the recurrent neural network model is proposed. Firstly， based on the actual passenger flow data of each rail transit station， the Pearson correlation coefficient is used to determine the influencing factors of short-term passenger flow of rail transit， such as the weather conditions， historical passenger flow， whether it is a peak time period， whether it is a working day， etc. Secondly， the K-means clustering algorithm is used to classify rail transit stations into three types： high， medium， and low passenger flow stations. Then the distribution of passenger flow for each station type in time and space is analyzed， to determine the peak period of passenger flow for each station type. Finally， two urban rail transit short-term passenger flow prediction methods based on long-short term memory neural network（LSTM） and gated recurrent unit（GRU） respectively are proposed to predict the passenger flow of each type of station in different time period. The experimental results show that 5 min is the best time granularity for short-term passenger flow prediction of the two models. In this time granularity， the overall performance of the GRU model is better than the LSTM model.

To clarify the catchment area and attraction intensity of urban rail transit stations， models of catchment area and attraction intensity of multi-connection modes of urban rail transit stations are proposed. Based on the main connection modes of urban rail transit， the catchment area of rail transit station is divided into direct catchment area and indirect catchment area. Considering the spatial characteristics around the station， lattice distribution algorithm based on threshold and grey distance delay model are proposed respectively to study the direct and indirect catchment area. On this basis， the attraction intensity models of multi-connection modes are constructed. The results indicate that： the catchment area of direct connection mode is not a regular circle， and the direct catchment area varies from stations. The attraction intensity of direct connection mode decreases gradually along the distance. For the indirect catchment area， the catchment area of indirect connection mode varies from stations， too. With the increases of the distance from the station， the indirect attraction intensity decreases first， then increases and then decreases.

By considering the flexible lane setting， the vehicle and pedestrian delay analysis is carried out for the bus priority intersection with left-turn phase， combined with the pre-signal system. The design is adopted at an intersection to analyze the arrival and departure process of vehicles at the main signal and pre-signal， and calculate the delay changes of buses， straight social vehicles and left-turn vehicles before and after flexible lane setting. The results show that after a flexible lane setting is adopted at an entrance of a bus priority intersection with a left-turn phase， when the intersection is saturated， the bus delay is significantly reduced. The delay of left-turn vehicles and the total delay of straight vehicles are both reduced.

Through the questionnaire， 2391 taxi drivers' personal attributes， physical fatigue perception， work load， risk driving behaviors and road traffic accident experience were collected. Multiple indicators and multiple causes（MIMIC） model is used to explore the induced effects of physical fatigue perception， risky driving behavior on traffic accidents， and to verify the effects of causal variable for gender， age and work pressure on physical fatigue perception and distracted driving behavior. Four machine learning algorithms， including logistic regression， naive Bayes， support vector machine and random forest， are selected to predict the taxi accident. The results show that： the increase of physical fatigue perception， errors driving behavior， radical driving behavior and distracted driving behavior can lead to the growth of accident rate， and gender, age and work load can affect the physical fatigue perception and frequency of distracted driving behavior. The accident prediction model based on machine learning is very effective， and the prediction effect of random forest is the highest， and the prediction accuracy of single characteristic variables 'risky driving behavior', 'distracted driving behavior' and 'physical fatigue perception' is acceptable. When the personal attribute indicators of 'working load'， 'age'， and 'gender' are introduced， the prediction accuracy can be further improved.

In order to plan a reasonable scheme for the fleet travel and minimize the energy consumption of electric vehicles during traffic congestion periods， while relieving the mileage anxiety of delivery personnel， The pure electric delivery vehicle routing problem considering time-varying traffic congestion was studied. Firstly， a novel mixed integer programming model（TD-EVRP） is established to step for its extended model of EVRP， which especially characterizes the time-varying traffic congestion state and measures the energy consumption of electric delivery vehicles. This model also refines its constraints including customer time window， vehicle load， and vehicle charging requirements. Secondly， a traffic congestion avoidance strategy is introduced into the model， which allows vehicles to park and charge at customer points during peak traffic periods in order to actively avoid unnecessary energy consumption caused by traffic congestion. Finally， the improved ant colony algorithm is designed to acquire the optimal solution. The comparison analysis shows that the TD-EVRP model with avoidance strategy constructed than without scenario can significantly reduce the energy consumption by 14.91% in small-scale case.

To consider worker's operational proficiency in the dual-resource constrained flexible job shop scheduling problem（DRCFJSP）， an improved Jaya algorithm is proposed to solve the DRCFJSP in this paper. Unlike the flexible job shop scheduling problem， the DRCFJSP deals with the three sub-problems machine allocation， job sequence and worker assignment. The standard Jaya algorithm is improved for solving the DRCFJSP problem with the minimization of makespan. The main improvements include introducing a three-dimensional vector representation coding scheme， using the properties of jobs， machines and workers to implement the population initialization， adopting the discretization characteristics of shop scheduling to update the iterative mechanism， and designing the neighborhood search operators and acceptance criteria based on the critical path， are presented. The test instances were generated based on the flexible job shop scheduling benchmark and were used to test the performance of the proposed algorithm. The computational results demonstrate that the improved Jaya algorithm is efficient and effective. Moreover， the improved Jaya algorithm can realize the reasonable allocation of personnel and rapid sequencing of jobs under the limited resources.

The motion differential equation of steel-concrete composite beams considering the shear slip effect was derived based on the Timoshenko beam theory. And the analytical expressions of eigenfrequencies and modes for simply-supported steel-concrete composite beams were obtained. The dynamic reduction coefficients which were more suitable for dynamic analysis of steel-concrete composite beams were proposed， and the analytical expressions of "stiffness reduction coefficient" and "frequency reduction coefficient" were given. Then the dynamic reduction coefficients were compared with the existing one based on Euler-Bernoulli beam theory. The effects of shear connector stiffness， shear deformation， rotary inertia， and span-to-depth ratio on the frequency reduction coefficient were discussed. The results show that the effect of the rotary inertia can be ignored， but that of the shear deformation must be considered in the dynamic analysis of steel-concrete composite beams with smaller span-to-depth ratios.

The crushing tests of 15 reinforced concrete beams were carried out by high-voltage pulse discharge（HVPD） in liquid. Based on the comparative analysis of the electrohydraulic effect and the copper wire explosion in liquid on the crushing test of concrete beams， the latter is determined as the optimal scheme. The parameters in the test include strength of concrete， borehole aperture， hole spacing， HVPD voltage and number of copper wires. The results show that the breaking effect of concrete beam increases obviously with the increase of discharge voltage and the number of copper wires， and slightly with the decrease of concrete strength. In the case of all the other parameters being equal， the concrete beam with a hole diameter of 40 mm is better than the concrete beam with a hole diameter of 50 mm. And the concrete beam with drilling spacing of 300 mm has a better crushing effect than the concrete beam with drilling spacing of 400 mm. According to the test results， a mathematical expression was established between the failure degree of concrete beam （i.e.， the crack width of concrete at beam side） and the output voltage， the number of copper wires discharged， the strength of concrete and other key parameters.

To study the seismic performance of prefabricated and pre-decorated modular building joints, two groups of double-wall joints and one group of four-wall joints were tested under low cyclic loading, and compared with cast-in-place shear walls in terms of failure mode, hysteretic performance, ductility and energy consumption. The results showed that when the axial compression ratio is 0.4， the bearing capacity of the double-split shear wall specimen is significantly improved but its ductility coefficient is only 58.53% of that of the cast-in-place shear wall specimen. The energy dissipation capacity and ductility of the double-split shear wall specimen are decreased by 5.5% and 3.4%， and the yield load and ultimate load are increased by 7.9% and 5.6% respectively， compared with the cast-in-place shear wall when the axial compression ratio is 0.1. Due to the existence of vertical and horizontal grouting joints， the four-split shear wall has poor integrity， fast cracking rate and poor bearing capacity.

In order to study the variation of vehicle-bridge coupling vibration response of curved bridge， based on the principles of modal synthesis， a 15-degree-of-freedom five-axis trailer space model was established. Taking a 50+100+50 m three-span curved continuous beam bridge （R=600 m） as a calculation example to obtain bridge modal information. Based on the wheel-bridge displacement relationship， a vehicle-bridge coupling numerical analysis method for curved Bridges is proposed， and the influence of vehicle speed， road unevenness， radius of curvature and vehicle load on the dynamic response of curved bridges was discussed. The results show that road roughness and vehicle load are the main sensitive parameters that affect the dynamic response of the bridge. The impact of vehicle speed on curved bridges is complex， and it is difficult to find its regularity. The change in the radius of curvature has a greater impact on the torsion angle of the birdge. The influence of the vertical displacement of the main girder is relatively small. The relevant research results can provide theoretical guidance for dynamic safety assessment of such bridges.

A two-dimensional heat transfer model was established using the finite element numerical simulation approach to investigate the heat transfer characteristics and its influence law of electric heating snow-melting asphalt pavement. The impact of heating system design parameters（layout spacing， embedding depth， and input power） and environmental parameters（wind speed， ambient temperature） on surface temperature distribution and heating rate of the pavement was investigated， and the accuracy and effectiveness of the model are verified by the measured data of the test section. The results revealed that the two-dimensional heat transfermodel can predict road surface temperature of electrothermal snowmelt asphalt pavement well in an accurate level， with an average maximum absolute error of 1.1 °C， an average accuracy of 88.1%， and a maximum accuracy of 96.6%. Among all the factors， layout spacing is the most important factor affecting the temperature uniformity of the road surface， followed by embedding depth. Input power and environmental factors mainly affect the temperature rising rate and the maximum temperature of electrothermal snowmelt asphalt pavement.

In order to determine the feasibility and optimal embedding conditions of conductive EPDM composite materials in pavement structure， the stress and strain curves of composite materials were studied by compression test， and a two-dimensional conductive EPDM composite pavement structure model was established by finite element software ANSYS. The mechanical response of the composite pavement structure with different embedding locations， embedding thicknesses and different loads of the composite material were analyzed and compared with the normal pavement structure respectively. The compression test results show that the composite material is mainly elastic in the presence of forces， with a modulus of elasticity of 30.6 MPa. The mechanical response results show that the composite material is feasible for use in pavement structures under reasonable assumptions and that the best location for embedding is under the asphalt concrete layer， with the smaller the embedment thickness the better. In the case of overloading， it is necessary to focus on the effect of shear stresses on the composite pavement structures. The results of this study can provide a reference for practical engineering applications of composite materials.

Based on the analysis of the stress mechanism and failure mode of stud shear connectors， the differential equation of the deflection line of stud shear connectors is established based on the theory of single pile under lateral load in elastoplastic foundation. According to mechanical characteristics in different stages of the stud shear connectors， the boundary conditions are introduced， the stud shear connectors of flexural differential is analyzed， and the calculation formulas of the shear bearing capacity， shear stiffness and elastic stage at the end of the slip value of the stud shear connectorscalculation formula are put forward. The load-slip tri-fold constitutive model of stud shear connectors is established, which has the characteristics of simple mathematical form and clear physical meaning. The results show that the calculated values of shear capacity， shear stiffness， elastic end slip and ultimate slip of stud shear connectors are 85.44 kN， 64.92 kN/mm， 0.64 mm and 6.13 mm respectively and the calculated values are in good agreement with the experimental values. The three-fold line load-slip relationship curve of stud shear connectors is close to the full-curve load-slip curve fitted by Buttry and Ollgaard， and it is in good agreement with the test load-slip curve of stud shear connectors.

A significant step of fault detection and isolation（FDI） in dynamic systems is to select a set of sensors that meets the fault detectability and isolability with the least cost. For uncertain systems， the quantification method of fault distinguishability is used to quantify the fault diagnosability performance. Since the search space of the sensor selection problem increases exponentially with the increase of the sensor scale， there is currently no research on complete method of sensor selection for quantifying the fault diagnosability in uncertain systems. This paper proposes a sensor selection algorithm based on the set blocking strategy to solve this problem： realizing subset blocking and superset blocking through BILP； reducing the nodes that need to be traversed through iterative blocking search space. The experimental results show that compared with the depth-first traversal method， the efficiency of the proposed method is increased by 4.41~103.37 times for most of the use cases in the experiment.

Due to the influence of image noise， the accuracy of image feature recognition is reduced. Therefore， an intelligent image feature recognition model based on multi-source big data analysis is proposed. According to the improved adaptive two-dimensional median filtering method， all pixels in the small window are checked and the noise is filtered. If the pixels that are not polluted by noise are detected， they are directly output. On the basis of single scale fuzzy mathematics， the three Gaussian model is used to iteratively process the uniform image brightness value， The gray value of the neighborhood center point is decomposed into the real part and the imaginary part， the local binary feature and the Brushlet feature are extracted， the gray value of the adjacent two pixels is subtracted， and the image features are intelligently recognized according to the heuristic factor and the maximum gray gradient. The simulation results show that the proposed model can adjust the resolution evenly when the image is unclear， and ensure the integrity of information， accurate and effective recognition results， and strong anti noise ability.

In order to better control the landslide hazard risk and reduce the disaster loss to the greatest extent， a prediction model and simulation of landslide displacement interval of accumulation layer based on CEEMDAN theory were proposed. The landslide information is collected by multi type sensors， the sample entropy is calculated by the adaptive complete set empirical mode decomposition algorithm， the noise range is obtained according to the correlation coefficient， and the signal noise is removed by the set reasonable threshold； Construct a distributed architecture of multi-sensor data fusion to improve the comprehensiveness of landslide data； The final prediction results are output through the construction of prediction model and sample training. The simulation results show that the proposed model has high prediction interval coverage and low interval width， ensure the prediction accuracy.

To construct a topological map with rooms as nodes and based on the object semantic grid map， this paper proposes a topological segmentation method with virtual door detection and a room generation method based on the inclusion relationship of rooms and objects. Firstly， detect the occupied grids that meet the width of the actual doors， then use geometric shapes and object semantics to optimize the detection results， and finally employ the objects contained in the topological nodes to generate the room concepts. Through testing in multiple scenarios， the effectiveness of the algorithm is verified.

The personalized transcranial magnetic stimulation can improve the treatment effect， and the reconstruction of the head model based on brain tissue segmentation helps to optimize the personalized treatment plan. Therefore， in order to achieve automated brain tissue segmentation and improve segmentation accuracy， a multi-scale attention network（MSAN） based on transfer learning are proposed. In the network， the method of transfer learning is used to solve the overfitting problem caused by small sample data， and the complementary information of the multi-modality data can be learned. At the same time， the extraction and fusion of multi-scale features are used to improve the performance of the network， and the attention mechanism is used to focus the important information after the fusion of multi-scale features， which improves the accuracy of brain tissue segmentation. Extensive experiments have been conducted on the database provided by the MRBrainS Challenge to verify the effectiveness of our model. Currently， this method ranks second in the MRBrainS Challenge and has achieved the best results in multiple indicators. MSAN can provide a better brain tissue segmentation result for the reconstruction of personalized head models， and optimize the stimulation effect of transcranial magnetic stimulation.

In this paper, the main purpose is to improve the curve detection effect, and considering the detection speed comprehensively, a curve lane detection method based on residual network is proposed. In this method， the residual network is used as the main framework and the curve enhancement is realized by adding the curve structure constraints to the loss function. On the other hand， in order to reduce the complexity of the model， the weights pruning technique is used to reduce the model. The result shows that the curve enhancement strategy proposed in this paper can not only effectively improve the algorithm performance in the curve scene， but also has little impact on the detection performance of straight lanes. After adding the sparse pruning strategy， the algorithm greatly reduces the computation time without significant performance degradation， which is more in line with actual production requirements.

In the actual localization scenarios， the receiving station is usually installed on the moving platform， which leads to the random error in its moving state information. However， the target source location accuracy is very sensitive to the location information of the receiving station， and the small error in the location of the receiving station will lead to a large error in the estimation of the target source position. Therefore， considering the random error of the location information of the receiving station， a solution to locate the mobile source using the measured values of the time difference of arrival and the frequency difference of arrival is proposes in this paper. Logistic chaotic mapping is introduced into sparrow search algorithm to locate and track the target. Logistic chaotic mapping can reduce the risk of the algorithm convergence to local optimal， so as to solve the problem of poor localization accuracy in the case of low sensor position error. The analysis of simulation results shows that the accuracy of the proposed algorithm is closer to the Cramer-Rao lower bound than that of semi-definite programming and reformulation linearization technique（SDP-RLT）， genetic algorithm， sparrow search algorithm and ant colony algorithm under the condition of low sensor position error.