The transmission power of mobile terminals is one of the key factors that affect the distances between 5G-R base stations.Selecting a reasonable terminal transmission power parameter is of great significance for expanding the distances between base stations,making full use of the resources in existing base station sites along the railway lines,and saving the deployment costs for 5G-R base stations.First of all,based on the 5G-R service and function requirements,the bandwidth requirements of different types of mobile users are obtained.Then,based on the uplink demand,the base station spacing of the 5G-R mobile terminal with 200 mW and 2 W transmission power is estimated by using theoretical calculation and simulation.The results show that,compared with 200 mW,the distance between 5G-R base stations is 1.79-2 times greater than that of the original 5G-R base stations when the terminal transmission power is 2 W.Using typical engineering parameters of high-speed railways,with an uplink transmission rate of no less than 1 Mb·s-1 at the edge of the community and a terminal transmission power of 2 W,the distance between 5G-R base stations in suburban open areas can reach 3.3-4.0 km(RRU on the tower).Mobile terminals should determine their appropriate power levels of uplink transmission according to their individual uplink traffic.It is suggested that the 5G-R communication module of the on-board integrated transmission equipment should use 2 W transmission power.
The existence of wireless network time delay is the key problem restricting the development of wireless communication network for high-speed train.Time delay samples are obtained by building a semi-physical network control test platform based on wireless communication.Aiming at the high complexity of time delay,a processing method combining complementary ensemble empirical mode decomposition(CEEMD)and singular value decomposition(SVD)is proposed.Sample entropy is introduced to reduce the amount of computation.Finally,LSSVM model optimized by chaotic particle swarm optimization is used to complete time delay prediction.Taking traction-brake module of high-speed train as the controlled object,a sliding mode controller based on multi-power reaching law is designed for time delay compensation.Chaotic particle swarm optimization algorithm is used to optimize the controller parameters intelligently.Integrated time and absolute error(ITAE)are improved into ITAE*for the fitness function to reduce the tracking error and improve the response speed.The results show that the sample entropy of the maximum component is reduced from'above 2.7'to'below 2.0'using CEEMD decomposition.After secondary treatment with SVD,the sample entropy of Intrinsic Mode Function 1(IMF1)component with rest of the maximum sample entropy reduced to 1.5,which effectively reduces the difficulty of prediction.After improvement,Mean Square Error(MSE),Mean Absolute Error(MAE)and Mean Absolute Percentage Error(MAPE)of delay prediction are reduced to 0.1396,0.2964 and 0.0083 respectively,and the prediction accuracy is improved.For longer time delay,ITAE*index is only 0.136％of that before compensation,and the cumulative absolute error of speed tracking is 0.696％of that before compensation.The speed and racking curve have no obvious vibration when the running state of the high-speed train changes and can quickly return to the stable state when the train encounters disturbance.The research results can provide a new approach to introducing wireless network control for high-speed trains.
In order to better coordinate the allocation of railway transport capacity,enhance the quality of passenger travel service,and objectively reflect the game relationship between railway enterprises and passengers,an optimization method of train stop schedule under elastic demand is proposed.Firstly,based on the consideration of elastic demand and strict train capacity constraints,the interrelationships among the train stop schedule,passenger travel cost,and elastic demand are analyzed,thereby formulating an optimization model of train stop schedule.Secondly,the linearization methods such as relaxation and outer approximation techniques are employed to reformulate the model into a linear programming for efficient solution.Finally,the effectiveness of the proposed method is verified by a small-scale example and a real example of Zhengzhou-Xi'an High-Speed Railway,respectively.The results show that the optimized stop schedule achieves a reasonable allocation of railway transport capacity.In comparison to the real stop schedule,the optimized one reduces 22 stops and decreases the total additional stop time of passengers by 76.14％with an increase of 0.30％in fare revenue,which realizes the consideration of both railway transportation efficiency and passenger service quality.
In response to the limitations of traditional structural configuration detection due to insufficient sample capacity in establishing comprehensive existing structural models,a neural network-based structural configuration inversion algorithm is proposed for identifying existing large-span station structures.Based on the linear elastic assumption of large-span spatial structures and the correlation between configuration deviations and differences in structural systems and stiffness,a structural configuration base vector space is constructed by superimposing equivalent random loads.Using configuration base vectors and measured configurations at sampled nodes as input,a backpropagation(BP)neural network is trained to obtain the actual overall structure configuration.The performance of the algorithm is validated and analyzed using the large-span spatial structure of the Pingshan high-speed railway station in Shenzhen as an example.The results indicate that the number of random loads and training samples are the primary factors influencing the accuracy and convergence speed of the inversion algorithm.For large-span spatial structures of the same type as the Pingshan station building,when the number of base vectors exceeds 200 and 300 sets of training samples are selected,the inversion effect with low error and fast convergence speed can be obtained.The proposed neural network-based structural configuration inversion algorithm can effectively infer the true overall configuration of the structure based on a small number of measured node configurations.
To study the influence laws of different levels of eddy current braking on rail temperature rise,based on the principle of linear eddy current braking,the theoretical formulae of eddy current braking force and rail temperature rise are derived using the law of energy conservation and the skin effect formula of conductors,with the combination of the data of braking force test and heat transfer theory.According to the theoretical formulae,the finite element method is used to establish a rail temperature rise model,and the temperature variation of different parts of the rail after a train applies emergency eddy current braking are analyzed.According to the minimum time interval between the designed train and the current train in operation of Beijing-Shanghai high-speed railway,the rail temperature variation of multiple trains under repeated eddy current braking is calculated,respectively.The results show that using the law of energy conservation to derive the theoretical formulae for eddy current braking force and the rail temperature rise not only ensures the calculation accuracy,but also simplifies the calculation process.After emergency braking of a train,the temperature rise at the top of the rail is the highest,while the temperature changes at the waist and the bottom of the rail are minimal during the braking process.When multiple trains repeatedly apply eddy current braking in the same section,the higher the braking level,the shorter the train interval,and the greater the cumulative temperature rise of rail.When the maximum service braking is repeatedly applied under the condition of minimum train interval,the maximum rail temperature rise meets the requirements of relevant standards for line maintenance.
The increase of train axle load puts forward higher requirements for rail cleanliness.In order to explore the influence laws on inclusions of rail service life,a finite element model of rails was established to analyze the influences of the inclusions'shape,depth and size on rail stress.The influence of inclusions on the fatigue life of rails was also analyzed by FE-SAFE fatigue simulation software and the"radius-length"curve of the rail fatigue damage caused by inclusions was calculated.The results show that smaller inclusions tended to cause local stress concentration in the rail;the larger the distance between the inclusions and the rail tread,the smaller the maximum equivalent stress of the rail;the larger the loaded area of inclusions,the more obvious the influence on rail stress;the inscribed radius of inclusions had more influence on the stress state of the rail.For 25 t axle load,the fatigue life of the rail was inversely proportional to the cube of the inscribed radius of the inclusion,and was inversely proportional to the square of the length of the inclusion.The inscribed radius of the inclusion had more influence on rail fatigue life than its length;the"radius-length"combination of the critical sizes of Al2O3 inclusions at 5 mm under the rail tread were"20 μm-1 620 μm","60 μm-395 μm"and"100 μm-207 μm".Thus,emphasis should be placed on the oversize large inclusions in rail inspection.
This study proposes using carbon dioxide phase transition power to drive the acceleration of the test car,aiming to solve the problem that the air cannon of the rail vehicle collision testbed cannot drive the test car to meet the requirements of higher test speed under existing equipment conditions.Based on the law of mass conservation,energy conservation,and the real gas properties of carbon dioxide,a numerical model of the work process of the driving system was developed.After solving the model,the calculation accuracy of the model was validated through experiments.Through numerical model,the power performance of air and carbon dioxide was compared,and the effects of initial pressure and temperature,as well as the initial volume of the cylinder on the power performance of carbon dioxide were analyzed.The results indicated that the model effectively simulated the power process of the test car when driven by the liquid-gas phase transition of carbon dioxide.Under the same structural conditions,the use of CO2 as a driving medium resulted in a 10.5％reduction in drive time,a 21.2％increase in final speed compared with that of air.This was observed with an initial pressure of 20 MPa and a driving medium mass of 170 kg.Additionally,increasing the initial pressure from 8 MPa to 20 MPa in the reservoir cylinder led to a 58.3％increase in the test car's final speed and a 32.2％decrease in acceleration time.The initial temperature has a significant effect on the phase state of CO2 in the drive cylinder,with lower temperature favoring a phase state closer to the gas-liquid saturation line.Moreover,the initial volume of the drive cylinder has great effect on the acceleration time,test car overload,and piston impact.Specifically,when the initial volume increases from 0.01 m3 to 0.20 m3,the maximum overload decreases by 23.07％,but the time spent increases by 20.9％.
To address the complexity challenges encountered in the development of the Train Autonomous Circumambulate System(TACS)modeling using the Event-B method,a methodology combining the instantiation of Abstract Data Types(ADT)with Event-B is proposed for the formal development and verification of TACS.Firstly,functional and safety requirements of TACS are extracted based on its requirements and relevant contents including case studies,which are described in non-formalized language.Subsequently,to avoid the difficulty in proving caused by complex modeling requirements in abstract modeling,a hierarchical design of the modeling refinement levels is conducted based on the TACS functional and safety requirements.Finally,the functional and safety requirements of TACS are formally modeled and verified using formal Event-B language.During the modeling process,the abstract concepts of ADT are employed to specify the essential attributes of complex system components,such as the track network,track regions,and Movement Authority(MA),in the initial model.Specific definitions of those components are introduced in subsequent refinement stages,reducing the complexity of system development and proof.The results indicate that the proposed approach helps reduce the complexity of intricate details in TACS during the early development stages,resulting in a 100％success rate in proving obligations.Furthermore,the success rate of automated proof accounts for 83％,while that of the manual proof is only 17％,simplifying the proof obligation while effectively increasing the proportion of automated proofs.
Crushed rock revetment embankment(CRRE)has been widely used in the construction and reinforcement of highways and railways in permafrost regions.However,there have been controversies in recent years on the cooling effectiveness of CREE.Especially,debates around whether the CREE built on warm permafrost could withstand adverse effects of future climate warming have been drawing attention in academia.Based on the field-observed data and numerical simulations along the Qinghai-Tibet Railway in the past 20 years,this study explores the long-term thermal regime of CRRE built on warm permafrost and the degradation process of warm permafrost in the context of climate warming.The results show that the CRRE can effectively raise the artificial permafrost table.After nearly 20 years of operation,the artificial permafrost table beneath the CRRE still rises by nearly 2 m compared with the original.However,during this period,there has been a warming process in the shallow permafrost subgrade.A warm permafrost layer with a thickness of 6-8 m developed beneath the artificial permafrost table.The compression and creep deformation of the layer can cause considerable embankment settlement.Till 30 years after the embankment construction,the artificial permafrost table under the embankment has declined to the same level as that of the natural permafrost table under the scenario of air temperature rise of 2.6 ℃ in 50 years.From then,the artificial permafrost table under the embankment and the natural permafrost table have declined almost synchronously,and the thawing of shallow permafrost subgrade will cause excessive embankment settlement.Therefore,for CRRE built in warm permafrost regions,it is necessary to timely adopt reinforcement measures such as thermosyphone,based on variations in the permafrost subgrade temperature and development of the embankment settlement.
The stability of the limestone collapse slope is high in its natural state.However,strong earthquake would cause rock fissure expansion which will lead to slope failure.Generally,the dynamic response,failure mode,and energy transfer of limestone collapse slope under strong earthquakes could be investigated with large-scale shaking table model.Consequently,the results show that the failure process of limestone collapse slope under earthquake activity could be divided into four stages-the tensile crack of the slope shoulder joints and the shear displacement of the top bedding planes,the expansion of the vertical joint fissure and the increase of shear displacement,the formation of stepped sliding surface as well as the instability and collapse of the rock mass.All the stages satisfy the failure mode of tensile-shear-slip failure.In addition,the acceleration response of the slope indicates both the elevation amplification effect and the surface effect and the PGA amplification factor grows approximately linearly below 0.55 times the slope height while it grows more rapidly above 0.55 times the slope height.Besides,when the input seismic wave amplitude is less than 0.4 g,the PGA amplification coefficient increases with the increase of the input wave amplitude and decreases when the amplitude exceeds 0.4 g.The identification of the marginal spectrum shows that the slope failure starts from the shoulder and gradually shifts downward,and the damage on the slope surface is greater than that on the inside of the slope.Moreover,the sensitivity of the marginal spectrum is better than the PGA amplification coefficient during damage identification.
Relying on the blasting engineering of the new constructed tunnel in Hainan Railway Renovation Project,the vibration influence of seismic wave generated by the drilling and blasting method during construction on the adjacent existing railway tunnel lining is studied.The on-site blasting vibration testing is conducted,a finite element numerical model for adjacent existing tunnel blasting based on the measured data is established,and the on-site monitoring method is adopted to verify the model.Taking measured vibration waves as blasting loads,the influence law of different surrounding rock levels and different tunnel clear distances on the blasting dynamic response of adjacent existing railway tunnel lining is analyzed.The results show that the blasting vibration superposition effect is easy to occur when 15 ms inter-hole initiation delay is adopted under such engineering geological environment.The results obtained by the two methods of modelling and monitoring have good agreement in terms of 3 aspects including vibration waveform,peak velocity of blasting vibration,and spectrum characteristics.The simulation results can characterize the response characteristics of blasting vibration monitored on-site.The model establishment method is feasible and the parameter settings are reliable.When the blasting parameters remain unchanged,the smaller the clear distance between tunnels and the better and more complete the quality of tunnel surrounding rock,the greater the blasting vibration speed of the existing railway tunnel lining,the more obvious the influence of blasting seismic wave on the lining,and the more prone to occur tensile stress and the more likely to cause tensile damage to the lining.When the adjacent existing railway tunnels are blasted,it is necessary to focus on the dynamic response of the lining under two conditions:the small clear distance with good and poor surrounding rock quality.
Rail profile deviation has a significant influence on the service stability of EMU.Aiming at this issue,firstly,the influence of wheel-rail profiles and wheel-rail geometric parameters on equivalent conicity is analyzed,and the dynamic simulation model of EMU trailer is established.Secondly,under single factor and multi-factor coupling,the variation laws of lateral vibration acceleration of the bogie frame on straight and curved lines are analyzed.Finally,the limit value of rail profile deviation is put forward.The results show that the nominal equivalent conicity increases with the increase of positive deviation of rail profile and inner distance of wheelset,and the decrease of track gauge for the wom wheels.The maximum nominal equivalent conicity reaches 0.59.When the train operation speed is less than 200 km·h-1,the influence of single factor on the lateral vibration acceleration of the bogie frame is slight.With the increase of speed and rail profile deviation,the influence of single factor such as gauge,wheelset inner distance,increases gradually.The multi-factor coupling effect of wheel-rail system makes the lateral vibration acceleration of the bogie frame further increase.When the rail profile deviation is less than 0.2 mm,the lateral vibration acceleration under different working conditions is less than the limit value.When the rail profile deviation is greater than 0.4 mm,the lateral vibration acceleration under multi-factor coupling approaches or exceeds the limit value.When the rail profile deviation is greater than 0.8 mm,the lateral vibration acceleration is prone to approach or exceed the limit value due to single factor or multi-factor coupling.Therefore,it is suggested to control the rail profile deviation by grades,and the recommended values of grade Ⅰ-Ⅲ management are 0.2 mm,0.4 mm and 0.8 mm respectively.
Aiming at the characteristics of high speed and precise localization for open-circuit fault diagnosis of traction inverter in rail transit industry,an open-circuit fault diagnosis method for inverter based on average voltage and extreme learning machine is proposed.Firstly,the open-circuit fault of three-phase two-level topology inverter is analyzed,and the characteristics of open-circuit fault are summarized.Secondly,the average voltage is extracted from the fault characteristics as a basis for fault detection,and the fault feature vector is constructed using the related parameters of stator current.Finally,the ELM fault diagnosis model is trained offline,the fault classifier is generated and input into the online diagnosis process,to achieve the framework construction of the proposed fault diagnosis method.Robustness test is carried out based on normal working data,210 sets of fault data are obtained by setting different fault time,different fault types,different speed and load conditions,and then online diagnostic test is carried out,which is compared with the test results of mSVM,DT and RF for verification.The results show that compared with the other three methods,the proposed method has higher robustness under various normal conditions.As for the online diagnostic test,only 1 group of data is misdiagnosed,with the test accuracy reaching 99.5％.The training time is 0.28 s and the fault diagnosis time is 21 ms,which are both the shortest of these methods.The proposed method is suitable for applications requiring fast diagnosis,precise localization and strong robustness.
High-speed rail construction effectively facilitates the flow of factors between different regions,however,it has different impacts on different regions.Taking the opening of high-speed railway in Hunan Province as a quasi-natural experiment,96 counties in four major regions of Hunan Province are taken as the research objects,and the effects of high-speed rail construction on the economy of different regions are assessed by the synthetic control method,using the balanced panel data of the county economy in each of the four years before and after high-speed rail impacts.The results show that the impact of high-speed rail construction on economic growth varies with the degree of regional development and has the characteristics of exogenous variables.Under the joint action of the dual effects,the economic effect of high-speed rail in the four major regions of Hunan Province are different,which is manifested in the fact that the siphon effect in north Hunan is the smallest,followed by south Hunan,the impact in central Hunan is larger,while the siphon effect in west Hunan is the largest.After the construction of high-speed rail,the economy of some counties shows positive growth;the economy of some regions is greatly affected by the siphon effect,which makes the labor force and other factors of production of these regions flow out,and the economy shows negative growth;in addition,the two economic effects of high-speed rail construction in a small number of counties are offset,and the economic growth rate has no obvious change.Based on this,only by giving full play to their own advantages,effectively suppressing the"siphon effect"of high-speed rail network on economic resources,and forming a virtuous circle of"outflow-inflow",can each region better promote economic growth.
In view of the fatigue propagation behavior of lining cracks under the influence of train load,the fatigue propagation reliability of cracked lining cracks was studied based on fatigue fracture mechanics and reliability theory.Firstly,the cracking mode of lining and the type of stress intensity factors affecting crack propagation were analyzed,and in accordance with the Paris formula,the limit state equation of lining based on the crack depth was established.Then,on the basis of an operating railway tunnel,the ANSYS finite element software was used to establish the numerical model of tunnel-ground coupled dynamic with cracked lining.The stress field and stress intensity factor at the crack tip under the influence of train load were calculated,and the reliability indexes of cracks with different depths during the service life of lining were solved through the Monte Carlo method.The results show that the train load has a significant effect on the stress field at the crack tip.When there is a 10 cm crack in the vault and the train load is not considered,the maximum principal stress at the crack tip of the vault lining is-3.47 MPa,whereas when the train load is considered,it is-7.26 MPa.The amplitudes of stress intensity factors at the crack tip do not refuse to obey the normal distribution law under different initial crack conditions,and their mean value and standard deviation increase with the increase of the crack depth.The reliability index of cracked lining decreases with the longer of service time,and the greater the initial crack depth,the smaller the lining reliability index and the shorter the service time.
To study the mechanical behavior of wheel/rail under the excitation of asymmetrically unsupported sleepers,a wheel/rail dynamic model is established based on the parameters of typical ballasted lines and in-service vehicles in China.The analytical solution of rail deflection displacement based on Winker elastic foundation assumption is used to verify the reliability of the simulated results.On this basis,the mapping relationship between the wheel/rail vertical force and the maximum rail deflection displacement and the depth of unsupported sleeper is simulated under the condition in which the vehicle runs through 1-3 asymmetrically unsupported sleepers at the speed of 200 km·h-1.The results show that critical unsupported capacity of the sleepers exists in all calculation conditions,the maximum deflection displacement of the rail increases with the depth of unsupported sleeper when the depth is less than the critical depth;and the increased slope is related to the number of unsupported sleepers.When the depth is greater than the critical depth,the maximum deflection displacement of the rail and the wheel/rail vertical force tend to be stable.There is the difference in equivalent support stiffness of the foundation under the rail in the zone of the asymmetrically unsupported sleepers,which will cause wheelset load increase or reduction.The rates of maximum wheel load decrease is about 9.54％,20.3％,28.2％,respectively under the conditions of 1-3 single-side unsupported sleepers.The rate of maximum wheel load decrease reduces but the maximum deflection displacement of rail increases in the case of double-side unsupported sleepers.The research results can provide reference for wheel/rail mechanical behavior,dynamic/static difference analysis of track irregularity and disease detection in zones with unsupported sleepers.
In order to further break through the bottleneck of heavy-haul railway transport capacity,a cooperative cruise control method of heavy-haul trains under virtually coupled train formation(VCTF)based on multi-agent theory is proposed.First,an improved potential energy function is proposed through the application of the multi-agent theory in order to solve the cooperative control of unit trains under VCTF by taking the safety protection control of relative braking distance and stable tracking operation control of unit trains as the core functions after the typical mode of heavy-haul train under VCTF is determined.Secondly,the controller is designed to realize cooperative cruise control of VCTF based on potential energy function and LaSalle invariance principle.The simulation results show that all 6 heavy-haul unit trains can converge to the VCTF target speed of 75 km·h-1 within 50 km after operation,the trains are under stable running condition,and the distance between trains is only slightly higher than the minimum distance between trains by 2 km.The transportation capacity of the VCTF mode is 18％-28％higher than that of the traditional mode when the train runs at speed of 40-120 km·h-1.In a word,cooperative cruise control method based on VCTF can ensure the safe,stable and efficient operation of long and heavy heavy-haul unit trains without changing the existing moving block signal system by replacing the physical coupler buffer device with VCTF.
Relying on the Jialing River Bridge of Chongqing Rail Transit Line 15,and considering the main stress state and mechanical characteristics of the structure formation process,the mechanical models of the main stress state are simplified,and the key design parameters,such as main beam diagonal brace fulcrum,diagonal brace angle and stiffness,pier height,are systematically studied.The results show that,the fulcrum of main beam diagonal brace has a great influence on the distribution of the structural bending moment.The fulcrum location should be designed according to the bearing pressure principle of the main beam bending diagonal brace.The length ratio of the upper chord beam segment to the midspan beam segment should be 0.41-0.53.When the location of the main beam diagonal brace fulcrum is reasonable,the change of diagonal brace stiffness and angle has little impact on the bending moment,but the angle of diagonal brace directly determines the axial force of the upper chord beam and diagonal brace.Considering the engineering economy,the angle of diagonal brace should be 25°-45°.When the location,the angle and stiffness of diagonal brace fulcrum are reasonable,the ratio of pier height to main span should be greater than 1/6.
In order to study the influences of intercity rail transit periodical ticket on commuters and its comprehensive effects on passenger flow and fare revenue,the relationship between the maximum number of available times K and discount coefficient β of periodical ticket within validity period and the average one-way ticket price for passengers is first analyzed.Then,a multinomial Logit model based on passenger travel frequency is established,the price elasticity coefficient of passenger flow and fare revenue under periodical ticket is introduced,and an optimal objective function of periodical ticket that consider both passenger flow and fare revenue of operators is set up.Finally,a case is designed to discuss the comprehensive effects of periodical ticket and its optimal strategy.The results show that with the increase of K,the influence of periodical ticket on the share rate of commuters with low frequency in intercity rail transit decreases,while the influence of that on the share rate of commuters with medium or high frequency increases first and then decreases;with the increase of β,the share rate of commuters shows continuous decrease.The implementation of periodical ticket system reduces the fare revenue of the operators,but increases passenger turnover.When the passengers with high travel frequency in the corridor account for a relatively large proportion,the implementation of periodical ticket is more effective.In three scenarios dominated by commuters with high,medium and low frequency,the objective function value under the optimal periodical ticket strategy is 2.20％,2.17％and 1.96％higher than that under the current periodical ticket strategy respectively.The effect of periodical ticket is better when the commuting distance is relatively long.For example,when the commuting distance is 50 km,the implementation of periodical ticket reduces the objective function value,while when the commuting distance is 90 km,it significantly increases the objective function value.
Crack detection and width identification is an important basis for the of maintenance and repair operations of ballastless track slab.To this end,a crack width measurement method based on improved Faster R-CNN and orthogonal projection is proposed.Based on the virtual model synthesis data,the depth network is fully trained to realize accurate detection of surface cracks on ballastless track slab in complex background in order to improve the reliability of crack geometric feature quantization.Firstly,a parametric 3D BIM model of ballastless track structure is established based on 2D CAD drawings,and a random fusion of real crack features and virtual track model and the rendering of real inspection scene are realized through UE5 physics engine.Then,the virtual camera output is configured to simulate virtual crack images of real inspection scenarios;the improved Faster R-CNN network is adequately trained and tested on the original images captured by the track inspection vehicle.Finally,the width of the cracks in the detection results is calculated pixel by pixel using the orthogonal projection method and compared with the manual point-taking measurement results.The results show that the average precision of the improved Faster R-CNN network for crack detection is increased by approximately 10％.The network performance varies with the proportion of virtual and real images of the training data and reaches saturation at 4:1,with an average precision of 95.12％.In addition,the network trained with the fusion crack dataset can achieve higher recall while maintaining high accuracy,effectively reducing the missing detection of cracks.Compared with manual measurement,the minimum and maximum crack widths measured by the orthogonal projection method are increased by 3.64％and 22.40％respectively,and the measurement results are more stable and are close to the real value,which have higher reliability.
