In recent years, the pace of bridge construction in China has been steadily accelerating, with the scale of projects and technical level reaching world-leading standards. The high-quality and innovative development of bridges is an important starting point and basic prerequisite for building China into a country with strong transportation network. To further enhance the strength of the bridge engineering discipline in China, promote the high-quality development of green low-carbon, sustainable and intelligent bridge engineering in China, and support the construction of a transportation powerhouse, this review, based on the analysis of the current industry development status and trends, systematically summarizes the latest scientific and technological innovation achievements in the field of bridge engineering in China in recent years and comprehensively sorts out the future key development directions covering four major themes: bridge engineering structural design and system innovation, disaster prevention and mitigation and structural safety, green construction and intelligent construction, and healthy operation and maintenance and longevity assurance. Specifically, it covers 21 hot research directions, including bridge function and analysis, high-performance materials, steel bridges and composite structure bridges, long-span bridge structures, innovative bridge foundation structures, new progress in bridge seismic resilience research, bridge wind, fire, and blast resistance, bridge engineering collision and protection, water resistance and resilience, multi-hazard coupling of bridges, high-quality bridge construction, green construction technologies and construction technologies, bridge monitoring and assessment, intelligent detection, on-bridge traffic safety, bridge life extension technologies, and integrated construction and maintenance platforms. The review provides guidance and reference for the development of the bridge engineering discipline in China and offers new perspectives and basic materials for researchers and technicians in the field of bridge engineering.

Given the substantial volume and significant aging of existing asphalt pavement structures, extending their service lives has become a strategic requirement for the sustainable development of the transportation industry. This study focuses on existing asphalt pavement structures in long-term service and provides a comprehensive review of the research progress in four key technologies for extending service life: health condition detection and assessment, performance evolution and life prediction, life extension design methods, and research and development of specialized structures and targeted materials. First, the current applications of nondestructive testing techniques, including ground-penetrating radar, are presented to evaluate the internal conditions and bearing capacities of pavement structures. Second, the advantages and limitations of full-scale testing, the finite element method, and the discrete element method for studying the mechanical performance responses and degradation of pavement structures are systematically summarized. Based on these results, design strategies and parameter acquisition methods for extending the life of existing asphalt pavement structures are reviewed. This paper also summarizes the verification models for fatigue and permanent deformation as well as life-extending structural combinations. Finally, life-extension materials and technologies for different pavement layers and functional requirements are summarized. Future research is recommended to integrate comprehensive evaluation indicators, analyze the performance degradation characteristics of existing asphalt pavement structures, establish remaining life prediction models and life extension design methods, and develop high-performance strengthening materials to extend the service lives of asphalt pavement structures.

To investigate the issue of arch expansion on semi-rigid base asphalt pavements in northwest China, the mechanical behavior of asphalt pavements under actual service conditions was analyzed via practical investigation and numerical simulation. A method based on temperature-stress was proposed to assess the degree of arch expansion. First, the temperature field of typical semi-rigid base-layer asphalt pavements in Xinjiang was measured. The spatial and temporal distribution characteristics of temperature at different pavement depths were analyzed, and a pavement temperature-field estimation model was established to determine the temperature load under actual service conditions. Second, a pavement thermal and mechanical coupled numerical model was established to analyze the effects of temperature difference, base-layer modulus, and expansion coefficient on the temperature-stress and vertical deformation of the asphalt pavement. Finally, based on in-service pavement projects in the Xinjiang region, a field investigation on the base layer condition and arch expansion was conducted. The relationship between the arch-expansion degree and the temperature-stress on the base layer was established, and a categorization standard for the degree of arch-expansion development was determined. The results show a significant phase difference in the temperature distribution at different pavement depths in Xinjiang, and that the daily temperature fluctuation becomes flatter as the depth increases. The differences in the temperature gradient and temperature change rate result in different stress states in different structural layers at various moments. Within 30 cm below the road surface, the structural layer of the pavement experiences alternating tensile and compressive forces throughout the day, and the base layer is more susceptible to stress changes, thus resulting in more frequent arch-expansion deformation on the base layer. The degree of arch-expansion development increases with the modulus of the base layer and can be categorized into four grades: no arching, slight arching, medium arching, and severe arching. The results of this study can provide a reference for the nondestructive assessment of arch expansion and the design of pavement anti-arch expansion.

To develop and verify the design principles and construction technology of long-life asphalt pavement, this study conducted 100 million accelerated loading tests and long-term, high-frequency tracking observations and experimental research of the full-scale test track (RIOHTrack). First, a set of “SIHUA” construction technologies with high engineering reliability and wide applicability, which can significantly improve the durability of asphalt pavement, is proposed and verified, namely “structural design functionalization, material design balancing, construction technology homogenization, and quality control proceduralization.” Then, five essential performance evolution laws of asphalt pavement in a wide range of base stiffness are revealed: the coupling effect of load and environment,structural dependence of service performance, inflection point of performance evolution and inverse “S”-shaped trend, the dual damage of pavement T-D cracking and rutting, and the negative thixotropic effect of FWD deflection. Based on the dual-life standards of the long-life pavement including the safety life of structure and functional life of the road surface, as well as the bidirectional fatigue damage mode, a structural system of long-life asphalt pavement in a wide range of base stiffness is established, and the corresponding design principles and indexes are clarified. Taking the structural rigidity and structural modulus as the core, a back-calculation method for the structural layer modulus of the multi-layer system is proposed. This method is based on the FWD deflection basin back-calculation of the four-layer system, and the principles are structural rigidity conservation and uniqueness of the structural layer rigidity response. Finally, based on the inverse “S” evolution trend of service performance, an external factor evolution model is constructed, where the structural rigidity is used as the internal factor index, and the structural modulus is used to determine the mechanical index to solve the structural dependence problem of service performance. As a result, a new model of structural performance evolution that is suitable for the structure in a wide range of base stiffness is proposed, and a design system for long-life asphalt pavement is initially formed.

With the large-scale maintenance period of Chinese roads, the service time of a large number of roads has entered the late stage of design life, but the existing maintenance system still lacks the standards of highway life extension, and the current life evaluation index is mainly for newly built roads. Taking into account the deterioration of road conditions due to environmental factors and load during service, a long-life evaluation method is proposed. This method combines the pavement's intrinsic health condition with external service environment factors, along with corresponding index thresholds linked to life expectancy. A road health index (H_PHI) was constructed to characterize the structural health level of the road using principal component analysis(PCA)The Deflection Basin Parameters (DBP) were used to include: the Surface Curvature Index (SCI), the Base Damage Index (BDI), the Base Curvature Index (BCI), and the Condition Index (CI).Based on the natural zoning of China's highways for wet coefficient (K), standard freezing depth (H₀), and freezing index (F) characterization of the natural environment, combined with the traffic load level, the jointly constructed environmental harsh index (E_EHI) describes the severity of the external service environment of the road. Based on H_PHI and E_EHI, the road health deterioration rate (R_HDR) was constructed as an evaluation index for the long life of roads in service. Preliminary research results show that H_PHI and R_HDR can jointly determine the life stage in the three-stage life theory. In the second life stage, R_HDR less than 0.2 is the long-life threshold for a road life of 30 years, less than 0.7 is the standard threshold for a road life of 15 years, and when the R_HDR increases suddenly and continuously, it characterizes the road life into the third stage as the warning threshold. In the theoretical domain, the proposed long-life evaluation index translates the abstract concept of road lifespan into a concrete and practical index. It converts the three-stage road theory into an actionable index, enabling long-term monitoring and accurate lifespan prediction of road conditions, thereby offering a robust theoretical foundation for analyzing the longevity of roads in service.

The aging of reclaimed asphalt under environmental climate conditions is the main internal factor contributing to the performance degradation of regenerated pavements. Accurately characterizing and predicting the aging characteristics of recycled asphalt with respect to temporal and spatial variations is of great significance for assessing the aging degree of asphalt and the service condition of pavements. Therefore, this study mainly investigated the variation rules of rubberized reclaimed asphalt in both time and space dimensions under accelerated aging of artificial climate. Specifically, the aging characteristics of asphalt extracted from an aged rubberized reclaimed asphalt mixture in layers were investigated. The spatiotemporal variation rules of aging characteristics and the effects of material composition were clarified through dynamic shear, multiple creep recovery, and Fourier transform infrared spectroscopy tests. Then, a spatiotemporal evolution model based on Fick's second law was established, and aging prediction based on this model was analyzed. The results indicate that the evolution of aging characteristic parameters, such as rutting factor, the reciprocal of the non-recoverable creep compliance, recovery rate, carbonyl index, aromatic ring index, and sulfoxide index, have an opposite trend in both the time and space dimensions, with the greatest reduction at 0-12.5 mm in depth, gradually approaching 0 thereafter. The incorporation of rubber powder, hard matrix asphalt, and reclaimed asphalt pavement (RAP) can increase the rutting factor, reciprocal of unrecoverable creep compliance, and recovery rate of rubberized reclaimed asphalt by at least 20%, while the addition of rubber powder has no significant effect on the functional group index of reclaimed asphalt. The spatiotemporal model effectively reflects the evolution of aging characteristic parameters, with the model's coefficient of determination mostly above 0.9. Based on the spatiotemporal evolution model, the predicted results show that when the relationship between the carbonyl index and time follows a square root function at the pavement surface, the milling depths for recycled pavement after 5 and 10 years of service are 0.027 m and 0.086 m, respectively. These results provide valuable references for determining maintenance periods and milling depths during pavement management and rehabilitation.

Deep learning combined with ground penetrating radar (GPR) imaging has become a popular research area for detecting subsurface distress in roads. However, due to the difficulty in obtaining ground truth samples, improving the accuracy of subsurface distress detection under small-sample conditions remains challenging. To address this issue, this paper proposes a self-adaptive curriculum learning model framework to enhance the detection accuracy of subsurface distress. The framework follows a progressive training strategy from easy to difficult, which can dynamically evaluate the difficulty of sample data and adaptively adjust the training order of data. Firstly, a teacher-student adaptive curriculum learning framework is constructed, which alternately optimizes the sample difficulty evaluation mechanism of the teacher network and subsurface distress detection performance of the student network. A soft-boundary constrained loss function is introduced to dynamically evaluate the difficulty of samples based on the prediction error of the detection model, and the adaptive scheduling of sample training order is realized by the backpropagation of weighted loss. To adapt to different scales and types of detection models, the EfficientNet series network is used to construct the feature extraction module. Based on actual engineering data, a dataset containing three types of road subsurface distress-reflective cracks, loose, and voids-is constructed with a total of 1 857 samples. Four strategies are designed in the experiment: adaptive curriculum learning, random training, specified order, and specified reverse order, which are verified by two subsurface distress detection algorithms, YOLOv9 and DETR. The results show that starting training from simple data samples and gradually increasing the difficulty of samples can effectively accelerate model convergence and improve detection accuracy. Compared with the other three methods, the proposed adaptive curriculum learning framework improved the detection accuracy of subsurface distress by 6.21%, 4.01%, and 10.64%, respectively. By comparing the difficulty evaluation of each sample, it is found that the detection difficulty of subsurface distress in urban road scenes is greater, but there is no significant difference in the difficulty level between different subsurface distress types. This study provides a new technical approach for the detection of road subsurface distress, which will help to improve the accuracy and efficiency of detection.

To improve the accuracy and reliability of the existing rutting prediction model, based on the RIOHTrack full-scale loop accelerated loading test, the rutting prediction model in the current code in China was corrected, and a model correction method is proposed to improve the reliability of the model and make it more suitable for rutting estimation in the region. It is found that the rutting model in the design code of asphalt pavement in China has a significant structural dependence, and the model has the highest estimation accuracy and smallest error for the semi-rigid base asphalt pavement structure with an asphalt concrete layer thickness of 12 cm, while the estimation accuracy of other structures is not high. To improve the accuracy and reliability of the rutting prediction model, a new model was established by introducing local correction coefficients into the existing model. After local correction, the accuracy of the rutting prediction model for all structures was greatly improved: the MAPE was reduced from 29.04% to 7.81% before correction for a single structure, and the coefficient of determination R² was greater than 0.82; the MAPE was lower (6.34% - 7.99%) and the coefficient of determination R² is above 0.87 when the correction is classified according to seven major structures. Based on the characteristics of different pavement structures and materials, as well as the influence of local climate and traffic loads, the modified model is more suitable for the rutting prediction of various pavement structures in the RIOHTrack area.

Coplanar array capacitance imaging technology acquires the permittivity distribution of tested objects by detecting variations in capacitance, providing a reliable, non-destructive method for detecting and identifying hidden damages in asphalt pavements. In this paper, a set of sensitivity field optimization strategies are proposed for obtaining a high imaging accuracy, approximate real, and uniform-sensitivity field distribution. This optimized sensitivity field is used for reconstructing the distribution images of hidden damages in asphalt mixtures. Firstly, the multi-layer sensitivity fields are fused using the wavelet transform to obtain the fused sensitivity field. Secondly, according to the measured dynamic coplanar capacitance distributions, the threshold optimization method is used to extract the feature sensitivity fields of different scanning steps, and then the real sensitive fields are obtained. Finally, based on non-local mean filtering, the sharp areas of the real sensitivity fields are smoothed to obtain homogenized sensitive fields. The hidden damages in asphalt mixtures is imaged based on the fusion, feature, and homogenized sensitivity fields. The fused sensitivity field distribution with clear contours is obtained by wavelet transform; the real sensitivity field, which can reflect the distribution of hidden damages is obtained based on the feature sensitivity field extracted by capacitance contribution; and the homogenized sensitivity field with uniform local distribution is obtained based on non-homogeneous mean filtering. The results show that the stability of the edges of the damages in the reconstructed images based on the fused sensitivity field is significantly improved; the artifacts around the damages in the reconstructed images based on the feature sensitivity field are basically eliminated; and the internal distribution of damages in the reconstructed images based on the homogenized sensitivity field is more uniform. From high to low, the degree of imaging quality improvement based on different sensitive field optimization methods isas follows feature sensitivity field> homogenized sensitivity field> fused sensitivity field.

At present, the research on the fatigue characteristics and damage evolution laws of asphalt mixtures in China and abroad is not sufficiently deep, resulting in inaccurate evaluation of the remaining life of structural layers. Therefore, dynamic load strength tests of asphalt mixtures under different temperatures and stress states were conducted, and the applicability of maximum tensile stress, Tresca, Mises, Drucker-Prager, Bresler-Pister, and simplified Desai strength models were analyzed based on this. The results show that the Bresler-Pister elliptical strength envelope and simplified Desai hexagonal envelope are in good agreement with the tensile, compressive, and splitting strength test points. Based on this, considering the effects of temperature and loading speed, the strength envelopes were obtained. Then, the fatigue life of the mixture was measured at different temperatures and stress levels. The specimens with loading times of 20%, 50%, 65%, and 80% fatigue life were selected for tensile, compressive, and splitting residual strength tests. Based on this, the Bresler-Pister and simplified Desai envelope were used for the first time to characterize the three-dimensional strength decay law. Furthermore, by defining the residual three-dimensional strength, loading times ratio, and other parameter groups, the internal unity of three-dimensional strength failure and fatigue failure was realized based on the damage evolution model established by the preferred Bresler-Pister criterion. The results of this study provide an important experimental and theoretical reference for predicting the residual strength of asphalt layers with different service lives, as well as fatigue load control in life extension design.

The digitalization of road engineering faces problems such as weak digital foundation, incomplete information acquisition methods, insufficient data sharing, unbalanced development, and insufficient business collaboration. Modern digital technologies, such as big data, cloud platforms, and mobile edge computing networks, can be integrated into the entire process of road design, construction, maintenance, and operation. An important strategy is to improve the levels of informatization and digitalization in the road engineering field and to empower the high-quality development of transportation. Based on this, digital technologies in the processes of smart road design, construction, maintenance, and operation were reviewed. The applications and advantages of digital technology in the process of smart road design, including road surveys, roadbeds, and pavement design, were summarized. The current research status of digital technology in road construction was summarized, mainly involving the intelligent control and precise management of various procedures in roadbeds and pavement construction. Strategies that use digital technology to improve vehicle-road coordination level, traffic management efficiency, and road traffic efficiency were introduced. Methods for improving road condition measurement efficiency and pavement maintenance strategies were presented to realize digitalization updates during road maintenance procedures. The development conditions and trends of digital technology applied to smart road design, construction, maintenance, and operation were summarized from multiple perspectives. Related work could provide guidance and suggestions for improving the intelligence level of road design, construction, maintenance, and operation and promote the high-quality development of the transportation industry.

As critical transportation infrastructures, roads are being developed to include precise pavement condition sensing, coordinated traffic management, and intelligent user-interactive services. Pavement dynamic response monitoring through embedded sensors offers real-time continuous data collection on traffic loads and pavement conditions, effectively minimizing interference from human and environmental factors. This is essential for the development of intelligent highways. However, challenges remain in embedded sensor installation techniques, such as durability and efficient processing of vast amounts of in situ monitoring data. This study introduces an innovative sensor-embedding method integrated with slipform paving for cement pavements, thereby enabling seamless sensor installation during pavement paving. Field tests were conducted using the advanced empirical mode decomposition method and continuous wavelet transform technique to analyze the time-frequency characteristics of the vibration signals at various vehicle speeds from an energy perspective. Our approach demonstrated high accuracy, with speed estimation errors consistently below 8%. This significant improvement in vibration-based speed estimation enhances the traffic monitoring capabilities of rigid pavements. By addressing the challenges of embedded sensor integration and signal processing, this study contributes to the advancement of more efficient and responsive smart highway technologies.

The skid resistance of pavement surfaces is influenced by both macro-texture and micro-texture.To achieve high-resolution 3D pavement texture acquisition using vehicle-mounted laser scanning equipment and enable continuous, non-contact microscale evaluation of pavement skid resistance, this study constructed a super-resolution network model based on self-supervised deep learning. The model recursively reconstructed low-resolution pavement textures to a 0.1 mm·pixel^-1 resolution in the driving direction. A total of 527 SMA-13 asphalt pavement textures with a resolution of 0.1 mm were prepared for training and testing. The nearest-neighbor method was used to downsample the textures by factors of 1/2, 1/4, 1/8, and 1/16 to simulate low-resolution textures captured at different vehicle speeds. Texture pairs with a two-fold resolution difference were used as input for the network model, and the dataset was randomly split into training and test sets in an 8∶2 ratio. The training set was augmented by small-scale segmentation. Under optimal weights, the model recursively reconstructed various low-resolution pavement textures to a 0.1 mm·pixel^-1 resolution. The reconstruction quality of the super-resolution textures was evaluated using Peak Signal-to-Noise Ratio (PSNR), Structural Similarity (SSIM), and the relative error of mean texture depth, with bicubic interpolation as a comparison. By integrating bicubic interpolation with the proposed super-resolution network, this study investigated the reconstruction performance of low-resolution pavement textures under non-predefined downsampling factors. Finally, using the OOA-LightGBM algorithm, feature parameters of the high-resolution and super-resolution reconstructed textures were extracted to build a skid resistance prediction model, exploring the feasibility of predicting skid resistance based on super-resolution reconstructed textures. Results show that the proposed super-resolution network enables the vehicle-mounted laser scanning device to collect 3D pavement textures with a resolution of 0.1 mm, a PSNR greater than 30 dB, an SSIM above 0.95, and an absolute relative error in mean texture depth (MTD) of less than 1%, within a speed of 128 km·h^-1. For pavement textures collected at constant speeds of 16 km·h^-1, 32 km·h^-1, and 64 km·h^-1, the average coefficient of determination (R²) for skid resistance prediction was 0.808.

Accurately predicting the potential changes in the control behavior of nearby vehicles is crucial for autonomous vehicles (AVs) to understand their driving environments, prevent safety hazards, and enhance lane-changing safety. Existing lane-changing trajectory-planning methods do not sufficiently consider the uncertainty of nearby vehicle control behaviors, which is particularly prominent in mixed traffic environments that include human-driven vehicles (HVs). Hence, a dynamic lane-changing trajectory-planning method that considers the interaction between HVs and AVs was proposed to improve the adaptability and safety of AV lane-changing behaviors. This method comprises three key models. First, a driving-style-identification model based on the LightGBM algorithm was established to accurately identify the driving styles of nearby vehicle drivers. Subsequently, the full velocity-difference model was used to simulate and analyze the dynamic control feedback of nearby vehicles under different driving styles during the lane-changing process of the AV. Finally, considering the interaction effects between an HV and AV, lane-changing trajectories and acceleration curves were dynamically generated to ensure the safety and smoothness of the operations. The results show that this method can effectively adjust lane-changing strategies based on real-time traffic conditions, thereby mitigating collision risks. Compared with conventional lane-changing trajectory planning methods, the proposed method demonstrates higher safety and adaptability in actual lane-changing scenarios.

Manholes are important ancillary facilities on urban roads and are prone to subsidence, tilting, and damage under the action of increasing traffic loads. Additionally, the pavement around manholes experiences various types of distress such as cracking, settlement, and loosening. When vehicles pass through distressed manhole covers-surrounding pavement, a decline in driving safety and comfort is noticeable and noise pollution is generated, which negatively affects the road performance and quality of life of neighboring residents. For these reasons, by collecting and sorting the relevant research results, the inducing mechanisms of distress for manhole covers-surrounding pavement were analyzed based on experimental, theoretical, and simulation methods. The development status of distress treatment technology was reviewed from the perspectives of structural optimization, material selection, and equipment innovation. Combined with the concept of intelligent transportation system construction, the achievements and limitations of the research on intelligent detection and supervision technology of manhole cover-surrounding pavement distress were evaluated, and the development trends in this field were investigated. The results show that the distress rate of the manhole cover-surrounding pavement is as high as 58.9% under dynamic driving loads, and the main distress types encompass structural settlement and pavement cracking. Specifically, the former primarily occurs within 3 cm, whereas the latter is concentrated in an area of 15-40 cm around the manholes. Various factors contribute to these distresses, including a crushed mortar cushion under the manhole cover, lack of manhole rings, substandard compaction of the subgrade and pavement, and low bearing capacity of the manhole cover. Successful methods have been proposed for effective distress delay and treatment, such as optimizing the structure of the manhole cover-surrounding pavement, improving the properties of materials used in the manhole cover cushion and surrounding pavement repair, and refining the approach leveling for manhole covers. The incorporation of data collected from mobile mapping systems, unmanned aerial vehicles, smartphones, and other equipment into deep learning algorithms allows for the automatic and efficient identification of manhole covers, the surrounding pavements, and their distress. Through technologies such as joint information processing, network transmission, and perception recognition, a relatively complete intelligent supervision system can be built to realize an integrated management of the construction, management, and maintenance for manhole covers-surrounding pavement.

Accurate and lightweight pavement distress detection can effectively reduce the hardware requirements of pavement inspection equipment and improve road inspection efficiency. This is crucial for high-quality development of smart road maintenance management. To address the issues of low accuracy in existing pavement distress detection methods, a lightweight road-disease detection method based on YOLOM combined with a state-space model (SSM) is proposed. First, based on the training characteristics of road disease detection tasks, a multiscan visual Mamba layer was designed, and the normalization methods were adjusted to quickly extract image features more suitable for small training batches. In addition, parallel computing units were added to accelerate the network computation speed and reduce the training time of the algorithm. Second, based on YOLOv9, the Mamba efficient layer aggregation network (MELAN) with SSM as the core mechanism and spatial pyramid MELAN (SPMELAN) were designed by integrating the attention hiding mechanism of SSM. The long-distance dependencies of the images were extracted using MELAN, and the global features of the disease images were mined. SPMELAN was used to fuse multiscale receptive field information and enhance both the full-size coverage capability of the receptive field of the high target detector and the model's ability to capture local details and global semantics. Consequently, YOLOM combined with SSM was proposed. Finally, comparison experiments were conducted using RDD2022. The results show that the F₁-score, mAP50, mAP50-95, and FLOPs of YOLOM are better than those of YOLOv9c. In comparison with the baseline model checking results, YOLOM achieves the highest detection accuracy and inference speed, with the smallest model size and complexity. YOLOM has significant advantages including low weight, low complexity, high detection accuracy, strong learning, and generalization capabilities, which can assist in intelligent road detection.

The timely and accurate detection of pavement cracks is crucial for extending the road service life and ensuring safe traffic. In response to the problems of segmentation discontinuity and false detection caused by weak cracks and background noise interference in existing methods, this study focuses on crack completeness and proposes a pavement crack detection method based on lightweight large kernels and low information waste to achieve efficient and accurate crack segmentation in a complex environment. First, a feature encoding structure based on multi-size large-kernel fusion was designed, which constructed a multiscale global receptive field to fully capture long-distance spatial dependencies for large-span pavement cracks. Second, innovative mechanisms such as importance-based pooling with forward-reverse complementary attention, diverse-branch decoding with training-inference decoupling, and hierarchical deep supervision with a reinforced misdetection penalty were constructed. These mechanisms aimed at alleviating the excessive information waste of crack details during both the encoding and decoding processes and improving the ability of the proposed model to capture and interpret fine-grained crack features. Further, this study combined structural reparameterization, partial convolution, and depthwise convolution to ensure effective control over the parameters while maintaining a highly expressive feature representation. Meanwhile, this strategy achieved an ultra-lightweight large-kernel structure and efficient model inference. The experimental results demonstrate that the proposed method can accurately segment pavement cracks in complex backgrounds with high crack completeness and low misdetection rates. In addition, it exhibits a lightweight parameter magnitude and real-time computational speed. The standard version of the proposed method can achieve an intersection-over-union rate of 77.17%, an F₁ score of 87.11%, and a recall rate of 87.41% for cracks using 3.67×10⁶ parameters. The mini-version achieves an intersection-over-union rate of 75.23%, an F₁ score of 85.63%, and a recall rate of 87.12% for cracks using only ultra-light 0.29×10⁶ parameters. Moreover, the proposed method demonstrates good adaptation to road conditions and high robustness to background noise while controlling the quantity deviation of the connected components at an extremely low level. Compared with the current mainstream methods, the proposed method exhibits a significant advantage in pavement crack detection tasks and can strongly support the monitoring, evaluation, and repair of pavement defects.

Real-time perception, accurate diagnosis, and evaluation of the service state of asphalt pavements provide the basis for the implementation of their active scientific maintenance, with embedded wireless sensing being the frontier technology in real-time pavement perception. Based on the road wireless self-powered smart aggregate (SmartRock), beam specimens of the SmartRock-embedded asphalt mixture were prepared with different crack depths, widths, and locations (horizontal distance of the crack from the center of the span). Repeated loading tests at different load levels were conducted under three-point bending using mechanics test systems (MTS). The piezoelectric output of SmartRock was tested under dynamic load excitation. The effects of load level and crack width, depth, and location on the piezoelectric output of SmartRock were studied. A size perception and positioning model for hidden cracks in asphalt pavement was constructed, and intelligent size sensing and intelligent positioning technologies for hidden cracks were proposed and verified. The results show that the peak value of the SmartRock piezoelectric output decreases with an increase in crack depth and width, and increases with an increase in the crack position offset distance. The peak value of the SmartRock piezoelectric output is sensitive to crack depth, width and position, indicating that the SmartRock piezoelectric output peak can be used to perceive hidden cracks. A hidden crack size perception model was established, and a crack size perception technology was proposed based on multi-objective programming. The perception accuracies of crack depth and width are 92.8% and 95.3%, respectively. A crack localization model of a hidden crack was established, and crack localization technology was proposed based on the three-point localization theory, with a localization accuracy of 95.1%. This achievement provides a feasible technology for the intelligent high-precision identification and precise positioning of hidden cracks in asphalt pavements.

The development of pavement crack segmentation technology is crucial for assessing the safety and durability of civil infrastructure. However, accurately segmenting cracks of irregular shapes in complex and dynamic background environments remains a challenging task. To improve the segmentation performance, we propose a CNN and scale adaptive fusion network-based pavement crack segmentation method. Specifically, for the dual-encoder based on CNN and Transformer, we utilized scale-adaptive Transformer blocks, which integrate scale-adaptive multi-head attention and a detail-enhanced feed-forward network to effectively capture multi-scale features and enhance detail information. Additionally, we employed a global-local feature fusion module to aggregate the intermediate features from the middle layers of the dual-encoder. For the decoder, we designed a large kernel dual-attention module to enhance the detailed boundaries and mitigate the influence of background noise, achieving highly accurate crack segmentation. Finally, we combined the cross-entropy segmentation and Dice losses to optimize the network training process. We conducted comprehensive comparison and ablation experiments on the DeepCrack, Crack500, and CFTR478 datasets to demonstrate the effectiveness of the proposed method. The experimental results show that our method is superior to other methods and outperforms DTrc-Net and FAT-Net on the CFTR478 validation set by 1.58% and 1.82% mIoU, respectively. Furthermore, in complex scenes with low-light, rainy, and slippery conditions and roads with different materials, our method can still effectively identify and accurately segment the crack regions, maintaining clear boundaries. Moreover, our method is applicable to pavement crack segmentation in real campus scenarios, obtaining high-quality segmentations of pavement cracks, and has good practical application prospects.