Ground penetrating radar road layering defect identification system based on deep learning

Abstract

The invention relates to the technical field of road defect recognition, in particular to a ground penetrating radar road layering defect recognition system based on deep learning, which comprises the following steps: analyzing the material compatibility poor degree of a filling material and an original road material based on road maintenance material attribute data and maintenance process data; the maintenance quality control risk in the maintenance process is analyzed in combination with the maintenance process data and the ground penetrating radar road hierarchical data; based on a material compatibility poor degree analysis result and a maintenance quality control risk analysis result, evaluating a pavement defect risk after maintenance; performing early warning on the road maintenance quality according to the road surface defect risk assessment result after maintenance; the reliability of road maintenance quality and road service safety can be effectively guaranteed, later-stage diseases caused by poor material compatibility and construction quality defects are reduced, and the safety accident risk is reduced.

Description

Technical Field

[0001] This invention relates to the field of road defect recognition technology, and in particular to a deep learning-based ground-penetrating radar road layer defect recognition system. Background Technology

[0002] As the core of the transportation system, the structural health of roads directly affects driving safety, operational efficiency, and maintenance costs. Delamination defects between the road surface and base layers, a type of highly concealed and hazardous condition, are a key focus and challenge in road maintenance management. These defects often remain hidden within the structure in their early stages, making them difficult to detect through surface observation. However, under continuous vehicle loads, they gradually develop into reflective cracks, potholes, and even localized collapses, significantly shortening road lifespan and posing a serious threat to road traffic safety. Traditional road inspection methods primarily rely on manual core drilling and conventional non-destructive testing techniques. The former is destructive, inefficient, and has limited representativeness, making large-scale surveys difficult. The latter, such as deflectometers, while indirectly reflecting overall load-bearing capacity, lacks sufficient accuracy and spatial positioning capabilities for identifying delamination defects such as interface voids and loosening, failing to meet the demands of modern precision road maintenance. Ground-penetrating radar (GPR) technology, with its high sensitivity to differences in dielectric properties, can non-destructively and rapidly acquire continuous cross-sectional images of the road's internal structure, providing a powerful technical means for identifying delamination defects.

[0003] However, current ground-penetrating radar (GPR) technology for road inspection typically stops at identifying and locating abnormal reflection signals in radar images. It fails to deeply correlate the detected layered defects with key process parameters and material properties during road maintenance. For example, when radar images show poor interface bonding in a repair area, without tracing the compatibility of the filler material with the old pavement and the strict control of compaction parameters, it's difficult to accurately determine the root cause, evolution trend, and actual impact on long-term road performance. This leads to a lack of targeted maintenance decisions, potentially resulting in over- or under-maintenance. Furthermore, current technology lacks the ability to dynamically predict post-repair pavement defect risks. Its analysis results are often based on static judgments from single-point-in-time radar images, failing to integrate multi-source information to assess the long-term reliability of repair quality. This leaves maintenance management in a passive, reactive mode, hindering risk-based preventative maintenance.

[0004] To address these issues, this application presents a deep learning-based ground-penetrating radar road layer defect identification system. Summary of the Invention

[0005] The purpose of this invention is to provide a deep learning-based ground-penetrating radar (GPR) road layer defect identification system. This system acquires GPR road layer data, road repair material attribute data, and repair process data for the repair area; then constructs a material incompatibility analysis model and a repair quality control risk analysis model; subsequently, based on the output results of the material incompatibility analysis model and the repair quality control risk analysis model, it constructs a defect risk assessment model to quantitatively assess the risk of road surface defects after repair. This ensures the reliability of road repair quality and the safety of road service, reduces subsequent damage caused by material incompatibility and construction quality defects, and lowers the risk of safety accidents.

[0006] This invention is implemented as follows: In a first aspect, this invention provides a deep learning-based ground-penetrating radar (GPR) road layering defect identification system, comprising a multi-source data integration module, a compatibility assessment module, a process quality correlation module, a risk quantification assessment module, and an intelligent decision-making and early warning module. The multi-source data integration module acquires GPR road layering data of the repair area during road maintenance, and simultaneously acquires road maintenance material attribute data and maintenance process data of the repair area. The compatibility assessment module constructs a material compatibility analysis model based on the road maintenance material attribute data and maintenance process data to analyze the degree of material incompatibility between the fill material and the original road material. The process quality correlation module combines maintenance process data and GPR road layering data to construct a maintenance quality control risk analysis model to analyze maintenance quality control risks during the maintenance process. The risk quantification assessment module constructs a post-repair pavement defect risk assessment model based on the material compatibility analysis results and the maintenance quality control risk analysis results to assess the post-repair pavement defect risk. The intelligent decision-making and early warning module provides early warnings regarding road maintenance quality based on the post-repair pavement defect risk assessment results.

[0007] As a preferred embodiment of the present invention, a material compatibility analysis model is constructed to analyze the degree of material incompatibility between the fill material and the original road material, including the following specific steps: S21. Extract the thermal expansion coefficients of the filler material and the original road material from the road maintenance material property data; and extract the construction ambient temperature and the standard applicable temperature range of the material from the maintenance process data. S22. Subtract the thermal expansion coefficient of the filling material from the thermal expansion coefficient of the original road material and take the absolute value as the absolute difference in thermal expansion coefficients. S23. Divide the absolute difference in thermal expansion coefficients with the thermal expansion coefficients of the original road material to obtain the thermal expansion coefficient mismatch of the filling material. S24. Take the difference between the ambient temperature of the construction environment and the median of the applicable temperature range of the material standard as the temperature deviation difference. Divide the absolute value of the temperature deviation difference by the length of the applicable temperature range of the material standard to obtain the deviation of the construction temperature condition of the filling material in the road maintenance process. S25. The arithmetic mean of the thermal expansion coefficient mismatch and the deviation of the construction temperature condition is used to obtain the poor material thermal compatibility level of the filling material during road maintenance.

[0008] As a preferred embodiment of the present invention, a material compatibility analysis model is constructed to analyze the degree of material incompatibility between the fill material and the original road material, and the following specific steps are also included: S26. Extract the ratio of the elastic modulus of the filler material to that of the original road material from the road maintenance material property data, and extract the ratio of the actual compaction degree to the standard compaction degree from the maintenance process data; calculate the inverse of the ratio of the actual compaction degree to the standard compaction degree to obtain the compaction insufficiency index of the filler material during the road maintenance process. S27. The ratio of the elastic modulus of the filling material to that of the original road material is added to the compaction insufficiency index to obtain the degree of modulus mismatch of the filling material during road maintenance. S28. The arithmetic mean of the material thermal compatibility poorness level and modulus mismatch degree is used as the material compatibility poorness degree of the filling material during road maintenance.

[0009] As a preferred embodiment of the present invention, a maintenance quality control risk analysis model is constructed to analyze the maintenance quality control risks during the maintenance process, including the following specific steps: S31. Extract the number of compaction passes of filling materials and the range of compaction passes required by process specifications from the maintenance process data, and extract the base dielectric constant of the maintenance area and the standard base dielectric constant from the ground penetrating radar road layer data; S32. The absolute difference between the number of compaction passes of the filling material during road maintenance and the midpoint of the compaction pass range required by the process specification shall be taken as the absolute compaction difference value. The absolute compaction difference value shall be divided by the length of the compaction pass range required by the process specification. The result of the division shall be taken as the deviation of the compaction process of the filling material during road maintenance. S33. The ratio of the dielectric constant of the base layer in the repair area to that of the standard base layer is calculated by reciprocal to obtain the base layer compaction anomaly index; the deviation of the compaction process of the filling material during road maintenance is calculated by arithmetic average with the base layer compaction anomaly index, and the result of the arithmetic average is taken as the base layer compaction anomaly in the repair area during road maintenance.

[0010] As a preferred embodiment of the present invention, a maintenance quality control risk analysis model is constructed to analyze the maintenance quality control risks during the maintenance process, and the following specific steps are also included: S34. Extract the actual paving thickness of the filling material at all equidistant road detection points within the maintenance area from the maintenance process data, and extract the amplitude value of the reflected signal at the junction of the filling layer and the old road surface at all equidistant road detection points within the maintenance area from the ground penetrating radar road layer data. S35. Obtain the arithmetic mean of the actual paving thickness of the filling material at all equidistant road inspection points, and calculate the standard deviation of the actual paving thickness of the filling material at all equidistant road inspection points. Divide the standard deviation of the actual paving thickness of the filling material by the arithmetic mean, and use the result of the division as the coefficient of variation of the paving thickness of the filling material in the maintenance area during the road maintenance process. S36. Obtain the arithmetic mean of the reflected signal amplitude values ​​at the junction of the fill layer and the old road surface at all equidistant road detection points, and calculate the standard deviation of the reflected signal amplitude values ​​at the junction of the fill layer and the old road surface at all equidistant road detection points. Divide the standard deviation of the reflected signal amplitude values ​​by the arithmetic mean, and use the result of the division operation as the coefficient of variation of the reflected signal intensity at the junction interface of the repair area during the road maintenance process. S37. The arithmetic mean of the coefficient of variation of the thickness of the filling material and the coefficient of variation of the intensity of the reflected signal at the interface is used to obtain the structural uniformity and abnormality of the maintenance area during the road maintenance process. S38. The arithmetic mean of abnormal compaction and structural uniformity of the base layer in the maintenance area during road maintenance is calculated, and the result of the arithmetic mean is used as the maintenance quality control risk during the maintenance process.

[0011] As a preferred embodiment of the present invention, a post-repair pavement defect risk assessment model is constructed to assess the risk of post-repair pavement defects, including the following specific steps: S41. Extract the analysis results of the material compatibility insufficiency of the filling materials during the road maintenance process, and at the same time extract the analysis results of the maintenance quality control risk during the maintenance process; S42. Perform a geometric mean calculation on the results of the material compatibility poorness analysis and the maintenance quality control risk analysis, and use the result of the geometric mean calculation as the result of the road surface defect risk assessment after maintenance in the maintenance area.

[0012] In a preferred embodiment of the present invention, an early warning is issued for road repair quality based on the risk assessment results of road surface defects after repair, including the following specific contents: Obtain the risk assessment results of road surface defects after repair in the repair area, and preset the emergency maintenance threshold for road surface defects; when the risk assessment results of road surface defects after repair in the repair area are greater than or equal to the emergency maintenance threshold for road surface defects, issue an emergency maintenance warning for the road surface after repair in the repair area; when the risk assessment results of road surface defects after repair in the repair area are less than the emergency maintenance threshold for road surface defects, the repair quality of the repair area is deemed to be qualified.

[0013] Secondly, the present invention provides a deep learning-based ground-penetrating radar method for identifying road layer defects, comprising the following specific steps: During road maintenance, obtain ground-penetrating radar road layering data of the maintenance area, and simultaneously obtain road maintenance material property data and maintenance process data of the maintenance area; Based on road maintenance material property data and maintenance process data, a material compatibility analysis model is constructed to analyze the degree of material incompatibility between the fill material and the original road material; By combining maintenance process data with ground-penetrating radar road stratification data, a maintenance quality control risk analysis model is constructed to analyze the maintenance quality control risks during the maintenance process. Based on the analysis results of poor material compatibility and the risk analysis results of maintenance quality control, a risk assessment model for pavement defects after maintenance is constructed to assess the risk of pavement defects after maintenance. Based on the risk assessment results of road surface defects after repair, early warnings are issued regarding the quality of road repairs.

[0014] Thirdly, the present invention provides an electronic device, comprising: a processor and a memory, wherein the memory stores a computer program that can be called by the processor, and the processor executes a deep learning-based ground-penetrating radar road layer defect identification method by calling the computer program stored in the memory.

[0015] Compared with the prior art, the present invention has the following advantages and beneficial effects: 1. This invention analyzes the degree of material incompatibility between the filling material and the original road material, ensuring the compatibility between the filling material and the original road material, reducing the later road surface defects caused by compatibility issues, and improving the reliability of road surface maintenance quality. 2. This invention analyzes the risks of maintenance quality control during the maintenance process and accurately identifies construction defects, thereby ensuring the quality of road maintenance and reducing the risk of structural damage to the road during its service life. 3. By assessing the risk of road surface defects after repair, this invention can reduce safety accidents caused by the failure to address potential road surface repair hazards in a timely manner, and ensure the long-term service quality and safety of the repaired road surface. Attached Figure Description

[0016] Other features, objects, and advantages of the invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 This is a schematic diagram of the overall process of the deep learning-based ground-penetrating radar road layer defect identification method of the present invention; Figure 2 This is a schematic diagram of the structure of the deep learning-based ground-penetrating radar road layer defect identification system of the present invention; Figure 3 This is a flowchart illustrating step S2 of the deep learning-based ground-penetrating radar road layer defect identification method of the present invention. Detailed Implementation

[0017] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations thereof. In the absence of conflict, the embodiments of the present invention and the technical features in the embodiments can be combined with each other.

[0018] Example 1 like Figure 1 As shown, this embodiment provides a deep learning-based ground-penetrating radar method for identifying road layer defects, including the following specific steps: S1. Obtain ground-penetrating radar road layering data of the maintenance area during road maintenance, and simultaneously obtain road maintenance material property data and maintenance process data of the maintenance area; S2. Based on road maintenance material property data and maintenance process data, construct a material compatibility analysis model to analyze the degree of material incompatibility between the fill material and the original road material; S3. Combining maintenance process data with ground-penetrating radar road stratification data, a maintenance quality control risk analysis model is constructed to analyze the maintenance quality control risks during the maintenance process. S4. Based on the analysis results of poor material compatibility and the risk analysis results of maintenance quality control, construct a risk assessment model for pavement defects after maintenance, and assess the risk of pavement defects after maintenance; S5. Based on the risk assessment results of road surface defects after repair, issue early warnings on the quality of road repair.

[0019] In this embodiment, as Figure 3 As shown, step S2 involves constructing a material compatibility analysis model to analyze the degree of material incompatibility between the fill material and the original road material. This includes the following specific steps: S21. Extract the thermal expansion coefficients of the filler material and the original road material from the road maintenance material property data; and extract the construction ambient temperature and the standard applicable temperature range of the material from the maintenance process data. S22. Subtract the thermal expansion coefficient of the filling material from the thermal expansion coefficient of the original road material and take the absolute value as the absolute difference in thermal expansion coefficients. S23. Divide the absolute difference in thermal expansion coefficients with the thermal expansion coefficients of the original road material to obtain the thermal expansion coefficient mismatch of the filling material. S24. Take the difference between the ambient temperature of the construction environment and the median of the applicable temperature range of the material standard as the temperature deviation difference. Divide the absolute value of the temperature deviation difference by the length of the applicable temperature range of the material standard to obtain the deviation of the construction temperature condition of the filling material in the road maintenance process. S25. The arithmetic mean of the thermal expansion coefficient mismatch and the deviation of the construction temperature condition is used to obtain the poor material thermal compatibility level of the filling material during road maintenance.

[0020] For example, this embodiment pre-quantifies the risk level of incompatibility between the repair material and the original road structure due to temperature changes, from the perspective of the compatibility between the material's thermophysical properties and the construction environment temperature. First, the coefficient of thermal expansion of the fill material and the original road material in this embodiment needs to be measured on material samples taken from the site using standard laboratory testing methods (such as a thermomechanical analyzer, TMA). This value directly describes the rate of length change of the material under unit temperature change and is an inherent physical property of the material. The construction environment temperature data needs to be continuously monitored and recorded by a network of temperature sensors deployed at the repair site to obtain temperature data for the entire paving construction period and calculate its representative value (e.g., the average temperature value for the entire paving construction period as the construction environment temperature). The standard applicable temperature range of the material is a process parameter provided by the material supplier based on extensive experimental and engineering experience. In this embodiment, the standard applicable temperature range of the material clearly defines the upper and lower limits of the environmental temperature at which the material can maintain optimal working performance and safe application. When calculating the thermal expansion coefficient mismatch, the absolute difference between the thermal expansion coefficients of the two materials is compared with the original material's thermal expansion coefficient. This eliminates the influence of the absolute value of the materials, resulting in a dimensionless relative difference index. This allows for a unified scale to be established for the compatibility evaluation of different material combinations, facilitating cross-project comparisons. Similarly, in this embodiment, when calculating the deviation of construction temperature conditions, the absolute value of the difference between the ambient temperature and the median of the standard temperature range is normalized relative to the entire standard range. This transforms the deviation of actual environmental conditions from ideal conditions into a dimensionless scalar between 0 and 1, thus providing a direct reflection of the degree of adverseness of the construction environment. In this embodiment, the inherent mismatch of materials and adverse external environmental conditions are independent and equally important risk contributing factors. Their combined effect determines the initial risk of delamination defects caused by the thermal stress of the filler material during road maintenance, i.e., the poor thermal compatibility level of the filler material during road maintenance analyzed in this embodiment. Therefore, this embodiment uses the arithmetic mean of the thermal expansion coefficient mismatch and the deviation of the construction temperature condition to obtain the material thermal compatibility poor level. This allows the model to quantitatively assess the long-term hidden dangers that temperature factors may bring during road maintenance, providing scientific data support for whether to take adaptive measures such as heat preservation, cooling or material replacement, and avoiding the blindness of making decisions based solely on experience.

[0021] In this embodiment, step S2, which involves constructing a material compatibility analysis model to analyze the compatibility between the fill material and the original road material, also includes the following specific steps: S26. Extract the ratio of the elastic modulus of the filler material to that of the original road material from the road maintenance material property data, and extract the ratio of the actual compaction degree to the standard compaction degree from the maintenance process data; calculate the inverse of the ratio of the actual compaction degree to the standard compaction degree to obtain the compaction insufficiency index of the filler material during the road maintenance process. S27. The ratio of the elastic modulus of the filling material to that of the original road material is added to the compaction insufficiency index to obtain the degree of modulus mismatch of the filling material during road maintenance. S28. The arithmetic mean of the material thermal compatibility poorness level and modulus mismatch degree is used as the material compatibility poorness degree of the filling material during road maintenance.

[0022] For example, this embodiment further evaluates the ability of the fill layer and the original base layer to work together under load from the perspective of material mechanical property matching and construction compaction quality, thereby constituting a comprehensive evaluation of material compatibility. In this embodiment, the elastic modulus is a key mechanical parameter characterizing the material's ability to resist elastic deformation. Its value needs to be obtained by measuring material samples through laboratory mechanical property tests (such as uniaxial compression tests). This embodiment performs dimensionless processing on the calculation of the elastic modulus ratio, so that combinations of high-modulus and low-modulus materials can be compared on the same benchmark. In this embodiment, the actual compaction degree is obtained based on on-site quality inspection. For example, this embodiment can further use equipment such as a nuclear density meter or a nuclear-free density meter to immediately conduct multi-point tests on the fill layer after the maintenance work is completed, calculate its average density, and then calculate the ratio with the maximum dry density determined by the material through a standard compaction test. The actual compaction degree calculated in this embodiment directly reflects the effect of the construction rolling process on the material's density. The inadequate compaction index is obtained by taking the reciprocal of the ratio of the actual compaction degree to the standard compaction degree. A higher inadequate compaction index indicates less compaction of the filling material. In this embodiment, the material stiffness mismatch (static property) and the initial insufficient compaction caused by construction (dynamic process defect) have a cumulative effect on the risk of interface stress concentration. Even if the material moduli are similar, insufficient compaction will lead to a decrease in the actual stiffness of the filler layer, exacerbating the mismatch with the base layer. Conversely, even with good compaction, a significant difference in material moduli still poses a high risk. Therefore, the modulus ratio and the inadequate compaction index are added together to obtain the degree of modulus mismatch, enabling analysis of the degree of modulus mismatch. Finally, this embodiment calculates the material thermal compatibility poor level and the modulus mismatch degree using an arithmetic mean to obtain the comprehensive material compatibility poor level. Expanding from a single temperature dimension to a mechanical dimension and incorporating the actual effects of construction techniques, this approach fully quantifies the degree of material incompatibility. It can provide more comprehensive early warning of delamination hazards caused by material property mismatch and initial construction defects, offering precise quantitative guidance for material selection and optimized compaction processes during subsequent emergency road maintenance.

[0023] In this embodiment, step S3 involves constructing a maintenance quality control risk analysis model to analyze maintenance quality control risks during the maintenance process, including the following specific steps: S31. Extract the number of compaction passes of filling materials and the range of compaction passes required by process specifications from the maintenance process data, and extract the base dielectric constant of the maintenance area and the standard base dielectric constant from the ground penetrating radar road layer data; S32. The absolute difference between the number of compaction passes of the filling material during road maintenance and the midpoint of the compaction pass range required by the process specification shall be taken as the absolute compaction difference value. The absolute compaction difference value shall be divided by the length of the compaction pass range required by the process specification. The result of the division shall be taken as the deviation of the compaction process of the filling material during road maintenance. S33. The ratio of the dielectric constant of the base layer in the repair area to that of the standard base layer is calculated by reciprocal to obtain the base layer compaction anomaly index; the deviation of the compaction process of the filling material during road maintenance is calculated by arithmetic average with the base layer compaction anomaly index, and the result of the arithmetic average is taken as the base layer compaction anomaly in the repair area during road maintenance.

[0024] For example, this embodiment uses ground-penetrating radar data to correlate and verify abstract construction process parameters with specific physical engineering quality, thereby assessing the actual quality and potential risks of base course compaction. In this embodiment, the number of compaction passes refers to the number of times construction machinery (e.g., a road roller) compacts a unit area. This number is automatically recorded and uploaded to the quality management platform via a GPS positioning system and pass counter installed on the road roller. The range of compaction passes required by the process specifications is the range of compaction passes explicitly specified in the construction plan, and is the direct target of construction control. When calculating deviations from the compaction process, the absolute difference between the actual number of passes and the value in the specification is normalized relative to the length of the specification range. The purpose is to unify different specification requirements (e.g., some construction process specifications require 6-10 compaction passes, while others require 8-10 passes) onto the same deviation evaluation scale, allowing for horizontal comparison of construction quality across different sections. Meanwhile, extracting the dielectric constant of the base layer in the maintenance area from ground-penetrating radar data is crucial in this embodiment. The dielectric constant of the base layer in the maintenance area is processed using an inversion algorithm on the collected raw radar waveform data to obtain a dielectric constant distribution map of the underground medium at different depths in the maintenance area. The dielectric constant of the base layer is closely related to factors such as the density and moisture content of the material. Under stable moisture conditions, the higher the dielectric constant, the greater the density. The dielectric constant of the base layer in the maintenance area is compared with a standard value representing a good compaction state, established through extensive non-destructive and destructive testing (e.g., core drilling). The ratio is calculated and its reciprocal is taken to obtain the base layer compaction anomaly index. This transforms the physical quantity detected by radar into a base layer compaction anomaly index characterizing deviations from the ideal state (the larger the base layer compaction anomaly index, the more severe the compaction anomaly). Finally, this embodiment performs an arithmetic mean between the compaction process deviation characterizing the construction process and the base layer compaction anomaly index characterizing the physical result to obtain the base layer compaction anomaly status. This allows the technical solution provided in this embodiment to no longer view process compliance or radar image features in isolation, but to combine the two for correlation analysis; when the process deviation is large and the radar display is abnormal, the base layer compaction is seriously abnormal; and it can significantly improve the accuracy of risk assessment.

[0025] In this embodiment, step S3, which involves constructing a maintenance quality control risk analysis model to analyze maintenance quality control risks during the maintenance process, also includes the following specific steps: S34. Extract the actual paving thickness of the filling material at all equidistant road detection points within the maintenance area from the maintenance process data, and extract the amplitude value of the reflected signal at the junction of the filling layer and the old road surface at all equidistant road detection points within the maintenance area from the ground penetrating radar road layer data. S35. Obtain the arithmetic mean of the actual paving thickness of the filling material at all equidistant road inspection points, and calculate the standard deviation of the actual paving thickness of the filling material at all equidistant road inspection points. Divide the standard deviation of the actual paving thickness of the filling material by the arithmetic mean, and use the result of the division as the coefficient of variation of the paving thickness of the filling material in the maintenance area during the road maintenance process. S36. Obtain the arithmetic mean of the reflected signal amplitude values ​​at the junction of the fill layer and the old road surface at all equidistant road detection points, and calculate the standard deviation of the reflected signal amplitude values ​​at the junction of the fill layer and the old road surface at all equidistant road detection points. Divide the standard deviation of the reflected signal amplitude values ​​by the arithmetic mean, and use the result of the division operation as the coefficient of variation of the reflected signal intensity at the junction interface of the repair area during the road maintenance process. S37. The arithmetic mean of the coefficient of variation of the thickness of the filling material and the coefficient of variation of the intensity of the reflected signal at the interface is used to obtain the structural uniformity and abnormality of the maintenance area during the road maintenance process. S38. The arithmetic mean of abnormal compaction and structural uniformity of the base layer in the maintenance area during road maintenance is calculated, and the result of the arithmetic mean is used as the maintenance quality control risk during the maintenance process.

[0026] For example, this embodiment further evaluates the quality control level of the entire maintenance area from a macroscopic perspective of structural uniformity by analyzing the spatial variability of the paving thickness and radar interface signal, thereby forming a global risk assessment of construction quality. In this embodiment, the actual paving thickness of the fill material is obtained from a continuous monitoring system during construction. For instance, this embodiment can use a non-contact laser rangefinder or ultrasonic sensor installed on the paver to measure and record the paved material thickness at equal height densities (e.g., one cross-section per meter), forming a thickness profile line. The amplitude value of the reflected signal at the junction of the fill layer and the old road surface is obtained by scanning along the detection line with ground-penetrating radar, and then automatically identified and extracted by data processing software from the maximum amplitude value of the reflected wave at the target interface (the junction of the fill layer and the old road surface) in each trace. The calculation of the coefficient of variation for both the paving thickness and the coefficient of variation for the reflected signal intensity employs the statistical method of the ratio of standard deviation to mean. The core advantage of using the coefficient of variation in this embodiment lies in its dimensionless nature, allowing for comparison of the uniformity of paving operations with different average thicknesses (e.g., 5 cm and 10 cm) or detection data with different absolute signal intensities under the same standard. A larger coefficient of variation indicates a higher degree of data dispersion and poorer uniformity. Furthermore, in this embodiment, the physical thickness uniformity of the paving is the foundation for good structural uniformity, while the interface-bonded signal intensity uniformity detected by radar is a direct manifestation of structural uniformity in its mechanical bonding state. Both reflect the inherent quality uniformity of the maintenance area from different perspectives but from the same source (both originating from construction quality control). Therefore, the structural uniformity anomaly obtained by arithmetically averaging these two coefficients of variation can fully reflect the anomaly of the intrinsic quality uniformity of the repair area. Finally, in this embodiment, the base compaction anomaly representing the deep compaction quality and the structural uniformity anomaly representing the overall structural uniformity are arithmetically averaged to obtain the repair quality control risk. This constructs a comprehensive quality evaluation system from point to surface, from deep to surface, and from process to entity. It can not only identify obvious local defects, but also keenly capture the dispersion trend of the quality of the entire repair area. The overall non-uniformity reflected by the structural uniformity anomaly of the repair area is often a precursor to large-scale damage during future use.

[0027] In this embodiment, step S4 involves constructing a post-repair pavement defect risk assessment model to assess the risk of post-repair pavement defects, including the following specific steps: S41. Extract the analysis results of the material compatibility insufficiency of the filling materials during the road maintenance process, and at the same time extract the analysis results of the maintenance quality control risk during the maintenance process; S42. Perform a geometric mean calculation on the results of the material compatibility poorness analysis and the maintenance quality control risk analysis, and use the result of the geometric mean calculation as the result of the road surface defect risk assessment after maintenance in the maintenance area.

[0028] For example, this embodiment integrates the material compatibility analysis results and maintenance quality control risk analysis results obtained from the aforementioned steps to arrive at a comprehensive and forward-looking judgment on the overall defect risk of the road surface after repair. Here, this embodiment chooses geometric mean calculation instead of simple arithmetic mean to integrate material compatibility and maintenance quality control risk, based on a deep understanding of road damage mechanisms. The characteristic of geometric mean in this embodiment is its extreme sensitivity to the synergistic or weakest link effects among evaluation factors. Specifically, in this embodiment, if material compatibility is extremely poor but construction quality control is excellent, the geometric mean result will not simply take an intermediate value like the arithmetic mean, but will tend towards a lower value, reflecting a certain degree of compensation for inferior materials by high-quality construction, consistent with the actual engineering situation where excellent workmanship can, to some extent, compensate for some material deficiencies. Conversely, if material compatibility is good but construction quality control is extremely poor, a relatively high risk of road surface defects after repair in the repaired area will be obtained, indicating that poor workmanship is enough to destroy the potential of good materials. In this embodiment, when both material risk and construction risk are at a high level, a very high risk of pavement defects after repair is obtained. This accurately characterizes the potential risk of pavement defects after repair due to the superposition of inherent deficiencies in material construction and acquired mismatches in construction techniques. This embodiment employs a nonlinear fusion method, which, compared to the traditional method using a linear superposition arithmetic mean, more accurately portrays the relationship between the coupling and non-independent effects of multiple risk factors in reality. Using the geometric mean calculation result as the assessment result of the pavement defect risk after repair ensures that the assessment result not only reflects the average level of risk but also the most unfavorable combination of risk factors.

[0029] In this embodiment, step S5 provides an early warning for road repair quality based on the risk assessment results of road surface defects after repair, including the following specific details: Obtain the risk assessment results of road surface defects after repair in the repair area, and preset the emergency maintenance threshold for road surface defects; when the risk assessment results of road surface defects after repair in the repair area are greater than or equal to the emergency maintenance threshold for road surface defects, issue an emergency maintenance warning for the road surface after repair in the repair area; when the risk assessment results of road surface defects after repair in the repair area are less than the emergency maintenance threshold for road surface defects, the repair quality of the repair area is deemed to be qualified. In this embodiment, the emergency maintenance threshold for road surface defects is obtained experimentally by those skilled in the art. The specific experimental method is as follows: Ground-penetrating radar (GPR) road layering data of the repair area during multiple historical road repairs is obtained, along with corresponding road repair material attribute data and repair process data for the repair area. The GPR road layering data, corresponding road repair material attribute data, and repair process data of the repair area are substituted into each step of this embodiment to obtain the road surface defect risk assessment results after repair in the repair area during multiple historical road repairs. The judgment results regarding whether defects will occur in the subsequent service life of the road surface after repair in the repair area during multiple historical road repairs are obtained. The road surface defect risk assessment results and the judgment results regarding whether defects will occur in the subsequent service life of the road surface after repair in the repair area during multiple historical road repairs, obtained from the steps of this embodiment, are imported into fitting software to output the emergency maintenance threshold value for road surface defects that meets the highest accuracy rate for emergency maintenance judgment.

[0030] Example 2 like Figure 2 As shown, this embodiment provides a deep learning-based ground-penetrating radar road layer defect identification system, including: a multi-source data integration module, a compatibility assessment module, a process quality correlation module, a risk quantification assessment module, and an intelligent decision-making and early warning module; The system comprises the following modules: a multi-source data integration module for acquiring ground-penetrating radar road layering data of the repair area during road maintenance, as well as road maintenance material attribute data and maintenance process data; a compatibility assessment module for constructing a material compatibility analysis model based on the road maintenance material attribute data and maintenance process data, analyzing the degree of material incompatibility between the fill material and the original road material; a process quality correlation module for constructing a maintenance quality control risk analysis model by combining maintenance process data and ground-penetrating radar road layering data, analyzing the maintenance quality control risks during the maintenance process; a risk quantification assessment module for constructing a post-repair pavement defect risk assessment model based on the material compatibility analysis results and the maintenance quality control risk analysis results, assessing the post-repair pavement defect risks; and an intelligent decision-making early warning module for providing early warnings on road maintenance quality based on the post-repair pavement defect risk assessment results.

[0031] The parameters and steps for implementing the corresponding functions of each unit module in the deep learning-based ground-penetrating radar road layer defect identification system of the present invention described above can be referred to the parameters and steps in the embodiments of the deep learning-based ground-penetrating radar road layer defect identification method above, and will not be repeated here.

[0032] Example 3 An electronic device according to an embodiment of the present invention includes a processor and a memory. The memory stores a computer program that can be called by the processor. The processor executes a deep learning-based ground-penetrating radar (GPR) road layering defect identification method by calling the computer program stored in the memory. It should be noted that all computer programs for the deep learning-based GPR road layering defect identification method are implemented using C language. The multi-source data integration module, compatibility assessment module, process quality correlation module, risk quantification assessment module, and intelligent decision-making early warning module are all controlled by a remote server. In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0033] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A deep learning-based ground-penetrating radar road layer defect identification system, characterized in that, The system includes a multi-source data integration module, a compatibility assessment module, a process quality correlation module, a risk quantification assessment module, and an intelligent decision-making and early warning module. The multi-source data integration module acquires ground-penetrating radar road layering data of the repair area during road maintenance, as well as road maintenance material attribute data and maintenance process data. The compatibility assessment module, based on the road maintenance material attribute data and maintenance process data, constructs a material compatibility analysis model to analyze the degree of material incompatibility between the fill material and the original road materials. The process quality correlation module, combining maintenance process data and ground-penetrating radar road layering data, constructs a maintenance quality control risk analysis model to analyze maintenance quality control risks during the maintenance process. The risk quantification assessment module, based on the material compatibility analysis results and the maintenance quality control risk analysis results, constructs a post-repair pavement defect risk assessment model to assess the post-repair pavement defect risk. The intelligent decision-making and early warning module provides early warnings about road maintenance quality based on the post-repair pavement defect risk assessment results.

2. The deep learning-based ground-penetrating radar road layering defect identification system according to claim 1, characterized in that, The constructed material compatibility analysis model analyzes the degree of material incompatibility between the fill material and the original road material, including the following specific steps: S21. Extract the thermal expansion coefficients of the filler material and the original road material from the road maintenance material property data; and extract the construction ambient temperature and the standard applicable temperature range of the material from the maintenance process data. S22. Subtract the thermal expansion coefficient of the filling material from the thermal expansion coefficient of the original road material and take the absolute value as the absolute difference in thermal expansion coefficients. S23. Divide the absolute difference in thermal expansion coefficients with the thermal expansion coefficients of the original road material to obtain the thermal expansion coefficient mismatch of the filling material. S24. Take the difference between the ambient temperature of the construction environment and the median of the applicable temperature range of the material standard as the temperature deviation difference. Divide the absolute value of the temperature deviation difference by the length of the applicable temperature range of the material standard to obtain the deviation of the construction temperature condition of the filling material in the road maintenance process. S25. The arithmetic mean of the thermal expansion coefficient mismatch and the deviation of the construction temperature condition is used to obtain the poor material thermal compatibility level of the filling material during road maintenance.

3. The deep learning-based ground-penetrating radar road layering defect identification system according to claim 2, characterized in that, The aforementioned material compatibility analysis model, which analyzes the degree of material incompatibility between the fill material and the original road material, also includes the following specific steps: S26. Extract the ratio of the elastic modulus of the filling material to that of the original road material from the road maintenance material property data, and extract the ratio of the actual compaction degree to the standard compaction degree from the maintenance process data. The inverse of the ratio of actual compaction to standard compaction is used to obtain the compaction insufficiency index of filler materials during road maintenance. S27. The ratio of the elastic modulus of the filling material to that of the original road material is added to the compaction insufficiency index to obtain the degree of modulus mismatch of the filling material during road maintenance. S28. The arithmetic mean of the material thermal compatibility poorness level and modulus mismatch degree is used as the material compatibility poorness degree of the filling material during road maintenance.

4. The deep learning-based ground-penetrating radar road layering defect identification system according to claim 3, characterized in that, The construction of the maintenance quality control risk analysis model, which analyzes the maintenance quality control risks during the maintenance process, includes the following specific steps: S31. Extract the number of compaction passes of filling materials and the range of compaction passes required by process specifications from the maintenance process data, and extract the base dielectric constant of the maintenance area and the standard base dielectric constant from the ground penetrating radar road layer data; S32. The absolute difference between the number of compaction passes of the filling material during road maintenance and the midpoint of the compaction pass range required by the process specification shall be taken as the absolute compaction difference value. The absolute compaction difference value shall be divided by the length of the compaction pass range required by the process specification. The result of the division shall be taken as the deviation of the compaction process of the filling material during road maintenance. S33. The ratio of the dielectric constant of the base layer in the repair area to that of the standard base layer is calculated by reciprocal to obtain the base layer compaction anomaly index; the deviation of the compaction process of the filling material during road maintenance is calculated by arithmetic average with the base layer compaction anomaly index, and the result of the arithmetic average is taken as the base layer compaction anomaly in the repair area during road maintenance.

5. The deep learning-based ground-penetrating radar road layering defect identification system according to claim 4, characterized in that, The construction of the maintenance quality control risk analysis model, which analyzes the maintenance quality control risks during the maintenance process, also includes the following specific steps: S34. Extract the actual paving thickness of the filling material at all equidistant road detection points within the maintenance area from the maintenance process data, and extract the amplitude value of the reflected signal at the junction of the filling layer and the old road surface at all equidistant road detection points within the maintenance area from the ground penetrating radar road layer data. S35. Obtain the arithmetic mean of the actual paving thickness of the filling material at all equidistant road inspection points, and calculate the standard deviation of the actual paving thickness of the filling material at all equidistant road inspection points. Divide the standard deviation of the actual paving thickness of the filling material by the arithmetic mean, and use the result of the division as the coefficient of variation of the paving thickness of the filling material in the maintenance area during the road maintenance process. S36. Obtain the arithmetic mean of the reflected signal amplitude values ​​at the junction of the fill layer and the old road surface at all equidistant road detection points, and calculate the standard deviation of the reflected signal amplitude values ​​at the junction of the fill layer and the old road surface at all equidistant road detection points. Divide the standard deviation of the reflected signal amplitude values ​​by the arithmetic mean, and use the result of the division operation as the coefficient of variation of the reflected signal intensity at the junction interface of the repair area during the road maintenance process. S37. The arithmetic mean of the variation coefficient of the thickness of the filling material and the variation coefficient of the intensity of the reflected signal at the interface is used to obtain the structural uniformity and abnormality of the maintenance area during the road maintenance process. S38. The arithmetic mean of abnormal compaction and structural uniformity of the base layer in the maintenance area during road maintenance is calculated, and the result of the arithmetic mean is used as the maintenance quality control risk during the maintenance process.

6. The deep learning-based ground-penetrating radar road layer defect identification system according to claim 5, characterized in that, The construction of a post-repair pavement defect risk assessment model, which assesses the risk of post-repair pavement defects, includes the following specific steps: S41. Extract the analysis results of the material compatibility insufficiency of the filling materials during the road maintenance process, and at the same time extract the analysis results of the maintenance quality control risk during the maintenance process; S42. Perform a geometric mean calculation on the results of the material compatibility poorness analysis and the maintenance quality control risk analysis, and use the result of the geometric mean calculation as the result of the road surface defect risk assessment after maintenance in the maintenance area.

7. The deep learning-based ground-penetrating radar road layering defect identification system according to claim 6, characterized in that, The method of issuing early warnings on road maintenance quality based on the risk assessment results of road surface defects after repair includes the following specific contents: Obtain the risk assessment results of road surface defects after repair in the repair area, and preset the emergency maintenance threshold for road surface defects; When the risk assessment result of road surface defects after repair in the repair area is greater than or equal to the emergency maintenance threshold for road surface defects, an emergency maintenance warning will be issued again for the road surface after repair in the repair area. When the risk assessment result of road surface defects after repair in the repair area is less than the emergency maintenance threshold for road surface defects, the repair quality of the repair area is deemed to be qualified.

8. A deep learning-based ground-penetrating radar (GPR) method for identifying road layer defects, implemented based on any one of claims 1-7, characterized in that, The specific steps include the following: During road maintenance, obtain ground-penetrating radar road layering data of the maintenance area, and simultaneously obtain road maintenance material property data and maintenance process data of the maintenance area; Based on road maintenance material property data and maintenance process data, a material compatibility analysis model is constructed to analyze the degree of material incompatibility between the fill material and the original road material; By combining maintenance process data with ground-penetrating radar road stratification data, a maintenance quality control risk analysis model is constructed to analyze the maintenance quality control risks during the maintenance process. Based on the analysis results of poor material compatibility and the risk analysis results of maintenance quality control, a risk assessment model for pavement defects after maintenance is constructed to assess the risk of pavement defects after maintenance. Based on the risk assessment results of road surface defects after repair, early warnings are issued regarding the quality of road repairs.

9. An electronic device, comprising: A processor and a memory, wherein the memory stores a computer program that can be called by the processor; characterized in that the processor executes the deep learning-based ground-penetrating radar road layer defect identification method as described in any one of claims 8 by calling the computer program stored in the memory.

Images