A method and system for detecting latent cracks in a road

Abstract

The application discloses a kind of road implicit crack detection method and system, the method includes: collecting the road surface temperature time series data of target section, road surface vertical strain time series data and traffic load statistical time series data;Road surface temperature time series data, road surface vertical strain time series data, traffic load statistical time series data are input into pre-trained temperature crack response baseline model, obtain the crack open-close index time series of each time period on the target road section;Crack open-close index time series is generated according to several candidate detection time slices of pre-set time step, and the average crack open-close index and predicted traffic flow corresponding to each candidate detection time slice are calculated;Different lane closure mode and driving speed control combination are enumerated in each candidate detection time slice, and the candidate combination meeting the preset constraint condition is screened out as recommended detection task;Significantly reduce long-term systematic missed detection risk.

Description

Technical Field

[0001] This invention relates to the field of data storage technology, specifically to a method and system for detecting hidden cracks in roads. Background Technology

[0002] In the operation and maintenance of highways in cold regions, road surface cracks, especially hidden cracks, are difficult to detect. The behavior of these cracks is significantly temperature-dependent: during the day, under the influence of sunlight, the road surface heats up, and the asphalt concrete material expands, causing the cracks to open; while at night, during the cooling phase, the material contracts, and the cracks close, making them difficult to detect by visual and conventional detection methods.

[0003] Currently, road internal condition detection mainly relies on ground-penetrating radar (GPR) vehicles. However, to minimize traffic disruption and ensure operational safety, maintenance units typically schedule detection operations during the early morning hours when traffic is lighter. This leads to a fatal misconception about the detection window: the detection vehicle always scans when cracks are closed due to low temperatures, resulting in false positives in the collected radar echo data, which fail to reflect the true health condition of the road surface.

[0004] This systematic blind spot leads to serious hidden risks: long-term undetected hidden cracks accumulate continuously within the material due to fatigue, and their damage only expands rapidly under the combined conditions of high daytime temperatures and heavy traffic. Once the cracks reach a critical state, they can easily cause sudden through-type damage or local collapse of the road surface without significant warning, directly posing a major driving safety hazard and potentially triggering serious chain-reaction traffic accidents.

[0005] Therefore, existing technologies suffer from systematic omissions due to the misalignment of detection timing with the critical period of cracks. Proposing a method and system for detecting hidden cracks in roads to address the difficulties of existing technologies is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0006] The purpose of this invention is to provide a method and system for detecting hidden cracks in roads, in order to address the shortcomings of the prior art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: A method for detecting hidden cracks in roads includes: Collect time-series data on pavement temperature, pavement vertical strain, and traffic load statistics for the target road section; The road surface temperature time series data, road surface vertical strain time series data, and traffic load statistical time series data are input into the pre-trained temperature crack response baseline model to obtain the crack opening and closing index time series for each time period on the target road section. The crack opening and closing index time series is used to generate several candidate detection time slices according to a preset time step. The average crack opening and closing index and the predicted traffic flow are calculated for each candidate detection time slice. Different lane closure methods and vehicle speed control combinations are enumerated within each candidate detection time slice. Candidate combinations that meet preset constraints are selected as recommended detection tasks. The constraints include at least crack detectability constraints.

[0008] Furthermore, when it is impossible to avoid the crack closure period by adjusting the detection time, a controlled thermal composite excitation is actively applied to the target road section to induce the latent cracks in the closed state to open. The system receives baseline radar echo data and excitation radar echo data collected from the same target road segment and the same driving trajectory. Based on time and space, the baseline radar echo data and excitation radar echo data are registered. Using the crack-sensitive waveform characteristics pre-calibrated when training the temperature crack response baseline model, point-by-point difference and envelope analysis are performed on the baseline radar echo data and excitation radar echo data to suppress uniform structure reflection signals and highlight the new reflection anomalies caused by crack opening.

[0009] Furthermore, the temperature crack response baseline model is trained based on crack training data and the time series of crack opening and closing indices corresponding to temperature crack features. The temperature crack response baseline model is a long short-term memory network.

[0010] Furthermore, the crack detectability constraint requires that the crack opening and closing index corresponding to the candidate time slice is not lower than a preset crack detectability threshold; the recommended detection task includes at least a detection start time and a detection end time.

[0011] Furthermore, the temperature crack response baseline model output also includes crack opening threshold temperature and crack closing threshold temperature; the crack opening threshold temperature is the lowest temperature value above which crack-related anomalous reflections begin to stably appear in the ground penetrating radar echo waveform, and the crack closing threshold temperature is the highest temperature value below which the anomalous reflections disappear and the waveform returns to a continuous homogeneous state.

[0012] Furthermore, the crack opening and closing index is a dimensionless value between zero and one. A crack opening and closing index close to zero indicates that the crack is closed or difficult to detect, while a crack opening and closing index close to one indicates that the crack is fully open and easily identified by ground-penetrating radar.

[0013] Furthermore, methods for combined thermal excitation include... Step 1: Based on the temperature crack response baseline model, determine the crack opening threshold temperature at the target road section, and calculate the required heating time based on the current measured ambient temperature and the initial road surface temperature. Step 2: Based on the current vehicle speed and the distance between the road surface heating device and the ground-penetrating radar array probe, calculate the length of the heating zone and the thermal action time window to obtain the start-stop control strategy associated with the vehicle position. Step 3: Before the inspection vehicle enters the target road section, start the road surface heating device to gradually increase the road surface temperature along the direction of travel to the crack opening threshold temperature. Step 4: After the temperature of the heated area reaches the crack opening threshold temperature, start the adjustable wheel load loading device to increase the tire ground pressure of the loading wheel group to the preset short-term high wheel load level, so that the heated area can simultaneously bear the superposition of temperature and load in the same spatial position and at the same time. Step 5: When the detection vehicle continues to move forward until the ground-penetrating radar array probe is above the thermal zone, the ground-penetrating radar is immediately triggered to collect radar echo data of the internal structure of the road surface after excitation.

[0014] Furthermore, point-by-point differential operation is performed on each pair of registered radar waveforms to obtain a differential radar waveform sequence; An analysis window of a set length is slid along the depth direction on the differential radar waveform. The envelope and its amplitude statistics of the differential signal within the window are calculated to form a differential envelope profile along the depth direction. The pre-calibrated crack-sensitive waveform characteristics include at least: the depth range where cracks are likely to occur, the polarity type of the differential waveform, the range of continuous sampling points of the abnormal waveform, and the amplitude amplification factor threshold compared with the reflection of a normal uniform structure. The crack-sensitive waveform features are compared point by point with the differential envelope profile. Sampling points that simultaneously meet the following conditions are marked as crack response candidate points: depth falls within the crack-sensitive range, differential envelope amplitude exceeds a preset threshold, and waveform polarity is consistent with expectations. New reflection anomalies related to crack opening are retained.

[0015] A road hidden crack detection system, used to implement the aforementioned road hidden crack detection method, the system comprising: The data acquisition module is used to collect time-series data of pavement temperature, pavement vertical strain, and traffic load statistics for the target road section. The data processing module is used to input road surface temperature time series data, road surface vertical strain time series data, and traffic load statistical time series data into the pre-trained temperature crack response baseline model to obtain the crack opening and closing index time series for each time period on the target road segment. The detection and reconstruction module generates several candidate detection time slices from the crack opening and closing index time series according to a preset time step, and calculates the corresponding average crack opening and closing index and predicted traffic flow for each candidate detection time slice; within each candidate detection time slice, it enumerates different lane closure methods and vehicle speed control combinations, and selects candidate combinations that meet preset constraints as recommended detection tasks; the constraints include at least crack detectability constraints.

[0016] Furthermore, it also includes: The equivalent excitation module is used to actively apply controlled thermal composite excitation to the target road section when it is impossible to avoid the crack closure period by adjusting the detection time, so as to induce the closed hidden cracks to open. The differential analysis module receives baseline radar echo data and excitation radar echo data collected from the same target road segment and the same driving trajectory. Based on time and space, it registers the baseline radar echo data and the excitation radar echo data. Using the crack-sensitive waveform characteristics pre-calibrated when training the temperature crack response baseline model, it performs point-by-point differential and envelope analysis on the baseline radar echo data and the excitation radar echo data to suppress uniform structure reflection signals and highlight the new reflection anomalies caused by crack opening.

[0017] The technical effects and advantages of the road hidden crack detection method and system provided by this invention are as follows: 1. By deploying road surface temperature sensors and vertical strain sensors on the target road section, and in conjunction with ground-penetrating radar baseline survey lines and traffic flow detection equipment, time-series data of road surface temperature, vertical strain, and traffic load statistics are collected. The above multi-source time-series data are input into a pre-trained temperature crack response baseline model to obtain crack opening threshold temperature, crack closing threshold temperature, and crack opening / closing index time series, which are used to determine whether the crack is in an open or closed state at different time periods. Finally, based on the crack opening / closing index time series, the time period in the traditional early morning detection period where the crack is in a closed state for a long time is identified and marked as a detection window error.

[0018] This enables a shift from experience-based early morning selection to window selection based on a combination of temperature, load, and radar, thus realigning the detection timing with the high-risk opening state of hidden cracks. This transforms the time window that was almost certain to be missed into a controlled period of priority avoidance or active stimulation, significantly reducing the risk of long-term systemic missed detections.

[0019] 2. By collecting baseline radar echo data before excitation and radar echo data after excitation on the same road segment and the same driving trajectory, and performing time and space registration according to track number, timestamp, and mileage coordinates; after performing DC removal, bandpass filtering, and amplitude normalization preprocessing on each pair of registered radar waveforms, point-by-point differential operation is performed, and the envelope and its amplitude statistics are calculated by sliding a window along the depth direction on the differential waveform to obtain the differential envelope profile; using the crack-sensitive waveform characteristics pre-calibrated by the temperature crack response baseline model, the sampling points in the differential envelope profile that simultaneously meet multiple criteria are marked as crack response candidate points; the candidate points are clustered on the mileage and depth planes to generate suspected crack zones, and a distribution map of suspected hidden crack locations is constructed.

[0020] It is evident that this invention does not rely solely on single radar amplitude anomalies to determine cracks. Instead, it utilizes a combination of pre- and post-excitation differential and mechanism calibration features to effectively suppress false anomalies caused by reflections from uniform structures such as interlayer interfaces. This significantly improves the signal-to-noise ratio and spatial positioning resolution of latent cracks, providing precise spatial guidance for subsequent borehole verification and local excavation. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0022] Figure 1 This is a schematic diagram of a road hidden crack detection system according to Embodiment 1 of the present invention; Figure 2 This is a schematic diagram of a road hidden crack detection system according to Embodiment 1 of the present invention; Figure 3 This is a flowchart of a method for detecting hidden cracks in roads according to Embodiment 3 of the present invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] It should be noted that when a component is said to be "fixed to" another component, it can be directly attached to the other component or there may be an intervening component. When a component is said to be "connected to" another component, it can be directly connected to the other component or there may be an intervening component.

[0025] Example 1 like Figure 1 As shown in the figure, this embodiment of a method and system for detecting hidden cracks in roads includes a data processing module and a detection and reconstruction module, and the modules are connected by wired and / or wireless means.

[0026] The data acquisition module is used to collect time-series data on pavement temperature at multiple points along the length of the target road segment, time-series data on pavement vertical strain at corresponding locations, and time-series data on traffic load statistics obtained from traffic flow detection equipment. Specifically, the pavement temperature time-series data reflects temperature changes at the surface and a certain depth; the pavement vertical strain time-series data reflects the combined deformation response of the structure under temperature and load conditions; and the traffic load statistics time-series data reflects the traffic volume and equivalent axle load intensity per unit time.

[0027] The data processing module is used to input pavement temperature time series data, pavement vertical strain time series data, and traffic load statistical time series data into a pre-trained temperature crack response baseline model to obtain the crack opening threshold temperature, crack closing threshold temperature, corresponding traffic load sensitive interval, and crack opening and closing index time series for each time period on the target road segment; and to determine whether there is a risk of systematic missed detection in the traditional predetermined detection time arrangement based on the crack opening and closing index time series.

[0028] The crack opening threshold temperature is defined as the lowest temperature value above which stable crack-related anomalous reflections begin to appear in the ground-penetrating radar echo waveform. The crack closing threshold temperature is defined as the highest temperature value below which the anomalous reflections essentially disappear and the waveform returns to a continuous and homogeneous state. The crack opening and closing index is a dimensionless value between zero and one. A crack opening and closing index close to zero indicates that the crack is completely closed or difficult to detect, while a crack opening and closing index close to one indicates that the crack is fully open and easily identifiable by ground-penetrating radar.

[0029] Training methods for the baseline model of temperature crack response include: V sets of crack training data were pre-collected, where V is an integer greater than 1. The crack training data included temperature crack features and the corresponding crack opening and closing index time series. The temperature crack features included pavement temperature time series data, pavement vertical strain time series data, and traffic load statistical time series data. The pre-collected data are as follows: The temperature crack features collected at the same spatial location over multiple day-night cycles correspond one-to-one with the ground-penetrating radar echo waveform segments at the same time. Crack-sensitive features such as reflection amplitude, phase change, and waveform discontinuity are extracted through waveform morphology analysis. These features are combined with the temperature crack features to form training data for modeling. By comparing whether stable crack-sensitive features appear at the same location under different temperature and load combinations, the crack opening threshold temperature and the crack closing threshold temperature are determined, and the variation law of the crack opening and closing index with respect to the temperature crack features is derived.

[0030] To ensure the quality of the temperature crack characteristics, the input data for the temperature crack response baseline model is collected as follows: Road surface temperature sensors and road surface strain sensors are buried every several tens of meters along the driving direction at representative lane wheel track locations on the target road section. The road surface temperature sensors are preferably buried in the upper part of the asphalt layer and at the interface of the base course, used to acquire surface and deep layer temperatures respectively. The road surface strain sensors are preferably arranged vertically along the road surface to directly measure the vertical strain response under temperature-load superposition. The traffic load statistical time-series data is collected through existing toll station lane counting equipment, inductive loop detectors, or dynamic weighing equipment, and the number of passing vehicles, vehicle type distribution, and estimated equivalent axle load counts are counted for each time period. The original waveform sequence of the ground-penetrating radar echo is obtained by scanning the ground-penetrating radar baseline line at a constant speed multiple times, recording precise mileage coordinates and timestamps for each scan, so as to perform temporal and spatial registration with the aforementioned temperature, strain, and traffic load data.

[0031] The preferred training method for the temperature crack response baseline model is a piecewise fitting method based on the temperature crack characteristics of the test section, specifically: On the target road section or a test road section with similar structure and climate conditions, several representative locations with obvious hidden cracks are selected. The location and morphology of the cracks are confirmed by core drilling, excavation inspection, or high-resolution imaging, and these locations are used as training points labeled with cracks. For each training point, the temperature crack characteristics and corresponding ground-penetrating radar echo data characteristics are extracted for all time moments within several 24-hour cycles during the monitoring period. Based on whether the waveform shows crack-sensitive characteristics, the temperature crack characteristics are divided into open and closed samples. Then, statistical analysis is performed on the temperature dimension and the traffic load dimension to determine the temperature threshold and load threshold that best distinguish the two types of samples. The segment point with the lowest misclassification rate under the least squares sense is preferably used as the crack opening threshold temperature and the crack closing threshold temperature. Subsequently, the crack opening and closing index is defined as a continuous function that gradually approaches zero below the crack closing threshold temperature, gradually approaches one above the crack opening threshold temperature, and interpolates linearly or nonlinearly with temperature and load in the intermediate interval, realizing the mapping from the original multi-source input data to the quantitative index of crack opening and closing status. By repeating the above process at multiple training points and spatially interpolating the results, the temperature crack response baseline model covering the entire target road segment can be obtained. The temperature crack response baseline model is a time series model such as a long short-term memory network.

[0032] The detection and reconstruction module generates several candidate detection time slices from the crack opening and closing index time series according to a preset time step, and calculates the corresponding average crack opening and closing index and predicted traffic flow for each candidate detection time slice. Within each candidate detection time slice, different lane closure methods and vehicle speed control combinations are enumerated, and candidate combinations that meet the "preset constraints" are selected as recommended detection tasks.

[0033] The preset constraints include: crack detectability constraints, traffic carrying capacity constraints, and construction safety boundary constraints. The crack detectability constraint requires that the crack opening and closing index corresponding to the candidate time slice is not lower than the preset crack detectability threshold, so that the hidden cracks are fully open and can be reliably identified by ground-penetrating radar within the time slice. The traffic carrying capacity constraint requires that the predicted traffic flow obtained based on the traffic flow prediction model does not exceed the acceptable traffic flow limit under the current lane closure mode. The acceptable traffic flow limit is pre-calibrated according to the number of remaining traffic lanes, the road design capacity, and the service level requirements to ensure that the overall traffic capacity of the road section is not excessively weakened during the detection period. The construction safety boundary constraint requires that the candidate time slice and the corresponding lane closure mode meet the construction safety boundary conditions pre-set by the operation and maintenance management department. The construction safety boundary conditions at least limit the minimum number of remaining traffic lanes, the allowable vehicle speed range inside and outside the detection work area, the minimum safe distance between the detection work area and the ramp entrances and exits of the service area, the lower limit of nighttime lighting and visibility, and the necessary safety isolation facility configuration standards.

[0034] The construction safety boundary conditions include at least the following: the maximum number of lanes that can be closed under different traffic flow ranges, the minimum number of lanes remaining open during the inspection operation, the upper and lower limits of vehicle operating speed, the minimum safe distance from key nodes such as ramps and service area entrances and exits, and nighttime lighting and visibility requirements.

[0035] Traffic flow prediction can be determined using historical data, or by extracting historical traffic flow time series data from toll station traffic records, inductive loop detectors, or dynamic weighing detection data. Using date type (weekday, weekend, holiday), time period (hourly or finer granular), and seasonal attributes as feature variables, cluster analysis is performed on traffic flow at different time periods to obtain several daily patterns with similar diurnal variation characteristics. Within each daily pattern, a time-series-based regression model or piecewise statistical model is used to fit a prediction function for traffic flow changes over time. This allows for the output of predicted traffic flow values ​​and vehicle composition ratios within a given target date and time period. Predicted traffic flow is used to determine whether detection operations can be scheduled within a candidate detection time period while ensuring road capacity and safety redundancy.

[0036] By stitching together several consecutive candidate time slices on the time axis, a detection window interval that meets the minimum continuous operation time requirement is formed. Each detection window interval is assigned a comprehensive score, which can be a weighted function of the average crack opening and closing index, detection duration, and traffic impact within the interval. Finally, several detection window intervals are selected as recommended detection tasks based on the comprehensive scores from high to low, including the ground penetrating radar detection plan. The start and end times, corresponding closed lanes, permitted lanes, suggested driving speed range, and safety warning information of each detection task are packaged and output to the detection vehicle dispatching system to guide the actual detection operation.

[0037] In one example, for a section of highway in a cold region, the detection and reconstruction module reads the aforementioned crack opening and closing index time series and traffic flow prediction results. It finds that the crack opening and closing index is consistently below 0.1 between 2:00 AM and 3:00 AM, and the predicted traffic flow is relatively low. However, between 5:00 AM and 6:00 AM, the crack opening and closing index can reach above 0.6, and the traffic flow is still below the preset upper limit. Considering the construction safety boundary condition stipulating "at least two lanes must be kept open," the detection and reconstruction module automatically abandons scheduling detection between 2:00 AM and 3:00 AM, shifting the original detection window entirely to 5:00 AM and 6:00 AM. It recommends a detection scheme that closes only the outer lane and restricts vehicle speed to no more than 60 kilometers per hour, thereby significantly increasing the proportion of time when hidden cracks are open without significantly reducing traffic capacity.

[0038] Example 2: like Figure 2 As shown, the system also includes an equivalent excitation module and a differential analysis module. When it's impossible to avoid the crack closure period by adjusting the detection time, a controlled thermal composite excitation is actively applied to the target road section. This causes the road surface to reach a thermal state comparable to the combined high temperature and heavy load of daytime traffic within a short period, thereby inducing the opening of latent cracks that are currently closed. To this end, a road surface heating device and an adjustable wheel load loading device are arranged at the front of the detection vehicle. The road surface heating device can be multiple sets of infrared heating units or hot air jet units, and the adjustable wheel load loading device is a loading wheel assembly that can actively adjust tire contact pressure. The detection vehicle is controlled to travel at a limited speed, causing the road surface heating device to form a high-temperature zone several meters long in front of the ground-penetrating radar array probe. Simultaneously, a short-term wheel load greater than the average daily traffic load is applied through the adjustable wheel load loading device, causing the road surface in this area to reach a thermal state comparable to the combined high temperature and heavy load of daytime traffic within a short period, thereby inducing the opening of latent cracks. This overcomes the limitations of traditional methods that can only passively wait for natural high-temperature periods, improving the visibility and detection rate of latent cracks.

[0039] Methods of combined thermal excitation include Step 1: Based on the temperature crack response baseline model, determine the crack opening threshold temperature at the target road section, and calculate the required heating time based on the current measured ambient temperature and the initial road surface temperature. Step 2: Based on the current speed of the detected vehicle and the distance between the road surface heating device and the ground-penetrating radar array probe, calculate the length of the heating zone and the thermal action time window to obtain the start-stop control strategy associated with the vehicle position. Step 3: Before the detection vehicle enters the target road section, the road surface heating device is activated to gradually raise the road surface temperature along the direction of travel to the crack opening threshold temperature, and closed-loop regulation is performed through the road surface temperature sensor during this process. Step 4: After the temperature of the heated area reaches the target range, the adjustable wheel load loading device is activated to increase the tire ground pressure of the loaded wheel set to the preset short-term high wheel load level, so that the heated area can simultaneously bear the superposition of temperature and load in the same spatial position and the same time period, thereby forming a local thermal state that simulates the high temperature and heavy load working conditions during the day. Step 5: When the detection vehicle continues to move forward until the ground-penetrating radar array probe is above the thermal zone, the ground-penetrating radar is immediately triggered to collect radar echo data of the internal structure of the road surface after being excited.

[0040] In one example, for a road section near a service area exit where the detection time cannot be adjusted, the temperature crack response baseline model indicates that a road surface temperature increase of at least 10 degrees Celsius and a wheel load amplification factor of approximately 1.5 times the daily average are required for the cracks in this section to open significantly. Maintenance personnel set the detection vehicle speed to 20 kilometers per hour and the distance between the road surface heating device and the ground-penetrating radar array probe to 10 meters. Under closed-loop control, the road surface temperature within this 10-meter range is increased by more than 10 degrees Celsius compared to the ambient temperature, and a short-term wheel load higher than that of normal traffic is applied through the adjustable wheel load loading device. When the detection vehicle reaches this area, the ground-penetrating radar array probe acquires radar echo data corresponding to the latent cracks that have opened under the equivalent excitation of high temperature and heavy load, achieving effective detection of latent cracks during periods when the cracks would normally be closed.

[0041] The differential analysis module receives baseline radar echo data (before excitation) and post-excitation radar echo data collected again on the same target road segment and the same driving trajectory. Based on time and space, the baseline and post-excitation radar echo data are registered. Using the crack-sensitive waveform characteristics pre-calibrated during the training of the temperature crack response baseline model, point-by-point differential and envelope analysis are performed on the two sets of data to suppress uniform structure reflection signals, highlight new reflection anomalies caused by crack opening, and generate a distribution map of the suspected locations of the hidden cracks. Specifically: Both the reference radar echo data and the excitation radar echo data are stored according to the ground-penetrating radar channel number, sampling time, sampling depth, and mileage coordinates. Each radar channel is accompanied by a collection timestamp and spatial location information corresponding to the mileage information of the detected vehicle. The differential analysis module first performs time-space registration on the two sets of data based on the timestamp and mileage coordinates. For radar channels at the same mileage position on the same driving trajectory, the pair of reference radar echo data and excitation radar echo data with the closest collection time are selected and combined into a one-to-one corresponding waveform pair. After registration, each pair of radar waveforms is processed according to preset filtering rules for DC removal, bandpass filtering, and amplitude normalization to reduce the influence of environmental noise and equipment drift, resulting in preprocessed reference radar waveforms and preprocessed excitation radar waveforms.

[0042] The differential analysis module performs point-by-point differential operations on each registered radar waveform to obtain a differential radar waveform sequence. Then, it slides an analysis window of a set length along the depth direction on the differential radar waveform, calculates the envelope and amplitude statistics of the differential signals within the window, and forms a differential envelope profile along the depth direction. The pre-calibrated crack-sensitive waveform features in the temperature crack response baseline model include at least: the depth range where cracks are likely to occur, the polarity type of the differential waveform, the range of continuous sampling points of the abnormal waveform, and the amplitude amplification factor threshold compared with the reflection of a normal uniform structure. The differential analysis module compares the above crack-sensitive waveform features with the differential envelope profile point by point, and marks the sampling points that simultaneously meet the conditions of depth falling within the crack-sensitive range, differential envelope amplitude exceeding the preset threshold, and waveform polarity consistent with expectations as crack response candidate points, thereby suppressing common components generated by interlayer interfaces and uniform structure reflections, and retaining only new reflection anomalies directly related to crack opening.

[0043] Furthermore, the differential analysis module converts the candidate crack response points on each radar track into spatial depth coordinates according to their mileage coordinates and corresponding propagation time, constructing a distribution map of suspected hidden crack locations in a planar coordinate system with mileage as the abscissa and depth as the ordinate. Multiple candidate points with spatial distances less than a preset neighborhood scale in both the mileage and depth directions are clustered and merged to generate suspected crack zones with continuous geometric shapes. Simultaneously, the average or maximum value of the differential envelope amplitude within each suspected crack zone is calculated as an indicator of crack anomaly intensity at that location. The distribution map of suspected hidden crack locations can be stored as two-dimensional raster data or a vector set of suspected crack zones, and can serve as a basis for crack risk classification.

[0044] In one example, for several radar tracks near a certain kilometer marker, the differential analysis module detected multiple candidate points with differential envelope amplitudes exceeding twice the normal background and consistent polarity within a depth range of 30 to 50 centimeters from the road surface. These candidate points were continuously distributed for approximately five meters along the driving direction. After clustering and projection, a continuous anomaly band located within this depth range was formed on the distribution map of suspected latent crack locations. This anomaly band was marked as a highly suspected latent crack location, and the corresponding anomaly intensity index was output, providing a basis for subsequent borehole verification and maintenance decisions.

[0045] Example 3 Please see Figure 3 As shown, this embodiment provides a method for detecting hidden cracks in roads. For details not covered, please refer to Embodiments 1 and 2. The method includes: Collect time-series data on pavement temperature, pavement vertical strain, and traffic load statistics for the target road section; The road surface temperature time series data, road surface vertical strain time series data, and traffic load statistical time series data are input into the pre-trained temperature crack response baseline model to obtain the crack opening and closing index time series for each time period on the target road section. The crack opening and closing index time series is used to generate several candidate detection time slices according to a preset time step. The average crack opening and closing index and the predicted traffic flow are calculated for each candidate detection time slice. Different lane closure methods and vehicle speed control combinations are enumerated within each candidate detection time slice. Candidate combinations that meet preset constraints are selected as recommended detection tasks. The constraints include at least crack detectability constraints.

[0046] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for detecting hidden cracks in roads, characterized in that, include: Collect time-series data on pavement temperature, pavement vertical strain, and traffic load statistics for the target road section; Road surface temperature time series data, road surface vertical strain time series data, and traffic load statistical time series data are input into a pre-trained temperature crack response baseline model to obtain the crack opening and closing index time series of the target road segment. The crack opening and closing index time series is formed by arranging the crack opening and closing indices of the target road segment in each time period in chronological order. The crack opening and closing index corresponding to each time period is defined as a continuous numerical index that gradually approaches zero below the crack closure threshold temperature of the corresponding time period, gradually approaches one above the crack opening threshold temperature of the corresponding time period, and is interpolated linearly or nonlinearly according to temperature and load in the intermediate interval. The crack opening and closing index time series is used to generate several candidate detection time slots according to a preset time step. For each candidate detection time slot, the corresponding average crack opening and closing index and predicted traffic flow are calculated. Within each candidate detection time slot, different lane closure methods and speed control combinations are enumerated to select candidate detection time slots and corresponding lane closure methods and speed control combinations that meet preset constraints. These preset constraints include crack detectability constraints, traffic carrying capacity constraints, and construction safety boundary constraints. The crack detectability constraint requires that the average crack opening and closing index corresponding to the candidate detection time slot is not lower than a preset crack detectability threshold. The traffic carrying capacity constraint requires that the predicted traffic flow corresponding to the candidate detection time slot... The traffic flow does not exceed the acceptable traffic flow limit under the current lane closure method; the construction safety boundary constraint requires that the candidate detection time slices and corresponding lane closure methods meet the pre-set construction safety boundary conditions; the construction safety boundary conditions at least limit the minimum number of remaining traffic lanes and the allowable vehicle speed range inside and outside the detection operation area; by splicing several consecutive candidate detection time slices that meet the preset constraints on the time axis, a detection window interval that meets the minimum continuous operation time requirement is formed, and each detection window interval is assigned a comprehensive score, which is a weighted function of the average crack opening and closing index, detection duration, and traffic impact degree within the interval; several detection window intervals are selected as recommended detection tasks in descending order of comprehensive score.

2. The method for detecting hidden cracks in roads according to claim 1, characterized in that, When it is impossible to avoid the crack closure period by adjusting the detection time, a controlled thermal composite excitation is actively applied to the target road section to induce the latent cracks in the closed state to open. It receives baseline radar echo data and excitation radar echo data collected on the same target road segment and the same driving trajectory; it registers the baseline radar echo data and excitation radar echo data according to time and space; it uses the crack-sensitive waveform characteristics pre-calibrated when training the temperature crack response baseline model to perform point-by-point difference and envelope analysis on the baseline radar echo data and excitation radar echo data, suppressing uniform structure reflection signals and highlighting the new reflection anomalies caused by crack opening.

3. The method for detecting hidden cracks in roads according to claim 1, characterized in that, The temperature crack response baseline model is trained based on the temperature crack characteristics and the crack opening and closing index time series corresponding to the temperature crack characteristics. The temperature crack response baseline model is a long short-term memory network. The temperature crack characteristics include pavement temperature time series data, pavement vertical strain time series data, and traffic load statistical time series data.

4. The method for detecting hidden cracks in roads according to claim 1, characterized in that, The recommended detection task should include at least the detection start time and the detection end time.

5. The method for detecting hidden cracks in roads according to claim 1, characterized in that, include: The output of the temperature crack response baseline model also includes crack opening threshold temperature and crack closing threshold temperature; the crack opening threshold temperature is the lowest temperature value above which crack-related anomalous reflections begin to appear stably in the ground penetrating radar echo waveform, and the crack closing threshold temperature is the highest temperature value below which the anomalous reflections disappear and the waveform returns to a continuous homogeneous state.

6. The method for detecting hidden cracks in roads according to claim 5, characterized in that, The crack opening and closing index is a dimensionless value between zero and one. A crack opening and closing index close to zero indicates that the crack is closed or difficult to detect, while a crack opening and closing index close to one indicates that the crack is fully open and easily identified by ground penetrating radar.

7. The method for detecting hidden cracks in roads according to claim 2, characterized in that, Methods of combined thermal excitation include: Step 1: Based on the temperature crack response baseline model, determine the crack opening threshold temperature at the target road section, and calculate the required heating time based on the current measured ambient temperature and the initial road surface temperature. Step 2: Based on the current vehicle speed and the distance between the road surface heating device and the ground-penetrating radar array probe, calculate the length of the heating zone and the thermal action time window to obtain the start-stop control strategy associated with the vehicle position. Step 3: Before the inspection vehicle enters the target road section, start the road surface heating device to gradually increase the road surface temperature along the direction of travel to the crack opening threshold temperature. Step 4: After the temperature of the heated area reaches the crack opening threshold temperature, start the adjustable wheel load loading device to increase the tire ground pressure of the loading wheel group to the preset short-term high wheel load level, so that the heated area can simultaneously bear the superposition of temperature and load in the same spatial position and at the same time. Step 5: When the detection vehicle continues to move forward until the ground-penetrating radar array probe is above the thermal zone, the ground-penetrating radar is immediately triggered to collect radar echo data of the internal structure of the road surface after excitation.

8. The method for detecting hidden cracks in roads according to claim 7, characterized in that, Perform point-by-point differential operation on each pair of registered radar waveforms to obtain differential radar waveform sequences; An analysis window of a set length is slid along the depth direction on the differential radar waveform. The envelope and its amplitude statistics of the differential signal within the window are calculated to form a differential envelope profile along the depth direction. The pre-calibrated crack-sensitive waveform characteristics include at least: the depth range where cracks are likely to occur, the polarity type of the differential waveform, the range of continuous sampling points of the abnormal waveform, and the amplitude amplification factor threshold compared with the reflection of a normal uniform structure. The crack-sensitive waveform features are compared point by point with the differential envelope profile. Sampling points that simultaneously meet the following conditions are marked as crack response candidate points: depth falls within the crack-sensitive range, differential envelope amplitude exceeds a preset threshold, and waveform polarity is consistent with expectations. New reflection anomalies related to crack opening are retained.

9. A road hidden crack detection system, characterized in that, For implementing the method for detecting hidden cracks in roads according to any one of claims 1-8, the system comprises: The data acquisition module is used to collect time-series data of pavement temperature, pavement vertical strain, and traffic load statistics for the target road section. The data processing module is used to input pavement temperature time series data, pavement vertical strain time series data, and traffic load statistical time series data into a pre-trained temperature crack response baseline model to obtain the crack opening and closing index time series of the target road segment. The crack opening and closing index time series is formed by arranging the crack opening and closing indices of the target road segment in each time period in chronological order. The crack opening and closing index corresponding to each time period is defined as a continuous numerical index that gradually approaches zero below the crack closure threshold temperature of the corresponding time period, gradually approaches one above the crack opening threshold temperature of the corresponding time period, and is linearly or nonlinearly interpolated by temperature and load in the intermediate interval. The detection and reconstruction module is used to generate several candidate detection time slices from the crack opening and closing index time series according to a preset time step, and calculate the corresponding average crack opening and closing index and predicted traffic flow for each candidate detection time slice. Within each candidate detection time slice, different lane closure methods and speed control combinations are enumerated, and candidate detection time slices and corresponding lane closure methods and speed control combinations that meet preset constraints are selected. The preset constraints include crack detectability constraints, traffic carrying capacity constraints, and construction safety boundary constraints. The crack detectability constraint requires that the average crack opening and closing index corresponding to the candidate detection time slice is not lower than a preset crack detectability threshold. The traffic carrying capacity constraint requires that the average crack opening and closing index corresponding to the candidate detection time slice is not lower than a preset crack detectability threshold. The predicted traffic flow does not exceed the acceptable flow limit under the current lane closure method; the construction safety boundary constraint requires that the candidate detection time slices and corresponding lane closure methods meet the pre-set construction safety boundary conditions; the construction safety boundary conditions at least limit the minimum number of remaining traffic lanes and the allowable vehicle speed range inside and outside the detection operation area; by splicing several consecutive candidate detection time slices that meet the preset constraints on the time axis, a detection window interval that meets the minimum continuous operation time requirement is formed, and each detection window interval is assigned a comprehensive score, which is a weighted function of the average crack opening and closing index, detection duration and traffic impact degree within the interval; several detection window intervals are selected as recommended detection tasks in descending order of comprehensive score.

10. A road hidden crack detection system according to claim 9, characterized in that, Also includes: The equivalent excitation module is used to actively apply controlled thermal composite excitation to the target road section when it is impossible to avoid the crack closure period by adjusting the detection time, so as to induce the closed hidden cracks to open. The differential analysis module receives baseline radar echo data and excitation radar echo data collected on the same target road segment and the same driving trajectory. Based on time and space, it registers the baseline radar echo data and the excitation radar echo data. Using the crack-sensitive waveform characteristics pre-calibrated when training the temperature crack response baseline model, it performs point-by-point differential and envelope analysis on the baseline radar echo data and the excitation radar echo data to suppress uniform structure reflection signals and highlight the new reflection anomalies caused by crack opening.

Images