A precise calculation method and system for the amount of emulsified asphalt used in pavement maintenance engineering

Abstract

The application relates to the technical field of road engineering, and discloses a precise emulsified asphalt dosage calculation method and system for pavement maintenance engineering, wherein the precise emulsified asphalt dosage calculation method comprises the following steps: collecting three-dimensional color point cloud data of a road surface, and performing pretreatment; identifying a road surface disease type and position; inverting road surface material pore characteristics; calculating emulsified asphalt penetration depth and determining regional penetration depth in combination with a disease identification result data set; predicting an emulsified asphalt dosage; optimizing construction process parameters; generating a construction operation guide book and performing electrification; monitoring construction quality in real time and performing deviation analysis, and adjusting parameters based on deviation analysis results; and the disease identification model can quickly adapt to disease characteristics under different regions and different conditions by introducing migration learning and incremental learning technology.

Description

Technical Field

[0001] This invention relates to the field of road engineering technology, and more specifically, to a method and system for accurately calculating the amount of emulsified asphalt used in road maintenance projects. Background Technology

[0002] Road maintenance is a crucial measure to ensure the safe operation and extend the service life of road infrastructure. Among preventative maintenance techniques, over-surfacing, micro-surfacing, slurry seal, sand-containing fog seal, and fog seal are widely used due to their economic and efficient characteristics. Emulsified asphalt, as a key binder in these techniques, directly affects the quality and cost of the maintenance project through accurate calculation of its dosage.

[0003] Traditional methods for calculating emulsified asphalt usage rely heavily on construction experience and simplified empirical formulas. A usage range is typically provided based on pavement type and severity of damage, within which construction personnel adjust according to site conditions. This method has several shortcomings: it struggles to accurately reflect the actual condition of pavement damage. Traditional manual surveys or two-dimensional image recognition can only capture surface features, failing to measure three-dimensional parameters such as depth and volume, resulting in inaccurate input data for usage calculations. It neglects the impact of pavement material pore structure on emulsified asphalt penetration. Different types and service lives of pavement materials exhibit significant differences in porosity, pore size distribution, and other microscopic characteristics, but existing methods typically use uniform empirical values, failing to reflect the actual differences in material properties. Furthermore, it fails to adequately consider the role of environmental factors. Environmental conditions such as temperature and humidity affect the viscosity and demulsification rate of emulsified asphalt, thus influencing penetration depth and actual usage, but existing methods lack quantitative consideration of these environmental factors.

[0004] In recent years, with the development of sensing and artificial intelligence technologies, some pavement distress detection methods based on 3D scanning and image recognition have emerged. However, most of these methods focus on distress identification itself and have not yet established a complete technology chain from distress identification to material usage calculation. Existing technologies still have shortcomings in data acquisition quality control, distress identification model generalization ability, on-site acquisition of material parameters, and dynamic verification of calculation results, which limit further improvement in the accuracy of usage calculation.

[0005] Therefore, there is an urgent need to develop a precise calculation method for emulsified asphalt dosage that integrates high-quality data acquisition, intelligent defect identification, accurate dosage calculation, and closed-loop quality control, in order to meet the higher requirements of modern road maintenance projects for quality and efficiency. Summary of the Invention

[0006] This invention provides a method and system for accurately calculating the amount of emulsified asphalt used in road maintenance projects, solving the technical problems of low accuracy in calculating the amount of emulsified asphalt used and serious material waste in related technologies.

[0007] This invention provides a method for accurately calculating the amount of emulsified asphalt used in road maintenance projects, including: Collect three-dimensional color point cloud data of the road surface, perform preprocessing, and obtain high-quality three-dimensional road surface data; Based on high-quality 3D pavement data, the types and locations of pavement defects are identified, and a defect identification result dataset is obtained. Based on the damage identification result dataset, the pavement material porosity characteristics are inverted to obtain the pavement material porosity characteristic parameter dataset; Based on the pavement material pore characteristic parameter dataset, the penetration depth of emulsified asphalt is calculated and combined with the disease identification result dataset to determine the penetration depth of different regions, thus obtaining the effective penetration depth distribution data of different regions. Based on the effective penetration depth distribution data and disease identification result dataset of different regions, the amount of emulsified asphalt used is predicted, and the distribution data of emulsified asphalt used in different road sections is obtained. Based on the distribution data of emulsified asphalt usage in different road sections, the construction process parameters are optimized to obtain the construction parameter scheme for each road section. Based on the construction parameter scheme for each road section and the defect identification result dataset, a construction operation instruction manual is generated and digitized to obtain an electronic instruction manual. Based on the segmented construction parameter scheme and electronic guide, the construction quality is monitored in real time and deviation analysis is performed. Based on the deviation analysis results, the parameters are adjusted to obtain the project construction quality assessment report and the verified set of usage calculation parameters.

[0008] In a preferred embodiment, the preprocessing of the collected three-dimensional color point cloud data of the road surface includes: Data acquisition is performed using a dedicated inspection vehicle equipped with a 3D laser scanner, a high-resolution industrial camera, and an inertial navigation system. The Kalman filter algorithm is used to perform motion compensation on the point cloud data. A statistical outlier removal algorithm is used to denoise point cloud data, and points whose average distance exceeds a preset standard deviation multiple are identified as outliers and removed. The point cloud data is discretized into a three-dimensional voxel grid by using a voxel grid filtering method to obtain regularized point cloud grid data. A spatiotemporal synchronization method is used to establish the correspondence between point clouds and images, and the image color information is mapped to the point cloud grid points to obtain three-dimensional color point cloud data.

[0009] In a preferred embodiment, identifying the type and location of pavement defects includes: The local reference plane is calculated using the least squares plane fitting method. The elevation deviation of each point cloud grid point relative to the local reference plane is calculated. The elevation deviation is decomposed into texture component and disease component using the bilateral filtering method. An adaptive threshold segmentation method is used to mark grid points whose absolute elevation deviation exceeds a preset threshold as candidate disease points; A density-based clustering algorithm is used to aggregate candidate points into disease clusters to obtain the disease area segmentation results; Based on color image patches of the diseased area, texture and edge features of the image are extracted to construct an image feature vector; Based on the elevation data of the diseased area, the geometric features of the disease are extracted, and a geometric feature vector is constructed.

[0010] In a preferred embodiment, identifying the type and location of pavement defects further includes: Based on image feature vectors and geometric feature vectors, a comprehensive feature vector is constructed. The comprehensive feature vector is then input into a pre-trained deep convolutional neural network classification model, which outputs the probability distribution of disease types. Transfer learning techniques are used to fine-tune the pre-trained model, resulting in a fine-tuned model that adapts to the current disease characteristics of the project. For each diseased area, the classification results of multiple models are voted on to obtain the disease type identification result and identification confidence level.

[0011] In a preferred embodiment, the inverted pavement material pore characteristics include: Ground-penetrating radar was used to continuously scan along the road section to obtain time-domain signal data of road surface reflected waves; The instantaneous amplitude envelope of the signal is extracted using Hilbert transform, the reflection peaks at the interface are identified, and the propagation time of the electromagnetic wave in the road surface material is calculated. Based on two-way travel time data and surface thickness, the electromagnetic wave propagation speed is calculated, and the relative permittivity of the surface material is calculated using the relationship between electromagnetic wave speed and the relative permittivity of the material. The porosity is estimated by inverting the porosity of the pavement material using the equivalent medium theory of composite materials.

[0012] In a preferred embodiment, the inverted pavement material pore characteristics further include: Based on the information on pavement material type, the pavement material property database is queried to obtain the prior estimate of porosity and the estimation uncertainty based on the material type. Based on pavement service life data and traffic load level data, a pavement material aging model is used to calculate the aging correction coefficient and obtain the estimated porosity considering the aging effect. Based on the porosity estimates obtained from ground-penetrating radar inversion, prior estimates of material type, and aging correction estimates, a Bayesian fusion method is adopted to use the porosity estimates from the three sources as observations, establish a probabilistic model between the observations and the true porosity, and use the maximum a posteriori estimation criterion to obtain the optimal porosity estimate and its estimation variance after fusion.

[0013] In a preferred embodiment, calculating the penetration depth of emulsified asphalt and determining the regional penetration depth by combining the disease identification result dataset includes: Based on the material property test data of emulsified asphalt, the rheological parameters of emulsified asphalt at standard temperature were measured, and a viscosity-temperature dependence model was established. The permeability coefficient of porous media is described by a permeability coefficient calculation equation. Based on the rheological parameters of emulsified asphalt and the permeability coefficient of pavement materials, an extended form of the permeability law is used to describe the permeation process of emulsified asphalt in pavement materials. An exponential growth model is used to describe viscosity evolution, resulting in a viscosity time-dependent model. Based on the emulsified asphalt permeation kinetics model and viscosity-time dependence model, the numerical integration method is used to discretize the permeation time into time steps. At each time step, the permeation rate is calculated based on the viscosity at the current moment. When the permeation rate decreases to below a preset threshold, the permeation process is considered to have ended, and the final theoretical value of the permeation depth is obtained.

[0014] In a preferred embodiment, the predicted emulsified asphalt dosage includes: Hourly meteorological data for the construction date is obtained based on data from meteorological monitoring stations in the construction area. Based on the temperature data during the construction period, the viscosity-temperature dependence model of emulsified asphalt was used to calculate the viscosity correction coefficient of emulsified asphalt at the construction temperature. Based on the relative humidity data during the construction period, the humidity dependence model of emulsified asphalt demulsification rate is used to calculate the humidity correction coefficient for demulsification rate. Calculate the comprehensive correction factor, multiply the theoretical infiltration depth by the comprehensive correction factor, and obtain the actual infiltration depth after considering environmental factors; For each affected area, calculate the effective amount of emulsified asphalt used in that area. The amount is equal to the area multiplied by the penetration depth multiplied by the porosity multiplied by the density divided by the solid content. Add the amount of emulsified asphalt used for disease maintenance and the amount of emulsified asphalt used for preventive maintenance to obtain the total amount of emulsified asphalt used in each road segment unit.

[0015] In a preferred embodiment, the real-time monitoring of construction quality and the deviation analysis, followed by parameter adjustment based on the deviation analysis results, include: Real-time measurement of emulsified asphalt flow rate, measurement of actual vehicle speed, and real-time acquisition of vehicle location coordinates; For each sampling moment, read the flow rate, speed and location data, determine the road segment unit currently under construction based on the location data, calculate the actual spraying amount per unit area at the current moment based on the flow rate and speed data, compare the actual spraying amount with the target amount, and calculate the deviation value and deviation percentage. When the actual spraying volume deviates from the preset allowable range, a warning message is displayed on the vehicle terminal screen, and an audible alarm is triggered at the same time; Based on the deviation cause analysis report, pavement characteristic data and actual usage data of the deviation road sections are collected as new correction samples to adjust parameters and improve usage calculation methods.

[0016] In a preferred embodiment, a system for accurately calculating the amount of emulsified asphalt used in road maintenance projects is used to execute the above-described method for accurately calculating the amount of emulsified asphalt used in road maintenance projects, including: The data acquisition module is used to collect three-dimensional color point cloud data of the road surface, perform preprocessing, and obtain high-quality three-dimensional road surface data. The pavement defect identification module identifies the types and locations of pavement defects based on high-quality 3D pavement data, and obtains a defect identification result dataset. The pore feature inversion module, based on the defect identification result dataset, inverts the pore features of the pavement material to obtain the pavement material pore feature parameter dataset; The penetration depth calculation module calculates the penetration depth of emulsified asphalt based on the pavement material pore characteristic parameter dataset and determines the regional penetration depth by combining the disease identification result dataset, thus obtaining the effective penetration depth distribution data of different regions. The usage prediction module predicts the usage of emulsified asphalt based on the effective penetration depth distribution data and the disease identification result dataset in different regions, and obtains the emulsified asphalt usage distribution data for different road sections. The construction optimization module optimizes construction process parameters based on the distribution data of emulsified asphalt usage in different road sections, resulting in construction parameter schemes for each road section. Based on the construction parameter schemes for each road section and the defect identification result dataset, it generates construction operation instructions and digitizes them to obtain electronic instructions. The quality monitoring module monitors construction quality in real time and performs deviation analysis based on the construction parameter scheme and electronic guide for each road segment. Based on the deviation analysis results, it adjusts the parameters to obtain a project construction quality assessment report and a verified set of usage calculation parameters.

[0017] The beneficial effects of this invention are as follows: This invention uses three-dimensional laser scanning and high-resolution image fusion technology to accurately obtain the three-dimensional spatial parameters of road surface defects. Combined with the porosity characteristics of road surface materials obtained by ground-penetrating radar inversion, a multivariate usage prediction model considering defect morphology, material properties and environmental factors is established, realizing the transformation from qualitative estimation to quantitative calculation.

[0018] A fully intelligent technology system has been constructed, encompassing data acquisition, disease identification, usage calculation, construction optimization, and quality monitoring. This system not only solves the problem of usage calculation itself, but more importantly, it establishes a complete technology chain from data input to construction output, achieving organic connection and collaborative optimization of each link.

[0019] By introducing transfer learning and incremental learning techniques, the disease identification model can quickly adapt to disease characteristics in different regions and under different conditions. Through real-time monitoring and feedback correction mechanisms during the construction process, the computational model can be continuously optimized and self-improved, thereby improving the applicability and robustness of the method. Attached Figure Description

[0020] Figure 1 This is a flowchart of the main process of a method for accurately calculating the amount of emulsified asphalt used in road maintenance engineering according to the present invention. Figure 2 This is a detailed flowchart of a method for accurately calculating the amount of emulsified asphalt used in road maintenance projects according to the present invention; Figure 3 This is a module diagram of a system for accurately calculating the amount of emulsified asphalt used in road maintenance projects, as described in this invention. Detailed Implementation

[0021] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, some features described in the examples may be combined in other examples.

[0022] At least one embodiment of the present invention discloses a method for accurately calculating the amount of emulsified asphalt used in road maintenance projects, such as... Figures 1 to 2 As shown, it includes: Step 1: Collect 3D color point cloud data of the road surface, perform preprocessing, and obtain high-quality 3D road surface data; Step 1.1, Multi-source data acquisition and preprocessing; Based on the start and end station numbers and lane information of the road section to be maintained, a special inspection vehicle equipped with a 3D laser scanner, a high-resolution industrial camera, and an inertial navigation system was used. The vehicle traveled at a constant speed of 50 km / h. The scanner's scanning frequency was set to 100 Hz, generating 100 scan lines per second. The camera's shooting frequency was set to 20 frames per second, with an image resolution of 4096 pixels × 2160 pixels. This yielded raw 3D point cloud data and road surface image sequences covering the entire road section.

[0023] Based on the vehicle position, attitude, and velocity data recorded by the inertial navigation system, the Kalman filter algorithm is used to perform motion compensation on the point cloud data to eliminate point cloud distortion caused by vehicle movement and bumps, and obtain corrected point cloud data with a spatial resolution of 2 mm in the horizontal direction, 5 mm in the vertical direction, and 1 mm in the elevation direction.

[0024] Step 1.2, Noise Reduction and Regularization of Point Cloud Data; Based on the spatial distribution characteristics of point cloud data, a statistical outlier removal algorithm is employed. For each point, the average distance to its 50 neighboring points is calculated. Points whose average distance exceeds three standard deviations of the overall average are identified as outliers and removed, resulting in denoised point cloud data. Based on the denoised point cloud data, a voxel grid filtering method is used to discretize the point cloud data into a three-dimensional voxel grid with a spatial resolution of 10 mm × 10 mm × 5 mm. Centroid averaging is performed on all points within each voxel to obtain regularized point cloud grid data, with approximately 10,000 grid points per square meter of road surface.

[0025] Step 1.3, Color point cloud fusion and quality control; Based on the spatial coordinates of point cloud grid data and the timestamps of image sequences, a spatiotemporal synchronization method is used to establish the correspondence between point clouds and images. The color information of each pixel in the image is mapped to the nearest point cloud grid point, resulting in 3D color point cloud data containing both spatial coordinates and color information. Based on this 3D color point cloud data, a point cloud density analysis method is used to calculate the number of point clouds per square meter. When the point cloud density is below 8000 points per square meter, the data quality of that area is deemed insufficient, a data quality assessment report is generated, and road sections requiring additional scanning are marked, resulting in high-quality 3D road surface data and a quality assessment report.

[0026] In some embodiments, since road maintenance work may be carried out at night to reduce the impact on traffic, insufficient nighttime lighting conditions can affect image quality. An active lighting system can be used in conjunction with a camera for data acquisition, with the aim of obtaining high-quality road surface images under low-light conditions. Specifically, a high-brightness LED array lighting is installed in front of the detection vehicle. The lighting has a spectral range of 400 nm to 700 nm and an illuminance of over 1000 lux at the road surface. The lighting is triggered synchronously with the camera, turning on at the moment of camera exposure and turning off after exposure to avoid glare interference to other road users. Using this method, road surface image data of comparable quality to daytime can be obtained at night.

[0027] Step 2: Based on high-quality 3D pavement data, identify the types and locations of pavement defects to obtain a defect identification result dataset; Step 2.1, Disease area detection and boundary determination; Based on the high-quality 3D pavement data output in step 1, a least-squares plane fitting method is used to calculate a local reference surface for each 10-meter × lane width road segment. This reference surface represents the pavement morphology under ideal, distress-free conditions for that segment. The elevation deviation of each point cloud grid point relative to the local reference surface is calculated, resulting in a pavement elevation deviation distribution map. Based on this distribution map, a bilateral filtering method is used to separate the pavement texture features at different scales. The spatial scale parameter of the filter is set to 30 mm, and the amplitude scale parameter is set to 3 mm. The elevation deviation is decomposed into a coarse aggregate texture component and a distress component, resulting in a distress elevation feature map after removing the coarse aggregate texture. Based on this distress elevation feature map, an adaptive threshold segmentation method is used to calculate the mean and standard deviation of the elevation deviation for each local region. Grid points with an absolute elevation deviation exceeding the mean plus twice the standard deviation are marked as candidate distress points, resulting in a set of candidate distress points. Based on the candidate disease point set, a density-based clustering algorithm is used to aggregate candidate points with a spatial distance of less than 50 mm into disease clusters. Small clusters containing fewer than 100 points are removed, resulting in preliminary disease region segmentation. Each disease region corresponds to a continuous disease cluster. Based on the disease region segmentation results and 3D color point cloud data, a region growing method is used, with disease clusters as seed points, to expand outward to locations where the elevation deviation gradient is less than a threshold, thus determining the precise boundary of the disease region. Simultaneously, a point cloud color image patch is extracted for each disease region. The size of the image patch is a rectangle extending 100 mm outward from the disease region, resulting in the boundary polygon of each disease region and its corresponding color image patch.

[0028] Step 2.2, Disease feature extraction and vector construction; Based on color image patches, the gray-level co-occurrence matrix method is used to extract the texture features of the image, including four feature parameters: contrast, correlation, energy, and homogeneity. The Canny edge detection operator is used to extract two feature parameters: edge intensity and edge density. Combined with two statistical feature parameters: the average gray level and gray level standard deviation, an 8-dimensional image feature vector is constructed to obtain the image feature vector of each diseased area.

[0029] Based on the elevation data of the diseased area, the geometric features of the disease are extracted, including five shape feature parameters: disease area, perimeter, equivalent diameter, aspect ratio, and compactness, as well as four depth feature parameters: maximum depth, average depth, depth standard deviation, and depth skewness. A nine-dimensional geometric feature vector is constructed to obtain the geometric feature vector of each diseased area.

[0030] Step 2.3, Deep Learning Classification and Transfer Learning; Based on image feature vectors and geometric feature vectors, a 17-dimensional comprehensive feature vector is constructed using a feature concatenation method. This comprehensive feature vector is then input into a pre-trained deep convolutional neural network classification model. This model contains 5 convolutional layers, 3 pooling layers, and 2 fully connected layers. It is pre-trained on a pavement distress dataset containing 10,000 labeled samples and outputs the probability distribution of four distress types: cracks, potholes, loose surfaces, and ruts. The type with the highest probability is selected as the recognition result, thus obtaining the preliminary classification result for each distress area.

[0031] Based on 100 typical disease samples obtained from the preliminary investigation of the current maintenance project, these samples were manually labeled by experienced engineers. Using transfer learning technology, the parameters of the first four convolutional layers of the pre-trained model were fixed, and only the fifth convolutional layer and the fully connected layer were fine-tuned. The number of training rounds was set to 50, the learning rate was set to 0.0001, and the cross-entropy loss function and Adam optimizer were used to obtain a fine-tuned model adapted to the disease characteristics of the current project.

[0032] Step 2.4, Ensemble learning and confidence assessment; Based on the fine-tuned model, all diseased areas are classified again. At the same time, three classification models with different network structures are trained, including residual network, dense connection network and attention network. For each diseased area, the classification results of the three models are voted on. When at least two models give the same category, the category is adopted as the final result. Otherwise, it is marked as needing manual review. The disease type identification result and identification confidence are obtained. The confidence is defined as the proportion of the number of models that give the final category to the total number of models.

[0033] Step 2.5, Quantification of disease parameters and construction of database; Based on the disease type identification results, for crack-type diseases, a skeleton extraction algorithm is used to obtain the centerline of the crack. A depth profile perpendicular to the centerline is extracted every 10 mm along the centerline. The maximum depth point is identified on each profile. The maximum depth, average depth and depth variation coefficient of the crack are calculated to obtain detailed depth parameters of the crack disease.

[0034] Based on the disease type identification results, for pothole-type diseases, the elevation deviation of all point cloud grid points in the pothole area is integrated to calculate the volume of the pothole. The volume is then divided by the pothole area to obtain the average depth of the pothole, thus obtaining the volume and average depth parameters of the pothole disease.

[0035] Based on the identification results of the disease type, for loose diseases, the distribution of reflection intensity of point cloud in the loose area is statistically analyzed. Low reflection intensity indicates serious material loss. The average value and standard deviation of the reflection intensity are calculated, and the looseness severity index is defined as the ratio of the reflection intensity of the intact pavement to the reflection intensity of the loose area to obtain the severity parameter of the loose disease.

[0036] Based on parameters such as type, location coordinates, area, depth, volume, and severity of all diseased areas, a disease database is constructed. The database is stored in the form of relational tables. Each disease record contains fields such as disease number, station number, lateral location, type, geometric parameters, and confidence level, resulting in a disease identification result dataset. The disease identification result dataset contains information such as the type, location, and geometric parameters of each disease, providing disease distribution data for subsequent pore feature inversion and dosage calculation.

[0037] In some embodiments, due to significant differences in road surface material types and climate conditions in different regions, the apparent characteristics of the same type of disease may differ, and a fixed classification model may make identification errors. An active learning strategy can be adopted to improve the generalization ability of the model. The aim is to continuously optimize the model during project implementation. Specifically, during the disease identification process, for disease areas with an identification confidence level below 0.6, their feature data are automatically extracted and saved to a sample library to be labeled. Every 50 samples to be labeled are manually labeled by engineers, and the labeled samples are added to the training set for retraining and fine-tuning the model to update the classification model. This method allows the model to continuously learn new disease characteristics during project implementation, and the identification accuracy gradually improves.

[0038] Step 3: Based on the damage identification result dataset, invert the pavement material pore characteristics to obtain the pavement material pore characteristic parameter dataset; Step 3.1, Ground Penetrating Radar Detection and Porosity Inversion; Based on the station number and lateral location of each defect in the defect identification result dataset output in step 2, ground-penetrating radar (GPR) is used to continuously scan along the road segment. The center frequency of the GPR is set to 1000 MHz, the time window to 20 nanoseconds, and the scanning interval to 50 mm. The GPR emits electromagnetic pulses and receives the echo signals reflected from the road surface, obtaining the time-domain signal data of the road surface reflection wave. Based on the time-domain signal data of the reflection wave, Hilbert transform is used to extract the instantaneous amplitude envelope of the signal, identify the reflection peaks at the air-to-road surface interface and the surface-to-base interface, measure the time difference between the air reflection peak and the surface reflection peak, and combine this with the propagation speed of electromagnetic waves in air (300,000 km / s) to calculate the propagation time of electromagnetic waves in the road surface material, obtaining the two-way travel time data of the surface material. Based on the two-way travel time data of the surface material and the known surface thickness (typically 40-60 mm, obtained through core sampling or design data), the electromagnetic wave propagation speed in the surface material is calculated using the relationship that the electromagnetic wave velocity is twice the thickness divided by the two-way travel time, yielding the electromagnetic wave velocity data for the surface material. Based on this electromagnetic wave velocity data, the relative permittivity of the surface material is calculated using the relationship between the electromagnetic wave velocity and the material's relative permittivity (equal to the speed of light in vacuum divided by the square of the electromagnetic wave velocity in the material), resulting in the permittivity distribution data for the surface material. Based on the dielectric constant distribution data of the surface material, the equivalent medium theory of composite materials is adopted. The road surface material is regarded as a three-phase mixture of mineral aggregate, asphalt, and air. According to the Lichtenecker-Rother mixing formula, the equivalent dielectric constant of the material satisfies that the cube root of the equivalent dielectric constant is equal to the volume fraction of mineral aggregate multiplied by the cube root of the dielectric constant of mineral aggregate plus the volume fraction of asphalt multiplied by the cube root of the dielectric constant of asphalt plus the volume fraction of air multiplied by the cube root of the dielectric constant of air. Among them, the dielectric constant of mineral aggregate is taken as 6.5, the dielectric constant of asphalt is taken as 3.0, and the dielectric constant of air is taken as 1.0. The volume fraction of air is the porosity. The porosity is solved by inversion to obtain the estimated value of the porosity of the road surface material based on ground penetrating radar measurement.

[0039] Step 3.2, Multi-source porosity estimation and Bayesian fusion; Based on the information on pavement material type, the pavement material property database is queried. The database contains the standard porosity range of commonly used pavement materials such as asphalt concrete, cement concrete, and cement-stabilized crushed stone in the newly constructed state. For dense-graded asphalt concrete, the standard porosity range is 3 to 5, and for open-graded asphalt concrete, the standard porosity range is 18 to 25. The a priori estimate of porosity and the estimation uncertainty based on material type are obtained.

[0040] Based on pavement service life data and traffic load level data, a pavement material aging model is adopted. This model describes the evolution of porosity over time. For asphalt pavement, under the influence of traffic load and environmental factors, the porosity of the surface material gradually decreases. The rate of decrease is related to traffic load and climate conditions. An aging correction coefficient is calculated, which represents the ratio of actual porosity to standard porosity. For pavements used for less than 5 years, the aging correction coefficient is taken as 0.9 to 1.0; for pavements used for 5 to 10 years, the aging correction coefficient is taken as 0.8 to 0.9; and for pavements used for more than 10 years, the aging correction coefficient is taken as 0.7 to 0.8. Porosity estimates considering the effects of aging are obtained.

[0041] Based on the porosity estimates obtained from ground-penetrating radar inversion, prior estimates of material type, and aging correction estimates, a Bayesian fusion method is adopted to use the porosity estimates from the three sources as observations. A probabilistic model between the observations and the true porosity is established. Assuming that the observation error follows a normal distribution, the maximum a posteriori estimation criterion is used to obtain the optimal estimate of the fused porosity and its estimation variance. The estimation variance reflects the degree of uncertainty of the porosity.

[0042] Step 3.3, Porosity correction and distribution map construction under the influence of disease; Based on the defect identification result dataset output in step 2, for crack-type defect areas, cracks damage the integrity of the material and increase the local pore space. The crack influence model is used to calculate the porosity increment caused by cracks, where the porosity increment is equal to the crack opening width multiplied by the crack depth divided by the area of ​​the calculation unit. The porosity increment is superimposed on the porosity of the intact area to obtain the effective porosity of the crack area.

[0043] Based on the disease identification dataset, for pothole-type disease areas, the potholes form open spaces that can be directly filled by emulsified asphalt without considering pore permeability. The effective porosity of the pothole area is set to 100, representing a completely open space, thus obtaining the effective porosity of the pothole area. For loose-type disease areas, the looseness of the material leads to increased porosity. Using the looseness influence model, the porosity increase coefficient is calculated based on the looseness severity index. For mild looseness, the increase coefficient is 1.2 to 1.5; for moderate looseness, it is 1.5 to 2.0; and for severe looseness, it is 2.0 to 3.0. The porosity of the intact area is multiplied by the increase coefficient to obtain the effective porosity of the loose area.

[0044] Based on the effective porosity of intact areas and various types of damaged areas, as well as the uncertainty of porosity, a spatial distribution map of pavement material porosity characteristics is constructed. This distribution map covers the entire maintenance section, and the spatial resolution is consistent with the grid resolution of the high-quality three-dimensional pavement data output in step 1, which is 10 mm × 10 mm, thus obtaining a dataset of pavement material porosity characteristic parameters.

[0045] In some embodiments, due to the limited detection depth of ground-penetrating radar (GPR), it is difficult to obtain dielectric constant information of the base course and subbase course for pavement structures with a thickness greater than 100 mm. Multi-frequency GPR can be used in combination to obtain material properties of pavement structural layers at different depths. Specifically, a low-frequency GPR with a center frequency of 400 MHz is used to detect the pavement base course and subbase course, with a detection depth of up to 500 mm. A high-frequency GPR with a center frequency of 1000 MHz is used to detect the pavement surface layer, with a detection depth of 100 mm but high resolution. The GPR data from the two frequencies are jointly inverted, and a layered medium model is used to solve for the dielectric constant and thickness of the surface layer, base course, and subbase course sequentially from top to bottom. This yields complete material property parameters of the multi-layered pavement structure, which are used to evaluate the impact of different depths of distress on the penetration of emulsified asphalt.

[0046] Step 4: Based on the pavement material pore characteristic parameter dataset, calculate the penetration depth of emulsified asphalt and combine it with the disease identification result dataset to determine the penetration depth of different regions, thereby obtaining the effective penetration depth distribution data of different regions. Step 4.1: Obtaining performance parameters of emulsified asphalt materials; Based on the material property test data of emulsified asphalt, the rheological parameters of emulsified asphalt, such as viscosity, density, surface tension, and solid content, were measured at a standard temperature of 25 degrees Celsius. Viscosity was measured using a rotational viscometer, with typical values ​​ranging from 10 to 50 mPa·s; density was measured using a densitometer, with typical values ​​ranging from 1000 kg / m³ to 1050 kg / m³; surface tension was measured using a surface tension meter, with typical values ​​ranging from 30 to 40 mN / m; and solid content was measured using an evaporation residue test, with typical values ​​ranging from 50 to 70. These were used to obtain the baseline rheological parameters of emulsified asphalt.

[0047] To establish the viscosity-temperature dependence, the viscosity of emulsified asphalt was measured using a rotational viscometer at different temperatures of 5°C, 15°C, 25°C, 35°C, and 45°C. Each temperature point was measured three times and the average value was taken to obtain experimental data points showing the viscosity change with temperature. Based on these experimental data points, a viscosity-temperature characteristic curve of the emulsified asphalt was plotted, which describes the relationship between viscosity and temperature. A logarithmic function was used for fitting, and the logarithm of viscosity is linearly related to the reciprocal of temperature. The fitting parameters were determined by the least squares method to obtain the viscosity-temperature dependence model.

[0048] Step 4.2, Permeability coefficient calculation and dynamic modeling; Based on the porosity, average pore size, and pore connectivity of the pavement material pore characteristic parameter dataset output in step 3, the Kozeny-Carman equation is used to describe the permeability coefficient of porous media. The permeability coefficient is equal to the cube of the porosity multiplied by the square of the average pore size divided by the pore tortuosity, then divided by 180, then divided by 1 minus the square of the porosity. The pore tortuosity reflects the degree of curvature of the pore channels, with typical values ​​ranging from 1.5 to 3.0, thus obtaining the permeability coefficient distribution of the pavement material. Based on the rheological parameters of emulsified asphalt and the permeability coefficient of the pavement material, an extended form of Darcy's law is used to describe the permeation process of emulsified asphalt in the pavement material. The permeation rate is equal to the permeability coefficient multiplied by the pressure gradient divided by the viscosity. Initially, the emulsified asphalt forms a liquid film on the pavement surface with a thickness equal to the spraying amount divided by the density. Under the action of gravity, the liquid film permeates into the pavement interior. The pressure gradient is equal to the gravitational acceleration multiplied by the density. A differential equation for the evolution of the permeation depth over time is established, resulting in the emulsified asphalt permeation kinetic model.

[0049] Step 4.3, Demulsification kinetic coupling and numerical solution; Based on the demulsification kinetics of emulsified asphalt, after the emulsified asphalt comes into contact with the road material, the emulsifier is adsorbed on the material surface, the emulsion droplets coalesce, and the asphalt phase precipitates, resulting in an increase in viscosity over time. An exponential growth model is used to describe the viscosity evolution, where viscosity is equal to the initial viscosity multiplied by the demulsification rate constant multiplied by the time exponent. The demulsification rate constant depends on the type of emulsifier, the surface properties of the material, and the ambient temperature, with a typical value of 0.001 to 0.01 seconds, thus obtaining a viscosity time-dependent model.

[0050] Based on the emulsified asphalt permeation kinetics model and viscosity-time dependence model, the numerical integration method is used to discretize the permeation time into 1-second time steps. At each time step, the permeation velocity is calculated based on the viscosity at the current moment, the permeation depth increment is calculated based on the permeation velocity, and the permeation depth is updated. At the same time, the viscosity is updated according to the demulsification model. When the permeation velocity drops below the threshold of 0.01 mm / s, the permeation process is considered to have ended, and the final theoretical value of the permeation depth is obtained.

[0051] Step 4.4, Uncertainty assessment and confidence interval calculation; Due to the uncertainty of pavement material porosity, a Monte Carlo simulation method was used to assess the uncertainty of permeation depth. 1000 samples were randomly selected from the probability distribution of porosity, and the permeation depth was calculated repeatedly for each sample, resulting in 1000 permeation depth values. The mean, standard deviation, and probability distribution of the permeation depth were statistically analyzed to obtain the statistical characteristics of the permeation depth. Based on the probability distribution of the permeation depth, a confidence interval method was used to calculate the permeation depth interval at a 95% confidence level. This interval represents a 95% probability that the actual permeation depth lies within this interval. The lower limit of this interval was taken as the design value for the permeation depth to ensure the conservatism of the dosage calculation, avoid insufficient dosage, and obtain a design value for the permeation depth with confidence guarantee.

[0052] Step 4.5: Determine the infiltration depth for each region; Based on the disease type and permeation depth design value in the disease identification result dataset output in step 2, for intact pavement areas, the permeation depth adopts the theoretically calculated design value, with a typical value of 3 to 8 mm. For cracked areas, the permeation depth is equal to the crack depth plus the permeation depth of the intact area. For pothole areas, the permeation depth is equal to the pothole depth. For loose areas, the permeation depth is equal to the permeation depth of the intact area multiplied by the square root of the ratio of the porosity of the loose area to the porosity of the intact area, thus obtaining the effective permeation depth distribution data for different areas.

[0053] In some embodiments, since the demulsification rate of emulsified asphalt is greatly affected by road surface humidity during actual construction, the demulsification rate slows down when the road surface is wet. A method of detecting road surface moisture content and dynamically correcting the demulsification model can be adopted to improve the applicability of penetration depth calculation under different humidity conditions. Specifically, during road surface data collection, an infrared thermal imager is used to measure the surface temperature distribution of the road surface. Wet road sections are identified based on temperature anomaly areas. For road sections identified as wet, a small number of road material samples are collected, and the moisture content is measured using the drying method. A relationship model between moisture content and demulsification rate constant is established. For every 1% increase in moisture content, the demulsification rate constant decreases by 10 to 20%. The corrected demulsification rate constant is applied to the penetration depth calculation to obtain a penetration depth value that takes into account the influence of road surface humidity.

[0054] Step 5: Based on the effective penetration depth distribution data and disease identification result dataset of different regions, predict the amount of emulsified asphalt used to obtain the distribution data of emulsified asphalt used in different road sections. Step 5.1, Environmental Factor Correction and Actual Permeability Measurement; Based on meteorological monitoring station data for the construction area, hourly meteorological data for the construction period were obtained, including air temperature, relative humidity, wind speed, air pressure, and solar radiation intensity. The measurement accuracy for air temperature was 0.1 degrees Celsius, relative humidity was 1 degree Celsius, and wind speed was 0.1 meters per second, resulting in a detailed meteorological data sequence for the construction period. Based on the air temperature data during the construction period, an emulsified asphalt viscosity temperature dependence model was used to calculate the viscosity of emulsified asphalt at the construction temperature. Compared to a reference temperature of 25 degrees Celsius, for every 10 degrees Celsius increase in temperature, the viscosity decreased by 0.5 to 0.7 times; for every 10 degrees Celsius decrease in temperature, the viscosity increased by 1.5 to 2.0 times, yielding a viscosity correction coefficient for emulsified asphalt at the construction temperature. Based on the relative humidity data during the construction period, an emulsified asphalt demulsification rate humidity dependence model was used. Increased relative humidity slows down water evaporation and reduces the demulsification rate; for every 10% increase in relative humidity, the demulsification rate constant decreased by 5 to 10. The impact of humidity on the demulsification rate was calculated, yielding a humidity correction coefficient for the demulsification rate. Based on wind speed data during construction, a pavement moisture evaporation model was used. Increased wind speed accelerates surface moisture evaporation and promotes demulsification. For every 1 meter per second increase in wind speed, the evaporation rate increases by 10 to 15. The impact of wind speed on moisture evaporation was calculated, yielding an evaporation rate wind speed correction coefficient. Based on solar radiation intensity data during construction, solar radiation heats the pavement surface, increasing the temperature of emulsified asphalt and the pavement itself. For every 100 watts per square meter increase in radiation intensity, the pavement surface temperature rises by approximately 2 to 3 degrees Celsius. The impact of solar radiation on pavement temperature was calculated, yielding a pavement temperature radiation correction coefficient. Based on viscosity correction coefficients, demulsification rate correction coefficients, evaporation rate correction coefficients, and pavement temperature correction coefficients, a multi-factor coupling method was used. The comprehensive correction coefficient is equal to the square root of the viscosity correction coefficient multiplied by the demulsification rate correction coefficient multiplied by the evaporation rate correction coefficient multiplied by the pavement temperature correction coefficient. This comprehensive correction coefficient reflects the combined influence of environmental factors on the penetration depth of emulsified asphalt. The theoretical penetration depth in the effective penetration depth distribution data of different areas output in step 4 was multiplied by the comprehensive correction coefficient to obtain the actual penetration depth considering environmental factors.

[0055] Step 5.2, calculating usage by region and summarizing by road segment unit; Based on the area of ​​each disease in the disease identification result dataset output in step 2, the actual penetration depth calculated in step 5.1, and the density and solid content of emulsified asphalt, for each disease area, the effective amount of emulsified asphalt used in that area is calculated. The amount is equal to the area multiplied by the penetration depth multiplied by the porosity in the pavement material pore characteristic parameter dataset output in step 3 multiplied by the density divided by the solid content. The division by the solid content is because emulsified asphalt is sprayed instead of pure asphalt. The solid content represents the mass percentage of actual asphalt in the emulsified asphalt, thus obtaining the amount of emulsified asphalt used in each disease area. The 3D pavement model constructed based on the high-quality 3D pavement data output in step 1 divides the entire maintenance section into standard road segment units of 10 meters × lane width. For each road segment unit, the amount of emulsified asphalt used in all the defective areas it contains is accumulated to obtain the amount of emulsified asphalt used for defective maintenance. For the intact areas in the road segment unit, the area of ​​the intact area is calculated, multiplied by the actual penetration depth, porosity, and density of the intact area, and then divided by the solid content to obtain the amount of emulsified asphalt used for preventive maintenance. The amounts of emulsified asphalt used for defective maintenance and preventive maintenance are added together to obtain the total amount of emulsified asphalt used in each road segment unit. The amount of emulsified asphalt used in all road segment units is summarized to obtain the distribution data of emulsified asphalt used in each road segment.

[0056] Step 5.3, visualization of usage distribution and total amount calculation; Based on the total emulsified asphalt usage and area of ​​each road segment unit, the usage per unit area is calculated in kilograms per square meter or liters per square meter, generating a usage distribution map for the entire road segment. This map uses color depth to represent usage volume, with darker colors for segments with higher usage and lighter colors for segments with lower usage, providing a visual representation of the spatial distribution of usage. Based on the emulsified asphalt usage of all road segment units, the theoretical total usage of emulsified asphalt for the entire maintenance project is summed. Considering material losses during construction, including equipment and pipeline residues, uneven spraying, and boundary overflows, a construction loss coefficient is used for correction. For experienced construction teams and high-precision equipment, the loss coefficient is 1.05; for moderately experienced teams and medium-precision equipment, it is 1.08; and for less experienced teams and lower-precision equipment, it is 1.10. Multiplying the theoretical total usage by the loss coefficient yields the actual total amount of emulsified asphalt procured, taking into account construction losses.

[0057] Step 5.4, usage statistics and composition analysis; Based on the usage data of each road segment, statistical analysis is performed to calculate the mean, standard deviation, maximum, minimum and coefficient of variation of the usage of the entire road segment. The coefficient of variation reflects the dispersion of the usage distribution. A large coefficient of variation indicates that the usage varies greatly between different road segments, requiring frequent adjustments to equipment parameters during construction. A small coefficient of variation indicates that the usage distribution is uniform and the construction parameters are relatively stable, thus obtaining the statistical characteristic parameters of the usage distribution.

[0058] Based on the correspondence between disease types and usage in the disease identification dataset output in step 2, the usage proportions of cracked areas, pothole areas, loose areas, and intact areas are statistically analyzed to determine the contribution of different disease types to the total usage. This provides data support for engineering decisions. Specifically, the usage composition analysis can identify the main disease patterns of the pavement and determine the disease types with the highest usage, providing a basis for subsequent maintenance priorities and resource allocation. At the same time, differentiated construction quality control measures are formulated based on the usage proportions of different disease types, and these data assets serve as a reference for cost estimation and maintenance strategy formulation for similar future projects, resulting in a usage composition analysis report categorized by disease type.

[0059] In some embodiments, due to the possibility of heavy traffic or special usage requirements on certain special road sections, higher maintenance quality is required, necessitating an increase in the amount of emulsified asphalt used to improve the durability of the maintenance layer. A differentiated adjustment method based on the importance of road sections can be adopted. The aim is to provide more reliable maintenance guarantees for critical road sections. Specifically, the importance coefficient of a road section is calculated based on parameters such as traffic volume level, truck ratio, slope, and curve radius. For heavy-load road sections with an average daily traffic volume exceeding 50,000 vehicles and a truck ratio exceeding 30%, the importance coefficient is taken as 1.15 to 1.25. For road sections with long longitudinal slopes or small-radius horizontal curves, the importance coefficient is taken as 1.10 to 1.20. For general road sections, the importance coefficient is taken as 1.0. The standard amount is multiplied by the importance coefficient to obtain the differentiated amount. This method can achieve optimized allocation of maintenance resources, ensuring overall maintenance effectiveness while prioritizing key road sections.

[0060] Step 6: Based on the distribution data of emulsified asphalt usage in different road sections, optimize the construction process parameters to obtain the construction parameter scheme for each road section; based on the construction parameter scheme for each road section and the defect identification result dataset, generate the construction operation instruction manual and digitize it to obtain the electronic instruction manual. Step 6.1: Obtaining equipment parameters and establishing response equations; Based on the selected construction equipment model, the equipment technical parameter database is queried to obtain parameters such as spraying width, flow rate adjustment range, travel speed range, and flow control accuracy. For commonly used emulsified asphalt distributors, the spraying width is adjustable from 2.5 meters to 4.5 meters, the flow rate adjustment range is 10 liters to 100 liters per minute, the travel speed range is 2 kilometers per hour to 10 kilometers per hour, and the flow control accuracy is ±3 of the set value. These are the technical performance parameters of the construction equipment. Based on these parameters, a mathematical relationship is established between the construction parameters and the spraying volume per unit area. The spraying volume per unit area equals the equipment flow rate divided by the spraying width and then divided by the travel speed. In this formula, the flow rate is in liters per minute, the spraying width is in meters, and the travel speed is in kilometers per hour (which needs to be converted to meters per minute by multiplying the speed value by 1000 and then dividing by 60). The spraying volume per unit area is in liters per square meter. This yields the construction parameter response equation.

[0061] Based on the target unit area usage of each road segment in the emulsified asphalt usage distribution data output in step 5, the reverse calculation method is used to calculate the combination of equipment flow rate and driving speed that satisfies the target usage from the construction parameter response equation. For a given spraying width, the flow rate is equal to the target usage multiplied by the spraying width multiplied by the driving speed. There are infinitely many combinations of flow rate and speed that satisfy the target usage, thus obtaining the feasible solution space of the construction parameters.

[0062] Step 6.2: Establishing constraints and solving multi-objective optimization problems; Based on the feasible solution space of construction parameters and equipment technical performance constraints, constraints are established: the flow rate must be within the equipment flow adjustment range, and the travel speed must be within the equipment speed range. Simultaneously, considering construction quality requirements, the travel speed should not be too fast to ensure spray uniformity; the upper speed limit is set to 6 km / h, and the lower speed limit is set to 3 km / h to ensure construction efficiency. Feasible solutions satisfying all constraints are screened, resulting in a set of constrained feasible solutions for the construction parameters. Based on this set of constrained feasible solutions, a multi-objective optimization model is established. The objective functions include construction efficiency and usage uniformity. The construction efficiency objective is represented by maximizing the travel speed, while the usage uniformity objective, based on the emulsified asphalt usage distribution data for each road segment output in step 5, is represented by minimizing the flow rate variation rate between adjacent road segment units. A weighted summation method is used to transform the two objective functions into a single objective function. The weight coefficients are determined according to the actual engineering needs; if efficiency is prioritized, the efficiency objective has a higher weight; if quality is prioritized, the uniformity objective has a higher weight. A typical weight configuration is 0.4 for efficiency and 0.6 for uniformity, resulting in a comprehensive objective function. Based on the comprehensive objective function and the set of feasible solutions, a genetic algorithm is used to solve for the optimal construction parameters. The population size of the genetic algorithm is set to 100, the number of generations is set to 500, the crossover probability is set to 0.8, the mutation probability is set to 0.1, and the fitness function is the negative value of the comprehensive objective function. Through iterative evolution, the individuals in the population gradually move closer to the optimal solution. When the change in the optimal fitness is less than 0.001 for 50 consecutive generations, the algorithm is considered to have converged, and the globally optimal combination of equipment flow and driving speed parameters is obtained.

[0063] Step 6.3, parameter smoothing and differential adjustment; Based on the optimized construction parameters, considering the continuity of parameters between adjacent road segment units and avoiding frequent parameter changes that could lead to construction difficulties, a moving average filtering method is adopted. For the construction parameters of each road segment unit, the parameters of the unit and the two units before and after it, for a total of five units, are averaged to obtain smoothed construction parameters. These parameters have good continuity while ensuring the required usage.

[0064] Based on the smoothed construction parameters and the emulsified asphalt usage distribution data for each road segment output in step 5, for road segments with high-density defects where the usage is higher than the average level, separate construction parameters are set. The driving speed is appropriately reduced to 2 to 3 kilometers per hour, and the flow rate is increased to the upper limit allowed by the equipment to ensure that the defective areas are fully cured. For general road segments where the usage is close to the average level, standard construction parameters are used, with a driving speed of 4 to 5 kilometers per hour and the flow rate adjusted according to the target usage, thus obtaining the road segment construction parameter scheme.

[0065] Step 6.4, Generation and electronic output of construction instructions; Based on the segmented construction parameter scheme, and combining the defect distribution information in the defect identification result dataset output in step 2 and the segmented emulsified asphalt usage distribution data output in step 5, a detailed construction operation guide is generated. The guide is in tabular form, with columns including segment number, starting station number, ending station number, segment length, lane location, main defect type, defect area percentage, target usage, spraying width, equipment flow rate, driving speed, construction duration, and precautions. Each segment unit corresponds to one row in the table, resulting in a structured construction guidance data table. Based on the construction guidance data table, a printable construction guidance document is generated in electronic document format. Simultaneously, the data is exported as an electronic file format readable by the equipment. For sprayers equipped with an onboard intelligent control system, the electronic file can be directly loaded into the onboard terminal, enabling automatic setting and real-time display of construction parameters, reducing human error, and resulting in an electronic guide.

[0066] In some embodiments, since multiple devices may operate simultaneously during actual construction, and the performance parameters of different devices may differ, it is necessary to rationally allocate construction tasks. A multi-device collaborative optimization method can be adopted to improve overall construction efficiency and balance the workload of each device. Specifically, for multiple devices, the technical performance parameters of each device are obtained. Based on the usage requirements of each road segment and the device performance, a task allocation optimization model is established. The goal is to minimize the total construction time and make the working time of each device as close as possible. Integer programming is used to solve the optimal task allocation scheme, determine the road segment range responsible for each device, optimize the construction parameters of each device, and generate a dedicated construction instruction manual for each device. This method can fully leverage the efficiency advantages of multi-device collaborative operation.

[0067] Step 7: Based on the construction parameter scheme for each road segment and the electronic guide, monitor the construction quality in real time and conduct deviation analysis. Adjust the parameters based on the deviation analysis results to obtain the project construction quality assessment report and the verified set of usage calculation parameters. Step 7.1, Real-time monitoring and early warning adjustments; Based on the construction equipment, an electromagnetic flow sensor is installed on the emulsified asphalt delivery pipeline of the sprayer truck. The sensor measures the emulsified asphalt flow rate through the pipeline in real time, with a measurement accuracy of ±0.5 times full scale and a sampling frequency of 10 times per second. A speed sensor is installed on the vehicle chassis to measure the actual vehicle speed, with a measurement accuracy of 0.1 kilometers per hour and a sampling frequency of 10 times per second. A high-precision GNSS positioning receiver is installed on the top of the vehicle to acquire the vehicle's position coordinates in real time, with a positioning accuracy of 0.1 meters and a sampling frequency of 5 times per second, thus obtaining a real-time monitoring data stream of the construction process. Based on the real-time monitoring data stream, a data processing program is established on the vehicle terminal. For each sampling moment, the flow rate, speed, and position data are read. Based on the position data, the current construction section unit is determined. Based on the flow rate and speed data, the actual spraying volume per unit area at the current moment is calculated. The actual spraying volume is equal to the flow rate divided by the spraying width and then divided by the speed. The actual spraying volume is compared with the target usage for that section unit in the road segment construction parameter scheme output in step 6 and the electronic guide, and the deviation value and deviation percentage are calculated to obtain real-time usage deviation data. Based on the dosage deviation data, the allowable deviation range is set to ±10% of the target dosage. When the actual spraying volume deviation exceeds the allowable range, a warning message is displayed on the vehicle terminal screen. The warning message includes the current station number, target dosage, actual spraying volume, deviation percentage, and adjustment suggestions. The adjustment suggestions are given according to the direction of the deviation. If the actual spraying volume is too small, it is recommended to increase the flow rate or decrease the speed; if the actual spraying volume is too large, it is recommended to decrease the flow rate or increase the speed. At the same time, an audible alarm is triggered to remind the operator to make timely adjustments and receive real-time warnings and adjustment guidance. Based on the operator's response to the warning, after the operator adjusts the equipment parameters, the monitoring system continuously tracks the change in the actual spraying volume to verify the adjustment effect. If the deviation falls within the allowable range after adjustment, the warning is lifted. If the deviation still exceeds the range after adjustment, further adjustments are prompted until the deviation meets the requirements. The time, location, and parameter changes of each warning and adjustment are recorded to obtain a warning and adjustment log of the construction process.

[0068] Step 7.2, Construction quality inspection and correction; Based on the completed road sections, a mobile infrared thermal imager was used to scan the road surface. The temperature distribution of the road surface after emulsified asphalt spraying differed from the surrounding unsprayed areas. The thermal imagery identified the sprayed coverage area and analyzed the uniformity of the spraying. For uniformly sprayed areas, the thermal imagery showed a uniform temperature distribution; for unevenly sprayed areas, obvious temperature stripes or patches appeared. The locations and areas of unevenly sprayed areas were identified, and their locations and areas were marked to obtain the spray uniformity assessment results. Based on the spray uniformity assessment results, for areas identified as insufficiently sprayed, respraying was arranged. The respraying amount was calculated as the difference between the target amount and the amount already sprayed in that area. For areas identified as oversprayed, if the overspray was not severe and did not affect the construction quality, it was retained and recorded. If the overspray was severe and might lead to bleeding or other defects, oil-absorbing materials were used for cleaning. The locations and amounts of respraying and cleaning were recorded to obtain a construction quality correction record.

[0069] Step 7.3, Construction data statistics and deviation analysis; Based on construction monitoring data for the entire road segment, the planned usage, actual usage, deviation values, and construction duration for each segment are summarized to calculate statistical indicators for usage across the entire road segment. These indicators include total planned usage, total actual usage, total deviation, average deviation, standard deviation of deviation, number and percentage of segments exceeding deviation limits, etc., generating a construction usage statistical report and obtaining quantified construction quality assessment indicators. Based on the construction usage statistical report, the spatial distribution characteristics and causes of deviations are analyzed. For road segments where deviations are concentrated, the pavement condition, defect types, and construction parameter settings are examined for any anomalies. The correlation between deviations and pavement characteristic parameters is analyzed to identify key factors leading to deviations, resulting in a deviation cause analysis report.

[0070] Step 7.4, Correction of dosage calculation method and optimization of parameters; Based on the deviation cause analysis report, for systematic deviations, i.e. deviations occurring in the same way across multiple road sections, it is determined that there are systematic errors in the usage calculation process from steps 2 to 5. This leads to a deviation between the output emulsified asphalt usage distribution data for each road section and the actual demand. The calculation method needs to be corrected to improve the usage prediction accuracy. Road surface characteristic data and actual usage data of the deviation road sections are collected and used as new correction samples. For the defect identification algorithm in step 2, the defect sample library is updated and the classification network is retrained to improve defect identification accuracy. For the pore feature inversion algorithm in step 3, the ground-penetrating radar inversion parameters and Bayesian fusion weights are updated. For the penetration depth calculation algorithm in step 4, the demulsification kinetic parameters and environmental correction coefficients are corrected. For the usage calculation formula in step 5, the calculation parameters and loss coefficients are optimized. An incremental learning method is used to adjust parameters and improve the method based on the original algorithm, resulting in the corrected usage calculation method. Based on the revised usage calculation method, the distribution data of emulsified asphalt usage for different road sections were recalculated. The calculation accuracy was evaluated using a cross-validation method. Historical project data and current project data were mixed and randomly divided into training and test sets. The calculation parameters were corrected on the training set, and the usage was calculated on the test set. The error between the predicted and actual usage was calculated to evaluate the accuracy and applicability of the calculation method. If the calculation accuracy met the requirements, the revised disease identification parameters, pore inversion parameters, permeability depth calculation parameters, environmental correction coefficients, usage calculation formulas, and other key parameters were standardized and organized to obtain a validated set of usage calculation parameters.

[0071] Step 7.5, Project Data Archiving and Report Generation; Based on all data from the construction process, including high-quality 3D pavement data from step 1, the defect identification result dataset from step 2, the emulsified asphalt usage distribution data for each road segment from step 5, the construction parameter scheme and electronic guidance manual for each road segment from step 6, and the monitoring data and quality assessment data from step 7, a project data archive is established. This data archive is stored in a relational database for easy querying and analysis. The data archive not only serves as a summary of the current project but also provides reference and guidance for similar projects in the future, resulting in a complete project data asset. Based on the project data archive, a project construction quality assessment report is generated. The report includes chapters such as project overview, pavement condition analysis, defect characteristic statistics, usage calculation results, construction process monitoring, quality inspection and evaluation, existing problem analysis, and improvement suggestions. It uses a combination of text descriptions and charts to comprehensively reflect the project implementation, resulting in a comprehensive project technical report.

[0072] In some embodiments, since the maintenance effect after construction takes time to manifest, the durability and long-term performance of the maintenance layer cannot be assessed in the short term. Therefore, accelerated aging tests can be used to predict the maintenance effect. The aim is to assess the maintenance quality and verify the rationality of the dosage calculation in a short period. Specifically, 7 days after construction, representative road sections are selected, and maintenance layer samples are obtained using core drilling. The sample diameter is 100 mm. The samples are then subjected to accelerated aging tests in the laboratory. An ultraviolet aging chamber is used to simulate long-term ultraviolet radiation, with a test duration of 168 hours, equivalent to 3 years in the natural environment. A freeze-thaw cycle test is used to simulate seasonal temperature changes, with 15 cycles, equivalent to 15 freeze-thaw seasons in the natural environment. After the test, the bond strength, permeability coefficient, and anti-skid performance of the samples are measured. The test results are compared with the design targets to assess the durability of the maintenance layer. If the performance indicators meet the requirements, the rationality of the dosage calculation is verified. If the performance indicators are insufficient, the reasons are analyzed and the dosage calculation model is adjusted. This method allows for the prediction of long-term performance at the time of maintenance project acceptance, providing a scientific basis for ensuring project quality.

[0073] A precise calculation system for emulsified asphalt usage in road maintenance projects, such as Figure 3 As shown, the method for accurately calculating the amount of emulsified asphalt used in road maintenance projects, as described above, includes: The data acquisition module is used to collect three-dimensional color point cloud data of the road surface, perform preprocessing, and obtain high-quality three-dimensional road surface data. The pavement defect identification module identifies the types and locations of pavement defects based on high-quality 3D pavement data, and obtains a defect identification result dataset. The pore feature inversion module, based on the defect identification result dataset, inverts the pore features of the pavement material to obtain the pavement material pore feature parameter dataset; The penetration depth calculation module calculates the penetration depth of emulsified asphalt based on the pavement material pore characteristic parameter dataset and determines the regional penetration depth by combining the disease identification result dataset, thus obtaining the effective penetration depth distribution data of different regions. The usage prediction module predicts the usage of emulsified asphalt based on the effective penetration depth distribution data and the disease identification result dataset in different regions, and obtains the emulsified asphalt usage distribution data for different road sections. The construction optimization module optimizes construction process parameters based on the distribution data of emulsified asphalt usage in different road sections, resulting in construction parameter schemes for each road section. Based on the construction parameter schemes for each road section and the defect identification result dataset, it generates construction operation instructions and digitizes them to obtain electronic instructions. The quality monitoring module monitors construction quality in real time and performs deviation analysis based on the construction parameter scheme and electronic guide for each road segment. Based on the deviation analysis results, it adjusts the parameters to obtain a project construction quality assessment report and a verified set of usage calculation parameters.

[0074] In one embodiment of the present invention, a specific example is provided: A 10-kilometer section of a highway from K100 to K110 is used as an application example. This section is a two-way four-lane road with a single lane width of 3.75 meters. The pavement type is dense-graded asphalt concrete, with a service life of 8 years. The average daily traffic volume is 35,000 vehicles per year, with 25% being trucks. The pavement has early defects such as network cracks, longitudinal cracks, and slight loosening. The pavement performance index is 78, which is in the transition stage from good to medium condition. Preventive maintenance is required to delay performance degradation.

[0075] The method of this invention is used to accurately calculate the amount of emulsified asphalt used in this road section. The specific implementation process is as follows: A detection vehicle equipped with a 3D laser scanner and a high-resolution camera was used to collect data on the road section at a speed of 50 km / h, obtaining 3D color point cloud data of the road surface. The data collection time was from 9:00 to 11:00 AM on June 15, 2024, with sunny weather, a temperature of 28 degrees Celsius, a relative humidity of 55%, and a wind speed of 2 meters per second. The total amount of data collected was approximately 500 GB.

[0076] Table 1 shows examples of pavement defect identification data for some road sections: Table 1: Examples of pavement defect identification data for some road sections;

[0077] Through the analysis of defects across the entire road section, a total of 568 network crack defects were identified, covering a total area of ​​6,850 square meters; 342 longitudinal crack defects, covering a total area of ​​1,820 square meters; and 215 minor loose defects, covering a total area of ​​2,650 square meters. The intact area was 63,680 square meters, with defects accounting for approximately 15% of the total area.

[0078] Ground-penetrating radar was used to invert the porosity characteristics of the road surface material. The average porosity of the intact area was 4.2, the effective porosity of the cracked area was 8.5 to 15.3, and the effective porosity of the loose area was 9.8 to 12.6.

[0079] Based on meteorological data of the construction environment, the planned construction period is from July 8 to July 12, 2024. The forecast average temperature is 32 degrees Celsius, the relative humidity is 60%, and the wind speed is 1.5 meters per second. The weather will be mainly sunny, which is suitable for construction. The calculated comprehensive environmental correction factor is 1.08.

[0080] Table 2 shows an example of the calculation results for emulsified asphalt usage in some road sections: Table 2: Examples of calculation results for emulsified asphalt usage in some road sections;

[0081] The calculation of usage across the entire road section yielded a theoretical total usage of 32,650 kg of emulsified asphalt. Considering a construction loss coefficient of 1.08, the actual total procurement amount was 35,262 kg. The average usage per unit area across the entire road section was 0.44 kg per square meter, and the coefficient of variation for the usage distribution was 0.18, indicating that the usage differences across different road sections were moderate.

[0082] The construction process parameters were optimized using the method of this invention. An emulsified asphalt sprayer with a spraying width of 3.5 meters was selected. The optimized construction plan is as follows: for road sections with dense defects, the driving speed is 3 kilometers per hour and the flow rate is 87.5 liters per minute; for general road sections, the driving speed is 4.5 kilometers per hour and the flow rate is 131.3 liters per minute; for intact road sections, the driving speed is 5 kilometers per hour and the flow rate is 145.8 liters per minute. The total construction time is expected to be 32 hours.

[0083] A real-time monitoring system was used during construction. The monitoring data showed that the actual amount of material used was 35,540 kg, which was 0.79 kg different from the planned amount. The deviation was within a reasonable range. Only 3 sections of the road had a usage deviation of more than 10 kg, accounting for 0.75% of the total, indicating that the construction quality was good.

[0084] Seven days after construction, the maintenance effect was tested. The pavement permeability increased from an average of 52 before construction to 68, indicating a significant improvement in anti-skid performance. The water permeability coefficient decreased from an average of 120 ml / min before construction to 15 ml / min, indicating a significant improvement in waterproof performance. The pavement smoothness remained stable, and the maintenance effect achieved the expected goals.

[0085] Compared with traditional empirical methods, which estimate the amount of emulsified asphalt needed for this road section based on empirical formulas as 41,500 kg, the actual amount calculated using the method of this invention is 35,262 kg, saving 15 kg of material and approximately 6,000 yuan in costs. At the same time, due to the accurate calculation of the amount, the problem of insufficient or excessive amount in some areas is avoided, resulting in more uniform maintenance quality. This application example fully verifies the effectiveness and practicality of the method of this invention.

Claims

1. A method for accurately calculating the amount of emulsified asphalt used in road maintenance projects, characterized in that, include: Collect three-dimensional color point cloud data of the road surface, perform preprocessing, and obtain high-quality three-dimensional road surface data; Based on high-quality 3D pavement data, the types and locations of pavement defects are identified, and a defect identification result dataset is obtained. Based on the damage identification result dataset, the pavement material porosity characteristics are inverted to obtain the pavement material porosity characteristic parameter dataset; Based on the pavement material pore characteristic parameter dataset, the penetration depth of emulsified asphalt is calculated and combined with the disease identification result dataset to determine the penetration depth of different regions, thus obtaining the effective penetration depth distribution data of different regions. Based on the effective penetration depth distribution data and disease identification result dataset of different regions, the amount of emulsified asphalt used is predicted, and the distribution data of emulsified asphalt used in different road sections is obtained. Based on the distribution data of emulsified asphalt usage in different road sections, the construction process parameters are optimized to obtain the construction parameter scheme for each road section. Based on the road segment construction parameter scheme and the defect identification result dataset, a construction operation instruction manual is generated and digitized to obtain an electronic instruction manual. Based on the segmented construction parameter scheme and electronic guide, the construction quality is monitored in real time and deviation analysis is performed. Based on the deviation analysis results, the parameters are adjusted to obtain the project construction quality assessment report and the verified set of usage calculation parameters. The porosity characteristics of the inverted pavement material include: Ground-penetrating radar was used to continuously scan along the road section to obtain time-domain signal data of road surface reflected waves; The instantaneous amplitude envelope of the signal is extracted using Hilbert transform, the reflection peaks at the interface are identified, and the propagation time of the electromagnetic wave in the road surface material is calculated. Based on two-way travel time data and surface thickness, the electromagnetic wave propagation speed is calculated, and the relative permittivity of the surface material is calculated using the relationship between electromagnetic wave speed and the relative permittivity of the material. The porosity is estimated by inverting and solving the equivalent medium theory of composite materials to obtain the porosity of the pavement material. The calculation of emulsified asphalt penetration depth and the determination of regional penetration depth based on the disease identification result dataset include: Based on the material property test data of emulsified asphalt, the rheological parameters of emulsified asphalt at standard temperature were measured, and a viscosity-temperature dependence model was established. The permeability coefficient of porous media is described by a permeability coefficient calculation equation. Based on the rheological parameters of emulsified asphalt and the permeability coefficient of pavement materials, an extended form of the permeability law is used to describe the permeation process of emulsified asphalt in pavement materials. An exponential growth model is used to describe viscosity evolution, resulting in a viscosity time-dependent model. Based on the emulsified asphalt permeation kinetics model and viscosity-time dependence model, the numerical integration method is used to discretize the permeation time into time steps. At each time step, the permeation rate is calculated based on the viscosity at the current moment. When the permeation rate decreases to below a preset threshold, the permeation process is considered to have ended, and the final theoretical value of the permeation depth is obtained.

2. The method for accurately calculating the amount of emulsified asphalt used in road maintenance engineering according to claim 1, characterized in that, The preprocessing of the collected three-dimensional color point cloud data of the road surface includes: Data acquisition is performed using a dedicated inspection vehicle equipped with a 3D laser scanner, a high-resolution industrial camera, and an inertial navigation system. The Kalman filter algorithm is used to perform motion compensation on the point cloud data. A statistical outlier removal algorithm is used to denoise point cloud data, and points whose average distance exceeds a preset standard deviation multiple are identified as outliers and removed. The point cloud data is discretized into a three-dimensional voxel grid by using a voxel grid filtering method to obtain regularized point cloud grid data. A spatiotemporal synchronization method is used to establish the correspondence between point clouds and images, and the image color information is mapped to the point cloud grid points to obtain three-dimensional color point cloud data.

3. The method for accurately calculating the amount of emulsified asphalt used in road maintenance engineering according to claim 1, characterized in that, The identification of pavement distress types and locations includes: The local reference plane is calculated using the least squares plane fitting method. The elevation deviation of each point cloud grid point relative to the local reference plane is calculated. The elevation deviation is decomposed into texture component and disease component using the bilateral filtering method. An adaptive threshold segmentation method is used to mark grid points whose absolute elevation deviation exceeds a preset threshold as candidate disease points; A density-based clustering algorithm is used to aggregate candidate points into disease clusters to obtain the disease region segmentation results; Based on color image patches of the diseased area, texture and edge features of the image are extracted to construct an image feature vector; Based on the elevation data of the diseased area, the geometric features of the disease are extracted, and a geometric feature vector is constructed.

4. The method for accurately calculating the amount of emulsified asphalt used in road maintenance engineering according to claim 3, characterized in that, The identification of pavement distress types and locations also includes: Based on image feature vectors and geometric feature vectors, a comprehensive feature vector is constructed. The comprehensive feature vector is then input into a pre-trained deep convolutional neural network classification model, which outputs the probability distribution of disease types. Transfer learning techniques are used to fine-tune the pre-trained model, resulting in a fine-tuned model that adapts to the current disease characteristics of the project. For each diseased area, the classification results of multiple models are voted on to obtain the disease type identification result and identification confidence level.

5. The method for accurately calculating the amount of emulsified asphalt used in road maintenance engineering according to claim 1, characterized in that, The inverted pavement material porosity characteristics also include: Based on the information on pavement material type, the pavement material property database is queried to obtain the prior estimate of porosity and the estimation uncertainty based on the material type. Based on pavement service life data and traffic load level data, a pavement material aging model is used to calculate the aging correction coefficient and obtain the estimated porosity considering the aging effect. Based on the porosity estimates obtained from ground-penetrating radar inversion, prior estimates of material type, and aging correction estimates, a Bayesian fusion method is adopted to use the porosity estimates from the three sources as observations, establish a probabilistic model between the observations and the true porosity, and use the maximum a posteriori estimation criterion to obtain the optimal porosity estimate and its estimation variance after fusion.

6. The method for accurately calculating the amount of emulsified asphalt used in road maintenance engineering according to claim 1, characterized in that, The predicted emulsified bitumen usage includes: Hourly meteorological data for the construction date is obtained based on data from meteorological monitoring stations in the construction area. Based on the temperature data during the construction period, the viscosity-temperature dependence model of emulsified asphalt was used to calculate the viscosity correction coefficient of emulsified asphalt at the construction temperature. Based on the relative humidity data during the construction period, the humidity dependence model of emulsified asphalt demulsification rate is used to calculate the humidity correction coefficient for demulsification rate. Calculate the comprehensive correction factor, multiply the theoretical infiltration depth by the comprehensive correction factor, and obtain the actual infiltration depth after considering environmental factors; For each affected area, calculate the effective amount of emulsified asphalt used in that area. The amount is equal to the area multiplied by the penetration depth multiplied by the porosity multiplied by the density divided by the solid content. Add the amount of emulsified asphalt used for disease maintenance and the amount of emulsified asphalt used for preventive maintenance to obtain the total amount of emulsified asphalt used in each road segment unit.

7. The method for accurately calculating the amount of emulsified asphalt used in road maintenance engineering according to claim 1, characterized in that, The real-time monitoring of construction quality and deviation analysis, and the parameter adjustment based on the deviation analysis results, include: Real-time measurement of emulsified asphalt flow rate, measurement of actual vehicle speed, and real-time acquisition of vehicle location coordinates; For each sampling moment, read the flow rate, speed and location data, determine the road segment unit currently under construction based on the location data, calculate the actual spraying amount per unit area at the current moment based on the flow rate and speed data, compare the actual spraying amount with the target amount, and calculate the deviation value and deviation percentage. When the actual spraying volume deviates from the preset allowable range, a warning message is displayed on the vehicle terminal screen, and an audible alarm is triggered at the same time; Based on the deviation cause analysis report, pavement characteristic data and actual usage data of the deviation road sections are collected as new correction samples to adjust parameters and improve usage calculation methods.

8. A precise calculation system for emulsified asphalt usage in road maintenance projects, characterized in that, A method for accurately calculating the amount of emulsified asphalt used in road maintenance engineering as described in any one of claims 1-7 includes: The data acquisition module is used to collect three-dimensional color point cloud data of the road surface, perform preprocessing, and obtain high-quality three-dimensional road surface data. The pavement defect identification module identifies the types and locations of pavement defects based on high-quality 3D pavement data, and obtains a defect identification result dataset. The pore feature inversion module, based on the defect identification result dataset, inverts the pore features of the pavement material to obtain the pavement material pore feature parameter dataset; The penetration depth calculation module calculates the penetration depth of emulsified asphalt based on the pavement material pore characteristic parameter dataset and determines the regional penetration depth by combining the disease identification result dataset, thus obtaining the effective penetration depth distribution data of different regions. The usage prediction module predicts the usage of emulsified asphalt based on the effective penetration depth distribution data and the disease identification result dataset in different regions, and obtains the emulsified asphalt usage distribution data for different road sections. The construction optimization module optimizes construction process parameters based on the distribution data of emulsified asphalt usage in different road sections, resulting in construction parameter schemes for each road section. Based on the construction parameter schemes for each road section and the defect identification result dataset, it generates construction operation instructions and digitizes them to obtain electronic instructions. The quality monitoring module monitors construction quality in real time and performs deviation analysis based on the construction parameter scheme and electronic guide for each road segment. Based on the deviation analysis results, it adjusts the parameters to obtain a project construction quality assessment report and a verified set of usage calculation parameters.

Images