Asphalt pavement aging degree multispectral imaging evaluation system and method thereof

Abstract

The invention relates to the field of road engineering, in particular to a multispectral imaging evaluation system and method for the aging degree of an asphalt pavement, and the method comprises the steps that a multispectral imaging module collects pavement spectral data; the topological feature extraction module is used for performing multi-scale feature decomposition on the spectral data and extracting topological structure features; the spectral line attention calculation module calculates weight coefficients of different spectral lines and generates weighted feature vectors; the time sequence feature learning module processes the time sequence change of the feature vector and captures an aging evolution rule; the life prediction module determines the current aging state based on the state vector, generates a plurality of possible aging evolution paths, and determines the pavement aging grade and the remaining life; the decision support module generates a maintenance decision suggestion according to the aging grade and the remaining life, and the multi-spectral imaging technology and the deep learning algorithm are innovatively combined, so that high-precision evaluation of the aging degree of the asphalt pavement and accurate prediction of the remaining life are realized, the scientificity of the maintenance decision is remarkably improved, and the maintenance cost is reduced.

Description

Technical Field

[0001] The present invention relates to the field of road engineering, and specifically to a multispectral imaging assessment system and method for the aging degree of asphalt pavement, and more particularly to a technical solution for implementing aging degree assessment and remaining life prediction of asphalt pavement using multispectral imaging technology combined with deep learning. Background Art

[0002] Asphalt pavement ages over time due to various factors, including UV radiation, temperature fluctuations, moisture erosion, and traffic loads. Pavement aging not only impacts driving comfort and safety, but also significantly shortens pavement service life and increases maintenance costs.

[0003] Existing methods for assessing asphalt pavement degradation primarily include manual visual inspection, physical sampling and analysis, and single-band image analysis. Manual visual inspection is highly subjective and inaccurate; physical sampling and analysis, while highly accurate, is highly destructive, inefficient, and costly; and single-band image analysis, while convenient, provides limited information and fails to fully reflect the pavement's degradation status. Furthermore, most existing assessment methods are limited to static assessments of pavement degradation, lacking the ability to dynamically analyze aging evolution and accurately predict remaining life.

[0004] With the development of multispectral imaging technology and artificial intelligence, it has become possible to use rich spectral information combined with deep learning algorithms to assess the degree of asphalt pavement degradation. However, there is currently a lack of comprehensive solutions that organically combine cutting-edge technologies such as multispectral imaging, topological feature extraction, time series learning, and multi-path prediction. Summary of the Invention

[0005] The main purpose of this invention is to provide a multispectral imaging assessment system and method for the degree of aging of asphalt pavement. By acquiring rich spectral information through multispectral imaging technology and combining innovative technologies such as topological feature extraction, attention mechanism, temporal learning and multipath prediction, the system can accurately assess the degree of aging of asphalt pavement and accurately predict the remaining life of the pavement.

[0006] The present invention proposes a multispectral imaging assessment system for asphalt pavement aging, comprising:

[0007] A multispectral imaging module, configured to collect spectral data of an asphalt pavement, wherein the spectral data includes a plurality of spectral lines, each of which includes spectral intensity values ​​in a plurality of wavelength ranges;

[0008] A topological feature extraction module, connected to the multispectral imaging module, is used to perform multi-scale feature decomposition on the spectral data, generate a feature matrix, and extract topological structure features from the feature matrix;

[0009] a spectral line attention calculation module, connected to the topological feature extraction module, for calculating weight coefficients of different spectral lines in the feature matrix, and performing weighted processing on the feature matrix based on the weight coefficients to generate a weighted feature vector;

[0010] A time series feature learning module, connected to the spectral line attention calculation module, is used to process the time series changes of the weighted feature vector, capture the time series evolution law of road aging, and output a state vector;

[0011] a life prediction module, connected to the time series feature learning module, configured to receive the state vector, determine the current aging state in a predefined feature space based on the state vector, generate multiple possible aging evolution paths, calculate the probability distribution of each path, and comprehensively determine the pavement aging grade and remaining life;

[0012] A decision support module is connected to the life prediction module and is used to generate maintenance decision suggestions based on the aging level and the remaining life.

[0013] Preferably, the multispectral imaging module comprises:

[0014] Hyperspectral camera, used to obtain spectral information of asphalt pavement in the wavelength range of 380-800nm;

[0015] a scanning mechanism, connected to the hyperspectral camera, and configured to control the hyperspectral camera to scan along a road surface;

[0016] An image preprocessing unit, connected to the hyperspectral camera, for performing noise removal and spectrum calibration on the collected spectral data;

[0017] A mobile terminal APP is connected to the image preprocessing unit, and is used to control the hyperspectral camera and the scanning mechanism, and display the spectral data.

[0018] Preferably, the topological feature extraction module includes:

[0019] A multi-scale decomposition unit, configured to decompose the spectral data according to different wavelength intervals to form low-frequency features, medium-frequency features, and high-frequency features;

[0020] A sliding window analysis unit, connected to the multi-scale decomposition unit, for sliding along the wavelength dimension to extract local features;

[0021] A topological structure extraction unit, connected to the sliding window analysis unit, for calculating the topological features of the spectral line data and generating a topological feature vector;

[0022] The feature fusion unit is connected to the multi-scale decomposition unit and the topological structure extraction unit, and is used to fuse features of different scales and topological features to generate a complete feature matrix.

[0023] Preferably, the spectral line attention calculation module includes:

[0024] The frequency domain correlation evaluation unit is used to analyze the richness of each band information and generate the frequency domain correlation weight;

[0025] a discrimination calculation unit connected to the frequency domain correlation evaluation unit, for calculating the significance of the spectral line differences between different aging levels and generating a discrimination weight;

[0026] A time series stability evaluation unit, connected to the discrimination calculation unit, for evaluating the stability of spectral line characteristics over time and generating a stability weight;

[0027] a weight fusion unit connected to the frequency domain correlation evaluation unit, the discrimination calculation unit, and the temporal stability evaluation unit, and configured to integrate the three weights to generate a final weight vector;

[0028] The weighted processing unit is connected to the weight fusion unit and is used to weight the feature matrix according to the weight vector and output a weighted feature vector.

[0029] Preferably, the temporal feature learning module includes:

[0030] A memory-enhanced network unit, configured to process temporal changes of the weighted feature vectors and filter and store key information through a multi-gating structure;

[0031] A multi-scale memory pool, connected to the memory-enhanced network unit, for simultaneously maintaining short-term memory, medium-term memory, and long-term memory;

[0032] A spatiotemporal feature interaction unit, connected to the multi-scale memory pool, for fusing spatial features and temporal features;

[0033] The feature pyramid structure is connected to the spatiotemporal feature interaction unit and is used to integrate feature information of different scales to generate a final state vector.

[0034] Preferably, the lifespan prediction module includes:

[0035] a manifold construction unit, configured to map the state vector to a predefined multi-dimensional feature space;

[0036] a multi-path generation unit, connected to the manifold construction unit, for generating multiple possible aging evolution paths based on the current state;

[0037] a path evaluation unit, connected to the multi-path generation unit, for calculating the probability weight of each path;

[0038] an aging grade judgment unit, connected to the path evaluation unit, for classifying the road surface aging degree into 0, 1, 2, and 3, wherein 3 represents the most severe road surface aging degree;

[0039] The remaining service life calculation unit is connected to the aging level judgment unit and is used to calculate the remaining service life based on the aging level and the path probability.

[0040] Preferably, the life prediction module further includes:

[0041] Uncertainty quantification unit, used to analyze measurement uncertainty, model uncertainty and prediction uncertainty;

[0042] The risk assessment unit is connected to the uncertainty quantification unit and is used to divide high-risk areas and low-risk areas and provide decision-making risk assessment.

[0043] Preferably, the decision support module includes:

[0044] Maintenance strategy generation unit, used to match the optimal maintenance measures according to the aging level and remaining life;

[0045] a cost-benefit analysis unit, connected to the maintenance strategy generation unit, for calculating the return on investment of different maintenance plans;

[0046] an implementation timing recommendation unit, connected to the cost-benefit analysis unit, for recommending an optimal maintenance timing based on the aging rate and budget constraints;

[0047] The feedback optimization unit is connected to the implementation timing recommendation unit and is used to collect maintenance effect data and adjust the prediction model parameters.

[0048] As an option, it also includes:

[0049] a database management module, connected to the topological feature extraction module, the temporal feature learning module and the life prediction module respectively, and used for storing feature data, state data and prediction results;

[0050] A communication interface module, connected to the decision support module, is used to implement data exchange with external systems and support data export and remote access;

[0051] The user interface module is connected to the decision support module and is used to display the assessment results and maintenance suggestions in the form of charts and reports.

[0052] The multispectral imaging method for evaluating the aging degree of asphalt pavement includes the following steps:

[0053] Collecting spectral data of an asphalt pavement, the spectral data including a plurality of spectral lines, each of the spectral lines including spectral intensity values ​​in a plurality of wavelength ranges;

[0054] Performing multi-scale feature decomposition on the spectral data to generate a feature matrix, and extracting topological structure features from the feature matrix;

[0055] Calculating weight coefficients of different spectral lines in the characteristic matrix, and performing weighted processing on the characteristic matrix based on the weight coefficients to generate a weighted characteristic vector;

[0056] Processing the time series changes of the weighted feature vector, capturing the time series evolution law of road aging, and outputting a state vector;

[0057] receiving the state vector, and determining a current aging state in a predefined feature space based on the state vector, generating multiple possible aging evolution paths, calculating a probability distribution of each path, and comprehensively determining a pavement aging grade and remaining life;

[0058] Based on the ageing level and the remaining life, maintenance decision recommendations are generated.

[0059] The present invention has the following beneficial effects:

[0060] 1. Improve assessment accuracy: Multispectral imaging acquires rich spectral information in the 380-800nm ​​wavelength range. Combined with topological feature extraction and attention mechanisms, it captures subtle changes in pavement aging, significantly improving the accuracy of aging assessment.

[0061] 2. Achieve dynamic prediction: Utilize the time series feature learning module to capture the evolution of aging, combined with a multi-path prediction framework, to achieve a technological leap from static assessment to dynamic prediction, providing a scientific basis for maintenance decisions.

[0062] 3. Reduce maintenance costs: Accurately identify the degree of pavement aging and predict the remaining lifespan, enabling precise maintenance, avoiding over- or under-maintenance, optimizing resource allocation, and reducing maintenance costs by approximately 20%.

[0063] 4. Extend service life: By promptly detecting early signs of aging and implementing preventive maintenance, the service life of the road surface can be extended by 15% to 25%, thereby improving the quality of road services.

[0064] 5. Achieve intelligent management: Provide a complete solution for pavement aging assessment and maintenance decision-making, promote the transformation of road maintenance from experience-based decision-making to data-based decision-making, and improve the scientific level of maintenance management. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] In order to more clearly illustrate the technical solution of the present invention, the following is a brief introduction to the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0066] Figure 1 This is the overall architecture diagram of the multispectral imaging assessment system for asphalt pavement aging degree of the present invention;

[0067] Figure 2 Schematic diagram of the structure of the multispectral imaging module of the present invention;

[0068] Figure 3 Schematic diagram of the topological feature extraction module of the present invention;

[0069] Figure 4 Schematic diagram of the structure of the spectrum line attention calculation module of the present invention;

[0070] Figure 5 Schematic diagram of the structure of the temporal feature learning module of the present invention;

[0071] Figure 6 This is a schematic diagram of the structure of the life prediction module of the present invention;

[0072] Figure 7 This is a schematic diagram of the structure of the decision support module of the present invention;

[0073] Figure 8 The figure is a flow chart of the multispectral imaging method for evaluating the aging degree of asphalt pavement according to the present invention. DETAILED DESCRIPTION

[0074] Please refer to Figure 1 - Figure 8 The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without inventive effort shall fall within the scope of protection of the present invention.

[0075] Reference Figure 1 The multispectral imaging evaluation system for asphalt pavement aging degree provided by the present invention includes a multispectral imaging module 1, a topological feature extraction module 2, a spectral line attention calculation module 3, a temporal feature learning module 4, a life prediction module 5 and a decision support module 6.

[0076] The multispectral imaging module 1 is used to collect spectral data of asphalt pavement. The spectral data includes multiple spectral lines, each of which includes spectral intensity values ​​in multiple wavelength ranges. Preferably, the spectral data collected by the multispectral imaging module 1 has a wavelength range of 380-800nm ​​and includes 400 different spectral lines.

[0077] The topological feature extraction module 2 is connected to the multispectral imaging module 1 and is used to perform multi-scale feature decomposition on the spectral data, generate a feature matrix, and extract topological structure features from the feature matrix.

[0078] The spectral line attention calculation module 3 is connected to the topological feature extraction module 2, and is used to calculate the weight coefficients of different spectral lines in the feature matrix, and perform weighted processing on the feature matrix based on the weight coefficients to generate a weighted feature vector.

[0079] The time series feature learning module 4 is connected to the spectral line attention calculation module 3 to process the time series changes of the weighted feature vector, capture the time series evolution law of road aging, and output the state vector.

[0080] The life prediction module 5 is connected to the time series feature learning module 4, and is used to receive the state vector, and determine the current aging state in the predefined feature space based on the state vector, generate multiple possible aging evolution paths, calculate the probability distribution of each path, and comprehensively determine the pavement aging grade and remaining life.

[0081] The decision support module 6 is connected to the life prediction module 5 and is used to generate maintenance decision suggestions based on the aging level and the remaining life.

[0082] In addition, the system also includes a database management module 7, a communication interface module 8 and a user interface module 9, which are used to support the data management, communication and user interaction functions of the system.

[0083] like Figure 2 As shown, the multispectral imaging module 1 includes a hyperspectral camera 11 , a scanning mechanism 12 , an image preprocessing unit 13 and a mobile terminal APP 14 .

[0084] The hyperspectral camera 11 is used to obtain spectral information of asphalt pavement within the wavelength range of 380-800 nm. In one embodiment of the present invention, the hyperspectral camera 11 has a resolution of 2048×1536 pixels, a sampling interval of 1 nm, and can simultaneously obtain 400 different spectral lines.

[0085] The scanning mechanism 12 is connected to the hyperspectral camera 11 and controls the hyperspectral camera 11 to scan the road surface. Preferably, the scanning mechanism 12 is mounted on a dedicated vehicle to ensure stable and comprehensive scanning. In one embodiment, the scanning vehicle, traveling at 30 km / h, can complete a full-coverage scan of 10 kilometers of road in one hour.

[0086] The image preprocessing unit 13 is connected to the hyperspectral camera 11 and is used to remove noise and calibrate the spectrum data collected. Specifically, the image preprocessing unit 13 uses a dark pixel correction method to calibrate the original spectrum data. The correction formula is:

[0087] ,

[0088] in: is the corrected reflectivity, is the original reflectivity, is the blackbody brightness (i.e. collecting background spectrum), is the brightness without black body under the road surface (i.e. collecting the reflection spectrum of the reference plate), To collect the brightness of the road surface at a certain point.

[0089] Mobile terminal app 14 is connected to image preprocessing unit 13 and is used to control hyperspectral camera 11 and scanning mechanism 12 and display spectral data. Mobile terminal app 14 includes three main functional modules: camera for controlling hyperspectral camera 11, photo for starting hyperspectral camera 11 to capture road surface images, and rubber plate for displaying aging spectrum lines.

[0090] like Figure 3 As shown, the topological feature extraction module 2 includes a multi-scale decomposition unit 21, a sliding window analysis unit 22, a topological structure extraction unit 23 and a feature fusion unit 24.

[0091] The multi-scale decomposition unit 21 is used to decompose the spectral data according to different wavelength ranges to form low-frequency features, medium-frequency features and high-frequency features. In one embodiment of the present invention, the multi-scale decomposition unit 21 divides the spectral range of 380-800nm ​​into three levels:

[0092] Low-frequency characteristics: 400-650nm range, corresponding to changes in material matrix;

[0093] Medium frequency characteristics: 380-420nm and 650-800nm ​​range, corresponding to the degree of oxidation;

[0094] High-frequency characteristics: specific sensitive bands, corresponding to microstructural damage;

[0095] The sliding window analysis unit 22 is connected to the multi-scale decomposition unit 21 and is used to extract local features by sliding along the wavelength dimension. Preferably, the sliding window analysis unit 22 uses a window size of 7-11 for scanning with a window step size of 2, which can effectively capture the spectral variation characteristics within the local wavelength range.

[0096] The topology extraction unit 23 is connected to the sliding window analysis unit 22 and is used to calculate the topological features of the spectral line data and generate a topological feature vector. Specifically, the topology extraction unit 23 constructs a multi-scale simple complex, calculates the persistence homology groups of different dimensions, identifies the main topological structure in the spectral line, generates a persistence graph, and extracts statistical features to form a topological feature vector.

[0097] The feature fusion unit 24 is connected to the multi-scale decomposition unit 21 and the topology structure extraction unit 23. It is used to fuse features of different scales and topological features to generate a complete feature matrix. Feature fusion adopts a hierarchical strategy, first horizontally fusing features of the same level, then vertically fusing features of different levels. The fused feature matrix has a dimension of m × 200, where m is the number of sampling points and 200 is the feature dimension after dimensionality reduction.

[0098] like Figure 4 As shown, the spectral line attention calculation module 3 includes a frequency domain correlation evaluation unit 31, a discrimination calculation unit 32, a temporal stability evaluation unit 33, a weight fusion unit 34 and a weighted processing unit 35.

[0099] The frequency domain correlation evaluation unit 31 is used to analyze the information richness of each band and generate frequency domain correlation weights. Specifically, the frequency domain correlation evaluation unit 31 evaluates the information richness by calculating the information entropy of the spectral lines in different bands. Bands with higher information entropy values ​​have greater weights. In one embodiment, the frequency domain correlation weights are normalized so that their sum is 1.

[0100] The discrimination calculation unit 32 is connected to the frequency domain correlation evaluation unit 31 and is used to calculate the significance of the spectral line differences between different aging levels and generate discrimination weights. Specifically, the discrimination calculation unit 32 uses the ratio of the between-class variance to the within-class variance as the discrimination index, with bands with higher discrimination having greater weights.

[0101] The time series stability evaluation unit 33 is connected to the discrimination calculation unit 32 and is used to evaluate the stability of the spectral line characteristics over time and generate a stability weight. Specifically, the time series stability evaluation unit 33 evaluates stability by calculating the coefficient of variation of multiple sampling results. The smaller the coefficient of variation, the greater the band weight.

[0102] The weight fusion unit 34 is connected to the frequency domain correlation evaluation unit 31, the discrimination calculation unit 32, and the temporal stability evaluation unit 33 to integrate the three weights and generate a final weight vector. In one embodiment of the present invention, weight fusion employs an adaptive weighting method, automatically adjusting the proportions of the three weights based on the current road surface state and environmental conditions. Typically, the initial ratio of the frequency domain correlation weight, the discrimination weight, and the temporal stability weight is 3:4:3.

[0103] The weighted processing unit 35 is connected to the weight fusion unit 34 and is used to weight the feature matrix according to the weight vector and output a weighted feature vector. The weighted processing is implemented using matrix multiplication to ensure that important features are strengthened and unimportant features are suppressed, thereby improving the efficiency and accuracy of subsequent processing.

[0104] like Figure 5 As shown, the temporal feature learning module 4 includes a memory-enhanced network unit 41, a multi-scale memory pool 42, a spatiotemporal feature interaction unit 43 and a feature pyramid structure 44.

[0105] Memory-enhanced network unit 41 processes temporal changes in weighted feature vectors and filters and stores key information through a multi-gating structure. Memory-enhanced network unit 41 is based on the LSTM architecture but has been specifically optimized to include four additional gating structures: an information entropy evaluation gate, a long-term dependency gate, a short-term change gate, and a feature fusion gate.

[0106] The multi-scale memory pool 42 is connected to the memory-enhanced network unit 41 to simultaneously maintain short-term memory, medium-term memory, and long-term memory. Specifically:

[0107] Short-term memory: stores the characteristic changes of the last 10 measurements and captures sudden aging characteristics;

[0108] Medium-term memory: stores seasonal change patterns with a period of 3 months;

[0109] Long-term memory: stores annual change trends with a cycle of 1 year;

[0110] The spatiotemporal feature interaction unit 43 is connected to the multi-scale memory pool 42 to fuse spatial and temporal features. Spatial features primarily reflect the distribution of the spectrum in the wavelength dimension, while temporal features reflect how the spectrum changes over time. This spatiotemporal feature interaction utilizes a bidirectional attention mechanism to achieve mutual enhancement of spatial and temporal features.

[0111] The feature pyramid structure 44 is connected to the spatiotemporal feature interaction unit 43 and is used to integrate feature information at different scales to generate the final state vector. The feature pyramid structure 44 consists of three layers: the bottom layer retains fine spectral features (wavelength granularity of 1 nm), the middle layer extracts regional features (wavelength interval of 10 nm), and the top layer captures full-spectral features (the entire 380-800 nm range). The feature pyramid achieves interactive fusion of features at different levels through upsampling and downsampling operations, ultimately outputting a 128-dimensional state vector.

[0112] like Figure 6 As shown, the life prediction module 5 includes a manifold construction unit 51, a multi-path generation unit 52, a path evaluation unit 53, an aging level judgment unit 54, a remaining life calculation unit 55, an uncertainty quantification unit 56 and a risk evaluation unit 57.

[0113] The manifold construction unit 51 is used to map the state vector to a predefined multidimensional feature space. Specifically, the manifold construction unit 51 uses nonlinear dimensionality reduction techniques to map the 128-dimensional state vector to a 5-dimensional Riemannian manifold. On this manifold, each point represents an aging state, and the geodesic distance between points indicates the difficulty of state transition.

[0114] The multi-path generation unit 52 is connected to the manifold construction unit 51 and is used to generate multiple possible aging evolution paths based on the current state. In one embodiment of the present invention, the multi-path generation unit 52 identifies eight typical evolution path patterns based on historical data and generates corresponding evolution trajectories based on the current state.

[0115] Path evaluation unit 53 is connected to multi-path generation unit 52 and is used to calculate the probability weight of each path. Path evaluation uses a Bayesian probability model, taking into account historical similarity, current state stability, and external factors (such as climate conditions and traffic load).

[0116] The aging grade determination unit 54 is connected to the path assessment unit 53 and is used to classify the degree of pavement aging into grades 0, 1, 2, and 3, with grade 3 representing the most severe aging. The aging grade is determined based on a comprehensive score of multiple indicators, including spectral feature similarity, topological structure changes, and predicted physical property values. Specifically, a score between 0 and 25 is classified as grade 0 (no noticeable aging), between 25 and 50 as grade 1 (mild aging), between 50 and 75 as grade 2 (moderate aging), and between 75 and 100 as grade 3 (severe aging).

[0117] The remaining service life calculation unit 55 is connected to the aging level judgment unit 54 and is used to calculate the remaining service life based on the aging level and the path probability. The remaining service life calculation adopts the weighted average method, which comprehensively considers the life prediction results of each possible path. The calculation formula is:

[0118] ,

[0119] in: is the predicted remaining life, is the probability weight of the i-th path, is the predicted lifespan when evolving along the i-th path, is the total number of paths.

[0120] Uncertainty quantification unit 56 is used to analyze measurement uncertainty, model uncertainty, and prediction uncertainty. Measurement uncertainty arises from instrument errors and environmental interference, model uncertainty arises from model structure and parameter selection, and prediction uncertainty arises from the inherent randomness of the aging process. Uncertainty quantification utilizes Monte Carlo methods, which perform multiple simulations to determine the statistical distribution of prediction results.

[0121] Risk assessment unit 57 is connected to uncertainty quantification unit 56 and is used to classify high-risk and low-risk areas, providing decision-making risk assessment. Risk assessment uses a risk matrix approach, with the horizontal axis representing probability of occurrence (scaled from 1 to 5) and the vertical axis representing severity of impact (scaled from 1 to 5). This generates a 5×5 risk matrix. Areas with a total score greater than 16 are considered high-risk, those with a total score less than 9 are considered low-risk, and those in between are considered medium-risk.

[0122] like Figure 7 As shown, the decision support module 6 includes a maintenance strategy generation unit 61 , a cost-benefit analysis unit 62 , an implementation timing recommendation unit 63 and a feedback optimization unit 64 .

[0123] The maintenance strategy generation unit 61 is used to match the optimal maintenance measures according to the aging level and the remaining life. Specifically, different maintenance strategies are adopted for different aging levels:

[0124] Level 0 (no obvious aging): routine monitoring, no special measures required;

[0125] Level 1 (mild aging): preventive maintenance, such as sealing and micro-surfacing;

[0126] Level 2 (moderate aging): functional repair, such as a thin layer of finishing;

[0127] Level 3 (severe aging): structural repair, such as milling and resurfacing;

[0128] The cost-benefit analysis unit 62 is connected to the maintenance strategy generation unit 61 and is used to calculate the return on investment of different maintenance plans. The investment return ratio calculation formula is:

[0129] ,

[0130] Among them, ROI is the return on investment, B is the benefit brought by maintenance (the value of life extension conversion), and C is the maintenance cost.

[0131] The implementation timing recommendation unit 63 is connected to the cost-benefit analysis unit 62 and is used to recommend the optimal maintenance timing based on the aging rate and budget constraints. This implementation timing recommendation utilizes a multi-objective optimization approach to balance technical and economic benefits. In one embodiment of the present invention, when the aging rate exceeds a critical value (e.g., an annual aging level increase of 0.5) and the return on investment is greater than 150%, immediate maintenance is recommended.

[0132] Feedback optimization unit 64 is connected to implementation timing recommendation unit 63 and is used to collect maintenance effect data and adjust the prediction model parameters. Feedback optimization adopts a closed-loop control concept, regularly comparing prediction results with actual results and updating model parameters through a backpropagation algorithm to achieve continuous model optimization.

[0133] The database management module 7 is connected to the topology feature extraction module 2, the time series feature learning module 4, and the lifespan prediction module 5, respectively, to store feature data, status data, and prediction results. The database management module 7 uses a hybrid storage strategy, storing recent data in a local SQLite database and archiving historical data to a cloud-based PostgreSQL database.

[0134] The database management module 7 includes the following main data tables:

[0135] Spectral data table: stores original spectral data and pre-processed spectral data;

[0136] Feature data table: stores the extracted topological features and weighted features;

[0137] Aging grade table: stores the aging grade assessment results of different road sections;

[0138] Life prediction table: stores remaining life prediction results and uncertainty analysis;

[0139] Maintenance record sheet: stores historical maintenance measures and their effects;

[0140] Communication interface module 8 connects to decision support module 6 to exchange data with external systems, enabling data export and remote access. Communication interface module 8 supports multiple communication protocols, including HTTP / HTTPS (web services), MQTT (IoT device communication), and WebSocket (real-time data transmission).

[0141] The main interfaces provided by the communication interface module 8 include:

[0142] Data acquisition interface: receiving spectral data collected on site;

[0143] Assessment result query interface: provides aging level and remaining life query functions;

[0144] Maintenance suggestion acquisition interface: obtain maintenance decision suggestions generated by the system;

[0145] Data export interface: supports data export in multiple formats such as CSV, Excel, PDF, etc.

[0146] The user interface module 9 is connected to the decision support module 6 and is used to display the assessment results and maintenance recommendations in the form of charts and reports. The user interface module 9 adopts a responsive design and supports dual-platform operation on PC and mobile terminals.

[0147] The user interface module 9 mainly includes the following functional areas:

[0148] Pavement condition visualization area: displays the distribution of pavement aging grades in the form of a heat map;

[0149] Spectral analysis area: displays characteristic spectral lines and their comparison with standard spectral lines;

[0150] Life prediction area: displays the remaining life prediction and uncertainty analysis in the form of charts;

[0151] Maintenance suggestion area: displays maintenance suggestions and cost-benefit analysis generated by the system;

[0152] like Figure 8 As shown, the multispectral imaging evaluation method for asphalt pavement aging degree of the present invention includes the following steps:

[0153] Step 1: Spectral data collection:

[0154] A vehicle-mounted multispectral imaging system was used to collect spectral data from asphalt pavement. First, the hyperspectral camera was controlled using the camera function of a mobile app. The vehicle was driven along the road at a steady speed of 30 km / h, collecting spectral data within the 380-800 nm wavelength range. During the collection process, 100% road coverage was ensured, meaning at least 10 sampling points were collected within any 1-square-meter area.

[0155] Step 2: Data preprocessing and feature extraction:

[0156] The collected spectral data were preprocessed, including noise removal, spectral calibration, and data normalization. Specifically, dark pixel correction was used to perform spectral calibration to eliminate the influence of ambient light, and then minimum-maximum normalization was performed to normalize all spectral data to the [0,1] interval.

[0157] Next, perform multi-scale feature decomposition on the pre-processed spectral data to generate a feature matrix. The specific decomposition methods include:

[0158] Low-frequency feature extraction: for the 400-650 nm range, the window size is 21 and the step size is 5;

[0159] IF feature extraction: for the 380-420nm and 650-800nm ​​ranges, the window size is 11 and the step size is 3;

[0160] High-frequency feature extraction: For sensitive bands, the window size is 7 and the step size is 1;

[0161] Then, the topological features of the spectral lines are calculated by the topological structure extraction unit to generate a topological feature vector. Finally, the features of different scales and the topological features are fused to generate a complete feature matrix with a dimension of m×200.

[0162] Step 3: Spectral line attention calculation:

[0163] Calculate the weight coefficients of different spectral lines in the feature matrix. First, the frequency domain correlation evaluation unit analyzes the richness of information in each band and generates frequency domain correlation weights. Second, the discrimination calculation unit calculates the significance of spectral line differences between different aging levels and generates discrimination weights. Third, the temporal stability evaluation unit evaluates the stability of spectral line features over time and generates stability weights. Finally, the weight fusion unit integrates the three weights to generate a final weight vector, which is used to weight the feature matrix and output a weighted feature vector.

[0164] Step 4: Time series feature learning:

[0165] The system processes the time series changes of weighted feature vectors to capture the temporal evolution of road aging. Specifically, it processes time series data using memory-enhanced network units, a multi-scale memory pool simultaneously maintains short-term, medium-term, and long-term memories, a spatiotemporal feature interaction unit fuses spatial and temporal features, and a feature pyramid structure integrates information at different scales, ultimately outputting a 128-dimensional state vector.

[0166] Step 5: Aging assessment and life prediction:

[0167] Based on the state vector, the current aging state is determined and future evolution is predicted. First, the manifold construction unit maps the state vector onto a 5-dimensional Riemannian manifold. The multi-path generation unit generates eight possible evolution paths, and the path evaluation unit calculates the probability weights for each path. Next, the aging grade determination unit classifies the pavement aging degree into levels 0-3. The remaining service life calculation unit calculates the remaining service life based on the aging grade and path probabilities. Simultaneously, the uncertainty quantification unit and the risk assessment unit analyze the reliability of the prediction results and the decision risk.

[0168] Step 6: Maintenance decision generation:

[0169] Maintenance decision recommendations are generated based on the aging level and remaining lifespan. The maintenance strategy generation unit matches the optimal maintenance measures based on the aging level. The cost-benefit analysis unit calculates the return on investment for each option. The implementation timing recommendation unit determines the optimal maintenance timing. Finally, the feedback optimization unit collects actual maintenance performance data to optimize the prediction model and decision-making strategy.

[0170] This example is applied to assessing the aging of asphalt pavement on a highway. The system uses a vehicle-mounted hyperspectral camera to scan along the G95 Capital Ring Expressway on a clear day. The hyperspectral camera has a resolution of 2048×1536 and a wavelength range of 200-2500nm. The actual analysis band used is 380-800nm.

[0171] The scanning vehicle travels at 30 km / h, collecting a complete set of spectral data every 10 meters. The raw spectral data undergoes dark pixel correction and normalization before being input into the topological feature extraction module. Multi-scale decomposition employs a three-layer structure, targeting low-frequency, mid-frequency, and high-frequency features. Topological feature extraction is based on persistent homology theory, constructing simple complexes, calculating persistent homology groups, and generating feature vectors.

[0172] The spectral attention calculation module analyzes the information richness, discrimination, and stability of each band and generates a weight vector. In this example, the 450-500nm band (corresponding to the degree of asphalt oxidation) and the 650-700nm band (corresponding to surface cracks) received higher weights (0.25 and 0.22, respectively).

[0173] After learning time series features and predicting lifespan, the system assessed the pavement aging degree of the G95 Capital Ring Expressway section from K50+000 to K55+000 as: K50+000 to K52+000 as Level 1 (mild aging), K52+000 to K53+500 as Level 2 (moderate aging), and K53+500 to K55+000 as Level 0 (no significant aging). For the Level 2 aging section, the system predicts a remaining service life of 2.5±0.3 years and recommends functional repair within the next maintenance cycle, with a return on investment of approximately 210%.

[0174] This embodiment is applied to the analysis of urban road aging evolution. The system scans the main urban roads regularly for 6 consecutive months, once a month, to form a complete time series data set.

[0175] The topological feature extraction module generates a feature matrix for each scan. The spectral line attention module dynamically adjusts weight coefficients based on six months of data, particularly strengthening the weights of bands sensitive to seasonal changes. The time series feature learning module processes six months of time series data using a memory-enhanced LSTM network to capture the evolution of aging.

[0176] Analysis results show that the aging rate of the city's main roads accelerated significantly between March and April, with the aging level rising from Level 1 to Level 2. Through multi-path prediction analysis, the system identified two main evolutionary paths: Path A (65% probability) predicts that the aging level will remain at Level 2 until the end of August; Path B (35% probability) predicts that under continued high temperatures and heavy traffic, the aging level may rise to Level 3 in September.

[0177] Based on this prediction, the decision support module recommended preventive maintenance in early July to prevent further deterioration. Specifically, it recommended the use of micro-surfacing technology, which has an estimated return on investment of 180% and can extend the remaining useful life from a predicted 3.2 years to 5.8 years.

[0178] This embodiment is applied to the prediction of asphalt pavement life under complex climate and traffic conditions. The test area is a connecting line of a coastal city. This section of road is affected by the marine environment, has large traffic volume fluctuations, and has complex aging conditions.

[0179] The system performed four complete scans over the course of a year: in winter, spring, summer, and autumn. The topological feature extraction module extracted stable and changing features from the data across the four seasons. The spectral line attention module dynamically adjusted weights based on environmental factors, particularly increasing the weighting of bands sensitive to salt erosion.

[0180] The time series feature learning module constructed a comprehensive model of seasonal changes and integrated it with historical meteorological and traffic data. The lifespan prediction module generated 12 possible evolution paths, covering various climate change and traffic growth scenarios. Through a multi-scenario weighted analysis, the system predicted the remaining lifespan of various sections of the road section to range from 1.8 to 4.5 years, with a 95% confidence interval.

[0181] Based on the prediction results, the decision support module recommends a segmented maintenance strategy: sections with a grade of deterioration of approximately 15% (Level 3) will undergo immediate structural repairs; sections with a grade of deterioration of approximately 40% (Level 2) will undergo functional repairs within six months; and sections with grades 1 and 0 (approximately 45%) will continue to be monitored. This maintenance strategy is expected to save 25% in maintenance costs while ensuring that the road's service level remains unchanged.

[0182] The multispectral imaging assessment system and method for asphalt pavement aging provided by this invention achieves precise assessment of asphalt pavement aging and accurate prediction of remaining life through the deep integration of multispectral imaging technology and advanced algorithms. It has the following significant advantages:

[0183] 1. Multi-dimensional data acquisition: Using multispectral imaging technology, rich spectral information in the 380-800nm ​​wavelength range is acquired to form a complete characteristic spectrum of pavement aging.

[0184] 2. High-precision feature extraction: Through topological feature extraction and attention mechanism, key features are extracted from massive spectral data to improve evaluation accuracy.

[0185] 3. Dynamic evolution prediction: Utilize time series feature learning and multi-path prediction to achieve dynamic analysis and accurate prediction of the pavement aging process.

[0186] 4. Scientific decision support: Based on accurate aging assessment and life prediction, provide scientific maintenance decision recommendations and optimize resource allocation.

[0187] The present invention is applicable to the aging assessment and maintenance management of various asphalt pavements such as highways, urban roads, and airport runways. It can significantly improve the scientificity and accuracy of maintenance decisions, reduce maintenance costs, and extend the service life of the pavement. It has broad application prospects.

[0188] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

Claims

1. Asphalt pavement aging degree multispectral imaging assessment system, characterized by: include: A multispectral imaging module, configured to collect spectral data of an asphalt pavement, wherein the spectral data includes a plurality of spectral lines, each of which includes spectral intensity values ​​in a plurality of wavelength ranges; A topological feature extraction module, connected to the multispectral imaging module, is used to perform multi-scale feature decomposition on the spectral data, generate a feature matrix, and extract topological structure features from the feature matrix; a spectral line attention calculation module, connected to the topological feature extraction module, for calculating weight coefficients of different spectral lines in the feature matrix, and performing weighted processing on the feature matrix based on the weight coefficients to generate a weighted feature vector; A time series feature learning module, connected to the spectral line attention calculation module, is used to process the time series changes of the weighted feature vector, capture the time series evolution law of road aging, and output a state vector; a life prediction module, connected to the time series feature learning module, configured to receive the state vector, determine the current aging state in a predefined feature space based on the state vector, generate multiple possible aging evolution paths, calculate the probability distribution of each path, and comprehensively determine the pavement aging grade and remaining life; A decision support module is connected to the life prediction module and is used to generate maintenance decision suggestions based on the aging level and the remaining life.

2. The multispectral imaging assessment system for asphalt pavement aging according to claim 1 is characterized in that: The multispectral imaging module includes: Hyperspectral camera, used to obtain spectral information of asphalt pavement in the wavelength range of 380-800nm; a scanning mechanism, connected to the hyperspectral camera, and configured to control the hyperspectral camera to scan along a road surface; An image preprocessing unit, connected to the hyperspectral camera, for performing noise removal and spectrum calibration on the collected spectral data; A mobile terminal APP is connected to the image preprocessing unit, and is used to control the hyperspectral camera and the scanning mechanism, and display the spectral data.

3. The multispectral imaging assessment system for asphalt pavement aging according to claim 1, characterized in that: The topological feature extraction module includes: A multi-scale decomposition unit, configured to decompose the spectral data according to different wavelength intervals to form low-frequency features, medium-frequency features, and high-frequency features; A sliding window analysis unit, connected to the multi-scale decomposition unit, for sliding along the wavelength dimension to extract local features; A topological structure extraction unit, connected to the sliding window analysis unit, for calculating the topological features of the spectral line data and generating a topological feature vector; The feature fusion unit is connected to the multi-scale decomposition unit and the topological structure extraction unit, and is used to fuse features of different scales and topological features to generate a complete feature matrix.

4. The multispectral imaging assessment system for asphalt pavement aging according to claim 1, characterized in that: The spectral line attention calculation module includes: The frequency domain correlation evaluation unit is used to analyze the richness of each band information and generate the frequency domain correlation weight; a discrimination calculation unit connected to the frequency domain correlation evaluation unit, for calculating the significance of the spectral line differences between different aging levels and generating a discrimination weight; A time series stability evaluation unit, connected to the discrimination calculation unit, for evaluating the stability of spectral line characteristics over time and generating a stability weight; a weight fusion unit connected to the frequency domain correlation evaluation unit, the discrimination calculation unit, and the temporal stability evaluation unit, and configured to integrate the three weights to generate a final weight vector; The weighted processing unit is connected to the weight fusion unit and is used to weight the feature matrix according to the weight vector and output a weighted feature vector.

5. The multispectral imaging assessment system for asphalt pavement aging according to claim 1, characterized in that: The temporal feature learning module includes: A memory-enhanced network unit, configured to process temporal changes of the weighted feature vectors and filter and store key information through a multi-gating structure; A multi-scale memory pool, connected to the memory-enhanced network unit, for simultaneously maintaining short-term memory, medium-term memory, and long-term memory; A spatiotemporal feature interaction unit, connected to the multi-scale memory pool, for fusing spatial features and temporal features; The feature pyramid structure is connected to the spatiotemporal feature interaction unit and is used to integrate feature information of different scales to generate a final state vector.

6. The multispectral imaging assessment system for asphalt pavement aging according to claim 1, characterized in that: The life prediction module includes: a manifold construction unit, configured to map the state vector to a predefined multi-dimensional feature space; a multi-path generation unit, connected to the manifold construction unit, for generating multiple possible aging evolution paths based on the current state; a path evaluation unit, connected to the multi-path generation unit, for calculating the probability weight of each path; an aging grade judgment unit, connected to the path evaluation unit, for classifying the road surface aging degree into 0, 1, 2, and 3, wherein 3 represents the most severe road surface aging degree; The remaining service life calculation unit is connected to the aging level judgment unit and is used to calculate the remaining service life based on the aging level and the path probability.

7. The multispectral imaging assessment system for asphalt pavement aging according to claim 6, characterized in that: The life prediction module also includes: Uncertainty quantification unit, used to analyze measurement uncertainty, model uncertainty and prediction uncertainty; The risk assessment unit is connected to the uncertainty quantification unit and is used to divide high-risk areas and low-risk areas and provide decision-making risk assessment.

8. The multispectral imaging assessment system for asphalt pavement aging according to claim 1, characterized in that: The decision support module includes: Maintenance strategy generation unit, used to match the optimal maintenance measures according to the aging level and remaining life; a cost-benefit analysis unit, connected to the maintenance strategy generation unit, for calculating the return on investment of different maintenance plans; an implementation timing recommendation unit, connected to the cost-benefit analysis unit, for recommending an optimal maintenance timing based on the aging rate and budget constraints; The feedback optimization unit is connected to the implementation timing recommendation unit and is used to collect maintenance effect data and adjust the prediction model parameters.

9. The multispectral imaging assessment system for asphalt pavement aging according to claim 1, characterized in that: Also includes: a database management module, connected to the topological feature extraction module, the temporal feature learning module and the life prediction module respectively, and used for storing feature data, state data and prediction results; A communication interface module, connected to the decision support module, is used to implement data exchange with external systems and support data export and remote access; The user interface module is connected to the decision support module and is used to display the assessment results and maintenance suggestions in the form of charts and reports.

10. A method for evaluating the degree of aging of an asphalt pavement using multispectral imaging, comprising: The following steps are involved: Collecting spectral data of an asphalt pavement, the spectral data including a plurality of spectral lines, each of the spectral lines including spectral intensity values ​​in a plurality of wavelength ranges; Performing multi-scale feature decomposition on the spectral data to generate a feature matrix, and extracting topological structure features from the feature matrix; Calculating weight coefficients of different spectral lines in the characteristic matrix, and performing weighted processing on the characteristic matrix based on the weight coefficients to generate a weighted characteristic vector; Processing the time series changes of the weighted feature vector, capturing the time series evolution law of road aging, and outputting a state vector; receiving the state vector, and determining a current aging state in a predefined feature space based on the state vector, generating multiple possible aging evolution paths, calculating a probability distribution of each path, and comprehensively determining a pavement aging grade and remaining life; Based on the ageing level and the remaining life, maintenance decision recommendations are generated.

Images