Pavement internal disease analysis method and system based on artificial intelligence

Abstract

The invention relates to the technical field of road maintenance and management, in particular to a pavement internal disease analysis method and system based on artificial intelligence. Five core processes of multi-layer sensor network real-time data acquisition, multi-modal data fusion and deep learning feature extraction, dynamic environment and use mode combined disease prediction and analysis, intelligent prediction and decision support system, and real-time disease condition evaluation and report generation are carried out. Through the technical means, comprehensiveness and timeliness of data acquisition, accuracy of disease identification and prediction, scientificity of a maintenance strategy and optimization of resource allocation are realized; the finally generated disease current situation report provides a detailed disease distribution diagram, disease type analysis and disease severity evaluation, comprehensive, accurate and timely disease information is provided for maintenance personnel, and the efficiency and effect of pavement maintenance and management are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of road maintenance and management, and in particular to an artificial intelligence-based road surface internal disease analysis method and system. Background Art

[0002] With the increase in traffic volume and heavy-loaded vehicles, pavement damage is becoming increasingly serious, which not only affects the service life of roads, but also poses a potential threat to traffic safety. There are many problems in the current pavement damage analysis, mainly including incomplete data collection, single data processing and analysis methods, inaccurate damage prediction and early warning, imperfect decision support, and untimely damage status assessment and report generation. These problems are mainly caused by reliance on a single type of sensor, lack of comprehensive analysis capabilities of multi-source data, lack of dynamic adjustment capabilities of prediction models, and lack of effective decision support systems.

[0003] The existing technology (Chinese invention patent, publication number: CN115755193A, name: pavement structure internal disease identification method) has the following deficiencies or defects in the face of these problems: single sensor type leads to incomplete data collection; single data processing and analysis methods cannot fully utilize multi-source data; disease prediction and early warning are inaccurate and lack dynamic adjustment capabilities; lack of effective decision support system, incomplete disease impact assessment; disease status assessment and report generation are not timely, and the report content is not detailed and comprehensive enough. These deficiencies and defects limit the comprehensiveness, accuracy and real-time performance of the existing technical solutions in disease identification, prediction, assessment and decision-making. Summary of the invention

[0004] In view of the many problems existing in the above-mentioned prior arts, the present invention provides an artificial intelligence-based pavement internal disease analysis method and system. The present invention utilizes a multi-level sensor network to collect pavement internal and external environmental data in real time, adopts multimodal data fusion and deep learning technology for data processing and feature extraction, and combines dynamic environment and usage pattern data to construct a disease cause model and a disease development trend prediction model; through an intelligent prediction system and a decision support system, real-time disease prediction and maintenance strategies are provided, and a detailed disease status report is generated. The present invention improves the accuracy and scientificity of disease identification, prediction and maintenance strategies by comprehensively utilizing multi-source data, and significantly improves the efficiency and effectiveness of pavement maintenance and management.

[0005] A method for analyzing internal pavement defects based on artificial intelligence comprises the following steps:

[0006] By applying a sensor network targeting the inside of the road surface, the internal structure and external environment data of the road surface are collected and preprocessed in real time to generate ground penetrating radar feature data, vibration feature data, environmental feature data and stress feature data;

[0007] After the ground penetrating radar feature data, vibration feature data, environmental feature data and stress feature data are synchronized in time and formatted uniformly, multimodal data fusion is performed using data-level, feature-level and decision-level fusion methods, and feature extraction is performed using a deep learning model to generate fused feature data and disease feature data.

[0008] Combine environmental characteristic data and usage pattern characteristic data to construct a disease cause model and a disease development trend prediction model, analyze the cause, development trend and impact of the disease, and generate disease cause analysis data, disease trend prediction data and disease impact assessment data; the usage pattern characteristic data is generated by time synchronization and format unification of traffic flow data, vehicle weight data and road use frequency data;

[0009] Based on the disease cause analysis data, disease trend prediction data and disease impact assessment data, build an intelligent prediction system and decision support system to provide real-time disease prediction and maintenance strategies, generate disease prediction data and maintenance strategy data, and optimize the intelligent prediction system and decision support system based on actual maintenance feedback;

[0010] Use real-time updated disease prediction data and disease characteristic data to assess current disease conditions and generate disease status reports.

[0011] Preferably, the sensor network includes: ground penetrating radar sensors, piezoelectric sensors, temperature and humidity sensors and accelerometers, arranged at different depths and areas, including:

[0012] By applying ground penetrating radar sensors to collect electromagnetic wave reflection data from the internal structure of the road surface;

[0013] Piezoelectric sensors are placed on and below the pavement surface to collect vibration and pressure data;

[0014] Temperature and humidity sensors are arranged around the road surface to collect ambient temperature and humidity data;

[0015] Accelerometers are placed in key areas of the road surface to collect dynamic stress data of the road surface.

[0016] Preferably, the multimodal data fusion method comprises:

[0017] Data-level fusion: combining raw data from different sensors at the data level to generate a fused data set;

[0018] Feature-level fusion: extract features of various types of data, and fuse the extracted features to generate fused feature data;

[0019] Decision-level fusion: After independently analyzing various types of data, the results are integrated to generate comprehensive decision-making data.

[0020] Preferably, the deep learning model includes a convolutional neural network and a graph convolutional network for feature extraction and disease identification, including:

[0021] Use convolutional neural network to extract features from fused feature data and generate preliminary disease feature data;

[0022] The preliminary disease feature data are further processed using a graph convolutional network to generate the final disease feature data.

[0023] Preferably, the disease cause model is calculated by the following formula:

[0024] C=f(E,U)=a·E+b·U+c

[0025] Among them, C represents disease cause analysis data, E represents environmental characteristic data, U represents usage pattern characteristic data, f represents the function of converting environmental characteristic data and usage pattern characteristic data into disease cause analysis data, a and b are weights, and c is the deviation parameter.

[0026] Preferably, the disease development trend prediction model is calculated by the following formula:

[0027] T=g(C,t)=α·C+β·t+γ

[0028] Among them, T represents disease trend prediction data, C represents disease cause analysis data, t represents time variable, g represents the function of converting disease cause analysis data and time variable into disease trend prediction data, α and β are weights, and γ is the deviation parameter.

[0029] Preferably, the intelligent prediction system uses a reinforcement learning algorithm to continuously optimize disease prediction and maintenance strategies, including:

[0030] Based on historical disease data and current disease data, the intelligent prediction model is trained using reinforcement learning algorithms;

[0031] According to the real-time updated disease data, the parameters of the intelligent prediction model are adjusted and optimized to generate real-time disease prediction data.

[0032] Preferably, the maintenance strategy data includes maintenance time, maintenance resources and specific operation suggestions, including:

[0033] Determine the areas that require priority maintenance and the specific maintenance time based on the disease impact assessment data;

[0034] Allocate maintenance resources based on disease severity and resource availability, and generate a maintenance resource allocation plan;

[0035] Provides specific operating recommendations, including maintenance tools to use, operating procedures, and safety precautions.

[0036] Preferably, the disease status report includes a disease distribution map, disease type analysis and disease severity assessment, and provides detailed information in the form of charts and text descriptions, including:

[0037] Disease distribution map: shows the specific location and distribution of diseases on the road surface;

[0038] Disease type analysis: Classify and describe different types of diseases and their characteristics;

[0039] Disease severity assessment: Quantitatively assess the severity of the disease and provide specific scores and recommendations.

[0040] A system for implementing the artificial intelligence-based pavement internal disease analysis method, comprising:

[0041] A multi-layer sensor network, including ground-penetrating radar sensors, piezoelectric sensors, temperature and humidity sensors, and accelerometers, is used to collect and pre-process the internal structure and external environment data of the road surface in real time;

[0042] The data processing module is used to synchronize the time and format of the ground penetrating radar feature data, vibration feature data, environmental feature data and stress feature data, fuse multimodal data using data-level, feature-level and decision-level fusion methods, extract features through a deep learning model, and generate fused feature data and disease feature data;

[0043] An analysis module is used to combine environmental characteristic data and usage pattern characteristic data to construct a disease cause model and a disease development trend prediction model, analyze the cause, development trend and impact of the disease, and generate disease cause analysis data, disease trend prediction data and disease impact assessment data. The usage pattern characteristic data is generated by time synchronization and format unification of traffic flow data, vehicle weight data and road use frequency data;

[0044] Intelligent prediction and decision support module, which is used to build intelligent prediction system and decision support system based on disease cause analysis data, disease trend prediction data and disease impact assessment data, provide real-time disease prediction and maintenance strategy, generate disease prediction data and maintenance strategy data, and optimize intelligent prediction system and decision support system according to actual maintenance feedback;

[0045] The condition assessment and report generation module is used to use real-time updated disease prediction data and disease characteristic data to assess the current disease condition and generate a disease status report.

[0046] Compared with the prior art, the advantages and beneficial effects of the present invention are:

[0047] The present invention realizes the comprehensiveness and timeliness of data collection through multi-level sensor network technology;

[0048] The present invention realizes the comprehensive analysis of multi-source data and the accuracy of disease identification through multi-modal data fusion and deep learning technology;

[0049] The present invention achieves accuracy and real-time performance of disease prediction and early warning through a disease prediction model technology that combines dynamic environment and usage mode;

[0050] The present invention realizes the scientific nature of maintenance strategy and optimization of resource allocation through intelligent prediction and decision support system technology;

[0051] The present invention realizes the timeliness and comprehensiveness of disease status assessment and report generation through real-time data analysis and report generation system technical means. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 It is a schematic diagram of the process of the present invention;

[0053] Figure 2 This is a flow chart of multimodal data fusion in the present invention;

[0054] Figure 3 It is a schematic diagram of the disease cause model and the disease development trend prediction model in the present invention;

[0055] Figure 4 This is the architecture diagram of the intelligent prediction and decision support system in the present invention;

[0056] Figure 5 It is a schematic diagram of the disease status report in the present invention;

[0057] Figure 6 It is a structural block diagram of the system of the present invention. DETAILED DESCRIPTION

[0058] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is obvious that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0059] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0060] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification, and should not be interpreted in an idealized or overly rigid manner.

[0061] In the case of using expressions such as "at least one of A, B, and C, etc.", it should generally be interpreted in accordance with the meaning of the expression generally understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.). In the case of using expressions such as "at least one of A, B, or C, etc.", it should generally be interpreted in accordance with the meaning of the expression generally understood by those skilled in the art (for example, "a system having at least one of A, B, or C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).

[0062] Some block diagrams and / or flow charts are shown in the accompanying drawings. It should be understood that some boxes or combinations thereof in the block diagrams and / or flow charts can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that these instructions can create a device for implementing the functions / operations described in these block diagrams and / or flow charts when executed by the processor. The technology of the present disclosure can be implemented in the form of hardware and / or software (including firmware, microcode, etc.). In addition, the technology of the present disclosure can take the form of a computer program product on a computer-readable storage medium storing instructions, which can be used by an instruction execution system or used in conjunction with an instruction execution system.

[0063] like Figure 1 As shown, a method for analyzing internal pavement defects based on artificial intelligence includes the following steps:

[0064] By arranging a sensor network for the inside of the road surface, the internal structure and external environment data of the road surface are collected and preprocessed in real time to generate ground penetrating radar feature data, vibration feature data, environmental feature data and stress feature data;

[0065] Preferably, the sensor network includes: ground penetrating radar sensors, piezoelectric sensors, temperature and humidity sensors and accelerometers, arranged at different depths and areas, including:

[0066] By applying ground penetrating radar sensors to collect electromagnetic wave reflection data from the internal structure of the road surface;

[0067] Piezoelectric sensors are placed on and below the pavement surface to collect vibration and pressure data;

[0068] Temperature and humidity sensors are arranged around the road surface to collect ambient temperature and humidity data;

[0069] Accelerometers are placed in key areas of the road surface to collect dynamic stress data of the road surface.

[0070] In the present invention, by arranging a variety of sensors, the data of the internal structure and external environment of the road surface can be comprehensively and accurately collected, providing detailed basic data for the analysis of internal road surface diseases. The ground penetrating radar sensor detects the changes in the internal structure of the road surface by emitting electromagnetic waves and receiving its reflected signals. The ground penetrating radar characteristic data generated by it can reflect the hierarchical structure of the road surface and the potential location of the disease. The piezoelectric sensor uses the piezoelectric effect. When the road surface is subjected to vehicle load or other pressure, the generated electrical signal reflects the vibration and pressure changes. In this way, the piezoelectric sensor can generate vibration characteristic data for monitoring the stress distribution on the road surface. The temperature and humidity sensor is used to collect environmental temperature and humidity data. Environmental conditions have an important influence on the formation and development of road surface diseases. The data provided by the temperature and humidity sensor can help understand the impact of environmental factors on the road surface. The accelerometer is arranged in the key area of ​​the road surface to collect dynamic stress data. When a vehicle passes by, the road surface will produce a dynamic response. The data of the accelerometer can reflect the dynamic stress changes of the road surface.

[0071] These sensors are preferably arranged at different depths and areas to ensure full coverage of the monitoring area and provide high-resolution data. By applying ground-penetrating radar sensors, their electromagnetic wave reflection data can penetrate the multi-layer structure of the road surface and provide detailed internal structural information. Piezoelectric sensors are arranged both on the surface and below the road surface to capture surface and internal vibrations and pressure changes. Temperature and humidity sensors are arranged in the environment around the road surface to ensure that the impact of changes in ambient temperature and humidity on the road surface can be accurately reflected. Accelerometers are arranged in key areas of the road surface, such as sections that are often crushed by heavy-loaded vehicles, to provide dynamic stress data in key areas.

[0072] Through the above arrangement and data acquisition method, the present invention can realize comprehensive monitoring of the internal structure and external environment of the road surface. The real-time collected data is preprocessed to generate ground penetrating radar feature data, vibration feature data, environmental feature data and stress feature data, which provide a reliable basis for subsequent multimodal data fusion and deep feature extraction. Specifically, the ground penetrating radar feature data is used to identify structural anomalies and potential disease locations inside the road surface, the vibration feature data is used to analyze the stress distribution on the road surface, the environmental feature data is used to evaluate the impact of environmental conditions on the road surface, and the stress feature data is used to monitor the response of the road surface under dynamic loads. These multimodal data provide comprehensive feature information for disease cause analysis and trend prediction through fusion and deep feature extraction, which helps to realize the accurate identification and prediction of road surface diseases, thereby formulating effective maintenance strategies, extending the service life of the road surface, and ensuring traffic safety.

[0073] like Figure 2 As shown, after the ground penetrating radar feature data, vibration feature data, environmental feature data and stress feature data are synchronized in time and formatted uniformly, multimodal data fusion is performed using data-level, feature-level and decision-level fusion methods, and feature extraction is performed using a deep learning model to generate fused feature data and disease feature data;

[0074] Preferably, the multimodal data fusion method comprises:

[0075] Data-level fusion: combining raw data from different sensors at the data level to generate a fused data set;

[0076] Feature-level fusion: extract features of various types of data, and fuse the extracted features to generate fused feature data;

[0077] Decision-level fusion: After independently analyzing various types of data, the results are integrated to generate comprehensive decision-making data.

[0078] Through time synchronization and format unification, data from different sensors can be compared and analyzed on the same timeline and in a consistent format for subsequent processing. Time synchronization is achieved by using timestamps or synchronization signals, and format unification is achieved through data normalization, converting data of different units and scales into a unified standard format.

[0079] Data-level fusion combines the raw data from different sensors at the data level to generate a fused data set. For example, the electromagnetic wave reflection data from the ground penetrating radar, the vibration data from the piezoelectric sensor, the environmental data from the temperature and humidity sensor, and the dynamic stress data from the accelerometer are combined into a comprehensive data set. This fused data set can provide more comprehensive road surface information and help discover disease characteristics that cannot be detected by a single sensor.

[0080] Feature-level fusion is to extract features from various types of data and fuse these features to generate fused feature data. Feature extraction refers to extracting information that can characterize the essence of data from raw data, such as extracting reflection intensity features from ground penetrating radar data, extracting frequency features from vibration data, extracting temperature and humidity change features from environmental data, and extracting stress distribution features from stress data. Fusion of these feature data can enhance the ability to identify road surface diseases. For example, by combining reflection intensity features and vibration frequency features, cracks inside the road surface can be more accurately identified.

[0081] Decision-level fusion is to analyze various types of data independently, fuse the analysis results, and generate comprehensive decision data. In this process, various types of data are processed by the deep learning model to generate preliminary disease identification results, and then these identification results are analyzed comprehensively. For example, ground penetrating radar data may identify cavities under the road surface, vibration data may identify cracks on the road surface, and environmental data may identify damage caused by temperature and humidity changes. Through decision-level fusion, these identification results can be combined to more comprehensively evaluate the health of the road surface and generate comprehensive disease characteristic data.

[0082] Through the multimodal data fusion method, the data advantages of different sensors can be fully utilized, the shortcomings of a single data source can be compensated, and the accuracy and robustness of disease identification can be improved. Combining deep learning models for feature extraction and decision-level fusion can not only automatically process large amounts of data, but also improve the accuracy and reliability of analysis results. In practical applications, this method can help maintenance personnel to promptly detect and deal with pavement diseases, extend the service life of the pavement, reduce maintenance costs, and ensure traffic safety. In this way, the present invention realizes a comprehensive, accurate and intelligent analysis of internal pavement diseases, effectively improving the efficiency of pavement maintenance and management.

[0083] Preferably, the deep learning model includes a convolutional neural network and a graph convolutional network for feature extraction and disease identification, including:

[0084] Use convolutional neural network to extract features from fused feature data and generate preliminary disease feature data;

[0085] The preliminary disease feature data are further processed using a graph convolutional network to generate the final disease feature data.

[0086] In the present invention, the deep learning model includes a convolutional neural network (CNN) and a graph convolutional network (GCN) for feature extraction and disease recognition. First, the convolutional neural network is used to extract features from the fused feature data to generate preliminary disease feature data; then, the graph convolutional network is used to further process the preliminary disease feature data to generate the final disease feature data.

[0087] Convolutional neural network (CNN) is a deep learning model widely used in the fields of image processing and feature extraction. Its core principle is to process the input data layer by layer through convolutional layers, pooling layers and fully connected layers to extract high-level features. In the present invention, CNN is used to process fused feature data and extract preliminary disease feature data. For example, when the input data is a data set that integrates ground penetrating radar features, vibration features, environmental features and stress features, CNN can extract features related to pavement diseases, such as crack morphology, abnormal stress distribution, etc. through convolution operations. The convolution operation can effectively capture local features, while the pooling operation can reduce the data dimension and prevent overfitting.

[0088] After the preliminary disease feature data is generated, it is further processed using a graph convolutional network (GCN). GCN is a deep learning model suitable for processing graph structured data. Its principle is to make full use of the connection relationship between nodes for feature extraction and information propagation through graph convolution operations. In the present invention, pavement disease can be regarded as a graph structure, in which each node represents the feature data at a different position, and the edges between nodes represent the correlation between these positions. For example, stress data and vibration data at different positions on the road surface may be correlated, and these related information can be captured through GCN. Through graph convolution operations, GCN can combine local features in the preliminary disease feature data with global information to generate more accurate and comprehensive disease feature data.

[0089] Combining the advantages of CNN and GCN, the present invention can effectively extract and identify road surface defects at different levels. CNN is good at processing structured grid data, such as image data, and can capture local features; while GCN can process unstructured graph data and capture complex correlation information between nodes. In practical applications, local features related to the defect are first extracted through CNN, and then these features are further integrated through GCN to generate the final defect feature data. This method not only improves the accuracy of feature extraction, but also enhances the robustness of defect identification.

[0090] In terms of implementation examples, for example, when detecting road cracks, the fused feature data includes information such as electromagnetic wave reflection intensity, vibration frequency, ambient temperature, and stress distribution. By processing this data, CNN can preliminarily identify the location and shape of possible cracks. Then, GCN uses these preliminarily identified feature data as input, combined with the correlation information between each location, to further confirm the actual situation of the cracks, identify the depth, length, and impact of the cracks on the road surface. The resulting disease feature data can be provided to maintenance personnel for formulating maintenance plans, repairing cracks in a timely manner, and ensuring road safety and durability.

[0091] In summary, the present invention can effectively extract and identify pavement disease features and generate accurate and comprehensive disease feature data by combining convolutional neural networks and graph convolutional networks. This method not only takes advantage of the advantages of different types of data, but also captures complex related information, improves the accuracy and reliability of pavement disease analysis, and thus provides strong support for pavement maintenance and management.

[0092] like Figure 3 As shown, combining environmental characteristic data and usage pattern characteristic data, constructing a disease cause model and a disease development trend prediction model, analyzing the cause, development trend and impact of the disease, and generating disease cause analysis data, disease trend prediction data and disease impact assessment data; the usage pattern characteristic data is generated by time synchronization and format unification processing of traffic flow data, vehicle weight data and road use frequency data;

[0093] Environmental characteristic data include environmental factors that affect road conditions, such as temperature, humidity, and rainfall. These data are collected in real time by temperature and humidity sensors arranged around the road surface, and after data preprocessing steps, they form standardized environmental characteristic data. These data are crucial to understanding the impact of environmental changes on road surface diseases. For example, frequent temperature changes may cause road surface materials to expand and contract, resulting in cracks.

[0094] Usage pattern characteristic data refers to the usage of the road, including traffic flow data, vehicle weight data and road surface usage frequency data. These data are collected in real time by sensors arranged on the road surface, and are generated after time synchronization and unified format processing. For example, traffic flow data can be obtained through traffic monitoring cameras and flow meters installed at the entrance of the road, vehicle weight data can be obtained through weighing sensors installed on the road surface, and road surface usage frequency data can be obtained through the statistics of the number of vehicle passes. These data can reflect the load of the road surface under different usage conditions and have a direct impact on the formation and development of diseases.

[0095] The disease cause model is used to analyze the impact of environmental characteristic data and usage pattern characteristic data on the formation of pavement diseases. The model is trained through a machine learning algorithm and uses historical data to establish the relationship between the environment and usage patterns and diseases. For example, the influence of factors such as ambient temperature, humidity changes, and traffic flow on pavement cracks, settlements, and other diseases can be trained through methods such as linear regression models and multi-layer perceptrons (MLP). The disease cause analysis data generated by the disease cause model can help maintenance personnel understand which factors are the main causes of diseases, so as to take targeted preventive measures.

[0096] The disease development trend prediction model is used to predict the development trend of diseases in the future. The model combines disease cause analysis data and time variables, and uses time series analysis methods (such as LSTM, ARIMA, etc.) to predict the development trajectory of the disease. For example, in the case of existing cracks, the model can predict the expansion speed and range of the cracks in the next few months, thereby providing a scientific basis for maintenance plans. With the disease development trend prediction data, maintenance personnel can formulate repair plans in advance to avoid further deterioration of the disease.

[0097] The damage impact assessment data is used to evaluate the impact of damage on pavement structure and traffic safety. This assessment combines damage cause analysis data and damage trend prediction data, and uses simulation technologies such as finite element analysis (FEA) to evaluate the impact of damage on the bearing capacity and service life of the pavement structure. For example, by simulating the stress of the pavement under different damage conditions, it is possible to evaluate the degree to which cracks weaken the overall strength of the pavement and the impact on the safety of heavy-loaded vehicles.

[0098] In practical applications, suppose that on a certain highway, environmental characteristic data shows that recent rainfall is frequent and the temperature changes are large, and the usage pattern characteristic data shows that the traffic volume on this section is large and there are many heavy-loaded vehicles. Through the disease cause model, it can be analyzed that rainfall and temperature changes cause fatigue of pavement materials, and heavy-loaded vehicles further aggravate the formation of cracks. Using the disease development trend prediction model, it can be predicted to what extent the cracks will expand in the next few months. Finally, through the disease impact assessment, it can be evaluated that if not repaired in time, the cracks may cause a decrease in pavement bearing capacity and potential safety hazards. These analysis results provide a scientific basis for highway management departments to help them formulate timely maintenance and repair plans.

[0099] In summary, the present invention combines environmental characteristic data and usage pattern characteristic data to construct a disease cause model and a disease development trend prediction model, thereby achieving a comprehensive analysis of the causes, development trends and impacts of pavement diseases, and providing important data support and decision-making basis for pavement maintenance and management.

[0100] Preferably, the disease cause model is calculated by the following formula:

[0101] C=f(E,u)=a·E+b·U+c

[0102] Among them, C represents disease cause analysis data, E represents environmental characteristic data, U represents usage pattern characteristic data, f represents the function of converting environmental characteristic data and usage pattern characteristic data into disease cause analysis data, a and b are weights, and c is the deviation parameter.

[0103] Environmental characteristic data (E) include environmental factors that affect the road surface conditions, such as temperature, humidity, rainfall, etc. These data are collected in real time by temperature and humidity sensors arranged around the road surface, and are processed through data preprocessing steps to form standardized environmental characteristic data. Environmental characteristic data reflects the impact of the natural environment on the road surface. For example, frequent temperature changes may cause the road surface material to expand and contract, resulting in cracks.

[0104] The usage pattern characteristic data (U) reflects the usage of the road, including traffic flow data, vehicle weight data and road surface usage frequency data. These data are collected in real time by sensors arranged on the road surface, and are generated after time synchronization and unified format processing. Traffic flow data can be obtained through traffic monitoring cameras and flow meters installed at the entrance of the road, vehicle weight data can be obtained through weighing sensors installed on the road surface, and road surface usage frequency data is obtained through the statistics of the number of vehicle passes. The usage pattern characteristic data can show the load of the road surface under different usage conditions, which has a direct impact on the formation and development of diseases.

[0105] The damage cause model converts environmental characteristic data and usage pattern characteristic data into damage cause analysis data (C) through the above formula. In this formula, a and b are weights, indicating the degree of influence of environmental factors and usage patterns on the damage causes, and c is a deviation term used to adjust the accuracy of the model output. The model is trained through a machine learning algorithm and uses historical data to establish the relationship between the environment and usage patterns and the damage. For example, the model can be trained to determine the degree of influence of factors such as temperature, humidity changes, and traffic flow on road cracks, settlement and other damages through methods such as linear regression, decision trees or multi-layer perceptrons.

[0106] The core of the disease cause model is to use a large amount of historical data to find the relationship between environmental characteristics and usage pattern characteristics and the cause of the disease through machine learning algorithms. Through this model, it is possible to accurately predict the type of road surface diseases and their severity under the current environment and usage pattern. Specifically, assuming that a road has experienced frequent rainfall and a large number of heavy-loaded vehicles in the past few months, the model can analyze the cumulative impact of these factors on the road surface and predict the most likely type of disease, such as crack expansion or road surface settlement.

[0107] For example, on a highway, the environmental characteristic data (E) shows that there has been frequent rainfall and large temperature changes recently, while the usage pattern characteristic data (U) indicates that the traffic volume on this section is large and there are many heavy-loaded vehicles. Through the disease cause model, it can be analyzed that rainfall and temperature changes cause fatigue of pavement materials, and heavy-loaded vehicles further aggravate the formation of cracks. The model uses historical data to determine the values ​​of a, b, and c, so that the formula can accurately calculate the disease cause analysis data (C) under this environment and usage pattern. This analysis can help highway management departments identify potential pavement diseases in advance and take corresponding preventive measures, such as timely repairing cracks or reinforcing the pavement, to extend the service life of the pavement and ensure driving safety.

[0108] In summary, the disease cause model comprehensively analyzes environmental characteristic data and usage pattern characteristic data, and uses machine learning algorithms to establish the relationship between the disease cause and these factors, thereby generating disease cause analysis data. This method not only improves the accuracy of disease identification, but also provides targeted preventive measures, significantly improving the efficiency and effectiveness of pavement maintenance and management.

[0109] Preferably, the disease development trend prediction model is calculated by the following formula:

[0110] T=g(C,t)=α·C+β·t+γ

[0111] Among them, T represents disease trend prediction data, C represents disease cause analysis data, t represents time variable, g represents the function of converting disease cause analysis data and time variable into disease trend prediction data, α and β are weights, and γ is the deviation parameter.

[0112] The disease cause analysis data (C) comes from the above-mentioned disease cause model and is generated by analyzing the environmental characteristic data and the usage pattern characteristic data. The disease cause analysis data reflects the formation of pavement diseases under the current environment and usage conditions, including information such as the disease type, location and severity. The time variable (t) represents the predicted time range, which can be the next few days, weeks or even months.

[0113] The core of the disease development trend prediction model is to use disease cause analysis data and time variables to predict the future development trajectory of the disease. The model is trained through a machine learning algorithm and combines historical data to establish the relationship between disease development and time. α and β in the formula are weights, which represent the degree of influence of disease cause analysis data and time variables on disease development trends, and γ is a deviation term used to adjust the accuracy of the model output.

[0114] The disease development trend prediction model can provide future predictions of pavement diseases and help managers develop preventive maintenance plans. For example, when the model inputs the current disease cause analysis data and the future time range, the model can output disease trend prediction data to show the possible development of the disease at a certain point in the future. This prediction can identify potential risk areas in advance, carry out maintenance and repair in time, and avoid further expansion of the disease.

[0115] For example, suppose that the current disease cause analysis data (C) of a road shows that there are many cracks, and the usage pattern characteristic data shows that the traffic volume on this road section is large and there are many heavy-loaded vehicles. Through the disease development trend prediction model, the development trend of these cracks can be predicted in the next few months. The model may show that under the current conditions, the cracks will expand to a larger area in the next few months and may affect the overall structure of the road surface. The prediction results (T) can help the highway management department to formulate a repair plan in advance, fill the cracks in time, and prevent the disease from further deteriorating.

[0116] In practical applications, disease development trend prediction models can be optimized by combining a variety of machine learning algorithms. For example, time series analysis methods such as long short-term memory networks (LSTM) and autoregressive integrated moving average models (ARIMA) can be used to capture the time dependence and periodic changes in disease development. In addition, regression algorithms such as decision trees and random forests can also be used for modeling and handling complex nonlinear relationships. Through the combination of these algorithms, disease development trend prediction models can provide more accurate and reliable prediction results.

[0117] In summary, the disease development trend prediction model comprehensively considers the disease cause analysis data and time variables, uses machine learning algorithms for training, establishes the relationship between disease development and time, and generates disease trend prediction data. This method can not only predict the future development trend of diseases, but also provide a scientific basis for pavement maintenance and management, helping managers to formulate maintenance plans in advance, prevent further expansion of diseases, and ensure the safety and durability of the road surface.

[0118] like Figure 4 As shown, based on the disease cause analysis data, disease trend prediction data and disease impact assessment data, an intelligent prediction system and decision support system are constructed to provide real-time disease prediction and maintenance strategies, generate disease prediction data and maintenance strategy data, and optimize the intelligent prediction system and decision support system based on actual maintenance feedback;

[0119] The disease cause analysis data is generated by the disease cause model, reflecting the impact of environmental characteristic data and usage pattern characteristic data on the formation of pavement diseases. The disease trend prediction data is generated by the disease development trend prediction model, predicting the development trajectory of the disease in the future. The disease impact assessment data evaluates the impact of the disease on the pavement structure and traffic safety, and is usually obtained through simulation technology such as finite element analysis (FEA).

[0120] The intelligent prediction system comprehensively analyzes this data and predicts the future development of pavement damage in real time. The system uses machine learning and artificial intelligence algorithms to combine historical data and real-time data to dynamically predict pavement damage. For example, using regression analysis, time series analysis or deep learning models, the intelligent prediction system can predict the development trend of pavement damage in the next few days, weeks or even months.

[0121] The decision support system provides specific maintenance strategies based on the output of the intelligent prediction system. The system combines the disease cause analysis data, disease trend prediction data and disease impact assessment data to provide scientific decision support for maintenance personnel. For example, the decision support system can recommend maintenance of a certain section of road within a specific time period to prevent further deterioration of the disease; or recommend the use of specific repair materials and methods to extend the service life of the road surface.

[0122] Through the collaborative work of the intelligent prediction system and the decision support system, specific disease prediction data and maintenance strategy data can be generated. Disease prediction data includes the type, location, severity and future development of the disease. Maintenance strategy data includes maintenance time, maintenance resource allocation, specific operation suggestions, etc. These data can help highway management departments take preventive maintenance measures in a timely manner, reduce the impact of diseases on traffic safety, and reduce maintenance costs.

[0123] In practical applications, the optimization of intelligent prediction systems and decision support systems depends on actual maintenance feedback. By regularly collecting and analyzing maintenance result data, the accuracy and effectiveness of predictions and decisions can be evaluated. For example, if a section of road surface is repaired according to the recommended maintenance strategy, but the disease still expands, it means that the system's predictions and decisions may be insufficient. By adjusting and optimizing model parameters, intelligent prediction systems and decision support systems can continuously improve their accuracy and reliability.

[0124] For example, cracks appeared in many places on a highway. Through the analysis of the disease cause model, it was found that frequent temperature changes and heavy vehicles were the main causes. The intelligent prediction system used the disease trend prediction data to predict that the cracks would further expand in the next few months. The decision support system recommended repairs before the cracks expanded and provided specific repair time and methods. After the actual maintenance, the management department reported that the repair effect was good. The system used this feedback to optimize the model and improve the accuracy of future predictions and decisions.

[0125] In summary, the present invention can provide real-time disease prediction and maintenance strategies by constructing an intelligent prediction system and a decision support system, generate specific disease prediction data and maintenance strategy data, and continuously optimize the system based on actual maintenance feedback. This method not only improves the accuracy of pavement disease prediction, but also enhances the scientific nature of maintenance decisions, effectively prolongs the service life of the pavement, and ensures traffic safety.

[0126] Preferably, the intelligent prediction system uses a reinforcement learning algorithm to continuously optimize disease prediction and maintenance strategies, including:

[0127] Based on historical disease data and current disease data, the intelligent prediction model is trained using reinforcement learning algorithms;

[0128] According to the real-time updated disease data, the parameters of the intelligent prediction model are adjusted and optimized to generate real-time disease prediction data.

[0129] The intelligent prediction system uses reinforcement learning algorithms to train intelligent prediction models based on historical and current disease data. Historical disease data includes past pavement disease conditions and their corresponding environmental and usage pattern feature data, which provide a large number of learning samples for the model. Current disease data is the pavement disease conditions collected in real time, including the latest ground penetrating radar feature data, vibration feature data, environmental feature data, and stress feature data.

[0130] During the training process, the reinforcement learning algorithm represents the interaction between the environment (i.e., pavement damage conditions) and the agent (i.e., prediction model) as a Markov decision process (MDP). At each time step, the agent observes the current state (such as the current damage cause analysis data and environmental feature data), selects an action (such as predicting future damage development or suggesting maintenance strategies), and then updates the strategy based on the feedback from the environment (such as the actual damage development) to maximize the cumulative reward (such as the accuracy of damage prediction or the effectiveness of maintenance strategies).

[0131] In actual applications, the intelligent prediction system will be optimized through the following steps:

[0132] Initialize the model: Based on the initial historical disease data, initialize the parameters of the intelligent prediction model and reinforcement learning algorithm.

[0133] Training model: The intelligent prediction model learns the best strategy by continuously interacting with the environment. For example, the model may try different disease prediction methods or maintenance strategies, observe their effects, and update the strategy based on reward signals (such as prediction accuracy or maintenance effect). The training process may include multiple rounds of iterations until the model converges.

[0134] Real-time optimization: When new disease data is updated in real time, the intelligent prediction system will adjust and optimize the model parameters based on this data. The reinforcement learning algorithm uses new data for online learning, allowing the model to adapt to new environmental changes and improve the accuracy of prediction and decision-making.

[0135] For example, the management department of a highway uses an intelligent prediction system to monitor and maintain the road surface. When the system detects cracks in a section of the road surface, it predicts the expansion trend of the cracks based on the current disease data and historical data, and recommends maintenance strategies. Reinforcement learning algorithms play a key role in this process: the system simulates different maintenance strategies (such as immediate repair or delayed repair), evaluates the effectiveness of each strategy, and then selects the optimal strategy and implements it. If the actual maintenance effect is not good, the system will further adjust the strategy based on the feedback data to continuously optimize the model.

[0136] In this way, the intelligent prediction system can generate real-time disease prediction data and maintenance strategy data. These data include the prediction results of disease type, location, severity and future development, as well as specific maintenance time, resource allocation and operation suggestions. Real-time disease prediction data helps management departments identify potential risk areas in advance, and maintenance strategy data provides a scientific basis for decision-making, ensuring that maintenance work is timely and effective.

[0137] In summary, the present invention constructs an intelligent prediction system by adopting a reinforcement learning algorithm, which enables it to continuously optimize disease prediction and maintenance strategies based on historical and current data. Through reinforcement learning, the intelligent prediction system not only improves the accuracy of disease prediction, but also enhances the scientific nature of maintenance decisions, effectively prolongs the service life of the road surface, and ensures traffic safety.

[0138] Preferably, the maintenance strategy data includes maintenance time, maintenance resources and specific operation suggestions, including:

[0139] Determine the areas that require priority maintenance and the specific maintenance time based on the disease impact assessment data;

[0140] Allocate maintenance resources based on disease severity and resource availability, and generate a maintenance resource allocation plan;

[0141] Provides specific operating recommendations, including maintenance tools to use, operating procedures, and safety precautions.

[0142] Based on the damage impact assessment data, determine the areas that need priority maintenance and the specific maintenance time. The damage impact assessment data is generated by the aforementioned damage cause analysis and damage trend prediction, and includes the type, location, severity of the damage and its potential impact on the pavement structure and traffic safety. By analyzing this data, the system can identify the areas with the most serious damage and the greatest impact on traffic, and formulate a detailed maintenance schedule based on the development trend of the damage and the actual pavement usage. For example, if a section of the road has serious cracks and the cracks expand rapidly, the system will prioritize the emergency repair of this section of the road and formulate a corresponding maintenance time to minimize interference with traffic.

[0143] Next, maintenance resources are allocated based on the severity of the damage and resource availability, and a maintenance resource allocation plan is generated. The severity of the damage is assessed through the damage cause model and the damage development trend prediction model, and resource availability is determined based on the actual situation of the current maintenance equipment, personnel, and material inventory. The system will reasonably allocate maintenance resources based on the severity and urgency of different damages. For example, for minor road damage, conventional maintenance equipment and personnel can be arranged; for serious damage, more resources may need to be deployed, including heavy machinery and professional maintenance personnel. The resource allocation plan should not only consider the urgency of the disease, but also the best use of maintenance resources to avoid waste of resources and duplication of maintenance work.

[0144] Finally, specific operational recommendations are provided, including the maintenance tools to be used, operating procedures, and safety precautions. Specific operational recommendations are generated based on maintenance strategy data and historical maintenance records. The system will recommend the most suitable maintenance tools and methods based on the type and severity of the disease. For example, for crack repair, the system may recommend the use of specific types of caulking materials and construction equipment, and provide detailed operating procedures, including crack cleaning, preparation of caulking materials, and construction process. At the same time, the system will also provide safety precautions to ensure the safety of maintenance work. For example, when repairing asphalt pavement in hot weather, it is necessary to prevent equipment overheating and construction workers from heatstroke; when constructing at night, sufficient lighting and warning signs need to be set up to ensure construction safety.

[0145] For example, the management department of a highway identified serious cracks in a section of pavement through an intelligent prediction system and decision support system, which may expand to a larger area in the next few weeks. Based on the disease impact assessment data, the system determined that the section of pavement needed priority maintenance and recommended that emergency repairs be arranged within the next week. Based on the severity of the disease and resource availability, the system generated a maintenance resource allocation plan and deployed sufficient maintenance equipment and professionals. At the same time, the system provided detailed operational suggestions, recommended the use of a certain efficient caulking material, and provided specific construction steps and safety precautions. In the actual construction process, the maintenance team followed the suggestions provided by the system and successfully repaired the pavement cracks, avoiding further expansion of the disease, and ensuring traffic safety and pavement durability.

[0146] Through the above steps, the generation and optimization of maintenance strategy data can effectively improve the efficiency and safety of road maintenance work. The present invention not only provides scientific disease prediction and maintenance strategies through intelligent prediction and decision support systems, but also ensures the efficient execution and safe operation of actual maintenance work. This method significantly improves the overall level of road maintenance and management, extends the service life of the road, reduces maintenance costs, and ensures the safety and smoothness of the road.

[0147] like Figure 5 As shown, the real-time updated disease prediction data and disease characteristic data are used to evaluate the current disease status and generate a disease status report.

[0148] Preferably, the disease status report includes a disease distribution map, disease type analysis and disease severity assessment, and provides detailed information in the form of charts and text descriptions, including:

[0149] Disease distribution map: shows the specific location and distribution of diseases on the road surface;

[0150] Disease type analysis: Classify and describe different types of diseases and their characteristics;

[0151] Disease severity assessment: Quantitatively assess the severity of the disease and provide specific scores and recommendations.

[0152] The disease prediction data is generated by the intelligent prediction system, which predicts the development trend of future diseases in real time based on historical disease data, current disease data and reinforcement learning algorithms. The disease feature data is extracted by multimodal data fusion and deep learning models, and contains the specific characteristics of pavement diseases, such as crack width, depth, length, surface deformation and stress distribution.

[0153] Using these real-time updated data, the system can comprehensively evaluate the current disease status and generate a detailed disease status report. The disease status report includes the following core parts:

[0154] Defect distribution map: The defect distribution map shows the specific location and distribution of defects on the road surface. By combining ground penetrating radar feature data, vibration feature data, environmental feature data, and stress feature data, the system can draw an accurate defect distribution map. For example, through ground penetrating radar data, cavities and cracks inside the road surface can be identified, and through vibration data, cracks and deformations on the surface can be identified. After these data are fused and processed, the generated defect distribution map can intuitively show the specific location and scope of the disease, helping maintenance personnel quickly locate the area that needs to be repaired.

[0155] Disease type analysis: Disease type analysis classifies and describes different types of diseases and explains their characteristics in detail. For example, the system can identify and classify different types of cracks (such as longitudinal cracks, transverse cracks, and mesh cracks), potholes, settlements, and surface wear. The characteristics of each disease type are described in detail, including its formation causes, development process, and impact on the pavement structure. Through this classification and analysis, maintenance personnel can better understand the nature and causes of diseases and formulate targeted maintenance strategies.

[0156] Disease severity assessment: Disease severity assessment quantitatively assesses the severity of the disease and provides specific scores and recommendations. The system combines disease feature data and prediction data, and uses specific assessment criteria and algorithms to quantitatively assess the severity of each disease. For example, by evaluating the width, depth and length of cracks, the severity score of the cracks can be calculated; by analyzing the magnitude and range of pavement settlement, the impact of settlement on traffic safety can be assessed. The assessment results are presented in the form of a score, accompanied by specific maintenance recommendations, such as immediate repair, regular monitoring or further investigation.

[0157] Through the generated disease status report, maintenance personnel can fully understand the current health status of the road surface and formulate a scientific maintenance plan accordingly. The charts and text descriptions in the report are intuitive, easy to understand and apply. For example, on a certain highway, the disease distribution map generated by the system shows the specific locations of multiple serious cracks; the disease type analysis further explains the types and causes of these cracks; the disease severity assessment provides the severity score of each crack and specific maintenance recommendations, such as some cracks need to be repaired immediately, while other cracks can be monitored regularly. Based on these detailed information, the maintenance team can take quick action to repair serious diseases, prevent potential risks, and ensure the safety and durability of the road surface.

[0158] In summary, the present invention generates a detailed disease status report, including a disease distribution map, disease type analysis, and disease severity assessment, by utilizing real-time updated disease prediction data and disease characteristic data. This report provides maintenance personnel with comprehensive and accurate disease information, helping them to formulate and implement effective maintenance strategies, significantly improving the efficiency and effectiveness of pavement maintenance and management.

[0159] like Figure 6 As shown, a system for implementing the artificial intelligence-based pavement internal disease analysis method comprises:

[0160] A multi-layer sensor network, including ground-penetrating radar sensors, piezoelectric sensors, temperature and humidity sensors, and accelerometers, is used to collect and pre-process the internal structure and external environment data of the road surface in real time;

[0161] The multi-level sensor network includes ground penetrating radar sensors, piezoelectric sensors, temperature and humidity sensors, and accelerometers. The ground penetrating radar sensor detects changes in the internal structure of the road surface by emitting electromagnetic waves and receiving their reflected signals to generate ground penetrating radar feature data. The piezoelectric sensor uses the piezoelectric effect to collect vibration and pressure data and generate vibration feature data. The temperature and humidity sensor collects environmental temperature and humidity data and generates environmental feature data. The accelerometer collects dynamic stress data of the road surface and generates stress feature data. These sensors are arranged at different depths and areas to ensure full coverage of the monitoring area and provide high-resolution data.

[0162] The data processing module is used to synchronize the time and format of the ground penetrating radar feature data, vibration feature data, environmental feature data and stress feature data, fuse multimodal data using data-level, feature-level and decision-level fusion methods, extract features through a deep learning model, and generate fused feature data and disease feature data;

[0163] The data processing module is used to synchronize the time and format the ground penetrating radar feature data, vibration feature data, environmental feature data and stress feature data. Time synchronization is achieved by using timestamps or synchronization signals, and format unification is achieved by data normalization, which converts data of different units and scales into a unified standard format. Then, multimodal data fusion is performed using data-level, feature-level and decision-level fusion methods. Data-level fusion combines the raw data of different sensors at the data level, feature-level fusion extracts the features of various types of data and fuses them, and decision-level fusion analyzes various types of data independently and then fuses the results. Feature extraction is performed through a deep learning model to generate fused feature data and disease feature data. The deep learning model includes a convolutional neural network (CNN) and a graph convolutional network (GCN), which are used for local feature extraction and global information integration, respectively.

[0164] An analysis module is used to combine environmental characteristic data and usage pattern characteristic data to construct a disease cause model and a disease development trend prediction model, analyze the cause, development trend and impact of the disease, and generate disease cause analysis data, disease trend prediction data and disease impact assessment data. The usage pattern characteristic data is generated by time synchronization and format unification of traffic flow data, vehicle weight data and road use frequency data;

[0165] The analysis module combines environmental characteristic data and usage pattern characteristic data to construct a disease cause model and a disease development trend prediction model. Environmental characteristic data include factors that affect road conditions, such as temperature, humidity, and rainfall. Usage pattern characteristic data is generated by time synchronization and format unification of traffic flow data, vehicle weight data, and road use frequency data. The disease cause model analyzes the impact of environmental characteristic data and usage pattern characteristic data on the formation of road diseases and generates disease cause analysis data; the disease development trend prediction model combines disease cause analysis data and time variables to predict the future development trajectory of diseases and generate disease trend prediction data and disease impact assessment data.

[0166] Intelligent prediction and decision support module, which is used to build intelligent prediction system and decision support system based on disease cause analysis data, disease trend prediction data and disease impact assessment data, provide real-time disease prediction and maintenance strategy, generate disease prediction data and maintenance strategy data, and optimize intelligent prediction system and decision support system according to actual maintenance feedback;

[0167] The intelligent prediction and decision support module builds an intelligent prediction system and a decision support system based on disease cause analysis data, disease trend prediction data, and disease impact assessment data. The intelligent prediction system uses reinforcement learning algorithms to continuously learn the best strategy through interaction with the environment to maximize the cumulative reward. The system performs model training and real-time optimization based on historical disease data and current disease data to generate real-time disease prediction data. The decision support system combines disease prediction data and impact assessment data to provide detailed maintenance strategies, including maintenance time, maintenance resource allocation, and specific operation recommendations, and optimizes the intelligent prediction system and decision support system based on actual maintenance feedback.

[0168] The condition assessment and report generation module is used to use real-time updated disease prediction data and disease characteristic data to assess the current disease condition and generate a disease status report.

[0169] The condition assessment and report generation module uses real-time updated disease prediction data and disease characteristic data to assess the current disease condition and generate a disease status report. The disease status report includes a disease distribution map, a disease type analysis, and a disease severity assessment. The disease distribution map shows the specific location and distribution of the disease on the road surface; the disease type analysis classifies and describes different types of diseases and their characteristics; the disease severity assessment quantitatively assesses the severity of the disease and provides specific scores and suggestions.

[0170] Through the collaborative work of the above modules, the present invention realizes comprehensive, real-time and accurate analysis of internal pavement diseases and formulation of maintenance strategies. The multi-level sensor network ensures the comprehensiveness and timeliness of data collection; the data processing module improves the accuracy of disease identification through multimodal data fusion and deep feature extraction; the analysis module combines environmental and usage pattern data to build a disease cause and trend prediction model, achieving accuracy and real-time disease prediction and early warning; the intelligent prediction and decision support module optimizes maintenance strategies and resource allocation through reinforcement learning algorithms; the condition assessment and report generation module provides detailed disease information, providing a scientific basis for maintenance and management. The overall effect significantly improves the efficiency and effectiveness of pavement maintenance and management, ensuring the safety and durability of the road.

[0171] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0172] The present application is described with reference to flowcharts and / or block diagrams of methods, devices (systems) and computer program products according to embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0173] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0174] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0175] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0176] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0177] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0178] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0179] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A method for analyzing internal pavement defects based on artificial intelligence, characterized in that: The following steps are involved: By arranging a sensor network for the inside of the road surface, the internal structure and external environment data of the road surface are collected and preprocessed in real time to generate ground penetrating radar feature data, vibration feature data, environmental feature data and stress feature data; After the ground penetrating radar feature data, vibration feature data, environmental feature data and stress feature data are synchronized in time and formatted uniformly, multimodal data fusion is performed using data-level, feature-level and decision-level fusion methods, and feature extraction is performed using a deep learning model to generate fused feature data and disease feature data. Combine environmental characteristic data and usage pattern characteristic data to construct a disease cause model and a disease development trend prediction model, analyze the cause, development trend and impact of the disease, and generate disease cause analysis data, disease trend prediction data and disease impact assessment data; the usage pattern characteristic data is generated by time synchronization and format unification of traffic flow data, vehicle weight data and road use frequency data; Based on the disease cause analysis data, disease trend prediction data and disease impact assessment data, build an intelligent prediction system and decision support system to provide real-time disease prediction and maintenance strategies, generate disease prediction data and maintenance strategy data, and optimize the intelligent prediction system and decision support system based on actual maintenance feedback; Use real-time updated disease prediction data and disease characteristic data to assess current disease conditions and generate disease status reports.

2. The method for analyzing internal pavement defects based on artificial intelligence according to claim 1 is characterized in that: The sensor network includes: ground penetrating radar sensors, piezoelectric sensors, temperature and humidity sensors and accelerometers, which are arranged at different depths and areas, including: By applying ground penetrating radar sensors to collect electromagnetic wave reflection data from the internal structure of the road surface; Piezoelectric sensors are placed on and below the pavement surface to collect vibration and pressure data; Temperature and humidity sensors are arranged around the road surface to collect ambient temperature and humidity data; Accelerometers are placed in key areas of the road surface to collect dynamic stress data of the road surface.

3. The method for analyzing internal pavement defects based on artificial intelligence according to claim 1 is characterized in that: The multimodal data fusion method comprises: Data-level fusion: combining raw data from different sensors at the data level to generate a fused data set; Feature-level fusion: extract features of various types of data, and fuse the extracted features to generate fused feature data; Decision-level fusion: After independently analyzing various types of data, the results are integrated to generate comprehensive decision-making data.

4. The method for analyzing internal road surface defects based on artificial intelligence according to claim 3 is characterized in that: The deep learning model includes a convolutional neural network and a graph convolutional network for feature extraction and disease identification, including: Use convolutional neural network to extract features from fused feature data and generate preliminary disease feature data; The preliminary disease feature data are further processed using a graph convolutional network to generate the final disease feature data.

5. The method for analyzing internal pavement defects based on artificial intelligence according to claim 1 is characterized in that: The disease cause model is calculated by the following formula: C=f(E,U)=a·E+b·U+c Among them, C represents disease cause analysis data, E represents environmental characteristic data, U represents usage pattern characteristic data, f represents the function of converting environmental characteristic data and usage pattern characteristic data into disease cause analysis data, a and b are weights, and c is the deviation parameter.

6. The method for analyzing internal road surface defects based on artificial intelligence according to claim 5 is characterized in that: The disease development trend prediction model is calculated by the following formula: T=g(C,t)=α·C+β·t+γ Among them, T represents disease trend prediction data, C represents disease cause analysis data, t represents time variable, g represents the function of converting disease cause analysis data and time variable into disease trend prediction data, α and β are weights, and γ is the deviation parameter.

7. The method for analyzing internal pavement defects based on artificial intelligence according to claim 1 is characterized in that: The intelligent prediction system uses reinforcement learning algorithms to continuously optimize disease prediction and maintenance strategies, including: Based on historical disease data and current disease data, the intelligent prediction model is trained using reinforcement learning algorithms; According to the real-time updated disease data, the parameters of the intelligent prediction model are adjusted and optimized to generate real-time disease prediction data.

8. The method for analyzing internal pavement defects based on artificial intelligence according to claim 1 is characterized in that: The maintenance strategy data includes maintenance time, maintenance resources and specific operation suggestions, including: Determine the areas that require priority maintenance and the specific maintenance time based on the disease impact assessment data; Allocate maintenance resources based on disease severity and resource availability, and generate a maintenance resource allocation plan; Provides specific operating recommendations, including maintenance tools to use, operating procedures, and safety precautions.

9. The method for analyzing internal road surface defects based on artificial intelligence according to claim 1, characterized in that: The disease status report includes a disease distribution map, disease type analysis and disease severity assessment, and provides detailed information in the form of charts and text descriptions, including: Disease distribution map: shows the specific location and distribution of diseases on the road surface; Disease type analysis: Classify and describe different types of diseases and their characteristics; Disease severity assessment: Quantitatively assess the severity of the disease and provide specific scores and recommendations.

10. A system for implementing the artificial intelligence-based pavement internal disease analysis method according to any one of claims 1 to 9, characterized in that: include: A multi-layer sensor network, including ground-penetrating radar sensors, piezoelectric sensors, temperature and humidity sensors, and accelerometers, is used to collect and pre-process the internal structure and external environment data of the road surface in real time; The data processing module is used to synchronize the time and format of the ground penetrating radar feature data, vibration feature data, environmental feature data and stress feature data, fuse multimodal data using data-level, feature-level and decision-level fusion methods, extract features through a deep learning model, and generate fused feature data and disease feature data; An analysis module is used to combine environmental characteristic data and usage pattern characteristic data to construct a disease cause model and a disease development trend prediction model, analyze the cause, development trend and impact of the disease, and generate disease cause analysis data, disease trend prediction data and disease impact assessment data. The usage pattern characteristic data is generated by time synchronization and format unification of traffic flow data, vehicle weight data and road use frequency data; Intelligent prediction and decision support module, which is used to build intelligent prediction system and decision support system based on disease cause analysis data, disease trend prediction data and disease impact assessment data, provide real-time disease prediction and maintenance strategy, generate disease prediction data and maintenance strategy data, and optimize intelligent prediction system and decision support system according to actual maintenance feedback; The condition assessment and report generation module is used to use real-time updated disease prediction data and disease characteristic data to assess the current disease condition and generate a disease status report.

Images