Dynamic Assessment Method and System for Road Multi-Element Collapse Risk Integrating Pavement Characterization

Abstract

An embodiment of the present invention discloses a method and system for dynamically evaluating the collapse risk of multiple road elements by integrating road surface characterization. The method includes: obtaining data collected by vehicle-mounted sensors and cameras to obtain real-time data, determining the actual area of the disease, the type of disease, and the road surface elevation to obtain the current road surface characterization data; determining historical data according to GPS data; performing normalization and encoding processing on the current road surface characterization data and the historical data; inputting the processed current road surface characterization data and the processed historical data into a road collapse risk assessment model to perform road collapse risk assessment to obtain an assessment result; sending the assessment result. By implementing the method of the embodiment of the present invention, it is possible to predict and identify the possible collapse risks underground to a great extent, improve the discrimination accuracy, reduce the operation cost, make up for the insufficient guarantee ability during the window period of road collapse risk assessment, and cope with the changing underground environment and construction conditions.

Description

Technical Field

[0001] The present invention relates to a method for evaluating road collapse, and more particularly to a dynamic evaluation method and system for multi-element collapse risk of roads integrating pavement characterization. Background Art

[0002] With the acceleration of urbanization, especially the development and utilization of urban underground space, it has reached an unprecedented breadth and depth. With the continuous expansion of urban transportation infrastructure, including the construction of roads, viaducts, expressways, and the large-scale development of the subway system, the utilization of underground space is also increasing continuously. Although these construction projects have greatly facilitated the lives of citizens and promoted the development of the urban economy.

[0003] The development of underground space faces various challenges. First, the geological conditions under municipal roads are complex and changeable, which makes the underground construction environment full of uncertainties; the layout of underground pipelines is intricate, increasing the difficulty and risk of construction. Existing methods for evaluating road collapse risk mainly rely on factors such as geological conditions, underground pipe network distribution, and construction activities for judgment. However, this kind of evaluation is usually carried out only once a year and fails to effectively consider the influence of dynamic factors such as rainfall during the rainy season and pavement diseases. In particular, when the rainfall increases during the rainy season, the geological environment may change, thus affecting the stability of the underground space.

[0004] Therefore, it is necessary to design a new method to achieve the prediction and identification of possible collapse risks underground to a great extent, improve the discrimination accuracy, reduce the operation cost, and make up for the lack of guarantee ability during the blank period of road collapse risk evaluation to cope with the changing underground environment and construction conditions. Summary of the Invention

[0005] The purpose of the present invention is to overcome the defects of the prior art and provide a dynamic evaluation method and system for multi-element collapse risk of roads integrating pavement characterization.

[0006] To achieve the above purpose, the present invention adopts the following technical solutions: A dynamic evaluation method for multi-element collapse risk of roads integrating pavement characterization, including:

[0007] Obtain the data collected by vehicle-mounted sensors and cameras to obtain real-time data, wherein the real-time data includes pavement videos, vibration sensor data, and GPS data;

[0008] Obtain the rainfall in the road area;

[0009] Determine the actual area, disease type, and pavement elevation of the disease according to the real-time data to obtain the current pavement characterization data;

[0010] Determine historical data based on the GPS data, where the historical data includes historical road surface characterization data corresponding to road areas, soil types, rainfall, construction records, and pipeline network information, and the historical road surface characterization data includes historical disease types, historical disease actual areas, and historical road surface elevations;

[0011] Perform normalization and encoding processing on the current road surface characterization data and the historical data to obtain the processed current road surface characterization data and the processed historical data;

[0012] Input the processed current road surface characterization data and the processed historical data into a road collapse risk assessment model to conduct road collapse risk assessment to obtain an assessment result;

[0013] Send the assessment result.

[0014] A further technical solution thereof is that the performing normalization and encoding processing on the current road surface characterization data and the historical data to obtain the processed current road surface characterization data and the processed historical data includes:

[0015] Perform normalization on the numerical data in the current road surface characterization data and the historical data, and perform one-hot encoding on the categorical data in the current road surface characterization data and the historical data to obtain the processed current road surface characterization data and the processed historical data.

[0016] A further technical solution thereof is that the inputting the processed current road surface characterization data and the processed historical data into a road collapse risk assessment model to conduct road collapse risk assessment to obtain an assessment result includes:

[0017] Input the processed current road surface characterization data and the processed historical data into a road collapse risk assessment model, and the road collapse risk assessment model processes the current road surface characterization data and the historical data through feature fusion, self-attention mechanism, and vector projection to obtain an assessment result.

[0018] A further technical solution thereof is that the inputting the processed current road surface characterization data and the processed historical data into a road collapse risk assessment model, and the road collapse risk assessment model processes the current road surface characterization data and the historical data through feature fusion, self-attention mechanism, and vector projection to obtain an assessment result includes:

[0019] Input the current road surface characterization data and the historical data into a road collapse risk assessment model, and the feature projection layer respectively extracts the severity features of the current road surface characterization data and the severity features of the historical data to obtain a current feature vector and a historical feature vector;

[0020] Calculate the differences of various factors in the historical data through the self-attention mechanism and determine the weight values of various factors;

[0021] Perform vector projection on the current feature vector and the historical feature vector to obtain a final feature vector representing the severity change of the current feature vector;

[0022] Pass the final feature vector through a fully connected layer and a Softmax layer to obtain a continuous risk score;

[0023] Divide the continuous risk score into discrete risk level regions to obtain an evaluation result.

[0024] Its further technical solution is: the current feature vector is a multi-dimensional vector, including the disease area corresponding to the current road surface characterization data, the one-hot encoding of the disease type, and the road surface elevation information; the historical feature vector is a multi-dimensional vector, including the historical disease area corresponding to the historical data, the one-hot encoding of the historical disease type, the historical road surface elevation information, the historical soil, rainfall, construction, and pipe network.

[0025] Its further technical solution is: the calculating the differences of various factors in the historical data through the self-attention mechanism and determining the weight values of various factors includes:

[0026] Calculate the differences of various factors in the historical data through the self-attention mechanism to obtain a historical factor change vector;

[0027] Perform an inner product calculation on the historical factor change vector to obtain an inner product result;

[0028] Calculate the weight values of the factors by passing the inner product result through a fully connected layer and a Softmax layer.

[0029] Its further technical solution is: the performing an inner product calculation on the historical factor change vector to obtain an inner product result includes:

[0030] Perform an inner product of the historical factor change vector and the transpose of the historical factor change vector to obtain an inner product result.

[0031] Its further technical solution is: the performing vector projection on the current feature vector and the historical feature vector to obtain a final feature vector representing the severity change of the current feature vector includes:

[0032] Project the current feature vector onto the historical feature vector to obtain an intermediate feature vector;

[0033] Determine the difference between the current feature vector and the intermediate feature vector, and project the current feature vector onto the difference to obtain a final feature vector.

[0034] Its further technical solution is that the intermediate feature vector includes the change amount of the disease area, the change situation of the disease type, the change situation of the road surface elevation information, and the change information of the historical data at a certain moment and the historical data at the previous moment.

[0035] The present invention also provides a dynamic evaluation system for the risk of multi-element collapse of roads integrating road surface characterization, including:

[0036] A data acquisition unit for acquiring the data collected by vehicle-mounted sensors and cameras to obtain real-time data, wherein the real-time data includes road surface videos, vibration sensor data, and GPS data;

[0037] A rainfall acquisition unit for acquiring the rainfall in the road area;

[0038] A current road surface characterization data determination unit for determining the actual area of the disease, the disease type, and the road surface elevation according to the real-time data to obtain the current road surface characterization data;

[0039] A historical data determination unit for determining historical data according to the GPS data, wherein the historical data includes historical road surface characterization data, soil type, rainfall, construction records, and pipeline network information corresponding to the road area, and the historical road surface characterization data includes historical disease types, historical actual disease areas, and historical road surface elevations;

[0040] A processing unit for encoding the current road surface characterization data and the historical data to obtain the processed current road surface characterization data and the processed historical data;

[0041] An evaluation unit for inputting the processed current road surface characterization data and the processed historical data into a road collapse risk evaluation model for road collapse risk evaluation to obtain an evaluation result;

[0042] A sending unit for sending the evaluation result.

[0043] The beneficial effects of the present invention compared with the prior art are as follows: The present invention collects road surface videos, vibration data, and GPS data in real time through vehicle-mounted sensors and cameras, and obtains the rainfall in the road area. According to these data, the disease area, unevenness, and defect type are determined. At the same time, GPS data is used to obtain relevant historical data. The processed current road surface characterization data and historical data are input into the road collapse risk evaluation model to generate and send risk evaluation results, realizing the prediction and identification of possible collapse risks underground to a great extent, improving the discrimination accuracy, reducing the operation cost, making up for the insufficient guarantee ability during the window period of road collapse risk evaluation, and coping with the changing underground environment and construction conditions.

[0044] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments. Description of the Drawings

[0045] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other accompanying drawings can be obtained based on these drawings without creative efforts.

[0046] Figure 1 It is a schematic flow chart of a method for dynamically evaluating the collapse risk of multiple road elements by fusing pavement characterization provided by an embodiment of the present invention;

[0047] Figure 2 It is a schematic sub - flow chart of a method for dynamically evaluating the collapse risk of multiple road elements by fusing pavement characterization provided by an embodiment of the present invention;

[0048] Figure 3 It is a schematic sub - flow chart of a method for dynamically evaluating the collapse risk of multiple road elements by fusing pavement characterization provided by an embodiment of the present invention;

[0049] Figure 4 It is a schematic sub - flow chart of a method for dynamically evaluating the collapse risk of multiple road elements by fusing pavement characterization provided by an embodiment of the present invention;

[0050] Figure 5 It is a schematic diagram showing the change of the disease area of the same disease within 12 weeks provided by an embodiment of the present invention;

[0051] Figure 6 It is a schematic diagram showing the change of the pavement elevation information of the same disease within 12 weeks provided by an embodiment of the present invention;

[0052] Figure 7 It is a schematic diagram showing the change of the disease type of the same disease within 12 weeks provided by an embodiment of the present invention;

[0053] Figure 8 It is a schematic diagram showing the change of the rainfall within 12 weeks provided by an embodiment of the present invention;

[0054] Figure 9 It is a schematic block diagram of a system for dynamically evaluating the collapse risk of multiple road elements by fusing pavement characterization provided by an embodiment of the present invention;

[0055] Figure 10 It is a schematic block diagram of a computer device provided by an embodiment of the present invention. Detailed Embodiments

[0056] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part rather than all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0057] It should be understood that when used in this specification and the appended claims, the terms "comprises" and "comprising" indicate the presence of the described features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0058] It should also be understood that the terms used in this specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in this specification of the present invention and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an", and "the" are intended to include the plural forms.

[0059] It should be further understood that the term " / and" used in this specification of the present invention and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0060] Please refer to Figure 1 , Figure 1 which is a schematic flowchart of a dynamic assessment method for the collapse risk of multiple road elements with integrated pavement characterization provided by an embodiment of the present invention. The dynamic assessment method for the collapse risk of multiple road elements with integrated pavement characterization is applied to a server. The server interacts with in-vehicle sensors, cameras, and rainfall detection devices, and monitors road diseases in real time through in-vehicle sensors, cameras, and rainfall data, and analyzes the actual area and defect types of the diseases. Combining GPS data to obtain historical information, such as soil type and construction records, and encoding and processing the data. Inputting the processed data into a risk assessment model, generating and sending a road collapse risk assessment result, which can achieve early screening and be supplemented by road disease review, can largely predict and identify possible collapse risks underground, can improve the discrimination accuracy, and reduce the operation cost.

[0061] The existing one-time road collapse risk assessment is usually carried out at a specific time point, which means that there may be a "window period", that is, an unmonitored time period, between the assessment and subsequent inspections. This assessment method cannot continuously track the changes in the road state, resulting in new diseases or risks that may occur during the window period not being discovered and processed in time, thus possibly increasing the risk of road collapse.

[0062] Relatively speaking, the method of this embodiment adopts an evaluation method that combines real-time monitoring with historical data, which can continuously track the road conditions and be updated in real time during the data collection process. This method can identify problems in advance before new diseases or risks occur, thus making up for the problem of insufficient guarantee ability during the empty window period in traditional evaluations, providing more timely and accurate risk warnings, and reducing potential safety hazards.

[0063] Figure 1 It is a schematic flow chart of a dynamic evaluation method for multi-element collapse risks of roads that integrates pavement characterization provided by an embodiment of the present invention. As Figure 1 shown, this method includes the following steps S110 to S170.

[0064] S110. Obtain the data collected by in-vehicle sensors and cameras to obtain real-time data, where the real-time data includes road surface videos, vibration sensor data, and GPS data.

[0065] In this embodiment, the road surface video refers to the real-time video images collected by in-vehicle cameras. It is used to capture and record the actual conditions of the road, including visual information such as cracks, potholes, and obstacles on the road surface. These video data can help evaluate the quality of the road and promptly detect areas that need repair.

[0066] The vibration sensor data is the data provided by in-vehicle vibration sensors, which records the vibrations and impacts received by the vehicle during driving. These sensors are usually installed in the vehicle's suspension system and are used to measure road unevenness, bumpiness, and the vehicle's dynamic response. The vibration sensor data can be used to evaluate the flatness of the road and potential structural problems.

[0067] The GPS data is the real-time position and motion data obtained through the Global Positioning System (GPS). The GPS data includes the precise position, speed, driving direction, and driving path of the vehicle. These data are used to determine the specific position of the vehicle on the road and can be associated with the road surface video and vibration sensor data to provide comprehensive road condition information.

[0068] S120. Obtain the rainfall in the road area.

[0069] In this embodiment, continuous rainfall is an important factor in the change of road diseases and has a certain impact on road collapse. The impact of the increase in rainfall on diseases varies in different time periods. As Figure 8 shown, an increase in rainfall in the 3 - 4th week and the 8 - 11th week may lead to significant changes in diseases. Generally, there is a positive correlation between rainfall and disease changes. Therefore, it is necessary to obtain the rainfall in the road area to be evaluated.

[0070] S130. Determine the actual area, disease type, and road surface elevation of the disease based on the real-time data to obtain the current road surface characterization data.

[0071] In this embodiment, the specific implementation steps for determining the actual area of the disease based on video data belong to the prior art. Specifically, reference can be made to the road element area calculation method, system, computer device, and storage medium disclosed in Chinese Patent CN117593355A, which will not be elaborated here.

[0072] The disease type refers to the defect type corresponding to the bumpy section.

[0073] The road elevation can be manifested as the unevenness of the road, that is, the undulation or unevenness of the road surface.

[0074] The specific implementation steps for determining the unevenness of the diseased road and the specific defect type of the bumpy section based on the vibration sensor data belong to the prior art. Specifically, reference can be made to the non-flat road quality detection method, system, computer device, and storage medium disclosed in Chinese Patent CN118310469A, which will not be elaborated here.

[0075] S140. Determine the historical data according to the GPS data, where the historical data includes the soil type, rainfall, construction records, and pipeline network information corresponding to the road area.

[0076] In this embodiment, the specific location of the current road surface characterization data is located according to the GPS data, and the historical data of the corresponding disease type, actual disease area, and road surface elevation is read. The historical data also includes data such as soil, rainfall, construction, and pipeline network at the corresponding location. The current road surface characterization data and the historical data are used as the severity evolution data.

[0077] The historical data can be retrieved from a preset historical database, which is a database established after collecting urban historical collapse hazard data and containing information such as soil quality, pipeline network, and road surrounding construction. By analyzing these historical data, it is found that there are usually some obvious characterizations on the road surface before a collapse occurs, such as cracks, net cracks, block cracks, subsidence, heaves, and potholes. These characterizations reveal the relationship between underground collapse hazards and road surface changes. This relationship indicates that by frequently monitoring and observing road surface changes, potential underground collapse risks can be predicted and identified to a certain extent.

[0078] Since different historical data (such as soil conditions, pipeline network status, construction history, and rainfall) have different effects on disease changes. For example:

[0079] Soil conditions: Rainfall causes soil erosion and a decline in soil bearing capacity, increasing the soil's sensitivity to disease changes, thereby reducing the bearing capacity of the roadbed. As the road usage time increases, this impact may become more significant.

[0080] Underground construction: Underground construction will disrupt the stress balance of the soil, leading to problems such as loose soil and cavities, and these factors may trigger changes in road surface diseases.

[0081] Generally speaking, factors such as soil, pipe networks, rainfall, underground construction, and road service life interact with each other and jointly affect the development of road diseases. Therefore, it is necessary to obtain data corresponding to each of the above factors for the road to be evaluated in order to more comprehensively and accurately evaluate the risk of road collapse.

[0082] S150. Encode the current road surface characterization data and the historical data to obtain the processed current road surface characterization data and the processed historical data.

[0083] In this embodiment, normalize the numerical data in the current road surface characterization data and the historical data, and perform one-hot encoding on the categorical data in the current road surface characterization data and the historical data to obtain the processed current road surface characterization data and the processed historical data. Among them, the numerical data includes the current disease area, the current road surface elevation, the historical disease area, the historical road surface elevation, soil, rainfall, construction, pipe networks, etc.; the categorical data includes the current disease type, the historical disease type, the soil type, and the construction type.

[0084] This step can be carried out by establishing a disease severity evolution model for normalization and one-hot encoding processing. Normalization makes all features on the same scale, avoiding the situation that some features have too much influence on the model due to their large value ranges. One-hot encoding is to convert categorical data into the form of binary vectors. Each category is represented by a binary bit, with only one position being 1 and the rest being 0. This encoding method can convert categorical data into a numerical format that the model can handle; one-hot encoding avoids imposing the sequential relationship of categorical data on the model, thus not introducing unnecessary sequential information; the categorical data is converted into a numerical format that the model can handle, enabling the model to use this information for training and prediction.

[0085] In this embodiment, the reason for considering the disease type, the actual area of the disease, the road surface elevation, the soil type, the rainfall, the construction record, and the pipe network information is that various factors (disease type, actual area, elevation change, soil type, rainfall, construction record, and pipe network information) interact with each other and jointly determine the stability and collapse risk of the road.

[0086] Specifically, as Figures 5 to 8As shown in Figure 1, during a 12-week continuous monitoring of a region, the following phenomena can be observed:

[0087] Changes in disease types and areas:

[0088] Disease changes: During the 9th to 11th week, monitoring data showed that the disease area increased significantly, and the disease type developed from "net cracks" to more serious "cracking". This change indicates that there may be serious cavity changes in the underground structure, and the potential risk of collapse has increased significantly.

[0089] Elevation changes: At the same time, the rate of decline of elevation data has also significantly accelerated, which further supports the possibility of underground cavity expansion and suggests that there is a high risk of collapse.

[0090] Effect of rainfall on disease changes:

[0091] Rainfall and Disease: There was a significant positive correlation between increased rainfall and disease changes. For example, during weeks 3 to 4 and 8 to 11, increased rainfall led to significant changes in disease conditions. This suggests that rainfall has a direct effect on disease changes.

[0092] The combined impact of historical data:

[0093] Soil: Rainfall causes soil erosion and a reduction in soil bearing capacity, which increases the soil's influence on disease changes and ultimately leads to a reduction in the bearing capacity of the roadbed. This effect becomes more significant as the road ages.

[0094] Pipeline network: Aging and damage of the pipeline network may lead to uneven distribution of groundwater, which in turn affects the stability of the road.

[0095] Underground construction: Underground construction may destroy the force balance of the soil, causing the soil to become loose and form cavities, which in turn may lead to disease changes.

[0096] Therefore, a comprehensive analysis of multiple factors such as damage type, rainfall, soil, pipeline network, construction records, etc. is needed to more comprehensively understand and predict changes in road damage and take effective measures to prevent potential collapse risks.

[0097] Due to the intertwined influence of multiple factors such as soil conditions, age of pipe network, rainfall, underground construction depth, and road aging, the model uses an attention mechanism to calculate the weight of these historical data on road diseases. The higher the weight value, the more significant the impact of the relevant data on the change of disease.

[0098] By comparing the differences in data before and after, it can be observed that the predicted subsidence risk assessment scores gradually increase, which means that the subsidence risk in this area has risen from a low level to a higher level. Using this high-frequency disease detection can effectively make up for the lack of guarantee of traditional detection methods during the empty window period, thereby improving the timeliness and accuracy of risk early warning.

[0099] S160. Input the processed current road surface characterization data and the processed historical data into the road subsidence risk assessment model for road subsidence risk assessment to obtain an assessment result.

[0100] In this embodiment, the assessment result refers to a number of discrete risk level areas.

[0101] The road subsidence risk assessment model includes a Transformer network, specifically including a feature projection layer; among them, the feature projection layer includes two feature extractors and , the feature extractor includes an input layer, a convolutional layer, and a fully connected layer for extracting the severity features of the current road surface characterization data; the feature extractor includes an input layer, a self-attention mechanism layer, and a fully connected layer for extracting the severity features of historical data at multiple time steps; the self-attention mechanism layer calculates the differences between various factors of the historical data to obtain a historical factor change vector. Calculate the inner product of these vectors to obtain the correlation between factors, calculate the weight values of the factors through the fully connected layer and Softmax, and combine the final feature vector obtained after projection of the current feature vector and the historical feature vector to determine a continuous risk score, which represents the road subsidence risk score following each input content (historical data and current road surface characterization data). Finally, divide the continuous risk scores on a road into several discrete risk level areas to form the final assessment result.

[0102] In one embodiment, please refer to Figure 3 , the above step S160 may include steps S161 to S165.

[0103] S161. Input the current road surface characterization data and the historical data into the road subsidence risk assessment model, and the feature projection layer respectively extracts the severity features of the current road surface characterization data and the severity features of the historical data to obtain a current feature vector and a historical feature vector.

[0104] In this embodiment, the current feature vector is a multi-dimensional vector, including the disease area corresponding to the current road surface characterization data, the one-hot encoding of the disease type, and the road surface elevation information; the historical feature vector is a multi-dimensional vector, including the historical disease area corresponding to the historical data, the one-hot encoding of the historical disease type, the historical road surface elevation information, the historical soil, rainfall, construction, and pipe network.

[0105] Specifically, through the feature extractor extracts the vector for the current moment , as the Q of the Transformer network, is a multi-dimensional vector. The first dimension of the vector is the disease area; the 2-8 dimensions are the one-hot encoding of the disease type; the 9-10 dimensions are the road surface elevation information. All historical data feature vectors pass through the feature extractor extracts the historical vector at the current moment , as the k of the Transformer network, The first dimension is the historical disease area; the 2-8 dimensions are the one-hot encoding of the historical disease type; the 9-10 dimensions are the historical road surface elevation information; the 13-20 dimensions are data such as historical soil, rainfall, construction, and pipe network.

[0106] The two feature extractors process the current road surface characterization data and the historical data separately. This separate processing method can more accurately extract the feature information at the current moment and the historical moment, helping the model to accurately capture the potential patterns and changes in the data, thereby improving the reliability of the evaluation results.

[0107] S162. Calculate the differences between various factors in the historical data through the self-attention mechanism, and determine the weight values of various factors.

[0108] In this embodiment, the weight value of each factor represents the degree of influence of the factor on the disease change.

[0109] The self-attention mechanism can calculate the differences and correlations between various factors in different time steps, thereby dynamically adjusting the weights of the factors. In this way, the model can better handle the non-linear relationships and time variations in the data, making the risk assessment more accurate.

[0110] In one embodiment, please refer to Figure 3 , the above step S162 may include steps S1621 to S1623.

[0111] S1621. Calculate the differences between various factors in the historical data through the self-attention mechanism to obtain the historical factor change vector.

[0112] In this embodiment, through the self-attention mechanism layer, calculate the historical data differences of various factors at multiple times to obtain a vector X composed of multiple historical factor change vectors.

[0113] Specifically, the self-attention mechanism layer is used to analyze and extract the change patterns of various factors at different times in the historical data, and then determine the difference in changes.

[0114] S1622. Perform an inner product calculation on the historical factor change vector to obtain an inner product result.

[0115] In this embodiment, the inner product result refers to the projection of one vector onto another vector. The larger the projection value, the higher the correlation between the vectors, which also represents the correlation of the mutual influence between various factors. Here, the vector refers to one of the historical factor change vectors.

[0116] Specifically, perform an inner product on the historical factor change vector X and the transpose X of the historical factor change vector T to obtain an inner product result. The inner product result is a matrix, and each element in the matrix reflects the degree of influence of a factor on another factor over time.

[0117] S1623. Calculate the weight values of the factors through the fully connected layer and the Softmax layer using the inner product result.

[0118] In this embodiment, the obtained inner product result is input into the fully connected layer, and different weight values are assigned to each factor through the Softmax layer. The sum of the weight values is 1. The larger the proportion of the weight corresponding to a factor, the more serious the influence of the factor on the disease change. The weight values are updated with each input of historical data.

[0119] Specifically, the inner product result is input into a fully connected layer. The role of the fully connected layer is to convert the inner product result into the weights of each factor.

[0120] Apply the Softmax function to the output of the fully connected layer to convert the result into weight values. The sum of these weight values is 1, indicating the relative influence degree of each factor on the disease change; the factors with larger weight values have a more significant influence on the disease change, while the factors with smaller weight values have a smaller influence.

[0121] Each time new historical data is input, recalculate the change vector of the factors, that is, the inner product result, and repeat the above steps. The weight values will be updated according to the new data, so that the weight values can reflect the latest factor correlation.

[0122] The self-attention mechanism is used to analyze and quantify the influence of each factor on the disease change at different time points. By calculating the difference in historical data and generating vectors, the inner product is used to evaluate the correlation between factors, and then the weights of each factor are obtained through the fully connected layer and the Softmax function, finally forming a comprehensive impact assessment of the disease change. As new data is input, the weights will be continuously updated to provide more accurate assessment results.

[0123] Through inner product calculation and the Softmax layer, the model can dynamically adjust the weights of various factors, enabling the evaluation results to reflect the latest data changes. This dynamic adjustment mechanism can make the risk assessment more timely and accurate, adapting to different environmental and conditional changes.

[0124] S163. Perform vector projection on the current feature vector and the historical feature vector to obtain a final feature vector representing the severity change of the current feature vector.

[0125] In one embodiment, refer to Figure 4 , the above step S163 may include steps S1631 to S1632.

[0126] S1631. Project the current feature vector onto the historical feature vector to obtain an intermediate feature vector.

[0127] In this embodiment, the intermediate feature vector includes the change amount of the disease area, the change situation of the disease type, the change situation of the road surface elevation information, and the change information of the historical data at a certain moment and the historical data at the previous moment.

[0128] S1632. Determine the difference between the current feature vector and the intermediate feature vector, and project the current feature vector onto the difference to obtain a final feature vector.

[0129] Specifically, project the current feature vector onto the historical feature vector to obtain an intermediate feature vector . Then, project the current feature vector onto the difference vector between the current feature vector and the intermediate feature vector to obtain the final feature vector .

[0130] As the v vector of the Transformer network, the specific meanings of each dimension of the final feature vector are as follows:

[0131] The 1st dimension: The change amount of the disease area;

[0132] The 2nd dimension: The change of the disease type (0 indicates no change, 1 indicates a change);

[0133] The 3rd - 4th dimensions: The change of the road surface elevation information;

[0134] The 5th - 12th dimensions: The change information of the historical data at time t and time t - 1.

[0135] The purpose of this processing method is to first limit the norm of the historical feature vector by projecting the current vector onto the historical vector so that the intermediate feature vector contains all necessary feature information. Then, by subtracting the current feature vector from the intermediate feature vector and projecting it, the finally obtained feature vector only contains the severity change information of the current road surface characterization data.

[0136] S164. Pass the finally obtained feature vector through a fully connected layer and a Softmax layer to obtain continuous risk scores.

[0137] In this embodiment, after passing through the fully connected layer, the output of the finally obtained feature vector is a vector with a length of 100 cells, and this vector represents the risk score range of each cell from 1 to 100. Then, through the Softmax layer, these risk scores are converted into a probability distribution, and the risk score corresponding to the cell with the highest probability is the output result at time point t. All these output results form a continuous risk score sequence, which is used to represent the road collapse risk score corresponding to each input.

[0138] S165. Divide the continuous risk scores into discrete risk level regions to obtain the evaluation result.

[0139] In this embodiment, the continuous risk scores on a road are divided into several discrete risk level regions.

[0140] During the processing, the finally obtained feature vector first limits the influence of historical features through projection, and then the feature vector obtained through the difference calculation focuses on the change information of the current road surface characterization data. This processing method ensures that the finally obtained feature vector accurately reflects the current severity change, which helps to improve the accuracy of risk assessment.

[0141] After the model processes the finally obtained risk scores through the fully connected layer and the Softmax layer, it can generate a detailed risk score sequence and divide it into discrete risk level regions. Such a refinement process can more specifically describe the risk level of the road, making risk management and decision-making more clear and targeted.

[0142] By evaluating the continuous risk scores, the model can provide real-time risk warnings. Identifying potential road collapse risks in advance helps to take preventive measures and reduce potential losses and safety hazards.

[0143] The model can process data of different types and sources, such as disease types, rainfall, construction, etc., enabling it to work effectively in a variety of practical application scenarios. This strong adaptability makes the model highly versatile.

[0144] S170, send the evaluation result.

[0145] In this embodiment, the evaluation result is sent to relevant departments and management personnel to take corresponding preventive measures to prevent the occurrence of accidents or mitigate the consequences of accidents.

[0146] The method of this embodiment evaluates the road collapse risk by combining current road surface characterization data and historical data. The current road surface characterization data includes features such as disease area, disease type, road surface elevation, etc., while the historical data covers disease development trends, historical disease types, soil conditions, rainfall, etc. This comprehensive consideration not only improves the accuracy of the evaluation but also enables a more comprehensive understanding of the impact of various factors on the risk.

[0147] Specifically, the collected road surface video, vibration sensor data, rainfall, and GPS data are combined with historical data for processing and then input into the road collapse risk assessment model. First, video data and vibration data are collected from the equipment on the vehicle to obtain road surface characterization disease data, including cracks, net cracks, block cracks, settlement, bumps, potholes, and road surface elevation, etc. After being processed, these data are used as the basic input of the model to judge the historical degree of various diseases on the road surface.

[0148] Secondly, rainfall data for the past week in the relevant area is obtained and used as an auxiliary input variable to consider the impact of rainfall on the road condition. Then, the GPS information of the current road surface characterization data is extracted, and the historical data at the corresponding location is found using this information.

[0149] After integrating this information into the model, the features of the current road surface characterization data and historical data are extracted through the feature projection layer in the model. Utilizing the time series processing ability of the Transformer network, the change amounts caused by time changes are calculated, including changes in disease area, disease type, road surface elevation information, and changes in historical data between the previous moment and the current moment. Finally, the model performs prediction calculations to generate a collapse risk assessment score and determine the collapse risk level.

[0150] The above-mentioned dynamic assessment method for multi-element collapse risks of roads with integrated pavement characterization collects road surface videos, vibration data, and GPS data in real time through in-vehicle sensors and cameras, and obtains the rainfall in the road area. Based on these data, the disease area, unevenness, and defect types are determined. At the same time, the GPS data is used to obtain relevant historical data. The processed current pavement characterization data and historical data are input into the road collapse risk assessment model to generate and send the risk assessment results, achieving the prediction and identification of possible underground collapse risks to a great extent, improving the discrimination accuracy, reducing the operation cost, and making up for the insufficient guarantee ability during the window period of road collapse risk assessment to cope with the changing underground environment and construction conditions.

[0151] Figure 9 It is a schematic block diagram of a dynamic assessment system 300 for multi-element collapse risks of roads with integrated pavement characterization provided by an embodiment of the present invention. As Figure 9 shown, corresponding to the above-mentioned dynamic assessment method for multi-element collapse risks of roads with integrated pavement characterization, the present invention also provides a dynamic assessment system 300 for multi-element collapse risks of roads with integrated pavement characterization. The dynamic assessment system 300 for multi-element collapse risks of roads with integrated pavement characterization includes units for executing the above-mentioned dynamic assessment method for multi-element collapse risks of roads with integrated pavement characterization, and this system can be configured in a server. Specifically, please refer to Figure 9 , the dynamic assessment system 300 for multi-element collapse risks of roads with integrated pavement characterization includes a data acquisition unit 301, a rainfall acquisition unit 302, a current pavement characterization data determination unit 303, a historical data determination unit 304, a processing unit 305, an assessment unit 306, and a sending unit 307.

[0152] A data acquisition unit 301 is configured to acquire data collected by vehicle-mounted sensors and cameras to obtain real-time data, where the real-time data includes road surface videos, vibration sensor data, and GPS data; a rainfall acquisition unit 302 is configured to acquire the rainfall in the road area; a current road surface characterization data determination unit 303 is configured to determine the actual area of the disease, the disease type, and the road surface elevation according to the real-time data to obtain current road surface characterization data; a historical data determination unit 304 is configured to determine historical data according to the GPS data, where the historical data includes historical road surface characterization data, soil type, rainfall, construction records, and pipe network information corresponding to the road area, and the historical road surface characterization data includes historical disease types, historical actual disease areas, and historical road surface elevations; a processing unit 305 is configured to encode the current road surface characterization data and the historical data to obtain processed current road surface characterization data and processed historical data; an evaluation unit 306 is configured to input the processed current road surface characterization data and the processed historical data into a road collapse risk assessment model to perform road collapse risk assessment to obtain an evaluation result; a sending unit 307 is configured to send the evaluation result.

[0153] In one embodiment, the processing unit 305 is configured to normalize the numerical data in the current road surface characterization data and the historical data, and perform one-hot encoding on the categorical data in the current road surface characterization data and the historical data to obtain processed current road surface characterization data and processed historical data.

[0154] In one embodiment, the evaluation unit 306 is configured to input the processed current road surface characterization data and the processed historical data into a road collapse risk assessment model, and the road collapse risk assessment model processes the current road surface characterization data and the historical data through feature fusion, self-attention mechanism, and vector projection to obtain an evaluation result.

[0155] In one embodiment, the evaluation unit 306 includes a feature extraction subunit, a weight calculation subunit, a vector projection subunit, a risk score determination subunit, and a division subunit.

[0156] A feature extraction sub-unit is configured to input the current road surface characterization data and the historical data into a road collapse risk assessment model, and the severity features of the current road surface characterization data and the severity features of the historical data are respectively extracted by a feature projection layer to obtain a current feature vector and a historical feature vector; a weight calculation sub-unit is configured to calculate the differences among various factors in the historical data through a self-attention mechanism and determine the weight values of the various factors; a vector projection sub-unit is configured to perform vector projection on the current feature vector and the historical feature vector to obtain a final feature vector representing the severity change of the current feature vector; a risk score determination sub-unit is configured to pass the final feature vector through a fully connected layer and a Softmax layer to obtain a continuous risk score; a division sub-unit is configured to divide the continuous risk score into discrete risk level regions to obtain an evaluation result.

[0157] In one embodiment, the weight calculation sub-unit includes a difference calculation module, an inner product module, and a weight value calculation module.

[0158] The difference calculation module is configured to calculate the differences among various factors in the historical data through a self-attention mechanism to obtain a historical factor change vector; the inner product module is configured to perform an inner product calculation on the historical factor change vector to obtain an inner product result; the weight value calculation module is configured to calculate the weight values of the factors by passing the inner product result through a fully connected layer and a Softmax layer.

[0159] In one embodiment, the inner product module is configured to perform an inner product of the historical factor change vector and the transpose of the historical factor change vector to obtain an inner product result.

[0160] In one embodiment, the vector projection sub-unit includes a first projection module and a re-projection module.

[0161] The first projection module is configured to project the current feature vector onto the historical feature vector to obtain an intermediate feature vector; the re-projection module is configured to determine the difference between the current feature vector and the intermediate feature vector and project the current feature vector onto the difference to obtain a final feature vector.

[0162] It should be noted that those skilled in the art can clearly understand that the specific implementation processes of the above-mentioned road multi-factor collapse risk dynamic assessment system integrating road surface characterization and each unit can refer to the corresponding descriptions in the foregoing method embodiments. For the sake of convenience and brevity of description, they are not elaborated herein.

[0163] The above-mentioned road multi-factor collapse risk dynamic assessment system 300 integrating road surface characterization can be implemented in the form of a computer program, and the computer program can run on a computer device as shown in Figure 10 shown.

[0164] Please refer to Figure 10 , Figure 10 which is a schematic block diagram of a computer device provided by an embodiment of the present application. The computer device 500 may be a server. Among them, the server may be an independent server or a server cluster composed of multiple servers.

[0165] Refer to Figure 10 , the computer device 500 includes a processor 502, a memory, and a network interface 505 connected through a system bus 501. Among them, the memory may include a non-volatile storage medium 503 and an internal memory 504.

[0166] The non-volatile storage medium 503 can store an operating system 5031 and a computer program 5032. The computer program 5032 includes program instructions. When the program instructions are executed, the processor 502 can be caused to execute a dynamic assessment method for the collapse risk of multiple road elements that integrates road surface characterization.

[0167] The processor 502 is used to provide computing and control capabilities to support the operation of the entire computer device 500.

[0168] The internal memory 504 provides an environment for the operation of the computer program 5032 in the non-volatile storage medium 503. When the computer program 5032 is executed by the processor 502, the processor 502 can be caused to execute a dynamic assessment method for the collapse risk of multiple road elements that integrates road surface characterization.

[0169] The network interface 505 is used for network communication with other devices. Those skilled in the art can understand that Figure 10 the structure shown in

[0170] is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device 500 to which the solution of the present application is applied. The specific computer device 500 may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0171] Obtain the data collected by in-vehicle sensors and cameras to obtain real-time data, where the real-time data includes road surface videos, vibration sensor data, and GPS data; obtain the rainfall in the road area; determine the actual area, disease type, and road surface elevation of the disease according to the real-time data to obtain the current road surface characterization data; determine historical data according to the GPS data, where the historical data includes historical road surface characterization data, soil type, rainfall, construction records, and pipeline network information corresponding to the road area, and the historical road surface characterization data includes historical disease types, historical actual disease areas, and historical road surface elevations; perform normalization and encoding processing on the current road surface characterization data and the historical data to obtain the processed current road surface characterization data and the processed historical data; input the processed current road surface characterization data and the processed historical data into a road collapse risk assessment model for road collapse risk assessment to obtain an assessment result; send the assessment result.

[0172] In one embodiment, when the processor 502 implements the step of performing normalization and encoding processing on the current road surface characterization data and the historical data to obtain the processed current road surface characterization data and the processed historical data, the specific implementation steps are as follows:

[0173] Perform normalization on the numerical data in the current road surface characterization data and the historical data, and perform one-hot encoding on the categorical data in the current road surface characterization data and the historical data to obtain the processed current road surface characterization data and the processed historical data.

[0174] In one embodiment, when the processor 502 implements the step of inputting the processed current road surface characterization data and the processed historical data into a road collapse risk assessment model for road collapse risk assessment to obtain an assessment result, the specific implementation steps are as follows:

[0175] Input the processed current road surface characterization data and the processed historical data into a road collapse risk assessment model, and the road collapse risk assessment model processes the current road surface characterization data and the historical data through feature fusion, self-attention mechanism, and vector projection to obtain an assessment result.

[0176] In one embodiment, when the processor 502 implements the step of inputting the processed current road surface characterization data and the processed historical data into a road collapse risk assessment model, and the road collapse risk assessment model processes the current road surface characterization data and the historical data through feature fusion, self-attention mechanism, and vector projection to obtain an assessment result, the specific implementation steps are as follows:

[0177] Input the current road surface characterization data and the historical data into the road collapse risk assessment model. The severity features of the current road surface characterization data and the historical data are respectively extracted by the feature projection layer to obtain a current feature vector and a historical feature vector. Calculate the differences between various factors in the historical data through the self-attention mechanism and determine the weight values of various factors. Perform vector projection on the current feature vector and the historical feature vector to obtain a final feature vector representing the severity change of the current feature vector. Pass the final feature vector through the fully connected layer and the Softmax layer to obtain a continuous risk score. Divide the continuous risk score into discrete risk level regions to obtain the evaluation result.

[0178] Among them, the current feature vector is a multi-dimensional vector, including the disease area corresponding to the current road surface characterization data, the one-hot encoding of the disease type, and the road surface elevation information. The historical feature vector is a multi-dimensional vector, including the historical disease area corresponding to the historical data, the one-hot encoding of the historical disease type, the historical road surface elevation information, the historical soil, rainfall, construction, and pipe network.

[0179] In one embodiment, when the processor 502 implements the step of calculating the differences between various factors in the historical data through the self-attention mechanism and determining the weight values of various factors, the specific implementation steps are as follows:

[0180] Calculate the differences between various factors in the historical data through the self-attention mechanism to obtain a historical factor change vector. Perform an inner product calculation on the historical factor change vector to obtain an inner product result. Calculate the weight values of the factors by passing the inner product result through the fully connected layer and the Softmax layer.

[0181] In one embodiment, when the processor 502 implements the step of performing an inner product calculation on the historical factor change vector to obtain an inner product result, the specific implementation steps are as follows:

[0182] Perform an inner product of the historical factor change vector and the transpose of the historical factor change vector to obtain an inner product result.

[0183] In one embodiment, when the processor 502 implements the step of performing vector projection on the current feature vector and the historical feature vector to obtain a final feature vector representing the severity change of the current feature vector, the specific implementation steps are as follows:

[0184] Project the current feature vector onto the historical feature vector to obtain an intermediate feature vector. Determine the difference between the current feature vector and the intermediate feature vector, and project the current feature vector onto the difference to obtain a final feature vector.

[0185] Among them, the intermediate feature vector includes the change amount of the disease area, the change situation of the disease type, the change situation of the road surface elevation information, and the change information of the historical data at a certain moment and the historical data at the previous moment.

[0186] It should be understood that in the embodiments of the present application, the processor 502 may be a central processing unit 305 (Central Processing Unit, CPU), and this processor 502 may also be other general-purpose processors, digital signal processors (Digital Signal Processor, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), field-programmable gate arrays (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or this processor may also be any conventional processor, etc.

[0187] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program includes program instructions, and the computer program can be stored in a storage medium, and this storage medium is a computer-readable storage medium. The program instructions are executed by at least one processor in the computer system to implement the process steps of the embodiments of the above methods.

[0188] Therefore, the present invention also provides a storage medium. This storage medium may be a computer-readable storage medium. The storage medium stores a computer program, and when the computer program is executed by a processor, the processor performs the following steps:

[0189] Obtain the data collected by the vehicle-mounted sensors and cameras to obtain real-time data, where the real-time data includes road surface videos, vibration sensor data, and GPS data; obtain the rainfall in the road area; determine the actual area, disease type, and road surface elevation of the disease according to the real-time data to obtain the current road surface characterization data; determine historical data according to the GPS data, where the historical data includes historical road surface characterization data, soil type, rainfall, construction records, and pipe network information corresponding to the road area, and the historical road surface characterization data includes historical disease type, historical disease actual area, and historical road surface elevation; perform normalization and encoding processing on the current road surface characterization data and the historical data to obtain the processed current road surface characterization data and the processed historical data; input the processed current road surface characterization data and the processed historical data into the road collapse risk assessment model for road collapse risk assessment to obtain an assessment result; send the assessment result.

[0190] In one embodiment, when the processor executes the computer program to implement the step of normalizing and encoding the current road surface characterization data and the historical data to obtain the processed current road surface characterization data and the processed historical data, the specific implementation steps are as follows:

[0191] Normalize the numerical data in the current road surface characterization data and the historical data, and perform one-hot encoding on the categorical data in the current road surface characterization data and the historical data, so as to obtain the processed current road surface characterization data and the processed historical data.

[0192] In one embodiment, when the processor executes the computer program to implement the step of inputting the processed current road surface characterization data and the processed historical data into a road collapse risk assessment model to perform road collapse risk assessment to obtain an assessment result, the specific implementation steps are as follows:

[0193] Input the processed current road surface characterization data and the processed historical data into a road collapse risk assessment model, and the road collapse risk assessment model processes the current road surface characterization data and the historical data through feature fusion, self-attention mechanism and vector projection to obtain an assessment result.

[0194] In one embodiment, when the processor executes the computer program to implement the step of inputting the processed current road surface characterization data and the processed historical data into a road collapse risk assessment model, and the road collapse risk assessment model processes the current road surface characterization data and the historical data through feature fusion, self-attention mechanism and vector projection to obtain an assessment result, the specific implementation steps are as follows:

[0195] Input the current road surface characterization data and the historical data into a road collapse risk assessment model. The feature projection layer respectively extracts the severity features of the current road surface characterization data and the severity features of the historical data to obtain a current feature vector and a historical feature vector; calculate the differences of various factors in the historical data through the self-attention mechanism and determine the weight values of various factors; perform vector projection on the current feature vector and the historical feature vector to obtain a final feature vector representing the severity change of the current feature vector; pass the final feature vector through a fully connected layer and a Softmax layer to obtain a continuous risk score; divide the continuous risk score into discrete risk level regions to obtain an assessment result.

[0196] Among them, the current feature vector is a multi-dimensional vector, including the disease area corresponding to the current road surface characterization data, the one-hot encoding of the disease type, and the road surface elevation information; the historical feature vector is a multi-dimensional vector, including the historical disease area corresponding to the historical data, the one-hot encoding of the historical disease type, the historical road surface elevation information, historical soil, rainfall, construction, and pipe network.

[0197] In one embodiment, when the processor executes the computer program to implement the step of calculating the differences between various factors in the historical data through the self-attention mechanism and determining the weight values of the various factors, the specific implementation is as follows:

[0198] Calculate the differences between various factors in the historical data through the self-attention mechanism to obtain a historical factor change vector; perform an inner product calculation on the historical factor change vector to obtain an inner product result; calculate the weight values of the factors through a fully connected layer and a Softmax layer for the inner product result.

[0199] In one embodiment, when the processor executes the computer program to implement the step of performing an inner product calculation on the historical factor change vector to obtain an inner product result, the specific implementation is as follows:

[0200] Perform an inner product of the historical factor change vector and the transpose of the historical factor change vector to obtain an inner product result.

[0201] In one embodiment, when the processor executes the computer program to implement the step of performing vector projection on the current feature vector and the historical feature vector to obtain a final feature vector representing the severity change of the current feature vector, the specific implementation is as follows:

[0202] Project the current feature vector onto the historical feature vector to obtain an intermediate feature vector; determine the difference between the current feature vector and the intermediate feature vector, and project the current feature vector onto the difference to obtain a final feature vector.

[0203] Among them, the intermediate feature vector includes the change amount of the disease area, the change situation of the disease type, the change situation of the road surface elevation information, and the change information of the historical data at a certain moment and the historical data at the previous moment.

[0204] The storage medium can be various computer-readable storage media such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a magnetic disk, or an optical disc that can store program codes.

[0205] Those of ordinary skill in the art will appreciate that the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. To clearly illustrate the interchangeability of hardware and software, the components and steps of the examples have been generally described in terms of function in the above description. Whether these functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0206] In several embodiments provided by the present invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For example, the division of each unit is only a logical function division, and there can be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed.

[0207] The steps in the method embodiments of the present invention can be adjusted, combined, and deleted according to actual needs. The units in the system embodiments of the present invention can be combined, divided, and deleted according to actual needs. In addition, in each embodiment of the present invention, the functional units can be integrated into one processing unit 305, or each unit can exist physically alone, or two or more units can be integrated into one unit.

[0208] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a terminal, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention.

[0209] As described above, the above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of various equivalent modifications or substitutions, and these modifications or substitutions should be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. A dynamic assessment method for road collapse risk with multiple factors integrating pavement representation, characterized in that: include: Acquire data collected by vehicle-mounted sensors and cameras to obtain real-time data, wherein the real-time data includes road surface video, vibration sensor data, and GPS data; Get the rainfall in the road area; Determine the actual area of ​​the damage, the type of damage and the elevation of the road surface according to the real-time data to obtain current road surface characterization data; Determine historical data based on the GPS data, wherein the historical data includes historical road surface characterization data, soil type, rainfall, construction records and pipe network information corresponding to the road area, and the historical road surface characterization data includes historical disease types, actual areas of historical diseases, and historical road surface elevations; Normalizing and encoding the current road surface representation data and the historical data to obtain processed current road surface representation data and processed historical data; Inputting the processed current road surface characterization data and the processed historical data into a road collapse risk assessment model, and extracting the severity features of the current road surface characterization data and the severity features of the historical data by a feature projection layer to obtain a current feature vector and a historical feature vector; The self-attention mechanism is used to calculate the differences of various factors in historical data and determine the weights of each factor. Projecting the current feature vector onto the historical feature vector to obtain an intermediate feature vector; Determine a difference between the current feature vector and the intermediate feature vector, and project the current feature vector onto the difference to obtain a final feature vector; Pass the final feature vector through a fully connected layer and a Softmax layer to obtain a continuous risk score; Divide the continuous risk score into discrete risk level areas to obtain the assessment results; sending the evaluation result; The current feature vector is a multidimensional vector, including the damage area, unique hot coding of the damage type, and road elevation information corresponding to the current road surface characterization data; the historical feature vector is a multidimensional vector, including the historical damage area, unique hot coding of the historical damage type, historical road elevation information, historical soil, rainfall, construction, and pipeline network corresponding to the historical data.

2. The method for dynamic assessment of road collapse risk integrating multiple factors of road surface representation according to claim 1 is characterized in that: The normalizing and encoding the current road surface representation data and the historical data to obtain the processed current road surface representation data and the processed historical data includes: The numerical data in the current road surface representation data and the historical data are normalized, and the categorical data in the current road surface representation data and the historical data are one-hot encoded to obtain processed current road surface representation data and processed historical data.

3. The method for dynamic assessment of road collapse risk integrating multiple factors of road surface representation according to claim 1 is characterized in that: The self-attention mechanism is used to calculate the differences of various factors in the historical data and determine the weight value of each factor, including: The difference of each factor in the historical data is calculated through the self-attention mechanism to obtain the historical factor change vector; Performing inner product calculation on the historical factor change vector to obtain an inner product result; The inner product result is passed through the fully connected layer and the Softmax layer to calculate the weight value of the factor.

4. The method for dynamic assessment of road collapse risk integrating multiple factors of road surface representation according to claim 3 is characterized in that: The inner product calculation of the historical factor change vector to obtain the inner product result includes: An inner product is performed on the historical factor change vector and the transpose of the historical factor change vector to obtain an inner product result.

5. The method for dynamic assessment of road collapse risk integrating multiple factors of road surface representation according to claim 4 is characterized in that: The intermediate feature vector includes the change in the damage area, the change in the damage type, the change in road elevation information, and the change information between the historical data at a certain moment and the historical data at the previous moment.

6. A dynamic assessment system for road collapse risk with multiple factors integrating road surface representation, characterized by: include: A data acquisition unit, used to acquire data collected by vehicle-mounted sensors and cameras to obtain real-time data, wherein the real-time data includes road surface video, vibration sensor data and GPS data; A rainfall acquisition unit, used for acquiring the rainfall in the road area; a current road surface characterization data determination unit, used to determine the actual area of ​​the damage, the type of damage and the road surface elevation according to the real-time data to obtain the current road surface characterization data; A historical data determination unit, used to determine historical data according to the GPS data, wherein the historical data includes historical road surface characterization data, soil type, rainfall, construction records and pipe network information corresponding to the road area, and the historical road surface characterization data includes historical disease type, actual area of ​​historical disease, and historical road surface elevation; A processing unit, configured to encode the current road surface representation data and the historical data to obtain processed current road surface representation data and processed historical data; An evaluation unit, used for inputting the processed current road surface representation data and the processed historical data into a road collapse risk evaluation model to perform road collapse risk evaluation to obtain an evaluation result; specifically, the processed current road surface representation data and the processed historical data are input into the road collapse risk evaluation model, and the road collapse risk evaluation model processes the current road surface representation data and the historical data through feature fusion, self-attention mechanism and vector projection to obtain an evaluation result; A sending unit, used for sending the evaluation result; The evaluation unit includes a feature extraction subunit, a weight calculation subunit, a vector projection subunit, a risk score determination subunit, and a division subunit; A feature extraction subunit is used to input the processed current road surface representation data and the processed historical data into the road collapse risk assessment model, and the feature projection layer extracts the severity features of the current road surface representation data and the severity features of the historical data respectively to obtain a current feature vector and a historical feature vector; a weight calculation subunit is used to calculate the difference of each factor in the historical data through a self-attention mechanism, and determine the weight value of each factor; A vector projection subunit is used to perform vector projection on the current feature vector and the historical feature vector to obtain a final feature vector representing the severity change of the current feature vector; specifically: projecting the current feature vector onto the historical feature vector to obtain an intermediate feature vector; determining a difference between the current feature vector and the intermediate feature vector, and projecting the current feature vector onto the difference to obtain a final feature vector; A risk score determination subunit, used for passing the final feature vector through a fully connected layer and a Softmax layer to obtain a continuous risk score; A division subunit is used to divide the continuous risk score into discrete risk level areas to obtain an assessment result; The current feature vector is a multidimensional vector, including the damage area, unique hot coding of the damage type, and road elevation information corresponding to the current road surface characterization data; the historical feature vector is a multidimensional vector, including the historical damage area, unique hot coding of the historical damage type, historical road elevation information, historical soil, rainfall, construction, and pipeline network corresponding to the historical data.

Images