A road surface peeling state evaluation method, electronic device and storage medium

Abstract

A method for evaluating the state of pavement spalling, an electronic device and a storage medium belong to the technical field of pavement disease detection. In order to timely discover and repair pavement damage, the present invention constructs a basic model and an enhanced model for pavement spalling identification; constructs an output layer loss function of the basic model and the enhanced model in the mathematical model for pavement spalling identification; constructs an inter-layer loss function of the basic model and the enhanced model in the mathematical model for pavement spalling identification; uses a data set to train the mathematical model for pavement spalling identification, obtains the trained mathematical model for pavement spalling identification, performs digital twin modeling, and obtains a digital twin model for pavement spalling identification; evaluates the state of pavement spalling according to the geometric characteristics of the spalling area, the speed of geometric feature change, the material and structural characteristics, the load characteristics, the environmental conditions, the maintenance and historical records, and the remaining service life indicators. The present invention realizes rapid and accurate identification of pavement spalling.

Description

Technical Field

[0001] The present invention belongs to the technical field of pavement disease detection, and in particular relates to a pavement spalling state evaluation method, electronic equipment and storage medium. Background Art

[0002] Pavement spalling refers to the shedding, breaking or decomposition of road surface materials due to various reasons. Common causes include material aging, traffic load pressure, unsuitable climatic conditions, poor construction quality, improper maintenance and other factors. Pavement spalling is not limited to the surface, but may also affect the structural layer of the road, thereby reducing the overall load-bearing capacity and service life of the road surface.

[0003] The occurrence of pavement spalling has a direct and serious impact on pavement use and traffic safety. First, the uneven road surface caused by spalling will reduce driving comfort and increase driving difficulty. When vehicles travel on damaged roads, they will experience increased vibration and noise, and vehicles that are affected for a long time will face more frequent maintenance and repair needs. Pavement spalling can cause drivers to suddenly change lanes to avoid damaged areas, increasing the risk of traffic accidents. In addition, the unevenness of the road surface may affect the driving efficiency of vehicles, increase fuel consumption and emissions, and impose an additional burden on the environment.

[0004] The study of pavement spalling status evaluation methods is of great significance for road maintenance and management. Through effective evaluation methods, pavement damage can be discovered and repaired in a timely manner to prevent damage from expanding into more serious structural problems, thereby greatly reducing the maintenance cost of the road throughout its life cycle.

[0005] The invention patent with the patent number of 202211109405.X and the invention name of "Method and system for detecting hidden peeling points of pavement based on deep image recognition" introduces a system based on deep image recognition for detecting hidden peeling points of pavement. The method first collects pavement images in a wet environment, preprocesses them to remove noise, and then marks the hidden peeling points. Next, the detection model is trained using a deep learning framework, and the model is used to identify peeling points in new images. This technology aims to increase the service life of roads and reduce maintenance costs, and provide efficient and intelligent road maintenance management. However, the lack of research on pavement peeling status evaluation methods will lead to the inability to detect and repair pavement damage in a timely manner, which will in turn aggravate further damage to the road and reduce the service life of the road. This not only increases the cost of road maintenance, but may also cause traffic accidents and driving inconveniences due to pavement quality problems, affecting traffic safety and efficiency. Summary of the invention

[0006] The problem to be solved by the present invention is to timely discover and repair road damage, and to propose a road peeling state evaluation method, electronic equipment and storage medium.

[0007] To achieve the above object, the present invention is implemented through the following technical solutions:

[0008] A method for evaluating pavement spalling state comprises the following steps:

[0009] S1. Use high-definition cameras to collect road images;

[0010] S2. For the road surface image collected in step S1, the images of road peeling, transverse cracks, cracks, and rutting are selected and manually annotated to construct a data set for a mathematical model for road peeling identification;

[0011] S3. Construct mathematical models for pavement spalling identification, including basic model and enhanced model;

[0012] S4. Construct the output layer loss function of the basic model and the enhanced model in the mathematical model for pavement spalling identification;

[0013] S5. Construct the inter-layer loss function of the basic model and the enhanced model in the mathematical model for pavement spalling identification;

[0014] S6. Using the data set obtained in step S2 to train the mathematical model for pavement peeling identification, a trained mathematical model for pavement peeling identification is obtained;

[0015] S7. Perform digital twin modeling based on the trained mathematical model for pavement spalling identification obtained in step S6 to obtain a digital twin model for pavement spalling identification;

[0016] S8. Based on the digital twin model for pavement spalling identification obtained in step S7, the pavement spalling status is evaluated according to the geometric characteristics of the spalling area, the speed of change of geometric characteristics, material and structural characteristics, load characteristics, environmental conditions, maintenance and historical records, and remaining service life indicators.

[0017] Further, the specific implementation method of step S2 includes the following steps:

[0018] S2.1. From the pavement images, manually select all images containing spalling, transverse cracks, fissures, and rutting;

[0019] S2.2. By manually labeling, the spalling area and the other three diseases are selected in the pavement image selected in step S1.1, and the spalling is marked as Spalling. The other three diseases, transverse cracks, cracks, and rutting, are marked as other1, other2, and other3 respectively, to construct a data set for the mathematical model of pavement spalling identification;

[0020] S2.3. Divide the data set of the mathematical model used for pavement spalling identification into a training set and a test set in a ratio of 8:2.

[0021] Furthermore, the specific implementation method of step S3 includes the following steps:

[0022] S3.1. Build a basic model. The basic model A consists of 1 input layer, 5 convolutional layers, 5 activation layers, 4 pooling layers, 1 global average pooling layer, and 1 output layer.

[0023] S3.2. Construct an enhanced model. The enhanced model B consists of 1 input layer, 10 convolutional layers, 10 activation layers, 5 pooling layers, 1 global average pooling layer, and 1 output layer.

[0024] Furthermore, the specific implementation method of step S4 includes the following steps:

[0025] S4.1. Construct the loss function of the basic model A and train the basic model A;

[0026] S4.1.1. The loss function of the basic model A is calculated as:

[0027] ;

[0028] in, is the loss function of the basic model A, is the total number of data samples in the basic model A, For the The true labels of samples, Based on the prediction of model A The probability that a sample is class 1;

[0029] The training set obtained in step S2 is input into the basic model A, and the weights and biases of the neural network units are calculated through the error back propagation algorithm to complete the training of the basic model A and obtain the trained basic model A;

[0030] S4.1.2. The node output values ​​of the four output units corresponding to the basic model A are , the node output values ​​of the four output units are initially converted to obtain the output value The corresponding probability value The calculation formula is:

[0031] ;

[0032] Then the node output values ​​of the four output units are further optimized and converted to obtain the output value The corresponding enhancement probability value The calculation formula is:

[0033] ;

[0034] in, G is the enhancement coefficient, and then the actual disease corresponding to each node is marked as ;

[0035] S4.1.3. Adoption The training set obtained in step S2 is combined with the error chain derivation rule to train the basic model A. After the training is completed, the output values ​​of the four types of diseases corresponding to the basic model A are obtained, which are recorded as ;

[0036] S4.2. Construct a loss function for the reinforcement model B and train the reinforcement model B;

[0037] S4.2.1. Input the training set obtained in step S2 into the enhanced model B, calculate the weights and biases of the neural network units through the error back propagation algorithm, complete the training of the enhanced model B, and obtain the trained enhanced model B;

[0038] The loss function of the enhanced model B is calculated as:

[0039] ;

[0040] in, To strengthen the loss function of model B; is the total number of data samples in the reinforcement model B; For the The true labels of samples, To strengthen the model B to predict the The probability that a sample is class 1;

[0041] S4.2.2. The node output values ​​of the four output units corresponding to the enhanced model B are b 1 , b 2 , b 3 , b 4 , the node output values ​​of the four output units are initially converted to obtain the output value The corresponding probability value The calculation formula is:

[0042] ;

[0043] Then the node output values ​​of the four output units are further optimized and converted to obtain the output value The corresponding enhancement probability value The calculation formula is:

[0044] ;

[0045] S4.2.3. Adoption The training set obtained in step S2 is combined with the error chain derivation rule to train the enhanced model B. After the training is completed, the disease output value corresponding to the enhanced model B is obtained as , among the 4 nodes in the output layer, the actual disease corresponding to each node is marked as , ;

[0046] S4.3. Construct output layer loss functions of the base model and enhanced model;

[0047] based on and , construct the loss function between the enhanced model B and the output layer of the basic model A , the calculation formula is:

[0048] ;

[0049] based on and , construct the output value The corresponding probability value The loss function of the base model A , the calculation formula is:

[0050] ;

[0051] based on and , establish the output layer loss function of the basic model and the enhanced model , the calculation formula is:

[0052] ;

[0053] in, for The weight coefficient of for The weight coefficient of .

[0054] Furthermore, the specific implementation method of step S5 includes the following steps:

[0055] S5.1. Assuming a small model a and large models b , build a small model a The second, third, fourth and output layers in the large model are bEstablish connections between the fourth, sixth, tenth layers and the output layer;

[0056] S5.2. For the second layer in the small model and the large model b The fourth layer in the small model establishes relationships a The output of the second layer is recorded as , the corresponding size of width × length × number of channels is Large Model b The output of the fourth layer is recorded as , the corresponding size of width × length × number of channels is ;

[0057] For size comparison, When , a convolution operation is added to the second layer output of the small model, the convolution kernel size is 1×1, and the number of convolution kernels is , the output is ,at this time, The corresponding output values ​​of each channel are , is the channel number, The corresponding output values ​​of each channel are ;

[0058] Build on Small model The second floor with the large model The loss function between the fourth layers , the calculation formula is:

[0059] ;

[0060] when When in the large model b In the output of the fourth layer, a convolution operation is added, the convolution kernel size is 1×1, and the number of convolution kernels is , the output is ;at this time, The corresponding output values ​​of each channel are: , The corresponding output values ​​of each channel are ;

[0061] Establish Small model The second floor with the large model The loss function between the fourth layers , the calculation formula is:

[0062] ;

[0063] S5.3. For the third layer in the small model and the large model The sixth layer in the relationship is established to obtain a small model The third floor and the large model Loss function between the sixth layers ;

[0064] S5.4. For the fourth layer in the small model and the large model The tenth layer in the relationship is established to obtain a small model Fourth floor and large model Loss function between the tenth layer ;

[0065] S5.5. When the small model is the base model A, the small model The second, third, and fourth layers in the graph are the output of pooling layer 1, pooling layer 2, and pooling layer 4 of the basic model A, respectively. When the large model is the enhanced model B, the large model The fourth, sixth, and tenth layers in the figure are the outputs of pooling layer 2, pooling layer 3, and pooling layer 5 of the enhanced model B, respectively.

[0066] Furthermore, the specific implementation method of step S6 includes the following steps:

[0067] S6.1. Based on the loss function between the output layer of the enhanced model B and the inter-layer loss function of the base model A, define the total loss function between the two models , the calculation formula is:

[0068] ;

[0069] in, for The corresponding weight coefficient is, for The corresponding weight coefficient is, for The corresponding weight coefficient;

[0070] S6.2. Using the data set obtained in step S2, based on the chain rule, the enhanced model B and the basic model A are trained by error back propagation. When the overall loss function When is the minimum, the corresponding model parameters are optimal, and the trained basic model A is the trained mathematical model for pavement spalling identification.

[0071] Furthermore, the specific implementation method of step S7 includes the following steps:

[0072] S7.1. Export the trained mathematical model for pavement spalling identification obtained in step S6 as a deployment file for independent services, including one of a pt file, a pth file, a json file, and a txt file;

[0073] S7.2. Construct a road information model;

[0074] S7.2.1. Collect information on the location, length, width, and layout of roads;

[0075] S7.2.2. The data obtained in step S7.2.1 is stored in a database, and a basic three-dimensional model of the road is constructed, wherein the basic three-dimensional model of the road includes geometric information of the road;

[0076] S7.2.3. Based on step S7.2.2, material properties, design specifications, and historical maintenance data are added to the basic three-dimensional model of the road to obtain a road information model;

[0077] S7.2.4. Export the road information model to a format file supported by the web application, including a json file and a js file;

[0078] S7.3. In the digital twin platform, import the deployment file of the trained mathematical model for pavement spalling identification obtained in step S7.1, and import the road information model format file obtained in step S7.2;

[0079] S7.4. Use Python to construct an API interface between image data and the trained mathematical model for pavement spalling identification for intelligent rutting identification. Use Python to construct an API interface between the trained mathematical model for pavement spalling identification and the road information model to complete the construction of the digital twin model for pavement spalling identification.

[0080] Furthermore, the specific implementation method of step S8 includes the following steps:

[0081] S8.1. Set the spalling area geometry, geometry change rate, material and structural characteristics, load characteristics, environmental conditions, maintenance and history records, and remaining service life as evaluation indicators for pavement spalling conditions;

[0082] S8.2. Use a high-definition camera to collect pavement image data of NR roads, transmit the pavement image data of each road to the digital twin platform of the digital twin model for pavement spalling identification constructed in step S7, intelligently identify the spalling area in the pavement image, and select the image containing the pavement spalling;

[0083] S8.3. The image containing the road peeling obtained in step S8.2 is scored using the expert scoring method, and the weight of each evaluation index set in step S8.1 relative to other evaluation indexes is scored to obtain the first The evaluation index is relative to Expert score of the weight of the evaluation indicators ;

[0084] S8.4. Based on the weighted expert scores of all evaluation indicators obtained in step S3, establish a road evaluation matrix , the calculation formula is:

[0085] ;

[0086] Among them, geometric characteristics, rate of change of geometric characteristics, material and structural characteristics, load characteristics, environmental conditions, maintenance and history records, and remaining service life are 1, 2, 3, 4, 5, 6, and 7 respectively;

[0087] S8.5. Calculate the evaluation matrix of the road based on the evaluation matrix of the road The weight vector , the calculation formula is:

[0088] ;

[0089] Then calculate the maximum eigenvector of the road evaluation matrix , the calculation formula is:

[0090] ;

[0091] Then calculate the maximum eigenvalue corresponding to the maximum eigenvector of the road evaluation matrix , the calculation formula is:

[0092] ;

[0093] in, represents the largest eigenvector The sum of all elements in , Represents the weight vector The sum of all elements in ;

[0094] S8.6. Calculate the maximum eigenvalue corresponding to all NR roads , sort all the maximum eigenvalues ​​from large to small. The larger the maximum eigenvalue, the greater the severity of the spalling in the corresponding road, and obtain the quantitative evaluation results of the spalling degree in different roads;

[0095] S8.7. Based on the maximum eigenvalue Perform data consistency analysis and set the consistency index as CS , the calculation formula is:

[0096] ;

[0097] Set the consistency ratio to Cb , the calculation formula is:

[0098] ;

[0099] in, RI It is the consistency index, which is obtained by querying the "Random Consistency Index RI Value Table";

[0100] By establishing a consistent ratio Cb The relationship with 0.1 determines the consistency of the relative interrelationships of the seven factors:

[0101] when When , it indicates that the consistency is acceptable;

[0102] when When , it indicates that the consistency is unacceptable. At this time, return to step S8.4 to readjust the evaluation matrix .

[0103] An electronic device comprises a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of a method for evaluating a pavement spalling state when executing the computer program.

[0104] A computer-readable storage medium stores a computer program, which implements a pavement spalling state evaluation method when executed by a processor.

[0105] Beneficial effects of the present invention:

[0106] The pavement spalling status evaluation method described in the present invention constructs a new neural network model. The model takes into account the output layer loss function and inter-layer loss function of the basic model and the enhanced model, forming a pavement spalling identification method based on digital twins to achieve rapid identification of pavement spalling. This method combines the lightweight characteristics of the small model and the accurate characteristics of the large model. By establishing the output layer loss function and inter-layer loss function of the small model and the large model, a lightweight model for rapid and accurate identification of pavement spalling is completed. The model has the highest level of lightweight and can be easily deployed in terminal devices such as edge devices to achieve rapid and accurate identification of pavement spalling.

[0107] The present invention discloses a method for evaluating the pavement spalling state, and proposes a method for constructing a road digital twin model and updating pavement spalling, so as to realize the construction of the digital twin model and the accurate updating of the spalling area.

[0108] The present invention provides a method for evaluating the pavement spalling state, which proposes a method for evaluating the pavement spalling state based on digital twins. The method can accurately quantify and evaluate the degree of damage and spalling of the pavement, detect problems in a timely manner, optimize the allocation of maintenance resources, and improve the pavement condition to improve driving comfort and traffic efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0109] Figure 1 The present invention is a flow chart of a method for evaluating pavement spalling status. DETAILED DESCRIPTION

[0110] In order to make the purpose, technical solution and advantages of the present invention more clear, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention, that is, the specific embodiments described are only part of the embodiments of the present invention, rather than all of the specific embodiments. The components of the specific embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations, and the present invention can also have other embodiments.

[0111] Therefore, the following detailed description of the specific embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents the selected specific embodiments of the present invention. Based on the specific embodiments of the present invention, all other specific embodiments obtained by those skilled in the art without making creative work are within the scope of protection of the present invention.

[0112] In order to further understand the content, features and effects of the present invention, the following specific implementation methods are given as examples, and the attached Figure 1 The detailed instructions are as follows:

[0113] Embodiment 1:

[0114] A method for evaluating pavement spalling state comprises the following steps:

[0115] S1. Use high-definition cameras to collect road images;

[0116] S2. For the road surface image collected in step S1, the images of road peeling, transverse cracks, cracks, and rutting are selected and manually annotated to construct a data set for a mathematical model for road peeling identification;

[0117] Furthermore, the specific implementation method of step S2 includes the following steps:

[0118] S2.1. From the pavement images, manually select all images containing spalling, transverse cracks, fissures, and rutting;

[0119] S2.2. By manually labeling, the spalling area and the other three diseases are selected in the pavement image selected in step S1.1, and the spalling is marked as Spalling. The other three diseases, transverse cracks, cracks, and rutting, are marked as other1, other2, and other3 respectively, to construct a data set for the mathematical model of pavement spalling recognition;

[0120] S2.3. Divide the data set of the mathematical model for pavement spalling identification into a training set and a test set in a ratio of 8:2;

[0121] S3. Construct mathematical models for pavement spalling identification, including basic model and enhanced model;

[0122] Pavement spalling is caused by many factors, including traffic load, temperature change, material aging, water damage, etc. The characteristics of pavement spalling images include irregular geometry, rough edges, and textures that are different from normal pavement. Images captured by on-board cameras are also affected by factors such as lighting, shadows, reflections, and road dirt. Therefore, the designed model structure includes input layer, convolution layer, activation layer, pooling layer, global average pooling layer, and output layer. The functions of each structural layer are as follows:

[0123] Input layer: Receives road image data collected by the vehicle camera. The input layer does not perform calculations, but only passes the pixel values ​​of the image to the next layer.

[0124] Convolutional layer: Responsible for extracting features from the image. The filters in the convolutional layer detect low-level features in the image, such as edges and textures, which are critical for identifying peeling areas. As the network goes deeper, subsequent convolutional layers are able to combine low-level features to detect more complex patterns, such as the specific shape and size of peeling areas.

[0125] Activation layer: helps introduce nonlinearity, which allows the model to learn and represent more complex feature patterns. In the spalling recognition task, nonlinearity is key to distinguish spalling from non-spalling areas.

[0126] Pooling layer: It reduces the number of parameters and computational complexity by reducing the spatial size of the feature map, and helps the model obtain a larger receptive field, thereby recognizing patterns in a larger area. This helps the model maintain important features while being more robust to small image changes and noise.

[0127] Global average pooling layer: It is a structured pooling method that averages the features of each feature map into a single value, thereby reducing the number of model parameters and helping to prevent overfitting. This is beneficial for processing large amounts of road image data, simplifying the model and speeding up training.

[0128] Output layer: used to integrate the features extracted by the previous layers and perform the final classification task.

[0129] Furthermore, the specific implementation method of step S3 includes the following steps:

[0130] S3.1. Build the basic model;

[0131] The basic model A consists of 1 input layer, 5 convolutional layers, 5 activation layers, 4 pooling layers, 1 global average pooling layer, and 1 output layer. The specific network structure is as follows:

[0132] Input layer: image size 227×227×3;

[0133] Convolution layer 1: kernel size 3×3; number of kernels 32; stride: 1; padding: same;

[0134] Activation layer 1: The activation function is ReLU;

[0135] Pooling layer 1: Type is maximum pooling; pooling size: 2×2; Stride: 2;

[0136] Convolutional layer 2: kernel size 3×3; number of kernels 64; Stride: 1; Padding: Same;

[0137] Activation layer 2: The activation function is ReLU;

[0138] Pooling layer 2: Type is maximum pooling; pooling size: 2×2; Stride: 2;

[0139] Convolution layer 3: kernel size 3×3; number of kernels 128; Stride: 1; Padding: Same;

[0140] Activation layer 3: The activation function is ReLU;

[0141] Pooling layer 3: Type is maximum pooling; pooling size: 2×2; Stride: 2;

[0142] Convolutional layer 4: kernel size 3×3; number of kernels 256; Stride: 1; Padding: Same;

[0143] Activation layer 4: The activation function is ReLU;

[0144] Convolution layer 5: kernel size 3×3; number of kernels 256; Stride: 1; Padding: Same;

[0145] Activation layer 5: The activation function is ReLU;

[0146] Pooling layer 4: Type is maximum pooling; pooling size: 2×2; Stride: 2;

[0147] Global average pooling layer 1;

[0148] Output layer: type is Softmax; number of neurons: 4;

[0149] S3.2. Constructing reinforcement models;

[0150] The enhanced model B consists of 1 input layer, 10 convolutional layers, 10 activation layers, 5 pooling layers, 1 global average pooling layer, and 1 output layer. The specific network structure is as follows:

[0151] Input layer: image size 227×227×3;

[0152] Convolution layer 1: kernel size 3×3; number of kernels 32; Stride: 1; Padding: Same;

[0153] Activation layer 1: The activation function is ReLU;

[0154] Convolutional layer 2: kernel size 3×3; number of kernels 64; Stride: 1; Padding: Same;

[0155] Activation layer 2: The activation function is ReLU;

[0156] Pooling layer 1: Type is maximum pooling; pooling size: 2×2; Stride: 2;

[0157] Convolution layer 3: kernel size 3×3; number of kernels 64; Stride: 1; Padding: Same;

[0158] Activation layer 3: The activation function is ReLU;

[0159] Convolutional layer 4: kernel size 3×3; number of kernels 64; Stride: 1; Padding: Same;

[0160] Activation layer 4: The activation function is ReLU;

[0161] Pooling layer 2: Type is maximum pooling; pooling size: 2×2; Stride: 2;

[0162] Convolution layer 5: kernel size 3×3; number of kernels 128; Stride: 1; Padding: Same;

[0163] Activation layer 5: The activation function is ReLU;

[0164] Convolution layer 6: kernel size 3×3; number of kernels 128; Stride: 1; Padding: Same;

[0165] Activation layer 6: The activation function is ReLU;

[0166] Pooling layer 3: Type is maximum pooling; pooling size: 2×2; Stride: 2;

[0167] Convolution layer 7: kernel size 3×3; number of kernels 256; Stride: 1; Padding: Same;

[0168] Activation layer 7: The activation function is ReLU;

[0169] Convolution layer 8: kernel size 3×3; number of kernels 256; Stride: 1; Padding: Same;

[0170] Activation layer 8: The activation function is ReLU;

[0171] Convolution layer 9: kernel size 3×3; number of kernels 256; Stride: 1; Padding: Same;

[0172] Activation layer 9: The activation function is ReLU;

[0173] Pooling layer 4: Type is maximum pooling; pooling size: 2×2; Stride: 2;

[0174] Convolution layer 10: kernel size 3×3; number of kernels 512; Stride: 1; Padding: Same;

[0175] Activation layer 10: The activation function is ReLU;

[0176] Pooling layer 5: Type: Max pooling; Pooling size: 2×2; Stride: 2;

[0177] Global average pooling layer 1;

[0178] Output layer: Type: Softmax; Number of neurons: 4.

[0179] S4. Construct the output layer loss function of the basic model and the enhanced model in the mathematical model for pavement spalling identification;

[0180] Furthermore, the specific implementation method of step S4 includes the following steps:

[0181] S4.1. Construct the loss function of the basic model A and train the basic model A;

[0182] S4.1.1. The loss function of the basic model A is calculated as:

[0183] ;

[0184] in, is the loss function of the basic model A, is the total number of data samples in the basic model A, For the The true labels of samples, Based on the prediction of model A The probability that a sample is class 1;

[0185] The training set obtained in step S2 is input into the basic model A, and the weights and biases of the neural network units are calculated through the error back propagation algorithm to complete the training of the basic model A and obtain the trained basic model A;

[0186] S4.1.2. The node output values ​​of the four output units corresponding to the basic model A are , the node output values ​​of the four output units are initially converted to obtain the output value The corresponding probability value The calculation formula is:

[0187] ;

[0188] Then the node output values ​​of the four output units are further optimized and converted to obtain the output value The corresponding enhancement probability value The calculation formula is:

[0189] ;

[0190] in, is the enhancement coefficient, and then the actual disease corresponding to each node is marked as , specifically, for pavement spalling 1, for other diseases is 0;

[0191] S4.1.3. Adoption The training set obtained in step S2 is combined with the error chain derivation rule to train the basic model A. After the training is completed, the output values ​​of the four types of diseases corresponding to the basic model A are obtained, which are recorded as ;

[0192] S4.2. Construct a loss function for the reinforcement model B and train the reinforcement model B;

[0193] S4.2.1. Input the training set obtained in step S2 into the enhanced model B, calculate the weights and biases of the neural network units through the error back propagation algorithm, complete the training of the enhanced model B, and obtain the trained enhanced model B;

[0194] The loss function of the enhanced model B is calculated as:

[0195] ;

[0196] in, To strengthen the loss function of model B; is the total number of data samples in the reinforcement model B; For the The true labels of samples, To strengthen model B, we predict ib The probability that a sample is class 1;

[0197] S4.2.2. The node output values ​​of the four output units corresponding to the enhanced model B are , the node output values ​​of the four output units are initially converted to obtain the output value The corresponding probability value The calculation formula is:

[0198] ;

[0199] Then the node output values ​​of the four output units are further optimized and converted to obtain the output value The corresponding enhancement probability value The calculation formula is:

[0200] ;

[0201] S4.2.3. Adoption The training set obtained in step S2 is combined with the error chain derivation rule to train the enhanced model B. After the training is completed, the disease output value corresponding to the enhanced model B is obtained as , among the 4 nodes in the output layer, the actual disease label corresponding to each node is , , specifically, for pavement spalling 1, for other diseases is 0;

[0202] S4.3. Construct output layer loss functions of the base model and enhanced model;

[0203] based on and , construct the loss function between the enhanced model B and the output layer of the basic model A , the calculation formula is:

[0204] ;

[0205] based on and , construct the output value The corresponding probability value The loss function of the base model A , the calculation formula is:

[0206] ;

[0207] based on and , establish the output layer loss function of the basic model and the enhanced model , the calculation formula is:

[0208] ;

[0209] in, for The weight coefficient of for The weight coefficient of .

[0210] Loss Function It is part of the loss function of the entire model. Together with the inter-layer loss function, it constitutes the loss function of the entire model, which is used to guide the update of model parameters and improve model performance. Generally speaking, in deep learning models, the loss function is used to measure the difference between the model's predicted output and the true label. The calculation of the loss function guides the update of network parameters. The more reasonable the loss function is, the better the performance of the model.

[0211] S5. Construct the inter-layer loss function of the basic model and the enhanced model in the mathematical model for pavement spalling identification;

[0212] Furthermore, the specific implementation method of step S5 includes the following steps:

[0213] S5.1. Assuming a small model and large models , build a small model The second, third, fourth and output layers in the large model are Establish connections between the fourth, sixth, tenth layers and the output layer;

[0214] S5.2. For the second layer in the small model and the large model The fourth layer in the small model establishes relationships The output of the second layer is recorded as , the corresponding size of width × length × number of channels is Large Model The output of the fourth layer is recorded as , the corresponding size of width × length × number of channels is ;

[0215] For size comparison, When , a convolution operation is added to the second layer output of the small model, the convolution kernel size is 1×1, and the number of convolution kernels is , the output is ,at this time, The corresponding output values ​​of each channel are , is the channel number, The corresponding output values ​​of each channel are ;

[0216] Build on Small model The second floor with the large model The loss function between the fourth layers , the calculation formula is:

[0217] ;

[0218] when When in the large model In the output of the fourth layer, a convolution operation is added, the convolution kernel size is 1×1, and the number of convolution kernels is , the output is ;at this time, The corresponding output values ​​of each channel are: , The corresponding output values ​​of each channel are ;

[0219] Establish Small model The second floor with the large model The loss function between the fourth layers , the calculation formula is:

[0220] ;

[0221] S5.3. For the third layer in the small model and the large model The sixth layer in the relationship is established to obtain a small model The third floor and the large model Loss function between the sixth layers ;

[0222] S5.4. For the fourth layer in the small model and the large model The tenth layer in the relationship is established to obtain a small model Fourth floor and large model Loss function between the tenth layer ;

[0223] S5.5. When the small model is the base model A, the small model The second, third, and fourth layers in the graph are the output of pooling layer 1, pooling layer 2, and pooling layer 4 of the basic model A, respectively. When the large model is the enhanced model B, the large model The fourth, sixth, and tenth layers in the figure are the outputs of pooling layer 2, pooling layer 3, and pooling layer 5 of the enhanced model B, respectively.

[0224] S6. Using the data set obtained in step S2 to train the mathematical model for pavement peeling identification, a trained mathematical model for pavement peeling identification is obtained;

[0225] Furthermore, the specific implementation method of step S6 includes the following steps:

[0226] S6.1. Based on the loss function between the output layer of the enhanced model B and the inter-layer loss function of the base model A, define the total loss function between the two models , the calculation formula is:

[0227] ;

[0228] in, for The corresponding weight coefficient is, for The corresponding weight coefficient is, for The corresponding weight coefficient;

[0229] S6.2. Using the data set obtained in step S2, based on the chain rule, the enhanced model B and the basic model A are trained by error back propagation. When the overall loss function When it is the smallest, the corresponding model parameters are optimal, and the trained basic model A is the trained mathematical model for pavement spalling identification;

[0230] S7. Perform digital twin modeling based on the trained mathematical model for pavement spalling identification obtained in step S6 to obtain a digital twin model for pavement spalling identification;

[0231] Furthermore, the specific implementation method of step S7 includes the following steps:

[0232] S7.1. Export the trained mathematical model for pavement spalling identification obtained in step S6 as a deployment file for independent services, including one of a pt file, a pth file, a json file, and a txt file;

[0233] S7.2. Construct a road information model;

[0234] S7.2.1. Collect information on the location, length, width, and layout of roads;

[0235] S7.2.2. The data obtained in step S7.2.1 is stored in a database, and a basic three-dimensional model of the road is constructed, wherein the basic three-dimensional model of the road includes geometric information of the road;

[0236] S7.2.3. Based on step S7.2.2, material properties, design specifications, and historical maintenance data are added to the basic three-dimensional model of the road to obtain a road information model;

[0237] S7.2.4. Export the road information model to a format file supported by the web application, including a json file and a js file;

[0238] S7.3. In the digital twin platform, import the deployment file of the trained mathematical model for pavement spalling identification obtained in step S7.1, and import the road information model format file obtained in step S7.2;

[0239] S7.4. Use Python to construct an API interface between image data and the trained mathematical model for pavement spalling identification for intelligent rutting identification. Use Python to construct an API interface between the trained mathematical model for pavement spalling identification and the road information model to complete the construction of the digital twin model for pavement spalling identification.

[0240] S8. Based on the digital twin model for pavement spalling identification obtained in step S7, evaluate the pavement spalling status according to the geometric characteristics of the spalling area, the speed of change of geometric characteristics, material and structural characteristics, load characteristics, environmental conditions, maintenance and historical records, and remaining service life indicators;

[0241] The specific implementation method of step S8 includes the following steps:

[0242] S8.1. Set the spalling area geometry, geometry change rate, material and structural characteristics, load characteristics, environmental conditions, maintenance and history records, and remaining service life as evaluation indicators for pavement spalling conditions;

[0243] Furthermore, the geometric characteristics comprehensively consider the area of ​​the spalling region, the average spalling depth, the maximum spalling depth, the spalling position, etc., wherein the spalling position considers the distance from the vehicle wheel track line.

[0244] The rate of change of geometric characteristics is a comprehensive consideration of the rate of change of indicators such as the spalling area, average spalling depth, maximum spalling depth, and spalling position.

[0245] Material and structural properties are a comprehensive consideration of pavement material type, material aging and fatigue conditions, pavement thickness, interlayer bonding status, foundation conditions, etc.

[0246] The load characteristics are a comprehensive consideration of the number and type of vehicles, the proportion of heavy-loaded vehicles, the frequency of heavy-loaded vehicles passing, etc.

[0247] Environmental conditions comprehensively consider temperature changes, rain, snow, ice and other climatic conditions and their impact on road conditions, while also considering road drainage conditions and water damage to the road surface.

[0248] Maintenance and history records take into account the number and quality of pavement repairs and the time since the last major repair.

[0249] The remaining service life takes into account the remaining service life of the road.

[0250] S8.2. Use a high-definition camera to collect pavement image data of NR roads, transmit the pavement image data of each road to the digital twin platform of the digital twin model for pavement spalling identification constructed in step S7, intelligently identify the spalling area in the pavement image, and select the image containing the pavement spalling;

[0251] Furthermore, the geometric feature data collection is based on all the identified pavement spalling areas, and through on-site review and measurement, the data such as the spalling area, the average spalling depth, the maximum spalling depth, and the spalling position are measured as the evaluation data of the geometric features;

[0252] The geometric feature change speed collection is to collect the pavement spalling area information at different times, compare the data such as the spalling area, the average spalling depth, the maximum spalling depth, the spalling position, etc., and calculate the change speed of the spalling area, the average spalling depth, the maximum spalling depth, the spalling position and other indicators as the evaluation data of the geometric feature change speed;

[0253] Material and structural property data are obtained from pavement design documents, historical maintenance, construction records, road structure and material performance test records, including pavement material type, material aging and fatigue condition, pavement thickness, interlayer bonding status, and foundation condition data.

[0254] The load characteristic data are the number and type of vehicles traveling on the road, the proportion of heavy-loaded vehicles, and the frequency of heavy-loaded vehicles passing by.

[0255] Environmental condition data are climatic conditions such as temperature changes, rain, snow, ice, etc., as well as road drainage conditions and water damage data of the road surface.

[0256] Maintenance and history data are the number and quality of pavement repairs and the time since the last major repair.

[0257] Remaining service life data is the remaining service life data of the road;

[0258] The above data are obtained from public channels.

[0259] S8.3. The image containing the road peeling obtained in step S8.2 is scored using the expert scoring method, and the weight of each evaluation index set in step S8.1 relative to other evaluation indexes is scored to obtain the first p The evaluation index is relative to q Expert score of the weight of the evaluation indicators ;

[0260] Furthermore, the expert scoring method sets scoring rules. For example, the scoring range is 0.1-1, where 0.1 means very unimportant and 1 means extremely important. Ten experts are required to score the proportion of geometric characteristics relative to the rate of change of the set characteristics based on the geometric feature data and the geometric feature change rate data. The average score of the ten experts is taken to obtain ;

[0261] S8.4. Based on the weighted expert scores of all evaluation indicators obtained in step S3, establish a road evaluation matrix , the calculation formula is:

[0262] ;

[0263] Among them, geometric characteristics, rate of change of geometric characteristics, material and structural characteristics, load characteristics, environmental conditions, maintenance and history records, and remaining service life are 1, 2, 3, 4, 5, 6, and 7 respectively;

[0264] S8.5. Calculate the evaluation matrix of the road based on the evaluation matrix of the road The weight vector , the calculation formula is:

[0265] ;

[0266] Then calculate the maximum eigenvector of the road evaluation matrix , the calculation formula is:

[0267] ;

[0268] Then calculate the maximum eigenvalue corresponding to the maximum eigenvector of the road evaluation matrix , the calculation formula is:

[0269] ;

[0270] in, represents the largest eigenvector The sum of all elements in , Represents the weight vector The sum of all elements in ;

[0271] S8.6. Calculate all The maximum eigenvalue corresponding to the road , sort all the maximum eigenvalues ​​from large to small. The larger the maximum eigenvalue, the greater the severity of the spalling in the corresponding road, and obtain the quantitative evaluation results of the spalling degree in different roads;

[0272] S8.7. Based on the maximum eigenvalue Perform data consistency analysis and set the consistency index as , the calculation formula is:

[0273] ;

[0274] Set the consistency ratio to , the calculation formula is:

[0275] ;

[0276] in, RI It is the consistency index, which is obtained by querying the "Random Consistency Index RI Value Table";

[0277] By establishing a consistent ratio The relationship with 0.1 determines the consistency of the relative interrelationships of the seven factors:

[0278] when When , it indicates that the consistency is acceptable;

[0279] when When , it indicates that the consistency is unacceptable. At this time, return to step S8.4 to readjust the evaluation matrix .

[0280] Embodiment 2:

[0281] An electronic device includes a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of a pavement spalling state evaluation method described in Example 1 when executing the computer program.

[0282] The computer device of the present invention may be a device including a processor and a memory, such as a single chip microcomputer including a central processing unit. Furthermore, the processor is used to implement the steps of the above-mentioned recommendation method based on CREO software that can modify the recommendation data driven by the relationship when executing the computer program stored in the memory.

[0283] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0284] The memory may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory may include a high-speed random access memory, and may also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0285] Embodiment 3:

[0286] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements a method for evaluating a pavement spalling state as described in Example 1.

[0287] The computer-readable storage medium of the present invention can be any form of storage medium that can be read by a processor of a computer device, including but not limited to non-volatile memory, volatile memory, ferroelectric memory, etc. A computer program is stored on the computer-readable storage medium. When the processor of the computer device reads and executes the computer program stored in the memory, the steps of the above-mentioned modeling method of modifiable relationship-driven modeling data based on CREO software can be implemented.

[0288] The computer program includes computer program code, which may be in source code form, object code form, executable file or some intermediate form, etc. The computer readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0289] The key points and intended protection points of the present invention are:

[0290] (1) A method for constructing a basic model and an output layer loss function of an enhanced model for intelligent identification of pavement spalling.

[0291] (2) A method for constructing inter-layer loss functions of the basic model and enhanced model for intelligent identification of pavement spalling.

[0292] (3) A road digital twin model construction and pavement spalling update method.

[0293] (4) A pavement spalling condition evaluation method based on digital twin.

[0294] It should be noted that relational terms such as "first" and "second" are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article 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, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0295] Although the present application has been described above with reference to specific embodiments, various modifications may be made thereto and parts thereof may be replaced with equivalents without departing from the scope of the present application. In particular, as long as there is no structural conflict, the various features in the specific embodiments disclosed in the present application may be used in combination with each other in any manner, and the fact that these combinations are not exhaustively described in this specification is only for the sake of omitting space and saving resources. Therefore, the present application is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A method for evaluating pavement spalling conditions, characterized in that: The steps include: S1. Use high-definition cameras to collect road images; S2. For the road surface images collected in step S1, images of road surface spalling, transverse cracks, fissures, and rutting are selected and manually annotated to construct a data set for a mathematical model for identifying road surface spalling; S3. Construct a mathematical model for pavement spalling identification, including a basic model and an enhanced model; S4. Construct the output layer loss function of the basic model and the enhanced model in the mathematical model for pavement spalling identification; S4.

1. Construct the loss function of the basic model A and train the basic model A; S4.1.

1. The loss function of the basic model A is calculated as: Among them, La is the loss function of the basic model A, Na is the total number of data samples in the basic model A, and y ia is the true label of the iath sample, p ia Predict the probability that the iath sample is class 1 based on the basic model A; The training set obtained in step S2 is input into the basic model A, and the weights and biases of the neural network units are calculated through the error back propagation algorithm to complete the training of the basic model A and obtain the trained basic model A; S4.1.

2. The node output values ​​of the four output units corresponding to the basic model A are a1, a2, a3, and a4. The node output values ​​of the four output units are initially converted to obtain the output value a i The corresponding probability value f ai The calculation formula is: Then the node output values ​​of the four output units are further optimized and converted to obtain the output value a i The corresponding enhancement probability value g ai The calculation formula is: Among them, G is the enhancement coefficient, and then the actual disease corresponding to each node is marked as h ai ; S4.1.

3. Use g ai The training set obtained in step S2 is combined with the error chain derivation rule to train the basic model A. After the training is completed, the output values ​​of the four types of diseases corresponding to the basic model A are obtained, which are recorded as j a1 , j a2 , j a3 , j a4 ; S4.

2. Construct the loss function of the reinforcement model B and train the reinforcement model B; S4.2.

1. Input the training set obtained in step S2 into the enhanced model B, calculate the weights and biases of the neural network units through the error back propagation algorithm, complete the training of the enhanced model B, and obtain the trained enhanced model B; The loss function of the enhanced model B is calculated as: Among them, Lb is the loss function of the enhanced model B; Nb is the total number of data samples in the enhanced model B; y ib is the true label of the ib-th sample, p ib Predict the probability that the ibth sample is category 1 for the enhanced model B; S4.2.

2. Strengthen the node output values ​​of the four output units corresponding to model B as b1, b2, b3, b4. Perform an initial transformation on the node output values ​​of the four output units to obtain the output value b i The corresponding probability value f bi The calculation formula is: Then the node output values ​​of the four output units are further optimized and converted to obtain the output value b i The corresponding enhancement probability value g bi The calculation formula is: S4.2.

3. Use g bi The training set obtained in step S2 is combined with the error chain derivation rule to train the enhanced model B. After the training is completed, the disease output value corresponding to the enhanced model B is j bi , among the 4 nodes in the output layer, the actual disease corresponding to each node is labeled as h bi , i=1,2,3,4; S4.

3. Construct output layer loss functions of the base model and enhanced model; Based on j bi and g ai , construct the loss function L between the enhanced model B and the output layer of the basic model A BA , the calculation formula is: Based on f ai and h ai , construct the output value a i The corresponding probability value f ai The loss function L of the base model A AA , the calculation formula is: Based on L BA and L AA , establish the output layer loss function L of the basic model and the enhanced model BAA , the calculation formula is: L BAA =c BA ·L BA +g AA ·L AA Among them, γ BA For L BA The weight coefficient, γ AA For L AA The weight coefficient of S5. Construct the inter-layer loss function of the basic model and the enhanced model in the mathematical model for pavement spalling identification; S6. Using the data set obtained in step S2 to train the mathematical model for pavement peeling identification, a trained mathematical model for pavement peeling identification is obtained; S7. Perform digital twin modeling based on the trained mathematical model for pavement spalling identification obtained in step S6 to obtain a digital twin model for pavement spalling identification; S8. Based on the digital twin model for pavement spalling identification obtained in step S7, the pavement spalling status is evaluated based on the geometric characteristics of the spalling area, the speed of change of geometric characteristics, material and structural characteristics, load characteristics, environmental conditions, maintenance and historical records, and remaining service life indicators.

2. A method for evaluating pavement spalling state according to claim 1, characterized in that: The specific implementation method of step S2 includes the following steps: S2.

1. From the road surface images, manually select all images containing spalling, transverse cracks, fissures, and rutting; S2.

2. By manually labeling, the spalling area and the other three diseases are selected in the pavement image selected in step S1.1, and the spalling is marked as Spalling. The other three diseases, transverse cracks, cracks, and rutting, are marked as other1, other2, and other3 respectively, to construct a data set for the mathematical model of pavement spalling recognition; S2.

3. Divide the data set of the mathematical model used for pavement spalling identification into a training set and a test set in a ratio of 8:

2.

3. A method for evaluating pavement spalling state according to claim 2, characterized in that: The specific implementation method of step S3 includes the following steps: S3.

1. Build a basic model. The basic model A consists of 1 input layer, 5 convolutional layers, 5 activation layers, 4 pooling layers, 1 global average pooling layer, and 1 output layer. S3.

2. Construct an enhanced model. The enhanced model B consists of 1 input layer, 10 convolutional layers, 10 activation layers, 5 pooling layers, 1 global average pooling layer, and 1 output layer.

4. A method for evaluating pavement spalling state according to claim 3, characterized in that: The specific implementation method of step S5 includes the following steps: S5.

1. Assume that there are a small model a and a large model b, establish the second, third, fourth layers and output layer of the small model a, and establish connections with the fourth, sixth, tenth layers and output layer of the large model b respectively; S5.

2. For the second layer in the small model and the fourth layer in the large model b, the output of the second layer of the small model a is recorded as Ma2, and the corresponding width × length × number of channels is na2 × pa2 × qa2; the output of the fourth layer of the large model b is recorded as Mb4, and the corresponding width × length × number of channels is nb4 × pb4 × qb4; For size comparison, when qa2<qb4, a convolution operation is added to the second layer output of the small model, the convolution kernel size is 1×1, the number of convolution kernels is qb4, and the output is Ma 2n At this time, Ma 2n The corresponding output values ​​of each channel are ki is the channel number, and the output value of each channel corresponding to Mb4 is Establish the loss function between the second layer of the small model a and the fourth layer of the large model b based on qa2<qb4 The calculation formula is: When qa2 ≥ qb4, a convolution operation is added to the output of the fourth layer of the large model b, with a convolution kernel size of 1×1 and a number of convolution kernels of qa2, and the output is Mb 4n ; At this time, the output values ​​of each channel corresponding to Ma2 are: Mb 4n The corresponding output values ​​of each channel are Establish the loss function between the second layer of small model a and the fourth layer of large model b when qa2≥qb4 The calculation formula is: S5.

3. Establish a relationship between the third layer in the small model and the sixth layer in the large model b, and obtain the loss function between the third layer of the small model a and the sixth layer of the large model b S5.

4. Establish a relationship between the fourth layer in the small model and the tenth layer in the large model b, and obtain the loss function between the fourth layer of the small model a and the tenth layer of the large model b S5.

5. When the small model is the basic model A, the second, third, and fourth layers in the small model a are the output of pooling layer 1, the output of pooling layer 2, and the output of pooling layer 4 of the basic model A, respectively; when the large model is the enhanced model B, the fourth, sixth, and tenth layers in the large model b are the output of pooling layer 2, the output of pooling layer 3, and the output of pooling layer 5 of the enhanced model B, respectively.

5. A method for evaluating pavement spalling state according to claim 4, characterized in that: The specific implementation method of step S6 includes the following steps: S6.

1. Based on the loss function between the output layer of the enhanced model B and the inter-layer loss function of the basic model A, define the total loss function L between the two models total , the calculation formula is: Among them, α 2-4 for The corresponding weight coefficient, α 3-6 for The corresponding weight coefficient, α 4-10 for The corresponding weight coefficient; S6.

2. Using the data set obtained in step S2, based on the chain rule, the enhanced model B and the basic model A are trained by error back propagation. When the overall loss function L total When is the minimum, the corresponding model parameters are optimal, and the trained basic model A is the trained mathematical model for pavement spalling identification.

6. A method for evaluating pavement spalling state according to claim 5, characterized in that: The specific implementation method of step S7 includes the following steps: S7.

1. Export the trained mathematical model for pavement spalling identification obtained in step S6 as a deployment file for independent services, including one of a pt file, a pth file, a json file, and a txt file; S7.

2. Construct a road information model; S7.2.

1. Collect information on the location, length, width, and layout of roads; S7.2.

2. The data obtained in step S7.2.1 is stored in a database, and a basic three-dimensional model of the road is constructed, wherein the basic three-dimensional model of the road includes geometric information of the road; S7.2.

3. Based on step S7.2.2, material properties, design specifications, and historical maintenance data are added to the basic three-dimensional model of the road to obtain a road information model; S7.2.

4. Export the road information model to a format file supported by the Web application, including a json file and a js file; S7.

3. In the digital twin platform, import the deployment file of the trained mathematical model for pavement spalling identification obtained in step S7.1, and import the road information model format file obtained in step S7.2; S7.

4. Use Python to construct an API interface between image data and the trained mathematical model for pavement spalling identification, which is used for intelligent identification of pavement spalling. Use Python to construct an API interface between the trained mathematical model for pavement spalling identification and the road information model, and complete the construction of a digital twin model for pavement spalling identification.

7. A method for evaluating pavement spalling state according to claim 6, characterized in that: The specific implementation method of step S8 includes the following steps: S8.

1. Set the geometric characteristics of the spalling area, the speed of change of geometric characteristics, material and structural characteristics, load characteristics, environmental conditions, maintenance and historical records, and remaining service life as the evaluation indicators of pavement spalling status; S8.

2. Use a high-definition camera to collect road surface image data of NR roads, transmit the road surface image data of each road to the digital twin platform of the digital twin model for road surface spalling identification constructed in step S7, intelligently identify the spalling area in the road surface image, and select the image containing the road surface spalling; S8.

3. The image containing road peeling obtained in step S8.2 is scored using the expert scoring method, and the weight of each evaluation index set in step S8.1 relative to other evaluation indexes is scored to obtain the expert score jz of the weight of the pth evaluation index relative to the qth evaluation index pq ; S8.

4. Based on the weighted expert scores of all evaluation indicators obtained in step S3, the road evaluation matrix JZ is established, and the calculation formula is: Among them, geometric characteristics, rate of change of geometric characteristics, material and structural characteristics, load characteristics, environmental conditions, maintenance and history records, and remaining service life are 1, 2, 3, 4, 5, 6, and 7 respectively; S8.

5. Based on the evaluation matrix of the road, calculate the weight vector QZ of the evaluation matrix JZ of the road, and the calculation formula is: Then calculate the maximum eigenvector TZ of the road evaluation matrix, the calculation formula is: Then calculate the maximum eigenvalue TZ corresponding to the maximum eigenvector of the road evaluation matrix max , the calculation formula is: Among them, SumTZ represents the sum of all elements in the maximum eigenvector TZ, and SumQZ represents the sum of all elements in the weight vector QZ; S8.

6. Calculate the maximum eigenvalue corresponding to all NR roads All the maximum eigenvalues ​​are sorted in descending order. The larger the maximum eigenvalue, the greater the severity of the spalling in the corresponding road, and the quantitative evaluation results of the spalling degree in different roads are obtained. S8.

7. Based on the maximum eigenvalue TZ max Perform data consistency analysis, set the consistency index to CS, and the calculation formula is: Set the consistency ratio to CSb, and the calculation formula is: Among them, RI is the consistency index, which is obtained by querying the "Random Consistency Index RI Value Table"; By establishing the relationship between the consistency ratio CSb and 0.1, the consistency of the relative relationships of the seven factors was determined: When CSb≤0.1, it indicates that the consistency is acceptable; When CSb>0.1, it indicates that the consistency is unacceptable. At this time, return to step S8.4 to readjust the evaluation matrix JZ.

8. An electronic device, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of a method for evaluating a pavement spalling state as described in any one of claims 1 to 7 when executing the computer program.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, a method for evaluating a pavement spalling state as described in any one of claims 1 to 7 is implemented.

Images