A method for analyzing factors of road bump formation, electronic equipment and storage medium

Abstract

A method for analyzing factors of road bump formation, an electronic device and a storage medium, belonging to the technical field of road engineering. In order to accurately analyze factors affecting road bumps, the present invention includes constructing a data set for rapid identification of road bumps; constructing a basic model, a first auxiliary model, a second auxiliary model and a reinforcement model for rapid identification of road bumps; based on the loss function between the output layers of the basic model, the first auxiliary model, the second auxiliary model and the reinforcement model for rapid identification of road bumps; constructing an inter-layer loss function of the reinforcement model and the basic model for rapid identification of road bumps; then the overall loss function of the rapid identification model of road bumps; improving the obtained rapid identification model of road bumps to obtain a lightweight rapid identification model of road bumps; constructing a digital twin model; and constructing a method for analyzing factors of road bump formation. The present invention can prevent the occurrence of road bumps during the design and construction stages, and effectively plan and execute maintenance work.

Description

Technical Field

[0001] The invention belongs to the technical field of road engineering, and in particular relates to a method for analyzing factors of road surface congestion, electronic equipment and storage medium. Background Art

[0002] Pavement congestion is a common road structural disease, usually manifested as irregular bulges or wavy protrusions on the road surface. This phenomenon is mostly caused by uneven materials under the road surface, poor drainage, instability of the base or subbase, and frequent heavy-loaded vehicles passing by.

[0003] Road bumps have a significant negative impact on road use and traffic. Road bumps can cause uneven driving, affecting driving comfort and vehicle stability. Especially for vehicles traveling at high speeds, road bumps can make vehicle control difficult and increase the risk of traffic accidents. Road bumps can also cause excessive wear on vehicles, especially putting greater pressure on the vehicle's suspension system and tires. In public transportation and freight transportation, this road condition can also cause transportation delays, affecting overall traffic efficiency and economic costs.

[0004] The invention patent with application number 201711445923.8 and invention name "A method for identifying multi-dimensional characteristic indicators of asphalt pavement bumps" (201711445923.8) proposes a method for accurately scanning and extracting features of asphalt pavement bumps using three-dimensional line laser detection equipment. The three-dimensional surface contour of the bump is reconstructed by acquiring and smoothing point cloud data, a three-dimensional coordinate system is established, and multi-dimensional characteristic indicators are extracted from it to analyze the severity and potential impact of the bump disease.

[0005] However, the above invention does not consider the analysis method of factors affecting road bumps, and cannot accurately identify and evaluate the various factors that cause road bumps, which will make it impossible for road maintenance teams to effectively prevent or promptly repair such diseases. This situation will not only shorten the service life of the road, but may also lead to increased maintenance costs. In addition, the safety of road users will also be threatened by the uneven road surface and potential risks of high-speed driving, which may lead to traffic accidents in serious cases. Summary of the invention

[0006] The problem to be solved by the present invention is to accurately analyze the factors affecting road congestion, and to propose a method for analyzing factors forming road congestion, an electronic device and a storage medium.

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

[0008] A method for analyzing factors of road bump formation includes the following steps:

[0009] S1. Using a vehicle-mounted three-dimensional structured light camera to collect road images, and then compressing the collected road images to obtain a compressed road image;

[0010] S2. Perform similar disease identification on the compressed road surface image obtained in step S1, extract all road surface images containing the bag and the disease type most similar to the image features of the bag, and construct a data set for rapid identification of road bag;

[0011] S3. Constructing a basic model, a first auxiliary model, a second auxiliary model and an enhanced model for rapid identification of road congestion; then constructing a loss function between the output layers of the basic model, the first auxiliary model, the second auxiliary model and the enhanced model for rapid identification of road congestion;

[0012] S4. Constructing the inter-layer loss function of the enhanced model and the basic model for rapid identification of road bumps; then constructing the overall loss function of the rapid identification model of road bumps;

[0013] S5. Using the data set obtained in step S2 for rapid identification of road congestion, the constructed basic model, the first auxiliary model, the second auxiliary model and the enhanced model are trained to obtain a trained basic model as a road congestion rapid identification model;

[0014] S6. The road congestion rapid identification model obtained in step S5 is improved to obtain a lightweight road congestion rapid identification model;

[0015] S7. Constructing a digital twin model based on the lightweight road congestion rapid identification model obtained in step S6;

[0016] S8. Based on the digital twin model obtained in step S7, a method for analyzing factors causing road congestion is constructed.

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

[0018] S2.1. Similar diseases of road bumps include longitudinal cracks, transverse cracks, cracks, rutting, potholes, spalling, oiling, subsidence, and undercutting;

[0019] S2.2. Manually select the processed road surface images obtained in step S1, select all road surface images including road bumps and similar diseases, and select the corresponding disease areas, and mark the road bumps, longitudinal cracks, transverse cracks, cracks, rutting, potholes, peeling, oil spills, subsidence, and edge gnawing areas in the road surface images in turn. PHO 1. PHO 2. PHO 3. PHO 4. PHO 5. PHO 6. PHO 7. PHO 8. PHO 9. PHO 10;

[0020] S2.3. Divide the pixel grayscale value 0~255 into 16 grayscale intervals, denoted as , =1,2,3,…,16;

[0021] S2.4. Count the number of pixels corresponding to each gray level in the road congestion Then, the number of pixels corresponding to each gray level of similar diseases of road bumps, such as longitudinal cracks, transverse cracks, cracks, rutting, potholes, peeling, oiling, subsidence, and gnawing, is counted. , , , , , , , , ;

[0022] S2.5. Standardize the number of pixels corresponding to each gray level to obtain the standard number of pixels, where the standard number of pixels corresponding to the road congestion is , the calculation formula is:

[0023] ;

[0024] The standard number of pixels corresponding to longitudinal cracks, transverse cracks, cracks, rutting, potholes, spalling, oil spills, subsidence, and edge gnawing is , , , , , , , , ;

[0025] S2.6. As a benchmark, the relationship function between pavement bumps and other diseases is established in turn, and the calculation formula is:

[0026] ;

[0027] Among them, ii represents the serial number of other similar diseases, ii=2,3,4,5…10, and we get , , , , , , , , They represent the relationship functions between road surface bumps and longitudinal cracks, transverse cracks, cracks, rutting, potholes, spalling, oiling, subsidence, and undercutting respectively;

[0028] S2.7. Comparison , , , , , , , , , and sort them in descending order, select the disease types corresponding to the top three relationship functions as the disease types most similar to the image features of road surface bumps, and then select all road surface images containing road surface bumps and the disease types most similar to the image features of road surface bumps;

[0029] S2.8. By manually labeling, the road surface images obtained in step S2.8 that contain road bumps and the types of defects that are most similar to the image features of the road bumps are framed to select the road pothole area and the other three defects, and the road bumps are labeled as upheaval, and the other three defects are labeled as other1, other2, and other3, respectively, to obtain a data set for rapid identification of road bumps;

[0030] S2.9. Divide the dataset for rapid identification of road congestion into a training set and a test set in a ratio of 8:2.

[0031] Furthermore, the specific implementation method of step S3 includes the following steps:

[0032] S3.1. Constructing a basic model for rapid identification of road congestion:

[0033] S3.1.1. The basic model A consists of 1 input layer, 9 convolutional layers, 9 activation layers, 3 pooling layers, 1 merging layer, 1 upsampling layer, 1 path aggregation network layer PANet, 1 pooling splicing layer, 1 spatial pyramid pooling layer SPP, and 1 output layer:

[0034] S3.1.2. Input the training set obtained in step S2 into the basic model A, calculate the weights and biases of the neural network units through the error back propagation algorithm, complete the training of the basic model A, and obtain the trained basic model A;

[0035] The calculation formula of the loss function of the basic model A is:

[0036] ;

[0037] in, is the loss function of the basic model A; is the total number of data samples in the basic model A; For the it The true labels of samples, Based on the prediction of model A it The probability that a sample is class 1;

[0038] S3.1.3. The node output values ​​of the four output units corresponding to the basic model A are a 1. a 2. a 3. a 4. Perform initial conversion on the node output values ​​of the four output units to obtain the output values The corresponding probability value The calculation formula is: ;

[0039] 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:

[0040] ;

[0041] in, G is the enhancement coefficient, and then the actual disease corresponding to each node is marked as ;

[0042] S3.1.4. 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 , , , ;

[0043] S3.2. Constructing the first auxiliary model for rapid identification of road congestion:

[0044] S3.2.1. The first auxiliary model B consists of 1 input layer, 9 convolutional layers, 9 activation layers, 5 pooling layers, 1 merging layer, 1 upsampling layer, 1 PANet layer, 1 pooling splicing layer, 1 SPP layer, and 1 output layer:

[0045] S3.2.2. Input the training set obtained in step S2 into the first auxiliary model B, calculate the weights and biases of the neural network units through the error back propagation algorithm, complete the training of the first auxiliary model B, and obtain the trained first auxiliary model B;

[0046] The calculation formula of the loss function of the first auxiliary model B is:

[0047] ;

[0048] in, is the loss function of the first auxiliary model B; is the total number of data samples in the first auxiliary model B; For the one The true labels of samples, The first auxiliary model B predicts one The probability that a sample is class 1;

[0049] S3.2.3. The node output values ​​of the four output units corresponding to the first auxiliary model B are b 1. b 2. b 3. b 4. Perform initial conversion on the node output values ​​of the four output units to obtain the output values b i The corresponding probability value The calculation formula is:

[0050] ;

[0051] 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 The calculation formula is:

[0052] ;

[0053] S3.2.4. Adoption The training set obtained in step S2 is combined with the error chain derivation rule to train the first auxiliary model B. After the training is completed, the disease output value corresponding to the first auxiliary model B is obtained as , among the 4 nodes in the output layer, the actual disease corresponding to each node is marked as , i =1,2,3,4;

[0054] S3.3. Constructing the second auxiliary model for rapid identification of road congestion:

[0055] S3.3.1. The second auxiliary model C consists of 1 input layer, 12 convolutional layers, 12 activation layers, 5 pooling layers, 1 merging layer, 1 upsampling layer, 1 PANet layer, 1 pooling splicing layer, 1 SPP layer, and 1 output layer:

[0056] S3.3.2. Input the training set obtained in step S2 into the second auxiliary model C, calculate the weights and biases of the neural network units through the error back propagation algorithm, complete the training of the second auxiliary model C, and obtain the trained second auxiliary model C;

[0057] The calculation formula of the loss function of the second auxiliary model C is:

[0058] ;

[0059] in, is the loss function of the second auxiliary model C; is the total number of data samples in the second auxiliary model C; For the ic The true labels of samples, The second auxiliary model C predicts ic The probability that a sample is class 1;

[0060] S3.3.3. The node output values ​​of the four output units corresponding to the second auxiliary model C are , the node output values ​​of the four output units are initially converted to obtain The corresponding probability value The calculation formula is:

[0061] ;

[0062] 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:

[0063] ;

[0064] in, G is the enhancement coefficient, and then the actual disease corresponding to each node is marked as ;

[0065] S3.3.4. Adoption The training set obtained in step S2 is combined with the error chain derivation rule to train the second auxiliary model C. After the training is completed, the output values ​​of the four types of diseases corresponding to the second auxiliary model C are obtained as follows: ;

[0066] S3.4. Construct an enhanced model for rapid identification of road congestion:

[0067] S3.4.1. The enhanced model D consists of 1 input layer, 14 convolutional layers, 14 activation layers, 5 pooling layers, 1 merging layer, 1 upsampling layer, 1 PANet layer, 1 pooling splicing layer, 1 SPP layer, and 1 output layer:

[0068] S3.4.2. Input the training set obtained in step S2 into the enhanced model D, calculate the weights and biases of the neural network units through the error back propagation algorithm, complete the training of the enhanced model D, and obtain the trained enhanced model D;

[0069] The loss function of the enhanced model D is calculated as:

[0070] ;

[0071] in, is the loss function of the enhanced model D; is the total number of data samples in the reinforcement model D; For the id The true labels of samples, To strengthen the model D to predict the id The probability that a sample is class 1;

[0072] S3.4.3. The node output values ​​of the four output units corresponding to the enhanced model D are , the node output values ​​of the four output units are initially converted to obtain The corresponding probability value The calculation formula is:

[0073] ;

[0074] 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:

[0075] ;

[0076] in, G is the enhancement coefficient, and then the actual disease corresponding to each node is marked as ;

[0077] S3.4.4. Adoption The training set obtained in step S2 is combined with the error chain derivation rule to train the enhanced model D. After the training is completed, the output values ​​of the four types of diseases corresponding to the enhanced model D are obtained as follows: ;

[0078] S3.5. Based on and , establish the loss function between the first auxiliary model B and the output layer of the basic model A, and the calculation formula is:

[0079] ;

[0080] based on and , construct the output value The corresponding probability value The loss function of the base model A , the calculation formula is:

[0081] ;

[0082] based on and , establish the relationship loss function between the output layer of the first auxiliary model B and the basic model A , the calculation formula is:

[0083]

[0084] in, for The weight coefficient of for The weight coefficient of

[0085] S3.6. Establish the relationship loss function between the output layer of the second auxiliary model C and the first auxiliary model B , the expression is:

[0086] ;

[0087] in, for The weight coefficient of for The weight coefficient of Based on and The loss function of Based on and The loss function of

[0088] The calculation formula is:

[0089] ;

[0090] The calculation formula is:

[0091] ;

[0092] S3.7. Establish the relationship loss function between the output layer of the second auxiliary model C and the basic model A , the expression is:

[0093] ;

[0094] in, for The weight coefficient of for The weight coefficient of Based on and The loss function is Based on and The loss function of

[0095] The calculation formula is:

[0096] ;

[0097] The calculation formula is:

[0098] ;

[0099] S3.8. Establish the relationship loss function between the output layer of the enhanced model D and the base model A , the expression is:

[0100] ;

[0101] in, for The weight coefficient of for The weight coefficient of Based on and The loss function of

[0102] The calculation formula is:

[0103] ;

[0104] S3.9. Establish the relationship loss function between the output layer of the reinforcement model D and the second auxiliary model C , the expression is:

[0105] ;

[0106] in, for The weight coefficient of for The weight coefficient of Based on and The loss function is Based on and The loss function of

[0107] The calculation formula is:

[0108] ;

[0109] The calculation formula is:

[0110] .

[0111] Furthermore, the specific implementation method of step S4 includes the following steps:

[0112] S4.1. Assuming a small model a and large models d , build a small model a The second, third, fourth and output layers in the large model are d Establish connections between the fourth, sixth, tenth layers and the output layer;

[0113] S4.2. For the second layer in the small model and the large model d 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 d The output of the fourth layer is recorded as , the corresponding size of width × length × number of channels is ;

[0114] 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 , k i is the channel number, The corresponding output values ​​of each channel are ;

[0115] Build on Small model a The second floor with the large modeld The loss function between the fourth layers , the calculation formula is:

[0116] ;

[0117] when When in the large model d 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 ;

[0118] Establish Small model a The second floor with the large model d The loss function between the fourth layers , the calculation formula is:

[0119] ;

[0120] S4.3. For the third layer in the small model and the large model b The sixth layer in the relationship is established to obtain a small model a The third floor and the large model b Loss function between the sixth layers ;

[0121] S4.4. For the fourth layer in the small model and the large model b The tenth layer in the relationship is established to obtain a small model a Fourth floor and large model b Loss function between the tenth layer ;

[0122] S4.5. When the small model is the base model A, the small model a 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 D, the large model d The fourth, sixth, and tenth layers in are the output of pooling layer 2, the output of pooling layer 3, and the output of pooling layer 5 of the enhanced model D respectively;

[0123] S4.6. Based on the loss function between the output layer of the enhanced model D and the inter-layer loss function of the base model A, define the total loss function between the two models , the calculation formula is:

[0124] ;

[0125] in, for The corresponding weight coefficient is, for The corresponding weight coefficient is, for The corresponding weight coefficient.

[0126] Furthermore, the specific implementation method of step S5 is to use the data set obtained in step S2 based on the chain derivative rule to train the basic model, the first auxiliary model, the second auxiliary model and the enhanced model through error back propagation. When the overall loss function When is minimum, the corresponding model parameters are optimal, and the trained basic model is obtained as a road congestion rapid recognition model.

[0127] Furthermore, the specific implementation method of step S6 includes the following steps:

[0128] S6.1. For the road congestion rapid identification model obtained in step S5, set the enhancement coefficient G The values ​​are: G =1,3,5,7,9,11,15,20,30,40,50;

[0129] S6.2. Based on different enhancement factors G 11 working conditions are obtained, and 11 models are obtained by training the road congestion rapid recognition model. The corresponding accuracy rates of the 11 road congestion rapid recognition models are as follows: , i=1,2,3,…,11, the calculation formula is:

[0130] ;

[0131] in, Represents the first i The number of samples that a model predicts as having a bag and actually has a bag when identifying the data in the test set; Represents the first i The number of samples that a model predicts as having a packet but is not actually having a packet when identifying the data in the test set;

[0132] Comparison The maximum value in ;

[0133] The corresponding road surface congestion rapid recognition model is the final lightweight road surface congestion rapid recognition model.

[0134] Furthermore, the specific implementation method of step S7 includes the following steps:

[0135] S7.1. The lightweight road congestion rapid identification model obtained in step S6 is exported as a deployment file for independent services, including one of a pt file, a pth file, a json file, and a txt file;

[0136] S7.2. Construct a road information model;

[0137] S7.2.1. Collect information on the location, length, width, and layout of roads;

[0138] 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;

[0139] 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;

[0140] 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;

[0141] S7.3. In the digital twin platform, import the deployment file of the lightweight road congestion rapid identification model obtained in step S7.1, and import the road information model format file obtained in step S7.2;

[0142] S7.4. Use Python to construct an API interface between image data and a lightweight road bump rapid identification model for intelligent rutting identification. Use Python to construct an API interface between a lightweight road bump rapid identification model and a road information model to complete the construction of a digital twin model.

[0143] Furthermore, the specific implementation method of step S8 includes the following steps:

[0144] S8.1. Use a 3D structured light camera to collect road surface data and upload it to the digital twin model established in step S7. Then use the lightweight road bump rapid identification model deployed in the system to intelligently identify the road bump area in the road image, select all images containing road bumps, conduct on-site review of the bumps, and measure the height GD of the road bumps.

[0145] S8.2. The factors affecting the formation of congestion include pavement material aging, ambient temperature, water damage, heavy-loaded vehicles, traffic load, pavement thickness, pavement temperature, and pavement compressive strength; pavement material aging is quantified by asphalt oxide content, water damage is quantified by pavement water seepage, heavy-loaded vehicles are quantified by the number of heavy-loaded vehicles, and traffic load is quantified by traffic volume;

[0146] S8.3. Select 10 time nodes t1, t2, t3, ... t10, collect data at each of the 10 time nodes, and obtain the values ​​of the corresponding quantitative indicators as follows:

[0147] Asphalt oxide content at 10 time points: LH1, LH2, LH3, LH4, LH5, LH6, LH7, LH8, LH9, LH10;

[0148] Ambient temperature at 10 time points: HW1, HW2, HW3, HW4, HW5, HW6, HW7, HW8, HW9, HW10;

[0149] Pavement seepage at 10 time points: SS1, SS2, SS3, SS4, SS5, SS6, SS7, SS8, SS9, SS10;

[0150] The number of heavy-loaded vehicles at 10 time points: ZZ1, ZZ2, ZZ3, ZZ4, ZZ5, ZZ6, ZZ7, ZZ8, ZZ9, ZZ10;

[0151] Traffic volume at 10 time nodes: JL1, JL2, JL3, JL4, JL5, JL6, JL7, JL8, JL9, JL10;

[0152] Pavement thickness at 10 time points: HD1, HD2, HD3, HD4, HD5, HD6, HD7, HD8, HD9, HD10;

[0153] Pavement temperature at 10 time points: LW1, LW2, LW3, LW4, LW5, LW6, LW7, LW8, LW9, LW10;

[0154] Pavement compressive strength at 10 time points: KY1, KY2, KY3, KY4, KY5, KY6, KY7, KY8, KY9, KY10;

[0155] Road congestion height at 10 time nodes: GD1, GD2, GD3, GD4, GD5, GD6, GD7, GD8, GD9, GD10;

[0156] S8.4. Calculate the average value of each quantitative index at 10 time points respectively, and get the average value of asphalt oxide content: The average ambient temperature is The average pavement seepage volume is The average number of heavy-loaded vehicles is The average traffic volume is The average pavement thickness is The average road surface temperature is The average compressive strength of the pavement is The average height of road bump is ;

[0157] S8.5. Calculate the correlation strength coefficient between each quantitative index and the road bump height , the calculation formula is as follows:

[0158] ;

[0159] in, For the I The quantitative indicator value at each time node, For the I The height of the road bump at each time node; ZB represents the quantitative evaluation index, including asphalt oxide content, ambient temperature, road water seepage, number of heavy-loaded vehicles, traffic volume, road thickness, road temperature, and road compressive strength; that is, the correlation strength coefficients between asphalt oxide content, ambient temperature, road water seepage, number of heavy-loaded vehicles, traffic volume, road thickness, road temperature, road compressive strength and road bump height are: , , , , , , , ;

[0160] S8.6. Arrange all the correlation strength coefficients obtained in step S8.5 in descending order and re-label them as TZN1 to TZN8, where the influencing factor of the congestion formation corresponding to TZN1 is the influencing factor with the greatest influence on the road congestion;

[0161] S8.7. Defining thresholds YZ ,calculate , it is Start counting from 1 and gradually increase. it is Value Satisfaction When , stop the calculation. At this time, it is The road congestion influencing factors corresponding to the values ​​are the main influencing factors that cause road congestion.

[0162] An electronic device includes a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of a method for analyzing factors of road congestion when executing the computer program.

[0163] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements a method for analyzing factors of road congestion.

[0164] Beneficial effects of the present invention:

[0165] The present invention discloses a method for analyzing factors of road surface bump formation, and proposes a method for determining similar diseases, thereby reducing the model's misjudgment of bumps.

[0166] The present invention provides a method for analyzing factors of road bump formation, and proposes a method for constructing a basic model, an auxiliary model, and an output layer loss function of an enhanced model to improve the model's ability to extract bump features.

[0167] The present invention provides a method for analyzing factors of road bump formation and proposes a lightweight intelligent road bump recognition model to improve the learning efficiency and performance of the model.

[0168] The present invention provides a method for analyzing factors of road bump formation, and proposes a basic model improvement method based on an enhancement coefficient to improve the generalization ability of the model.

[0169] The present invention discloses a method for analyzing factors that cause road bumps, and proposes a method for constructing a road digital twin model and updating road bump damage, thereby realizing the construction of a digital twin model and accurate updating of road bump areas.

[0170] The present invention provides a method for analyzing factors affecting road bumps, and proposes a method for analyzing factors affecting road bumps based on digital twins, which can help engineers prevent the occurrence of road bumps during the design and construction stages, plan and execute maintenance work more effectively, and solve problems in a targeted manner, thereby improving the overall quality and service life of roads. This analysis method also helps to improve road safety and protect the lives of road users by reducing the risk of accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0171] Figure 1 The present invention is a flow chart of a method for analyzing factors of road congestion. DETAILED DESCRIPTION

[0172] 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.

[0173] 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.

[0174] 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:

[0175] Embodiment 1:

[0176] A method for analyzing factors of road bump formation includes the following steps:

[0177] S1. Using a vehicle-mounted three-dimensional structured light camera to collect road images, and then compressing the collected road images to obtain a compressed road image;

[0178] S1.1. Use a vehicle-mounted 3D structured light camera to collect road images, and control the vehicle speed within 70 km / h during the collection process;

[0179] S1.2. compressing the collected road surface image by using Huffman coding to obtain a compressed road surface image;

[0180] S1.3. The compressed road surface image is projected on a plane, and the projected road surface image is the compressed road surface image;

[0181] S2. Perform similar disease identification on the compressed road surface image obtained in step S1, extract all road surface images containing the bag and the disease type most similar to the image features of the bag, and construct a data set for rapid identification of road bag;

[0182] Furthermore, the specific implementation method of step S2 includes the following steps:

[0183] S2.1. Similar diseases of road bumps include longitudinal cracks, transverse cracks, cracks, rutting, potholes, spalling, oiling, subsidence, and undercutting;

[0184] S2.2. Manually select the processed road surface images obtained in step S1, select all road surface images including road bumps and similar diseases, and select the corresponding disease areas, and mark the road bumps, longitudinal cracks, transverse cracks, cracks, rutting, potholes, peeling, oil spills, subsidence, and edge gnawing areas in the road surface images in turn. PHO 1. PHO 2. PHO 3. PHO 4. PHO 5. PHO 6. PHO 7. PHO 8. PHO 9. PHO 10;

[0185] S2.3. Divide the pixel grayscale value 0~255 into 16 grayscale intervals, denoted as , =1,2,3,…,16;

[0186] S2.4. Count the number of pixels corresponding to each gray level in the road congestion Then, the number of pixels corresponding to each gray level of similar diseases of road bumps, such as longitudinal cracks, transverse cracks, cracks, rutting, potholes, peeling, oiling, subsidence, and gnawing, is counted. , , , , , , , , ;

[0187] S2.5. Standardize the number of pixels corresponding to each gray level to obtain the standard number of pixels, where the standard number of pixels corresponding to the road congestion is , the calculation formula is:

[0188] ;

[0189] The standard number of pixels corresponding to longitudinal cracks, transverse cracks, cracks, rutting, potholes, spalling, oil spills, subsidence, and edge gnawing is , , , , , , , , ;

[0190] S2.6. As a benchmark, the relationship function between pavement bumps and other diseases is established in turn, and the calculation formula is:

[0191] ;

[0192] Among them, ii represents the serial number of other similar diseases, ii=2,3,4,5…10, and we get , , , , , , , , They represent the relationship functions between road surface bumps and longitudinal cracks, transverse cracks, cracks, rutting, potholes, spalling, oiling, subsidence, and undercutting respectively;

[0193] S2.7. Comparison , , , , , , , , , and sort them in descending order, select the disease types corresponding to the top three relationship functions as the disease types most similar to the image features of road surface bumps, and then select all road surface images containing road surface bumps and the disease types most similar to the image features of road surface bumps;

[0194] S2.8. By manually labeling, the road surface images obtained in step S2.8 that contain road bumps and the types of defects that are most similar to the image features of the road bumps are framed to select the road pothole area and the other three defects, and the road bumps are labeled as upheaval, and the other three defects are labeled as other1, other2, and other3, respectively, to obtain a data set for rapid identification of road bumps;

[0195] S2.9. Divide the dataset for rapid identification of road congestion into a training set and a test set in a ratio of 8:2.

[0196] S3. Constructing a basic model, a first auxiliary model, a second auxiliary model and an enhanced model for rapid identification of road congestion; then constructing a loss function between the output layers of the basic model, the first auxiliary model, the second auxiliary model and the enhanced model for rapid identification of road congestion;

[0197] Road bumps are local bulges caused by insufficient compaction of asphalt pavement materials or instability of the base layer, usually manifested as bumps and deformations on the road surface. This disease appears in the image as irregular hemispherical or circular raised areas with shadows and texture changes. Considering the characteristics of bumps, the model structure is designed to include input layer, convolution layer, activation layer, pooling layer, merging layer, upsampling layer, PANet layer, pooling splicing layer, SPP layer (spatial pyramid pooling), and output layer. The functions of each structural layer are as follows:

[0198] The input layer receives road images captured by the vehicle-mounted camera, which contain the local ridge features of the bump and its surrounding environment information.

[0199] The convolutional layer extracts features from the image through a series of filters, such as the contours of the bump, its texture, and the difference from the normal road surface.

[0200] The activation layer introduces nonlinearity, allowing the model to capture the complex pattern differences between congested and normal roads.

[0201] The pooling layer helps the model maintain sensitivity to the features of the packed image while reducing the amount of computation by reducing the spatial size, and improves tolerance to position changes.

[0202] The merging layer is used to fuse features from different convolutional layers to enhance the model's ability to recognize the shape and size of packets.

[0203] The upsampling layer restores the spatial resolution of the enveloping features so that the location of the enveloping features can be accurately located in the output layer.

[0204] The PANet layer improves the recognition accuracy of the geometric and structural features of the package by optimizing the feature pyramid and enhancing feature transfer.

[0205] The pooling concatenation layer is used to integrate the features of different pooling operations before the SPP layer to capture multi-scale encapsulation information.

[0206] The SPP layer processes the enveloping features to preserve the key information of enveloping images of different sizes and provide robustness to the input image size.

[0207] The output layer maps the extracted features into the final judgment output, such as the probability distribution map of the congested area, to achieve accurate identification of road congestion.

[0208] Furthermore, the specific implementation method of step S3 includes the following steps:

[0209] S3.1. Constructing a basic model for rapid identification of road congestion:

[0210] S3.1.1. The basic model A consists of 1 input layer, 9 convolutional layers, 9 activation layers, 3 pooling layers, 1 merging layer, 1 upsampling layer, 1 path aggregation network layer PANet, 1 pooling splicing layer, 1 spatial pyramid pooling layer SPP, and 1 output layer. The specific network structure is the following structure connected in sequence:

[0211] Input layer: image size 227×227×3;

[0212] Convolutional layer 1: convolution kernel size 3×3; number of convolution kernels 32; stride: 1; batch normalization;

[0213] Activation layer 1: The activation function is LeakyReLU;

[0214] Convolutional layer 2: kernel size 3×3; number of kernels 64; Stride: 1; batch normalization;

[0215] Activation layer 2: The activation function is LeakyReLU;

[0216] Convolution layer 3: convolution kernel size 3×3; number of convolution kernels 128; stride: 1; batch normalization;

[0217] Activation layer 3: The activation function is LeakyReLU;

[0218] Convolutional layer 4: convolution kernel size 3×3; number of convolution kernels 256; stride: 1; batch normalization;

[0219] Activation layer 4: The activation function is LeakyReLU;

[0220] SPP layer: extracts features of different scales and fuses them to increase the receptive field of the model;

[0221] Pooling layer 1: Type is global average pooling; pooling size: 5×5; Stride: 1;

[0222] Pooling layer 2: Type is global average pooling; pooling size: 9×9; Stride: 1;

[0223] Pooling layer 3: Type is global average pooling; pooling size: 13×13; Stride: 1;

[0224] Pooling concatenation layer: connect the outputs of the three pooling layers;

[0225] PANet layer: aggregates features at different levels and implements feature pyramids through upsampling and downsampling operations;

[0226] Convolution layer 5: kernel size 3×3; number of kernels 768; stride: 2; batch normalization;

[0227] Activation layer 5: The activation function is LeakyReLU;

[0228] Convolutional layer 6: kernel size 3×3; number of kernels 768; stride: 2; batch normalization;

[0229] Activation layer 6: The activation function is LeakyReLU;

[0230] Upsampling layer: uses the nearest neighbor interpolation method to upsample to the same size as the pooling splicing layer;

[0231] Merging layer: The output feature map of the downsampling layer is merged with the output feature map of the upsampling layer;

[0232] Convolutional layer 7: kernel size 3×3; number of kernels 256; stride: 1; batch normalization;

[0233] Activation layer 7: The activation function is LeakyReLU;

[0234] Convolutional layer 8: kernel size 3×3; number of kernels 128; Stride: 1; batch normalization;

[0235] Activation layer 8: The activation function is LeakyReLU;

[0236] Convolution layer 9: kernel size 3×3; number of kernels 64; stride: 1; batch normalization;

[0237] Activation layer 9: The activation function is LeakyReLU;

[0238] Output layer: type is normalized exponential function Softmax; number of neurons: 4;

[0239] S3.1.2. Input the training set obtained in step S2 into the basic model A, calculate the weights and biases of the neural network units through the error back propagation algorithm, complete the training of the basic model A, and obtain the trained basic model A;

[0240] The calculation formula of the loss function of the basic model A is:

[0241] ;

[0242] in, is the loss function of the basic model A; is the total number of data samples in the basic model A; For the it The true labels of samples, Based on the prediction of model A it The probability that a sample is class 1;

[0243] S3.1.3. The node output values ​​of the four output units corresponding to the basic model A are a 1. a 2. a 3. a 4. Perform initial conversion on the node output values ​​of the four output units to obtain the output values The corresponding probability value The calculation formula is:

[0244] ;

[0245] 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:

[0246] ;

[0247] in, G is the enhancement coefficient, and then the actual disease corresponding to each node is marked as Specifically, for road congestion 1, for other diseases is 0;

[0248] S3.1.4. 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 , , , ;

[0249] S3.2. Constructing the first auxiliary model for rapid identification of road congestion:

[0250] S3.2.1. The first auxiliary model B consists of 1 input layer, 9 convolutional layers, 9 activation layers, 5 pooling layers, 1 merging layer, 1 upsampling layer, 1 PANet layer, 1 pooling splicing layer, 1 SPP layer, and 1 output layer. The specific network structure is the following structure connected in sequence:

[0251] Input layer: image size 227×227×3;

[0252] Convolutional layer 1: convolution kernel size 3×3; number of convolution kernels 32; Stride: 1; batch normalization;

[0253] Activation layer 1: The activation function is LeakyReLU;

[0254] Convolutional layer 2: kernel size 3×3; number of kernels 64; Stride: 1; batch normalization;

[0255] Activation layer 2: The activation function is LeakyReLU;

[0256] Convolution layer 3: convolution kernel size 3×3; number of convolution kernels 128; stride: 1; batch normalization;

[0257] Activation layer 3: The activation function is LeakyReLU;

[0258] Convolutional layer 4: convolution kernel size 3×3; number of convolution kernels 256; stride: 1; batch normalization;

[0259] Activation layer 4: The activation function is LeakyReLU;

[0260] SPP layer: extracts features of different scales and fuses them to increase the receptive field of the model;

[0261] Pooling layer 1: Type is global average pooling; pooling size: 5×5; Stride: 1;

[0262] Pooling layer 2: Type is global average pooling; pooling size: 7×7; Stride: 1;

[0263] Pooling layer 3: Type is global average pooling; pooling size: 9×9; Stride: 1;

[0264] Pooling layer 4: Type is global average pooling; pooling size: 11×11; Stride: 1;

[0265] Pooling layer 5: Type is global average pooling; pooling size: 13×13; Stride: 1;

[0266] Pooling concatenation layer: connect the outputs of the five pooling layers;

[0267] PANet: Aggregates features at different levels and implements feature pyramids through upsampling and downsampling operations;

[0268] Convolution layer 5: kernel size 3×3; number of kernels 1280; stride: 2; batch normalization;

[0269] Activation layer 5: The activation function is LeakyReLU;

[0270] Convolutional layer 6: kernel size 3×3; number of kernels 1280; stride: 2; batch normalization;

[0271] Activation layer 6: The activation function is LeakyReLU;

[0272] Upsampling layer: uses the nearest neighbor interpolation method to upsample to the same size as the pooling splicing layer;

[0273] Merging layer: The output feature map of the downsampling layer is merged with the output feature map of the upsampling layer;

[0274] Convolutional layer 7: kernel size 3×3; number of kernels 512; stride: 1; batch normalization;

[0275] Activation layer 7: The activation function is LeakyReLU;

[0276] Convolutional layer 7: kernel size 3×3; number of kernels 256; stride: 1; batch normalization;

[0277] Activation layer 7: The activation function is LeakyReLU;

[0278] Convolutional layer 8: kernel size 3×3; number of kernels 128; Stride: 1; batch normalization;

[0279] Activation layer 8: The activation function is LeakyReLU;

[0280] Convolution layer 9: kernel size 3×3; number of kernels 64; stride: 1; batch normalization;

[0281] Activation layer 9: The activation function is LeakyReLU;

[0282] Output layer: type is Softmax; number of neurons: 4;

[0283] S3.2.2. Input the training set obtained in step S2 into the first auxiliary model B, calculate the weights and biases of the neural network units through the error back propagation algorithm, complete the training of the first auxiliary model B, and obtain the trained first auxiliary model B;

[0284] The calculation formula of the loss function of the first auxiliary model B is:

[0285] ;

[0286] in, is the loss function of the first auxiliary model B; is the total number of data samples in the first auxiliary model B; For the one The true labels of samples, The first auxiliary model B predicts one The probability that a sample is class 1;

[0287] S3.2.3. The node output values ​​of the four output units corresponding to the first auxiliary model B are b 1. b 2. b 3. b 4. Perform initial conversion on the node output values ​​of the four output units to obtain the output values b i The corresponding probability value The calculation formula is:

[0288] ;

[0289] 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 The calculation formula is:

[0290] ;

[0291] S3.2.4. Adoption The training set obtained in step S2 is combined with the error chain derivation rule to train the first auxiliary model B. After the training is completed, the disease output value corresponding to the first auxiliary model B is obtained as , among the 4 nodes in the output layer, the actual disease corresponding to each node is marked as , i =1,2,3,4, specifically, for road congestion 1, for other diseases is 0;

[0292] S3.3. Constructing the second auxiliary model for rapid identification of road congestion:

[0293] S3.3.1. The second auxiliary model C consists of 1 input layer, 12 convolutional layers, 12 activation layers, 5 pooling layers, 1 merging layer, 1 upsampling layer, 1 PANet layer, 1 pooling splicing layer, 1 SPP layer, and 1 output layer. The specific network structure is the following structure connected in sequence:

[0294] Input layer: image size 227×227×3;

[0295] Convolutional layer 1: convolution kernel size 3×3; number of convolution kernels 32; Stride: 1; batch normalization;

[0296] Activation layer 1: The activation function is LeakyReLU;

[0297] Convolutional layer 2: kernel size 3×3; number of kernels 64; Stride: 1; batch normalization;

[0298] Activation layer 2: The activation function is LeakyReLU;

[0299] Convolution layer 3: convolution kernel size 3×3; number of convolution kernels 128; stride: 1; batch normalization;

[0300] Activation layer 3: The activation function is LeakyReLU;

[0301] Convolutional layer 4: convolution kernel size 3×3; number of convolution kernels 128; stride: 1; batch normalization;

[0302] Activation layer 4: The activation function is LeakyReLU;

[0303] Convolution layer 5: kernel size 3×3; number of kernels 256; stride: 1; batch normalization;

[0304] Activation layer 5: The activation function is LeakyReLU;

[0305] Convolutional layer 6: kernel size 3×3; number of kernels 256; stride: 1; batch normalization;

[0306] Activation layer 6: The activation function is LeakyReLU;

[0307] SPP layer: extracts features of different scales and fuses them to increase the receptive field of the model;

[0308] Pooling layer 1: Type is global average pooling; pooling size: 5×5; Stride: 1;

[0309] Pooling layer 2: Type is global average pooling; pooling size: 7×7; Stride: 1;

[0310] Pooling layer 3: Type is global average pooling; pooling size: 9×9; Stride: 1;

[0311] Pooling layer 4: Type is global average pooling; pooling size: 11×11; Stride: 1;

[0312] Pooling layer 5: Type is global average pooling; pooling size: 13×13; Stride: 1;

[0313] Pooling concatenation layer: connect the outputs of the three pooling layers;

[0314] PANet: Aggregates features at different levels and implements feature pyramids through upsampling and downsampling operations;

[0315] Convolutional layer 7: kernel size 3×3; number of kernels 1280; stride: 2; batch normalization;

[0316] Activation layer 7: The activation function is LeakyReLU;

[0317] Convolutional layer 8: kernel size 3×3; number of kernels 1280; stride: 2; batch normalization;

[0318] Activation layer 8: The activation function is LeakyReLU;

[0319] Upsampling layer: uses the nearest neighbor interpolation method to upsample to the same size as the pooling splicing layer;

[0320] Merging layer: The output feature map of the downsampling layer is merged with the output feature map of the upsampling layer;

[0321] Convolutional layer 9: kernel size 3×3; number of kernels 512; Stride: 1; batch normalization;

[0322] Activation layer 9: The activation function is LeakyReLU;

[0323] Convolution layer 10: kernel size 3×3; number of kernels 256; stride: 1; batch normalization;

[0324] Activation layer 10: The activation function is LeakyReLU;

[0325] Convolutional layer 11: kernel size 3×3; number of kernels 128; Stride: 1; batch normalization;

[0326] Activation layer 11: The activation function is LeakyReLU;

[0327] Convolution layer 12: kernel size 3×3; number of kernels 64; stride: 1; batch normalization;

[0328] Activation layer 12: The activation function is LeakyReLU;

[0329] Output layer: type is normalized exponential function Softmax; number of neurons: 4;

[0330] S3.3.2. Input the training set obtained in step S2 into the second auxiliary model C, calculate the weights and biases of the neural network units through the error back propagation algorithm, complete the training of the second auxiliary model C, and obtain the trained second auxiliary model C;

[0331] The calculation formula of the loss function of the second auxiliary model C is:

[0332] ;

[0333] in, is the loss function of the second auxiliary model C; is the total number of data samples in the second auxiliary model C; For the ic The true labels of samples, The second auxiliary model C predicts ic The probability that a sample is class 1;

[0334] S3.3.3. The node output values ​​of the four output units corresponding to the second auxiliary model C are , the node output values ​​of the four output units are initially converted to obtain The corresponding probability value The calculation formula is:

[0335] ;

[0336] 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:

[0337] ;

[0338] in, G is the enhancement coefficient, and then the actual disease corresponding to each node is marked as Specifically, for road congestion 1, for other diseases is 0;

[0339] S3.3.4. Adoption The training set obtained in step S2 is combined with the error chain derivation rule to train the second auxiliary model C. After the training is completed, the output values ​​of the four types of diseases corresponding to the second auxiliary model C are obtained as follows:

[0340] S3.4. Construct an enhanced model for rapid identification of road congestion:

[0341] S3.4.1. The enhanced model D consists of 1 input layer, 14 convolutional layers, 14 activation layers, 5 pooling layers, 1 merging layer, 1 upsampling layer, 1 PANet layer, 1 pooling splicing layer, 1 SPP layer, and 1 output layer. The specific network structure is the following structure connected in sequence:

[0342] Input layer: image size 227×227×3;

[0343] Convolutional layer 1: convolution kernel size 3×3; number of convolution kernels 32; Stride: 1; batch normalization;

[0344] Activation layer 1: The activation function is LeakyReLU;

[0345] Convolutional layer 2: convolution kernel size 3×3; number of convolution kernels 32; stride: 1; batch normalization;

[0346] Activation layer 2: The activation function is LeakyReLU;

[0347] Convolution layer 3: convolution kernel size 3×3; number of convolution kernels 64; stride: 1; batch normalization;

[0348] Activation layer 3: The activation function is LeakyReLU;

[0349] Convolutional layer 4: convolution kernel size 3×3; number of convolution kernels 64; stride: 1; batch normalization;

[0350] Activation layer 4: The activation function is LeakyReLU;

[0351] Convolution layer 5: kernel size 3×3; number of kernels 128; stride: 1; batch normalization;

[0352] Activation layer 5: The activation function is LeakyReLU;

[0353] Convolutional layer 6: kernel size 3×3; number of kernels 128; Stride: 1; batch normalization;

[0354] Activation layer 6: The activation function is LeakyReLU;

[0355] Convolutional layer 7: kernel size 3×3; number of kernels 256; stride: 1; batch normalization;

[0356] Activation layer 7: The activation function is LeakyReLU;

[0357] Convolutional layer 8: kernel size 3×3; number of kernels 256; stride: 1; batch normalization;

[0358] Activation layer 8: The activation function is LeakyReLU;

[0359] SPP layer: extracts features of different scales and fuses them to increase the receptive field of the model;

[0360] Pooling layer 1: Type is global average pooling; pooling size: 5×5; Stride: 1;

[0361] Pooling layer 2: Type is global average pooling; pooling size: 7×7; Stride: 1;

[0362] Pooling layer 3: Type is global average pooling; pooling size: 9×9; Stride: 1;

[0363] Pooling layer 4: Type is global average pooling; pooling size: 11×11; Stride: 1;

[0364] Pooling layer 5: Type is global average pooling; pooling size: 13×13; Stride: 1;

[0365] Pooling concatenation layer: connect the outputs of the three pooling layers;

[0366] PANet: Aggregates features at different levels and implements feature pyramids through upsampling and downsampling operations;

[0367] Convolution layer 9: kernel size 3×3; number of kernels 1280; stride: 2; batch normalization;

[0368] Activation layer 9: The activation function is LeakyReLU;

[0369] Convolution layer 10: kernel size 3×3; number of kernels 1280; stride: 2; batch normalization;

[0370] Activation layer 10: The activation function is LeakyReLU;

[0371] Upsampling layer: uses the nearest neighbor interpolation method to upsample to the same size as the pooling splicing layer;

[0372] Merging layer: The output feature map of the downsampling layer is merged with the output feature map of the upsampling layer;

[0373] Convolutional layer 11: kernel size 3×3; number of kernels 512; Stride: 1; batch normalization;

[0374] Activation layer 11: The activation function is LeakyReLU;

[0375] Convolutional layer 12: kernel size 3×3; number of kernels 256; stride: 1; batch normalization;

[0376] Activation layer 12: The activation function is LeakyReLU;

[0377] Convolution layer 13: kernel size 3×3; number of kernels 128; stride: 1; batch normalization;

[0378] Activation layer 13: The activation function is LeakyReLU;

[0379] Convolutional layer 14: kernel size 3×3; number of kernels 64; Stride: 1; batch normalization;

[0380] Activation layer 14: the activation function is LeakyReLU;

[0381] Output layer: type is normalized exponential function Softmax; number of neurons: 4;

[0382] S3.4.2. Input the training set obtained in step S2 into the enhanced model D, calculate the weights and biases of the neural network units through the error back propagation algorithm, complete the training of the enhanced model D, and obtain the trained enhanced model D;

[0383] The loss function of the enhanced model D is calculated as:

[0384] ;

[0385] in, is the loss function of the enhanced model D; is the total number of data samples in the reinforcement model D; For the id The true labels of samples, To strengthen the model D to predict the id The probability that a sample is class 1;

[0386] S3.4.3. The node output values ​​of the four output units corresponding to the enhanced model D are , the node output values ​​of the four output units are initially converted to obtain The corresponding probability value The calculation formula is:

[0387] ;

[0388] 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:

[0389] ;

[0390] in, G is the enhancement coefficient, and then the actual disease corresponding to each node is marked as Specifically, for road congestion 1, for other diseases is 0;

[0391] S3.4.4. Adoption The training set obtained in step S2 is combined with the error chain derivation rule to train the enhanced model D. After the training is completed, the output values ​​of the four types of diseases corresponding to the enhanced model D are obtained as follows: ;

[0392] S3.5. Based on and , establish the loss function between the first auxiliary model B and the output layer of the basic model A, and the calculation formula is:

[0393] ;

[0394] based on and , construct the output value The corresponding probability value The loss function of the base model A , the calculation formula is:

[0395] ;

[0396] based on and , establish the relationship loss function between the output layer of the first auxiliary model B and the basic model A , the calculation formula is:

[0397] ;

[0398] in, for The weight coefficient of for The weight coefficient of

[0399] S3.6. Establish the relationship loss function between the output layer of the second auxiliary model C and the first auxiliary model B , the expression is:

[0400] ;

[0401] in, for The weight coefficient of for The weight coefficient of Based on and The loss function of Based on and The loss function of

[0402] The calculation formula is:

[0403] ;

[0404] The calculation formula is:

[0405] ;

[0406] S3.7. Establish the relationship loss function between the output layer of the second auxiliary model C and the basic model A , the expression is:

[0407] ;

[0408] in, for The weight coefficient of for The weight coefficient of Based on and The loss function is Based on and The loss function of

[0409] The calculation formula is:

[0410] ;

[0411] The calculation formula is:

[0412] ;

[0413] S3.8. Establish the relationship loss function between the output layer of the enhanced model D and the base model A , the expression is:

[0414] ;

[0415] in, for The weight coefficient of for The weight coefficient of Based on and The loss function of

[0416] The calculation formula is:

[0417] ;

[0418] S3.9. Establish the relationship loss function between the output layer of the reinforcement model D and the second auxiliary model C , the expression is:

[0419] ;

[0420] in, for The weight coefficient of for The weight coefficient of Based on and The loss function is Based on and The loss function of

[0421] The calculation formula is:

[0422] ;

[0423] The calculation formula is:

[0424] .

[0425] S4. Constructing the inter-layer loss function of the enhanced model and the basic model for rapid identification of road bumps; then constructing the overall loss function of the rapid identification model of road bumps;

[0426] Furthermore, the specific implementation method of step S4 includes the following steps:

[0427] S4.1. Assuming a small model a and large models d , build a small model a The second, third, fourth and output layers in the large model are d Establish connections between the fourth, sixth, tenth layers and the output layer;

[0428] S4.2. For the second layer in the small model and the large model d 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 d The output of the fourth layer is recorded as , the corresponding size of width × length × number of channels is ;

[0429] 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 , k i is the channel number, The corresponding output values ​​of each channel are ;

[0430] Build on Small model a The second floor with the large model d The loss function between the fourth layers , the calculation formula is:

[0431] ;

[0432] when When in the large model d 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 ;

[0433] Establish Small model a The second floor with the large model d The loss function between the fourth layers , the calculation formula is:

[0434] ;

[0435] S4.3. For the third layer in the small model and the large model b The sixth layer in the relationship is established to obtain a small model a The third floor and the large model b Loss function between the sixth layers ;

[0436] S4.4. For the fourth layer in the small model and the large model b The tenth layer in the relationship is established to obtain a small model a Fourth floor and large model b Loss function between the tenth layer ;

[0437] S4.5. When the small model is the base model A, the small model a 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 D, the large model dThe fourth, sixth, and tenth layers in are the output of pooling layer 2, the output of pooling layer 3, and the output of pooling layer 5 of the enhanced model D respectively;

[0438] S4.6. Based on the loss function between the output layer of the enhanced model D and the inter-layer loss function of the base model A, define the total loss function between the two models , the calculation formula is:

[0439] ;

[0440] in, for The corresponding weight coefficient is, for The corresponding weight coefficient is, for The corresponding weight coefficient.

[0441] S5. Using the data set obtained in step S2 for rapid identification of road congestion, the constructed basic model, the first auxiliary model, the second auxiliary model and the enhanced model are trained to obtain a trained basic model as a road congestion rapid identification model;

[0442] Furthermore, the specific implementation method of step S5 is to use the data set obtained in step S2 based on the chain derivative rule to train the basic model, the first auxiliary model, the second auxiliary model and the enhanced model through error back propagation. When the overall loss function When is minimum, the corresponding model parameters are optimal, and the trained basic model is obtained as a road congestion rapid recognition model.

[0443] S6. The road congestion rapid identification model obtained in step S5 is improved to obtain a lightweight road congestion rapid identification model;

[0444] There are many reasons for the formation of congestion, such as traffic density, parking behavior, road design, etc. Therefore, the forms of congestion are diverse and vary greatly from each other, and the characteristics are more complex. At the same time, the form of congestion will also change over time, which puts higher requirements on the model. Through the basic model improvement method based on the enhancement coefficient, the diverse feature information in the congestion can be better captured and the model recognition accuracy can be improved.

[0445] The specific implementation method of step S6 includes the following steps:

[0446] S6.1. For the road congestion rapid identification model obtained in step S5, set the enhancement coefficient G The values ​​are: G =1,3,5,7,9,11,15,20,30,40,50;

[0447] S6.2. Based on different enhancement factors G 11 working conditions are obtained, and 11 models are obtained by training the road congestion rapid recognition model. The corresponding accuracy rates of the 11 road congestion rapid recognition models are as follows: , i=1,2,3,…,11, the calculation formula is:

[0448] ;

[0449] in, Represents the first i The number of samples that a model predicts as having a bag and actually has a bag when identifying the data in the test set; Represents the first i The number of samples that a model predicts as having a packet but is not actually having a packet when identifying the data in the test set;

[0450] Comparison The maximum value in ;

[0451] The corresponding road surface bump quick identification model is the final lightweight road surface bump quick identification model;

[0452] S7. Constructing a digital twin model based on the lightweight road congestion rapid identification model obtained in step S6;

[0453] Furthermore, the specific implementation method of step S7 includes the following steps:

[0454] S7.1. The lightweight road congestion rapid identification model obtained in step S6 is exported as a deployment file for independent services, including one of a pt file, a pth file, a json file, and a txt file;

[0455] S7.2. Construct a road information model;

[0456] S7.2.1. Collect information on the location, length, width, and layout of roads;

[0457] 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;

[0458] 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;

[0459] 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;

[0460] S7.3. In the digital twin platform, import the deployment file of the lightweight road congestion rapid identification model obtained in step S7.1, and import the road information model format file obtained in step S7.2;

[0461] S7.4. Use Python to construct an API interface between image data and a lightweight road bump rapid identification model for intelligent rutting identification. Use Python to construct an API interface between a lightweight road bump rapid identification model and a road information model to complete the construction of a digital twin model.

[0462] S8. Based on the digital twin model obtained in step S7, a method for analyzing factors causing road congestion is constructed.

[0463] Furthermore, the specific implementation method of step S8 includes the following steps:

[0464] S8.1. Use a 3D structured light camera to collect road surface data and upload it to the digital twin model established in step S7. Then use the lightweight road bump rapid identification model deployed in the system to intelligently identify the road bump area in the road image, select all images containing road bumps, conduct on-site review of the bumps, and measure the height GD of the road bumps.

[0465] S8.2. The factors affecting the formation of congestion include pavement material aging, ambient temperature, water damage, heavy-loaded vehicles, traffic load, pavement thickness, pavement temperature, and pavement compressive strength; pavement material aging is quantified by asphalt oxide content, water damage is quantified by pavement water seepage, heavy-loaded vehicles are quantified by the number of heavy-loaded vehicles, and traffic load is quantified by traffic volume;

[0466] (1) Aging of pavement materials: Over time, pavement materials may lose elasticity and bonding ability due to aging, oxidation, etc. This can be characterized by testing the asphalt oxide content LH.

[0467] (2) Ambient temperature: The increase or decrease in temperature will affect the properties of the asphalt pavement, causing expansion or contraction, which may cause pavement congestion. This can be characterized by testing the ambient temperature HW.

[0468] (3) Water damage: Water can penetrate into the roadbed and pavement structure, causing the material to lose stability. This can be characterized by testing the pavement water seepage SS.

[0469] (4) Heavy-loaded vehicles: Heavy-loaded vehicles will generate greater stress on the road surface, especially in the area where the wheels contact the road surface. Since the stress of the heavy load exceeds the load expected during design, this will cause excessive deformation of the road surface material. This can be characterized by testing the number of heavy-loaded vehicles ZZ.

[0470] (5) Traffic load: High traffic volume will cause the pavement material to continuously bear dynamic loads. This repeated load will accelerate the aging and damage of the pavement. This can be characterized by testing the traffic volume JL.

[0471] (6) Pavement thickness: Pavement thickness affects the distribution of vehicle load. If the pavement layer is thin, the bearing capacity of the roadbed may be insufficient and the vehicle load cannot be effectively dispersed, resulting in roadbed deformation and congestion. This can be characterized by testing the pavement thickness HD.

[0472] (7) Pavement temperature: Pavement temperature has a significant impact on the performance of asphalt pavement. Pavement temperature sensors can be used to monitor the pavement temperature LW.

[0473] (8) Pavement compressive strength: A pavement with high compressive strength can better disperse the pressure from vehicle tires and reduce the occurrence of bumps. This can be characterized by testing the pavement compressive strength KY.

[0474] S8.3. Select 10 time nodes t1, t2, t3, ... t10, collect data at each of the 10 time nodes, and obtain the values ​​of the corresponding quantitative indicators as follows:

[0475] Asphalt oxide content at 10 time points: LH1, LH2, LH3, LH4, LH5, LH6, LH7, LH8, LH9, LH10;

[0476] Ambient temperature at 10 time points: HW1, HW2, HW3, HW4, HW5, HW6, HW7, HW8, HW9, HW10;

[0477] Pavement seepage at 10 time points: SS1, SS2, SS3, SS4, SS5, SS6, SS7, SS8, SS9, SS10;

[0478] The number of heavy-loaded vehicles at 10 time points: ZZ1, ZZ2, ZZ3, ZZ4, ZZ5, ZZ6, ZZ7, ZZ8, ZZ9, ZZ10;

[0479] Traffic volume at 10 time nodes: JL1, JL2, JL3, JL4, JL5, JL6, JL7, JL8, JL9, JL10;

[0480] Pavement thickness at 10 time points: HD1, HD2, HD3, HD4, HD5, HD6, HD7, HD8, HD9, HD10;

[0481] Pavement temperature at 10 time points: LW1, LW2, LW3, LW4, LW5, LW6, LW7, LW8, LW9, LW10;

[0482] Pavement compressive strength at 10 time points: KY1, KY2, KY3, KY4, KY5, KY6, KY7, KY8, KY9, KY10;

[0483] Road congestion height at 10 time nodes: GD1, GD2, GD3, GD4, GD5, GD6, GD7, GD8, GD9, GD10;

[0484] The above data are obtained from public channels;

[0485] S8.4. Calculate the average value of each quantitative index at 10 time points respectively, and get the average value of asphalt oxide content: The average ambient temperature is The average pavement seepage volume is The average number of heavy-loaded vehicles is The average traffic volume is The average pavement thickness is The average road surface temperature is The average compressive strength of the pavement is The average height of road bump is ;

[0486] S8.5. Calculate the correlation strength coefficient between each quantitative index and the road bump height , the calculation formula is as follows:

[0487] ;

[0488] in, For the I The quantitative indicator value at each time node, For the I The height of the road bump at each time node; ZB represents the quantitative evaluation index, including asphalt oxide content, ambient temperature, road water seepage, number of heavy-loaded vehicles, traffic volume, road thickness, road temperature, and road compressive strength; that is, the correlation strength coefficients between asphalt oxide content, ambient temperature, road water seepage, number of heavy-loaded vehicles, traffic volume, road thickness, road temperature, road compressive strength and road bump height are: , , , , , , , ;

[0489] S8.6. Arrange all the correlation strength coefficients obtained in step S8.5 in descending order and re-label them as TZN1 to TZN8, where the influencing factor of the congestion formation corresponding to TZN1 is the influencing factor with the greatest influence on the road congestion;

[0490] S8.7. Defining thresholds YZ ,calculate , it is Start counting from 1 and gradually increase. it is Value Satisfaction When , stop the calculation. At this time, it is The road congestion influencing factors corresponding to the values ​​are the main influencing factors that cause road congestion.

[0491] Embodiment 2:

[0492] An electronic device includes a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of a method for analyzing factors of road congestion formation described in Example 1 when executing the computer program.

[0493] 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, etc. Moreover, the processor is used to implement the steps of the above-mentioned method for analyzing factors of road congestion when executing the computer program stored in the memory.

[0494] 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.

[0495] 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.

[0496] Embodiment 3:

[0497] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements a method for analyzing factors of road congestion formation as described in Example 1.

[0498] The computer-readable storage medium of the present invention can be any form of storage medium that can be read by the processor of a computer device, including but not limited to non-volatile memory, volatile memory, ferroelectric memory, etc. The computer-readable storage medium stores a computer program. When the processor of the computer device reads and executes the computer program stored in the memory, the steps of the above-mentioned method for analyzing factors of road congestion formation can be implemented.

[0499] 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. It should be noted that the content contained in the computer readable medium may be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction.

[0500] The key points and protection points of the present invention are:

[0501] (1) A lightweight intelligent road congestion recognition model;

[0502] (2) A basic model improvement method based on enhancement coefficient;

[0503] (3) A method for constructing a road digital twin model and updating pavement congestion and damage.

[0504] (4) A method for analyzing factors affecting road congestion based on digital twins.

[0505] 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.

[0506] 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 analyzing factors of road congestion, characterized in that: The steps include: S1. Using a vehicle-mounted three-dimensional structured light camera to collect a road surface image, and then compressing the collected road surface image to obtain a compressed road surface image; S2. Perform similar disease recognition on the compressed road surface image obtained in step S1, extract all road surface images containing bumps and the disease types that are most similar to the image features of bumps, and construct a data set for rapid recognition of road bumps; S3. Construct a basic model, a first auxiliary model, a second auxiliary model and an enhanced model for rapid identification of road congestion; then construct a loss function between the output layers of the basic model, the first auxiliary model, the second auxiliary model and the enhanced model based on rapid identification of road congestion; The specific implementation method of step S3 includes the following steps: S3.

1. Construct a basic model for rapid identification of road congestion: S3.1.

1. The basic model A consists of 1 input layer, 9 convolutional layers, 9 activation layers, 3 pooling layers, 1 merging layer, 1 upsampling layer, 1 path aggregation network layer PANet, 1 pooling splicing layer, 1 spatial pyramid pooling layer SPP, and 1 output layer; S3.1.

2. Input the training set obtained in step S2 into the basic model A, calculate the weights and biases of the neural network units through the error back propagation algorithm, complete the training of the basic model A, and obtain the trained basic model A; The calculation formula of the loss function of the basic model A is: 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; 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; S3.1.

3. 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 ; S3.1.

4. 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 ; S3.

2. Construct the first auxiliary model for rapid identification of road congestion: S3.2.

1. The first auxiliary model B consists of 1 input layer, 9 convolutional layers, 9 activation layers, 5 pooling layers, 1 merging layer, 1 upsampling layer, 1 PANet layer, 1 pooling splicing layer, 1 SPP layer, and 1 output layer; S3.2.

2. Input the training set obtained in step S2 into the first auxiliary model B, calculate the weights and biases of the neural network units through the error back propagation algorithm, complete the training of the first auxiliary model B, and obtain the trained first auxiliary model B; The calculation formula of the loss function of the first auxiliary model B is: Where Lb is the loss function of the first auxiliary model B; Nb is the total number of data samples in the first auxiliary model B; y ib is the true label of the ibth sample, p ib Predict the probability that the ibth sample is category 1 for the first auxiliary model B; S3.2.

3. The node output values ​​of the four output units corresponding to the first auxiliary model B are b1, b2, b3, and b4. The node output values ​​of the four output units are initially converted 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 enhanced probability value g bi The calculation formula is: S3.2.

4. Use g bi The training set obtained in step S2 is combined with the error chain derivation rule to train the first auxiliary model B. After the training is completed, the disease output value corresponding to the first auxiliary 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; S3.

3. Construct the second auxiliary model for rapid identification of road congestion: S3.3.

1. The second auxiliary model C consists of 1 input layer, 12 convolutional layers, 12 activation layers, 5 pooling layers, 1 merging layer, 1 upsampling layer, 1 PANet layer, 1 pooling splicing layer, 1 SPP layer, and 1 output layer; S3.3.

2. Input the training set obtained in step S2 into the second auxiliary model C, calculate the weights and biases of the neural network units through the error back propagation algorithm, complete the training of the second auxiliary model C, and obtain the trained second auxiliary model C; The calculation formula of the loss function of the second auxiliary model C is: Where Lc is the loss function of the second auxiliary model C; Nc is the total number of data samples in the second auxiliary model C; y ic is the true label of the icth sample, p ic Predict the probability that the icth sample is class 1 for the second auxiliary model C; S3.3.

3. The node output values ​​of the four output units corresponding to the second auxiliary model C are c i , the node output values ​​of the four output units are initially converted to obtain c i The corresponding probability value f ci The calculation formula is: Then the node output values ​​of the four output units are further optimized and converted to obtain the output value c i The corresponding enhancement probability value g ci The calculation formula is: Among them, G is the enhancement coefficient, and then the actual disease corresponding to each node is marked as h ci ; S3.3.

4. Use g ci The training set obtained in step S2 is combined with the error chain derivation rule to train the second auxiliary model C. After the training is completed, the output values ​​of the four types of diseases corresponding to the second auxiliary model C are j ci ; S3.

4. Construct an enhanced model for rapid identification of road congestion: S3.4.

1. The enhanced model D consists of 1 input layer, 14 convolutional layers, 14 activation layers, 5 pooling layers, 1 merging layer, 1 upsampling layer, 1 PANet layer, 1 pooling splicing layer, 1 SPP layer, and 1 output layer; S3.4.

2. Input the training set obtained in step S2 into the enhanced model D, calculate the weights and biases of the neural network units through the error back propagation algorithm, complete the training of the enhanced model D, and obtain the trained enhanced model D; The loss function of the enhanced model D is calculated as: Where Ld is the loss function of the enhanced model D; Nd is the total number of data samples in the enhanced model D; y id is the true label of the idth sample, p id Predict the probability that the idth sample is class 1 for the enhanced model D; S3.4.

3. The node output values ​​of the four output units corresponding to the enhanced model D are d i , the node output values ​​of the four output units are initially converted to obtain d i The corresponding probability value f di The calculation formula is: Then the node output values ​​of the four output units are further optimized and converted to obtain the output value d i The corresponding enhanced probability value g di The calculation formula is: Among them, G is the enhancement coefficient, and then the actual disease corresponding to each node is marked as h di ; S3.4.

4. Use g di The training set obtained in step S2 is combined with the error chain derivation rule to train the enhanced model D. After the training is completed, the output values ​​of the four types of diseases corresponding to the enhanced model D are j di ; S3.

5. Based on j bi and g ai , establish the loss function between the first auxiliary model B and the output layer of the basic model A, and 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 relationship loss function L between the output layer of the first auxiliary model B and the basic model A 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 S3.

6. Establish the relationship loss function L between the output layer of the second auxiliary model C and the first auxiliary model B CBB , the expression is: L CBB =c CB ·L CB +g BB ·L BB Among them, γ CB For L CB The weight coefficient, γ BB For L BB The weight coefficient, L CB Based on j ci and g bi The loss function of L BB Based on f bi and h bi The loss function of L CB The calculation formula is: L BB The calculation formula is: S3.

7. Establish the relationship loss function L between the output layer of the second auxiliary model C and the basic model A CAA , the expression is: L CAA =c CA ·L CA +g AA ·L AA Among them, γ CA For L CA The weight coefficient, γ AA For L AA The weight coefficient, L CA Based on j ci and g ai The loss function, L AA Based on f ai and h ai The loss function of L CA The calculation formula is: L AA The calculation formula is: S3.

8. Establish the relationship loss function L between the output layer of the enhanced model D and the basic model A DAA , the expression is: L DAA =c DA ·L DA +g AA ·L AA Among them, γ DA For L DA The weight coefficient, γ AA For L AA The weight coefficient, L DA Based on j di and g ai The loss function of L DA The calculation formula is: S3.

9. Establish the relationship loss function L between the output layer of the reinforcement model D and the second auxiliary model C DCC , the expression is: L DCC =c DC ·L DC +g CC ·L CC Among them, γ DC For L DC The weight coefficient, γ CC For L CC The weight coefficient, L DC Based on j di and g ci The loss function, L CC Based on f ci and h ci The loss function of L DC The calculation formula is: L CC The calculation formula is: S4. construct the inter-layer loss function of the enhanced model and the basic model for rapid identification of road congestion; then construct the overall loss function of the rapid identification model of road congestion; S5. Using the data set for rapid identification of road congestion obtained in step S2, the constructed basic model, the first auxiliary model, the second auxiliary model and the enhanced model are trained to obtain the trained basic model as a rapid identification model for road congestion; S6. Improve the road congestion rapid identification model obtained in step S5 to obtain a lightweight road congestion rapid identification model; S7. Constructing a digital twin model based on the lightweight road congestion rapid identification model obtained in step S6; S8. Based on the digital twin model obtained in step S7, a method for analyzing factors causing road congestion is constructed.

2. A method for analyzing factors of road congestion according to claim 1, characterized in that: The specific implementation method of step S2 includes the following steps: S2.

1. Similar diseases of road bumps include longitudinal cracks, transverse cracks, cracks, rutting, potholes, spalling, oil spills, subsidence, and undercuts; S2.

2. Manually select the processed road surface images obtained in step S1, select all road surface images including road surface bumps and similar diseases, and frame the corresponding disease areas, and mark the road surface bumps, longitudinal cracks, transverse cracks, cracks, rutting, potholes, spalling, oil spills, subsidence, and edge gnawing areas in the road surface images as PHO1, PHO2, PHO3, PHO4, PHO5, PHO6, PHO7, PHO8, PHO9, and PHO10 in sequence; S2.

3. Divide the pixel grayscale value 0 to 255 into 16 grayscale intervals, denoted as QJ nu , nu=1,2,3,…,16; S2.

4. Count the number of pixels corresponding to each gray level in the road congestion bag Then, the number of pixels corresponding to each gray level of similar diseases such as longitudinal cracks, transverse cracks, cracks, rutting, potholes, peeling, oiling, subsidence, and gnawing of the road surface is counted. S2.

5. Standardize the number of pixels corresponding to each gray level to obtain the standard number of pixels, where the standard number of pixels corresponding to the road congestion is The calculation formula is: The standard number of pixels corresponding to longitudinal cracks, transverse cracks, cracks, rutting, potholes, spalling, oil spills, subsidence, and edge gnawing is S2.

6. As a benchmark, the relationship function between pavement bumps and other diseases is established in turn, and the calculation formula is: Where ii represents the serial number of other similar diseases, ii = 2, 3, 4, 5…10, and GBQ is obtained. 1-2 、GBQ 1-3 、GBQ 1-4 、GBQ 1-5 、GBQ 1-6 、GBQ 1-7 、GBQ 1-8 、GBQ 1-9 、GBQ 1-10 They represent the relationship functions between road surface bumps and longitudinal cracks, transverse cracks, cracks, rutting, potholes, spalling, oiling, subsidence, and undercutting respectively; S2.

7. Comparison with GBQ 1-2 、GBQ 1-3 、GBQ 1-4 、GBQ 1-5 、GBQ 1-6 、GBQ 1-7 、GBQ 1-8 、GBQ 1-9 、GBQ 1-10 , and sort them in descending order, select the disease types corresponding to the top three relationship functions as the disease types most similar to the image features of road surface bumps, and then select all road surface images containing road surface bumps and the disease types most similar to the image features of road surface bumps; S2.

8. By means of manual labeling, the road surface images obtained in step S2.7 that contain all road surface bumps and the types of damage that are most similar to the image features of the road surface bumps are framed to select the road surface pothole area and the other three types of damage, and the road surface bump is marked as upheaval, and the other three types of damage are marked as other1, other2, and other3, respectively, to obtain a data set for rapid identification of road surface bumps; S2.

9. Divide the dataset for rapid identification of road congestion into a training set and a test set in a ratio of 8:

2.

3. A method for analyzing factors of road congestion according to claim 2, characterized in that: The specific implementation method of step S4 includes the following steps: S4.

1. Assuming that there are a small model a and a large model d, 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 d respectively; S4.

2. Establish a relationship between the second layer in the small model and the fourth layer in the large model d. 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 d is recorded as Md4, and the corresponding width × length × number of channels is nd4 × pd4 × qd4; For size comparison, when qa2<qd4, 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 qd4, 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 Md4 is Establish the loss function between the second layer of the small model a and the fourth layer of the large model d based on qa2<qd4 The calculation formula is: When qa2 ≥ qd4, a convolution operation is added to the output of the fourth layer of the large model d, with a convolution kernel size of 1×1 and a number of convolution kernels of qa2, and the output is Md 4n ; At this time, the output values ​​of each channel corresponding to Ma2 are: The output values ​​of each channel corresponding to Md4 are Establish the loss function between the second layer of the small model a and the fourth layer of the large model d when qa2≥qd4 The calculation formula is: S4.

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 S4.

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 S4.

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 D, the fourth, sixth, and tenth layers in the large model d are the output of pooling layer 2, the output of pooling layer 3, and the output of pooling layer 5 of the enhanced model D, respectively; S4.

6. Based on the loss function between the output layer of the enhanced model D and the base model A and the inter-layer loss function, 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.

4. A method for analyzing factors of road congestion according to claim 3, characterized in that: The specific implementation method of step S5 is to use the data set obtained in step S2 based on the chain derivative rule to train the basic model, the first auxiliary model, the second auxiliary model and the enhanced model through error back propagation. When the overall loss function L total When is minimum, the corresponding model parameters are optimal, and the trained basic model is obtained as a road congestion rapid recognition model.

5. A method for analyzing factors of road congestion according to claim 4, characterized in that: The specific implementation method of step S6 includes the following steps: S6.

1. For the road congestion rapid identification model obtained in step S5, the enhancement coefficient G is taken as follows: G = 1, 3, 5, 7, 9, 11, 15, 20, 30, 40, 50; S6.

2. Based on different enhancement coefficients G, 11 working conditions are obtained, and 11 models are trained to obtain the road surface bump rapid identification model. The corresponding accuracy rates of the 11 road surface bump rapid identification models are set as AACa i' , i'=1,2,3,…,11, the calculation formula is: Among them, TP i' It indicates the number of samples predicted as possessing packets and actually possessing packets when the i'th model trained with the training set recognizes the data in the test set; FP i' It indicates the number of samples predicted as possessing packets but not actually possessing packets when the i'th model trained with the training set identifies the data in the test set; Compare to AACa i' The maximum value of AACa max =max[AACa i′ ]; AAC max The corresponding road surface congestion rapid recognition model is the final lightweight road surface congestion rapid recognition model.

6. A method for analyzing factors of road congestion according to claim 5, characterized in that: The specific implementation method of step S7 includes the following steps: S7.

1. The lightweight road congestion rapid identification model obtained in step S6 is exported 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 lightweight road congestion rapid identification model obtained in step S7.1, and import the road information model format file obtained in step S7.2; S7.

4. Use Python to build an API interface between image data and a lightweight road surface bump rapid identification model for intelligent rutting identification; use Python to build an API interface between a lightweight road surface bump rapid identification model and a road information model to complete the construction of a digital twin model.

7. A method for analyzing factors of road congestion according to claim 6, characterized in that: The specific implementation method of step S8 includes the following steps: S8.

1. Use a 3D structured light camera to collect road surface data and upload it to the digital twin model established in step S7. Then use the lightweight road bump rapid identification model deployed in the system to intelligently identify the road bump area in the road image, select all images containing road bumps, conduct on-site review of the bumps, and measure the height GD of the road bumps; S8.

2. The factors affecting the formation of congestion include aging of pavement materials, ambient temperature, water damage, heavy-loaded vehicles, traffic load, pavement thickness, pavement temperature, and pavement compressive strength; aging of pavement materials is quantified by the asphalt oxide content, water damage is quantified by the pavement seepage, heavy-loaded vehicles are quantified by the number of heavy-loaded vehicles, and traffic load is quantified by the traffic volume; S8.

3. Select 10 time nodes t1, t2, t3, ... t10, collect data at each of the 10 time nodes, and obtain the values ​​of the corresponding quantitative indicators as follows: Asphalt oxide content at 10 time points: LH1, LH2, LH3, LH4, LH5, LH6, LH7, LH8, LH9, LH10; Ambient temperature at 10 time points: HW1, HW2, HW3, HW4, HW5, HW6, HW7, HW8, HW9, HW10; Pavement seepage at 10 time points: SS1, SS2, SS3, SS4, SS5, SS6, SS7, SS8, SS9, SS10; The number of heavy-loaded vehicles at 10 time points: ZZ1, ZZ2, ZZ3, ZZ4, ZZ5, ZZ6, ZZ7, ZZ8, ZZ9, ZZ10; Traffic volume at 10 time nodes: JL1, JL2, JL3, JL4, JL5, JL6, JL7, JL8, JL9, JL10; pavement thickness at 10 time nodes: HD1, HD2, HD3, HD4, HD5, HD6, HD7, HD8, HD9, HD10; Pavement temperature at 10 time points: LW1, LW2, LW3, LW4, LW5, LW6, LW7, LW8, LW9, LW10; Pavement compressive strength at 10 time points: KY1, KY2, KY3, KY4, KY5, KY6, KY7, KY8, KY9, KY10; Road congestion height at 10 time nodes: GD1, GD2, GD3, GD4, GD5, GD6, GD7, GD8, GD9, GD10; S8.

4. Calculate the average values ​​of each quantitative index at 10 time points respectively, and obtain the average value of asphalt oxide content: The average ambient temperature is The average pavement seepage volume is The average number of heavy-loaded vehicles is The average traffic volume is The average pavement thickness is The average road surface temperature is The average compressive strength of the pavement is The average height of road bump is ; S8.

5. Calculate the correlation strength coefficient QD between each quantitative index and the road bump height ZB-GD , the calculation formula is as follows: Among them, ZBii is the quantitative index value at the ii-th time node, GDii is the road surface bump height at the ii-th time node; ZB represents the quantitative evaluation index, including asphalt oxide content, ambient temperature, road surface water seepage, number of heavy-loaded vehicles, traffic volume, road surface thickness, road surface temperature, and road surface compressive strength; that is, the correlation intensity coefficients between asphalt oxide content, ambient temperature, road surface water seepage, number of heavy-loaded vehicles, traffic volume, road surface thickness, road surface temperature, road surface compressive strength and road surface bump height are: QD LH-GD , QD HW-GD , QD SS-GD , QD ZZ-GD , QD JL-GD , QD HD-GD , QD LW-GD , QD KY-GD ; S8.

6. Arrange all the correlation strength coefficients obtained in step S8.5 in descending order and re-label them as TZN1 to TZN8, wherein the influencing factor of the congestion formation corresponding to TZN1 is the influencing factor with the greatest influence on the road congestion; S8.

7. Define the threshold YZ and calculate ni starts from 1 and gradually increases. When , the calculation stops. At this time, the road congestion influencing factor corresponding to the ni value is the main influencing factor causing road congestion.

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 analyzing factors of road congestion formation 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 analyzing factors causing road congestion is implemented as described in any one of claims 1 to 7.

Images