WO2024139287A1 - Road disease prediction method and apparatus, electronic device and storage medium - Google Patents

Abstract

A road disease prediction method, comprising: acquiring historical driving vehicle data and historical road environment data of each road section; determining a road disease influence factor according to the historical driving vehicle data and the historical road environment data; and predicting a road disease of a target road section according to the road disease influence factor to obtain a road disease prediction result of the target road section. The road disease influence factor is determined by means of the historical driving vehicle data and the historical road environment data of each road section, and the road disease of the target road section is predicted by means of the road disease influence factor, so that the road disease of the target road section can be predicted in advance, and relevant management departments can carry out targeted maintenance work on the target road section, thereby prolonging the service life of the target road section.

Description

Road damage prediction method, device, electronic device and storage medium Technical Field

The present invention relates to the field of smart cities, and in particular to a road damage prediction method, device, electronic equipment and storage medium.

This application claims the priority of the Chinese patent application filed with the China Patent Office on December 31, 2022, with application number 202211739057.4 and invention name “Road Defect Prediction Method, Device, Electronic Device and Storage Medium”, all contents of which are incorporated by reference in this application.

Background technique

Road CT is a method of scanning roads and identifying road defects through scanned images to obtain the internal road defect conditions. However, road CT scans existing defects in the road and cannot predict road defects that will occur in the future. It is also impossible to perform targeted maintenance on the road in advance to extend the road life.

Technical Solutions

The embodiment of the present invention provides a road disease prediction method, which aims to solve the problem that the existing technology can only scan existing diseases in the road, but cannot predict road diseases that will occur in the future. The road disease influencing factors are determined by the historical driving vehicle data and historical road environment data of each road section, and the road diseases of the target road section are predicted by the road disease influencing factors. It can be predicted in advance what kind of road diseases will occur in the target road section, so that the relevant management department can carry out targeted maintenance work on the target road section, thereby increasing the service life of the target road section.

In a first aspect, an embodiment of the present invention provides a road damage prediction method, the method comprising:

Obtain historical vehicle driving data and historical road environment data for each road section;

Determining road disease influencing factors based on the historical vehicle driving data and historical road environment data;

The road damage of the target road section is predicted according to the road damage influencing factor to obtain a road damage prediction result of the target road section.

Optionally, determining the road disease influencing factor according to the historical vehicle driving data and the historical road environment data includes:

Determine, according to the historical vehicle driving data and the historical road environment data, the impact data of road diseases, wherein the impact data includes a first number of data dimensions;

Performing dimensionality reduction processing on the influencing data to obtain sample dimensionality reduction data, wherein the sample dimensionality reduction data includes a second number of data dimensions, and the second number is smaller than the first number;

Based on the sample dimension reduction data, the road disease influencing factors are determined.

Optionally, the influencing data is subjected to dimensionality reduction processing to obtain sample dimensionality reduction data, wherein the sample dimensionality reduction data includes a second number of data dimensions, including:

Determine a first influence matrix according to the influence data, wherein the first influence matrix includes a first number of data dimensions;

Performing dimensionality reduction processing on the first influence matrix to obtain a second influence matrix;

The sample dimension reduction data is determined based on the first influence matrix and the second influence matrix.

Optionally, performing dimensionality reduction processing on the first influence matrix to obtain a second influence matrix includes:

determining a covariance matrix of the first influence matrix;

Based on the covariance matrix of the first influence matrix, dimension reduction processing is performed on the first influence matrix to obtain a second influence matrix.

Optionally, the performing dimensionality reduction processing on the first influence matrix based on the covariance matrix of the first influence matrix to obtain a second influence matrix includes:

Calculating a covariance matrix of the first influence matrix;

Calculate the eigenvalues of the covariance matrix and the eigenvectors corresponding to the eigenvalues;

The intermediate matrix is obtained according to a second number of the eigenvectors with the largest eigenvalues, and a second influence matrix is determined according to the first influence matrix and the intermediate matrix.

Optionally, obtaining the intermediate matrix according to a second number of the eigenvectors with the largest eigenvalues, and determining the second influence matrix according to the first influence matrix and the intermediate matrix include:

The first influence matrix is multiplied by the intermediate matrix to obtain a second influence matrix.

Optionally, predicting the road damage of the target road section according to the road damage influencing factor to obtain the road damage prediction result of the target road section includes:

Acquiring driving vehicle data and road environment data of the target road section;

Determining a target impact matrix of the target road section according to the road disease impact factor;

Based on the target impact matrix, the road damage of the target road section is predicted to obtain the road damage prediction result of the target road section.

In a second aspect, an embodiment of the present invention provides a road damage prediction device, the device comprising:

An acquisition module is used to acquire historical vehicle driving data and historical road environment data of each road section;

A determination module, used to determine the road disease impact factor based on the historical vehicle driving data and the historical road environment data;

The prediction module is used to predict the road damage of the target road section according to the road damage influencing factor to obtain the road damage prediction result of the target road section.

In a third aspect, an embodiment of the present invention provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the road damage prediction method provided in the embodiment of the present invention when executing the computer program.

In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps in the road damage prediction method provided in the embodiment of the invention are implemented.

In the embodiment of the present invention, the historical driving vehicle data and the historical road environment data of each road section are obtained; the road disease influencing factor is determined according to the historical driving vehicle data and the historical road environment data; the road disease of the target road section is predicted according to the road disease influencing factor, and the road disease prediction result of the target road section is obtained. The road disease influencing factor is determined by the historical driving vehicle data and the historical road environment data of each road section, and the road disease of the target road section is predicted by the road disease influencing factor. It is possible to predict in advance what kind of road disease will occur in the target road section, so that the relevant management department can carry out targeted maintenance work on the target road section, thereby increasing the service life of the target road section. .

BRIEF DESCRIPTION OF THE DRAWINGS

FIG1 is a flow chart of a road damage prediction method provided by an embodiment of the present invention;

FIG2 is a schematic diagram of the structure of a road damage prediction device provided by an embodiment of the present invention;

FIG3 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention.

Embodiments of the present invention

Please refer to FIG. 1 , which is a flow chart of a road damage prediction method provided by an embodiment of the present invention. As shown in FIG. 1 , the road damage prediction method includes the following steps:

101. Obtain historical vehicle driving data and historical road environment data for each road section.

In the embodiment of the present invention, the above-mentioned road sections refer to a plurality of sample road sections, and the above-mentioned sample road sections include road sections where road damage has occurred and road sections where road damage has not occurred.

The above historical driving vehicle data may be data related to the driving vehicle, such as the type of driving vehicle, the size of the driving vehicle, the speed of the driving vehicle, the number of driving vehicles, etc. The above historical driving vehicle data may be determined by image data collected by traffic cameras installed at each road section. Specifically, traffic cameras are installed at each road section, and the traffic images of the corresponding road section are collected by the traffic cameras. Vehicles are identified on the traffic images to obtain driving vehicle data of the corresponding road section, and the driving vehicle data is stored in the server for reuse.

The above historical road environment data may be weather data, road humidity data, road temperature data, road scanning data, road disease data, etc. The above weather data may be obtained from the weather department where the road is located, the above road scanning data may be obtained from the data scanned by the road scanning vehicle, and the above road disease data may be obtained from the road scanning data or from the records of road inspectors.

102. Determine the factors affecting road hazards based on historical vehicle driving data and historical road environment data.

In an embodiment of the present invention, the impact data of road diseases can be determined based on historical vehicle data and historical road environment data, and the road disease impact factor can be determined based on the impact data of road diseases. Specifically, the impact data of road diseases can be subjected to dimensionality reduction processing to obtain the road disease impact factor, and the above-mentioned dimensionality reduction processing can be a plurality of impact factors that have the greatest impact on road diseases among all the impact factors. By performing dimensionality reduction processing on the above-mentioned impact data, dimensionality reduction data with a lower data dimension can be obtained. For example, the impact data of road diseases are the type of traveling vehicles, the size of traveling vehicles, the speed of traveling vehicles, the number of traveling vehicles, weather data, road humidity data, road temperature data, etc., and the road disease impact factor can be the size of traveling vehicles, the speed of traveling vehicles, and the road temperature data.

103. The road damage of the target section is predicted based on the road damage influencing factors to obtain the road damage prediction result of the target section.

In the embodiment of the present invention, after obtaining the road disease influencing factor, the data corresponding to the road disease influencing factor can be screened in the data of the target road section by the road disease influencing factor, and the road disease of the target road section is predicted based on the screened data to obtain the road disease prediction result of the target road section. For example, if the road influencing factor is the size of the traveling vehicle, the speed of the traveling vehicle, and the road temperature data, the size of the traveling vehicle, the speed of the traveling vehicle, and the road temperature data can be screened in the data of the target road section, and the type of road disease that will occur in the target road section can be predicted based on the size of the traveling vehicle, the speed of the traveling vehicle, and the road temperature data.

Specifically, after the data corresponding to the road disease influencing factors are screened out, the data corresponding to the road disease influencing factors can be input into a trained time series model for prediction processing, and the time series model outputs the road disease prediction result of the target road section.

In the embodiment of the present invention, historical vehicle driving data and historical road environment data of each road section are obtained; road disease impact factors are determined based on the historical vehicle driving data and historical road environment data; road diseases of the target road section are predicted based on the road disease impact factors to obtain road disease prediction results of the target road section. By determining the road disease impact factors based on the historical vehicle driving data and historical road environment data of each road section, and predicting the road diseases of the target road section based on the road disease impact factors, it is possible to predict in advance what kind of road diseases will occur in the target road section, thereby enabling the relevant management department to carry out targeted maintenance work on the target road section, thereby increasing the service life of the target road section.

Optionally, in the step of determining the influencing factors of road hazards based on historical vehicle driving data and historical road environment data, the influencing data of road hazards can be determined based on the historical vehicle driving data and historical road environment data, the influencing data including a first number of data dimensions; the influencing data is subjected to dimensionality reduction processing to obtain sample reduced dimensionality data, the sample reduced dimensionality data including a second number of data dimensions, the second number being smaller than the first number; and the influencing factors of road hazards are determined based on the sample reduced dimensionality data.

In the embodiment of the present invention, the above-mentioned road sections include road sections where road damage has occurred and road sections where road damage has not occurred. After obtaining the historical driving vehicle data and historical road environment data of each road section, the historical driving vehicle data and historical road environment data of each road section are sorted to obtain the impact data of road damage. The impact data of the above-mentioned road damage is used to illustrate that the data has more or less an impact relationship with the occurrence of road damage. The impact data of the above-mentioned road damage can be the type of driving vehicle, the size of the driving vehicle, the speed of the driving vehicle, the number of driving vehicles, weather data, road humidity data, road temperature data, etc., and the road damage influencing factors can be the size of the driving vehicle, the speed of the driving vehicle, and the road temperature data. The above-mentioned road diseases mainly include road diseases corresponding to asphalt pavements, road diseases corresponding to cement pavements and road diseases corresponding to facilities along the line. Among them, the types of road diseases corresponding to asphalt pavements may include potholes, cracks, block cracks, cracks (transverse cracks, longitudinal cracks, irregular diagonal cracks), subsidence, rutting, poor road frame, debris, water accumulation, etc.; the types of road diseases corresponding to cement pavements may include panel breakage, potholes, broken plate corners, linear cracks, etc.; the types of road diseases corresponding to facilities along the line may include damaged guardrails, poor road frames, damaged manhole covers, etc.

The above-mentioned road disease impact data has a first number of data dimensions, and the above-mentioned first number of data dimensions can be understood as having a first number of candidate road disease impact factors. After obtaining the road disease impact data, the road disease impact data can be subjected to dimensionality reduction processing to obtain sample title dimensionality reduction data, and the sample dimensionality reduction data has a second number of data dimensions, and the above-mentioned second number of data dimensions can be understood as having a second number of road disease impact factors.

The above-mentioned dimensionality reduction processing can be performed by dimensionality reduction methods such as principal component analysis (PCA) and linear discriminant analysis (LDA). After obtaining the sample dimensionality reduction data, the road disease influencing factors can be determined according to the data dimensions in the sample dimensionality reduction data, and each data dimension corresponds to a road disease influencing factor.

Optionally, in the step of performing dimensionality reduction processing on the influence data to obtain sample dimensionality reduction data, wherein the sample dimensionality reduction data includes a second number of data dimensions, a first influence matrix can be determined based on the influence data, wherein the first influence matrix includes a first number of data dimensions; performing dimensionality reduction processing on the first influence matrix to obtain a second influence matrix; and determining the sample dimensionality reduction data based on the first influence matrix and the second influence matrix.

In an embodiment of the present invention, the number of all road sections is n, and the first number is m, then the impact data of m data dimensions corresponding to the n road sections can be constructed into a first impact matrix of n×m. In the first impact matrix, each column corresponds to a candidate impact factor of road hazards, and each row corresponds to the impact data of a road section.

The first influence matrix can be reduced in dimension by a principal component analysis algorithm or a linear discriminant analysis algorithm to obtain a second influence matrix, wherein the second influence matrix is n×k, k is a second number, that is, the data dimension of the sample reduced dimension data is k. The k can be preset. In the second influence matrix, the first column corresponds to a road disease influencing factor, and the first row corresponds to the influencing data of a road section.

Optionally, in the step of performing dimensionality reduction processing on the first influence matrix to obtain the second influence matrix, the covariance matrix of the first influence matrix can be determined; based on the covariance matrix of the first influence matrix, the first influence matrix is subjected to dimensionality reduction processing to obtain the second influence matrix.

In an embodiment of the present invention, for the first influence matrix, the average value of each column of performance indicator data is calculated, and the average value is subtracted from each column of performance indicator data to obtain the standard deviation corresponding to each performance indicator data, and the covariance corresponding to each performance indicator data is obtained by the square of the standard deviation, and an n×m covariance matrix is obtained based on the covariance.

By extracting the variation information of the performance index data from the covariance matrix, and then reducing the dimension of the covariance matrix through the variation information, it is possible to extract sample reduced dimension data with a high degree of variation, thereby ensuring the high influence of the sample reduced dimension data on road diseases.

Optionally, in the step of performing dimensionality reduction processing on the first influence matrix based on the covariance matrix of the first influence matrix to obtain the second influence matrix, the covariance matrix of the first influence matrix can be calculated; the eigenvalues of the covariance matrix and the eigenvectors corresponding to the eigenvalues are calculated; the intermediate matrix is obtained based on the second number of eigenvectors with the largest eigenvalues, and the second influence matrix is determined based on the first influence matrix and the intermediate matrix.

In an embodiment of the present invention, after obtaining the covariance matrix of the first influence matrix, the eigenvalues and corresponding eigenvectors of the covariance matrix are calculated; the eigenvectors are sorted according to the size of the eigenvalues; the eigenvectors with the largest eigenvalues of the first k are retained to obtain an m×k intermediate matrix. The second influence matrix is determined based on the first influence matrix and the intermediate matrix. The degree of variation of the performance index data is determined by the eigenvalues, so that sample dimensionality reduction data with a high degree of variation can be selected to ensure the high influence of the sample dimensionality reduction data on road diseases.

Each column in the above intermediate matrix corresponds to a road disease influencing factor, and each row corresponds to a eigenvector.

Optionally, in the step of predicting the road damage of the target road section according to the road damage influencing factors to obtain the road damage prediction result of the target road section, the first influence matrix and the intermediate matrix can be multiplied to obtain the second influence matrix.

In the embodiment of the present invention, after obtaining the first impact matrix and the intermediate matrix, matrix calculation is performed on the first impact matrix and the intermediate matrix to obtain the second impact matrix. Specifically, assuming that the first impact matrix is H=n×m and the intermediate impact matrix is D=m×k, then the second impact matrix is G=HD=n×k. In this way, the first impact matrix can be reduced from m-dimension to k-dimension, removing the data dimension that has little impact on road diseases.

Through matrix multiplication, dimensionality reduction processing is performed to obtain sample dimensionality reduction data with a high degree of variation, ensuring the high influence of sample dimensionality reduction data on road diseases.

After the second impact matrix is obtained, the data dimensions corresponding to each column in the second impact matrix may be determined as corresponding road hazard impact factors.

Optionally, in the step of predicting road damages of a target section according to the road damage influencing factors and obtaining a road damage prediction result for the target section, driving vehicle data and road environment data of the target section can be obtained; a target influence matrix of the target section can be determined according to the road damage influencing factors; and road damages of the target section can be predicted based on the target influence matrix to obtain a road damage prediction result for the target section.

In an embodiment of the present invention, the driving vehicle data of the target road section includes the type of driving vehicle, the size of driving vehicle, the speed of driving vehicle, the number of driving vehicles and other data related to driving vehicles. The driving vehicle data of the target road section can be determined by the image data collected by the traffic camera set on the target road section. Specifically, a traffic camera is set on the target road section, and the traffic image of the target road section is collected by the traffic camera, and the vehicle recognition is performed on the traffic image to obtain the driving vehicle data of the target road section, and the driving vehicle data is stored in the server for reuse.

The road environment data may be weather data of the target road section, road humidity data, road temperature data, road scanning data, road disease data, etc. The weather data may be obtained from the weather department of the target road, the road scanning data may be obtained from the data obtained when a road scanning vehicle scans the target road, and the road disease data may be obtained from the road scanning data of the target road section or from the records of road inspectors of the target road section.

After obtaining the driving vehicle data and road environment data of the target road section, the data related to the road disease factor is screened out to construct a target impact matrix G = 1 × k × T. In the above target impact matrix, k represents k road disease impact factors, and T represents the time dimension. After obtaining the target impact matrix, the target impact matrix is input into the time series model for prediction to obtain the road disease prediction result of the target road section.

The above-mentioned time series model can be a time series model based on a recurrent neural network RNN or a long short-term memory network LSTM. When training a time series model, a data set and an initial time series model can be obtained. The above-mentioned data set includes a sample influence matrix and a true disease label. The above-mentioned sample influence matrix and the target influence matrix are obtained by the same processing method. The sample influence matrix is input into the initial time series model to obtain the sample disease prediction result, and the error loss between the sample disease prediction result and the true disease label is calculated. With the minimum error loss as the optimization goal, the parameters of the initial time series model are adjusted, and the above-mentioned parameter adjustment process is iterated until the number of iterations reaches the preset number, or the initial time series model converges at the minimum error loss, the training is stopped, and the trained time series model is obtained.

After obtaining the trained time series model, the target impact matrix is input into the trained time series model, and the road damage prediction result of the target road section is output through the trained time series model. The road damage prediction result of the judge includes the road damage type and the time when the road damage occurs.

It should be noted that the road damage prediction method provided in the embodiment of the present invention can be applied to smart cameras, smart phones, computers, servers and other devices that can perform the road damage prediction method.

Optionally, please refer to FIG. 2 , which is a schematic diagram of the structure of a road disease prediction device provided by an embodiment of the present invention. As shown in FIG. 2 , the device includes:

An acquisition module 201 is used to acquire historical vehicle driving data and historical road environment data of each road section;

A determination module 202, for determining a road disease impact factor based on the historical vehicle driving data and the historical road environment data;

The prediction module 203 is used to predict the road damage of the target road section according to the road damage influencing factor to obtain the road damage prediction result of the target road section.

Optionally, the determining module 202 includes:

A first determination submodule is used to determine the impact data of road diseases according to the historical vehicle driving data and the historical road environment data, wherein the impact data includes a first number of data dimensions;

a processing submodule, configured to perform dimensionality reduction processing on the influencing data to obtain sample dimensionality reduction data, wherein the sample dimensionality reduction data includes a second number of data dimensions, and the second number is smaller than the first number;

The second determination submodule is used to determine the road disease influencing factors based on the sample dimension reduction data.

Optionally, the processing submodule includes:

A first determining unit, configured to determine a first influence matrix according to the influence data, wherein the first influence matrix includes a first number of data dimensions;

A processing unit, configured to perform dimensionality reduction processing on the first influence matrix to obtain a second influence matrix;

The second determining unit is used to determine the sample dimension reduction data based on the first influence matrix and the second influence matrix.

Optionally, the processing unit includes:

A determination subunit, configured to determine a covariance matrix of the first influence matrix;

A processing subunit is used to perform dimensionality reduction processing on the first influence matrix based on the covariance matrix of the first influence matrix to obtain a second influence matrix.

Optionally, the processing sub-unit is also used to calculate the covariance matrix of the first influence matrix; calculate the eigenvalues of the covariance matrix and the eigenvectors corresponding to the eigenvalues; obtain the intermediate matrix based on the second number of eigenvectors with the largest eigenvalues, and determine the second influence matrix based on the first influence matrix and the intermediate matrix.

Optionally, the processing subunit is further used to multiply the first influence matrix and the intermediate matrix to obtain a second influence matrix.

Optionally, the prediction module 203 includes:

An acquisition submodule, used to acquire the driving vehicle data and road environment data of the target road section;

A third determination submodule is used to determine the target impact matrix of the target road section according to the road disease impact factor;

The prediction submodule is used to predict the road damage of the target road section based on the target impact matrix to obtain the road damage prediction result of the target road section.

It should be noted that the road damage prediction device provided in the embodiment of the present invention can be applied to smart cameras, smart phones, computers, servers and other devices that can perform road damage prediction methods.

The road damage prediction device provided in the embodiment of the present invention can implement each process implemented by the road damage prediction method in the above method embodiment and can achieve the same beneficial effects. To avoid repetition, it will not be described here.

Referring to FIG. 3 , FIG. 3 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. As shown in FIG. 3 , the electronic device includes: a memory 302, a processor 301, and a computer program of a road damage prediction method stored in the memory 302 and executable on the processor 301, wherein:

The processor 301 is used to call the computer program stored in the memory 302 and execute the following steps:

Obtain historical vehicle driving data and historical road environment data for each road section;

Determining road disease influencing factors based on the historical vehicle driving data and historical road environment data;

The road damage of the target road section is predicted according to the road damage influencing factor to obtain a road damage prediction result of the target road section.

Optionally, the determining of the road disease impact factor according to the historical vehicle driving data and the historical road environment data performed by the processor 301 includes:

Determine, according to the historical vehicle driving data and the historical road environment data, the impact data of road diseases, wherein the impact data includes a first number of data dimensions;

Performing dimensionality reduction processing on the influencing data to obtain sample dimensionality reduction data, wherein the sample dimensionality reduction data includes a second number of data dimensions, and the second number is smaller than the first number;

Based on the sample dimension reduction data, the road disease influencing factors are determined.

Optionally, the processor 301 performs dimensionality reduction processing on the impact data to obtain sample dimensionality reduction data, where the sample dimensionality reduction data includes a second number of data dimensions, including:

Determine a first influence matrix according to the influence data, wherein the first influence matrix includes a first number of data dimensions;

Performing dimensionality reduction processing on the first influence matrix to obtain a second influence matrix;

The sample dimension reduction data is determined based on the first influence matrix and the second influence matrix.

Optionally, the processor 301 performs dimensionality reduction processing on the first influence matrix to obtain a second influence matrix, including:

determining a covariance matrix of the first influence matrix;

Based on the covariance matrix of the first influence matrix, dimension reduction processing is performed on the first influence matrix to obtain a second influence matrix.

Optionally, the processor 301 performs dimensionality reduction processing on the first influence matrix based on the covariance matrix of the first influence matrix to obtain a second influence matrix, including:

Calculating a covariance matrix of the first influence matrix;

Calculate the eigenvalues of the covariance matrix and the eigenvectors corresponding to the eigenvalues;

The intermediate matrix is obtained according to a second number of the eigenvectors with the largest eigenvalues, and a second influence matrix is determined according to the first influence matrix and the intermediate matrix.

Optionally, the processor 301 executes the step of obtaining the intermediate matrix according to the second number of eigenvectors with the largest eigenvalues, and determining the second influence matrix according to the first influence matrix and the intermediate matrix, including:

The first influence matrix is multiplied by the intermediate matrix to obtain a second influence matrix.

Optionally, the processor 301 predicts the road damage of the target road section according to the road damage influencing factor to obtain the road damage prediction result of the target road section, including:

Acquiring driving vehicle data and road environment data of the target road section;

Determining a target impact matrix of the target road section according to the road disease impact factor;

Based on the target impact matrix, the road damage of the target road section is predicted to obtain the road damage prediction result of the target road section.

The electronic device provided in the embodiment of the present invention can implement each process implemented by the road damage prediction method in the above method embodiment and can achieve the same beneficial effect. To avoid repetition, it will not be described here.

An embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the various processes of the road damage prediction method provided by the embodiment of the present invention are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The storage medium can be a disk, an optical disk, a read-only storage memory (RON) or a random access memory (RAN).

Claims (20)

1. [Incorporated by reference (Rule 20.6) 12.12.2023]
   A road damage prediction method, characterized in that it comprises the following steps:

   Obtain historical vehicle driving data and historical road environment data for each road section;

   Determining road disease influencing factors based on the historical vehicle driving data and historical road environment data;

   The road damage of the target road section is predicted according to the road damage influencing factor to obtain a road damage prediction result of the target road section.
2. [Incorporated by reference (Rule 20.6) 12.12.2023]
   The road damage prediction method according to claim 1, characterized in that the road damage influencing factor is determined based on the historical vehicle driving data and the historical road environment data, comprising:

   Determine, according to the historical vehicle driving data and the historical road environment data, the impact data of road diseases, wherein the impact data includes a first number of data dimensions;

   Performing dimensionality reduction processing on the influencing data to obtain sample dimensionality reduction data, wherein the sample dimensionality reduction data includes a second number of data dimensions, and the second number is smaller than the first number;

   Based on the sample dimension reduction data, the road disease influencing factors are determined.
3. [Incorporated by reference (Rule 20.6) 12.12.2023]
   The road damage prediction method according to claim 2, characterized in that the dimension reduction processing is performed on the impact data to obtain sample dimension reduction data, and the sample dimension reduction data includes a second number of data dimensions, including:

   Determine a first influence matrix according to the influence data, wherein the first influence matrix includes a first number of data dimensions;

   Performing dimensionality reduction processing on the first influence matrix to obtain a second influence matrix;

   The sample dimension reduction data is determined based on the first influence matrix and the second influence matrix.
4. [Incorporated by reference (Rule 20.6) 12.12.2023]
   The road damage prediction method according to claim 3, characterized in that the step of performing dimensionality reduction processing on the first influence matrix to obtain the second influence matrix comprises:

   determining a covariance matrix of the first influence matrix;

   Based on the covariance matrix of the first influence matrix, dimension reduction processing is performed on the first influence matrix to obtain a second influence matrix.
5. [Incorporated by reference (Rule 20.6) 12.12.2023]
   The road damage prediction method according to claim 4, characterized in that the covariance matrix based on the first influence matrix performs dimensionality reduction processing on the first influence matrix to obtain the second influence matrix, comprising:

   Calculating a covariance matrix of the first influence matrix;

   Calculate the eigenvalues of the covariance matrix and the eigenvectors corresponding to the eigenvalues;

   The intermediate matrix is obtained according to a second number of the eigenvectors with the largest eigenvalues, and a second influence matrix is determined according to the first influence matrix and the intermediate matrix.
6. [Incorporated by reference (Rule 20.6) 12.12.2023]
   The road damage prediction method according to claim 5 is characterized in that the step of obtaining the intermediate matrix according to the second number of eigenvectors with the largest eigenvalues, and determining the second influence matrix according to the first influence matrix and the intermediate matrix comprises:

   The first influence matrix is multiplied by the intermediate matrix to obtain a second influence matrix.
7. [Incorporated by reference (Rule 20.6) 12.12.2023]
   The road damage prediction method according to claim 6 is characterized in that the road damage of the target road section is predicted according to the road damage influencing factor to obtain the road damage prediction result of the target road section, comprising:

   Acquiring driving vehicle data and road environment data of the target road section;

   Determining a target impact matrix of the target road section according to the road disease impact factor;

   Based on the target impact matrix, the road damage of the target road section is predicted to obtain the road damage prediction result of the target road section.
8. [Incorporated by reference (Rule 20.6) 12.12.2023]
   A road damage prediction device, characterized in that the device comprises:

   An acquisition module is used to acquire historical vehicle driving data and historical road environment data of each road section;

   A determination module, used to determine the road disease impact factor based on the historical vehicle driving data and the historical road environment data;

   The prediction module is used to predict the road damage of the target road section according to the road damage influencing factor to obtain the road damage prediction result of the target road section.
9. [Incorporated by reference (Rule 20.6) 12.12.2023]
   An electronic device, characterized in that it comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps in the road damage prediction method as described in any one of claims 1 to 7 are implemented.
10. [Incorporated by reference (Rule 20.6) 12.12.2023]
    A computer-readable storage medium, characterized in that a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the steps in the road damage prediction method as described in any one of claims 1 to 7 are implemented.
11. [Incorrect filing (Rule 20.5bis)]
    A method for identifying road damage, applicable to a device, is characterized by comprising the following steps:

    Extracting a feature map of a road damage image, wherein the road damage image includes a target object;

    Inputting the feature map into a preset target neural network prediction model for region segmentation, and calculating weight values according to the region pixel values of the segmented region;

    Based on the weight value of the segmented region and the pixel value of the region, a downsampled feature map is obtained, wherein the downsampled feature map includes the target object;

    Performing feature information enhancement on the target object in the downsampled feature map to obtain a target feature map, wherein the target feature map includes a confidence level of a disease category of the target object;

    The disease category to which the target object belongs is determined based on the confidence level.
12. [Incorrect filing (Rule 20.5bis)]
    The method according to claim 1, characterized in that the step of inputting the feature map into a preset target neural network prediction model for region segmentation, and calculating weight values according to regional pixel values of the segmented regions, comprises:

    After preprocessing, the feature graph is input into the preset target neural network prediction model for training, wherein the preset target neural network prediction model includes a backbone network model;

    Based on the backbone network model, the road damage image is segmented into regions according to a preset size, and each region obtained by segmentation corresponds to a pixel value of the region;

    The segmented regions are normalized by a normalized exponential function, and the region pixel value of each region is converted into the weight value.
13. [Incorrect filing (Rule 20.5bis)]
    The method according to claim 1, characterized in that the step of obtaining a downsampled feature map based on the weight value of the segmented region and the pixel value of the region comprises:

    For each region obtained by segmentation, a region pixel weight of the region is calculated based on the weight value of the region and the region pixel value;

    Based on the calculated regional pixel weights of all regions, the downsampled feature map is obtained.
14. [Incorrect filing (Rule 20.5bis)]
    The method according to claim 2, characterized in that the step of enhancing feature information of the target object features in the downsampled feature map to obtain a target feature map comprises:

    Performing a convolution operation on the downsampled feature map to obtain a first convolution sub-feature map, a second convolution sub-feature map, and a third convolution sub-feature map;

    Transpose the second convolution sub-feature map, and calculate the dot product with the first convolution sub-feature map to obtain a feature convolution matrix;

    Normalizing the feature convolution matrix based on the normalized exponential function to obtain a pixel weight value corresponding to each pixel in the downsampled feature map;

    The feature map to be decomposed is obtained based on the dot product calculation of the pixel weight value corresponding to each pixel in the downsampled feature map and the third convolution sub-feature map;

    The feature map to be decomposed is decomposed and normalized to obtain the target feature map.
15. [Incorrect filing (Rule 20.5bis)]
    The method according to claim 4, characterized in that the decomposing and normalizing the feature map to be decomposed to obtain the target feature map comprises:

    Decomposing the to-be-decomposed characteristic graph into a first decomposition characteristic graph, a second decomposition characteristic graph, and a third decomposition characteristic graph;

    The first decomposition feature graph, the second decomposition feature graph and the third decomposition feature graph are respectively encoded and decoded in sequence to obtain a decoded first decomposition feature graph, a decoded second decomposition feature graph and a decoded third decomposition feature graph;

    The decoded first decomposition feature graph, the decoded second decomposition feature graph and the decoded third decomposition feature graph are connected and subjected to standard normalization processing to obtain the target feature graph.
16. [Incorrect filing (Rule 20.5bis)]
    The method according to claim 2, characterized in that the preprocessing of the feature map comprises:

    The feature map of the road damage image is scaled according to a preset scaling ratio to obtain a preset standard input size of the target neural network prediction model.
17. [Incorrect filing (Rule 20.5bis)]
    A road disease re-identification method, applicable to a platform end, is characterized by comprising the following steps:

    The platform end obtains a road disease image with road diseases uploaded by the device end, and based on the position coordinates of the road disease image, collects a plurality of re-identified images of the same category under different time and space conditions;

    Re-identify the plurality of re-identified images using a feature loss function, and calculate feature distances on the output multi-dimensional features;

    If the characteristic distance meets the preset distance threshold, it is determined that the road damage image and the re-identified image are of the same damage category.
18. [Incorrect filing (Rule 20.5bis)]
    A road disease identification device, applicable to a device end, characterized by comprising:

    A first acquisition module is used to extract a feature map of a road damage image, wherein the road damage image includes a target object;

    A segmentation module, used for inputting the feature map into a preset target neural network prediction model for region segmentation, and calculating weight values according to the region pixel values of the segmented region;

    A second acquisition module is used to acquire a down-sampled feature map based on the weight value of the segmented region and the pixel value of the region, wherein the down-sampled feature map includes the target object;

    A feature enhancement module, used for enhancing feature information of the target object in the downsampled feature map to obtain a target feature map, wherein the target feature map includes a confidence level of a disease category of the target object;

    The first judgment module is used to judge the disease category of the target object based on the confidence level.
19. [Incorrect filing (Rule 20.5bis)]
    An electronic device, characterized in that it comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements a method for identifying road damage as described in any one of claims 1 to 6 and the steps in a method for identifying road damage as described in claim 7.
20. [Incorrect filing (Rule 20.5bis)]
    A computer-readable storage medium, characterized in that a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, it implements the method for identifying road damage as described in any one of claims 1 to 6 and the steps in the method for identifying road damage as described in claim 7.