A data-driven method and system for predicting the remaining life of a road surface

Abstract

The application discloses a data-driven pavement residual life prediction method and system, and steps are as follows: step S1: collecting historical pavement data of a to-be-predicted pavement, wherein the historical pavement data comprises historical traffic data and historical weather data; step S2: training a pavement residual life prediction model, establishing a mapping relationship between the historical pavement data and current pavement residual life based on data driving, and obtaining the pavement residual life prediction model; and step S3: collecting current pavement data of the to-be-predicted pavement, merging the current pavement data into the historical pavement data to update the historical pavement data, inputting the updated historical pavement data into the pavement residual life prediction model, and obtaining pavement residual life of the to-be-predicted pavement. The application has the advantages that the historical traffic data and the historical weather data are used to predict the pavement residual life, and the pavement residual life can be predicted in real time and efficiently.

Description

Technical Field

[0001] This invention relates to the field of pavement life prediction technology, specifically to a data-driven method and system for predicting the remaining life of a pavement. Background Technology

[0002] With the continuous expansion of highway density and maintenance mileage, the demand for highway maintenance will increase significantly. To ensure timely and accurate maintenance, it is necessary to predict the remaining life of highway pavements, identify maintenance priorities based on the remaining life of different pavement types, and improve maintenance efficiency. Furthermore, predicting the remaining life of pavements is crucial for developing short- and long-term road maintenance plans, assessing road recovery times, and formulating reconstruction plans.

[0003] Currently, manual inspection is the most accurate method for predicting pavement life, but it requires on-site sample collection and laboratory analysis, which is inefficient and cannot provide real-time prediction of remaining pavement life. Existing automated methods for predicting remaining pavement life typically require the deployment of complex sensor equipment and data analysis of large amounts of pavement data. However, this approach faces numerous challenges in practical applications, including difficulties in sensor installation on roads already in use and excessively high equipment and retrofitting costs.

[0004] In summary, there is an urgent need for a data-driven method and system for predicting the remaining life of road surfaces to address the problems of complex sensor deployment and inability to make real-time predictions in existing technologies. Summary of the Invention

[0005] The purpose of this invention is to provide a data-driven method and system for predicting the remaining life of road surfaces, the specific technical solution of which is as follows:

[0006] This invention discloses a data-driven method for predicting the remaining life of a road surface, characterized by the following steps:

[0007] Step S1: Collect historical road surface data of the road surface to be predicted. The historical road surface data includes historical traffic data and historical weather data. The historical traffic data includes the cumulative mass of vehicles passing through, and the historical weather data includes the historical cumulative temperature difference, historical cumulative sunshine duration, historical cumulative snow and ice coverage duration, historical cumulative rainfall duration, and historical cumulative precipitation.

[0008] Step S2: Train the pavement remaining life prediction model. Specifically, establish a mapping relationship between the historical pavement data in Step S1 and the current remaining life of the pavement based on data-driven methods. The mapping relationship is the original pavement remaining life prediction model.

[0009] Step S3: Real-time prediction of road surface remaining life. Specifically, collect the road surface data of the road surface to be predicted for the current day. The road surface data of the current day includes the traffic data and the weather data of the current day. Merge the road surface data of the current day into the historical road surface data to obtain the updated historical road surface data. Input the updated historical road surface data into the road surface remaining life prediction model in step S2 to obtain the road surface remaining life of the road surface to be predicted.

[0010] The daily traffic data includes the cumulative vehicle mass that passed through that day, and the daily weather data includes the cumulative temperature difference, cumulative sunshine hours, cumulative snow and ice coverage time, cumulative rain time, and cumulative precipitation that day.

[0011] In a preferred embodiment, in step S1, at least three months of historical road surface data for the road surface to be predicted are collected.

[0012] In a preferred embodiment, in step S1, the method for obtaining the cumulative passing vehicle mass is as follows:

[0013] A vehicle type database is established using vehicle model as an index, which allows querying the vehicle characteristics corresponding to each vehicle model. These vehicle characteristics include vehicle type, vehicle curb weight, vehicle load capacity, and passenger capacity, and are updated regularly.

[0014] Images of vehicles traveling on the road surface to be predicted are captured, and image recognition is performed through a convolutional neural network to obtain the vehicle model. Vehicle feature information is extracted from the vehicle type database to identify the vehicle type, which includes non-van trucks, van trucks, buses, cars, and other vehicles.

[0015] For non-box trucks, they are divided into two categories according to whether they are carrying cargo: empty and fully loaded. If there is cargo in the cargo box, it is considered fully loaded; if there is no cargo, it is considered empty. Empty load is calculated based on the curb weight, and fully loaded load is calculated based on the maximum weight. For box trucks, they are all calculated as fully loaded.

[0016] For buses and cars, passenger mass is calculated as half of their authorized passenger capacity, with all passenger mass calculated accordingly; the mass of buses and cars is calculated based on their curb weight.

[0017] For other vehicles, the calculation is based on their full load weight.

[0018] The sum of the masses of vehicles that travel on the predicted road surface is the cumulative mass of passing vehicles.

[0019] In a preferred embodiment, in step S2, the road surface type includes asphalt pavement and cement concrete pavement.

[0020] In a preferred embodiment, in step S2, the mapping relationship is established based on a convolutional neural network.

[0021] In addition, the present invention also includes a data-driven pavement remaining life prediction system, which is used to implement the pavement remaining life prediction method as described above. The system includes a data acquisition unit, a data transmission unit, a data processing unit, and a database.

[0022] The data acquisition unit is used to acquire raw data;

[0023] The data transmission unit is used to transmit the raw data to the data processing unit;

[0024] The data processing unit is used to parse the raw data, obtain historical road surface data and current road surface data, train a road surface remaining life prediction model based on historical road surface data, merge the current road surface data and historical road surface data to obtain updated historical road surface data, and input it into the road surface remaining life prediction model to obtain the road surface remaining life.

[0025] The database is used to store vehicle type data, historical road surface data, and current road surface data. The current road surface data is incorporated into the historical road surface data the following day, and the historical road surface data is updated the following day.

[0026] In a preferred embodiment, the raw data includes road surface images, and the data processing unit obtains the cumulative mass of passing vehicles by parsing the road surface images.

[0027] In a preferred embodiment, the road surface image is acquired via a camera.

[0028] In a preferred embodiment, the data transmission unit transmits the cumulative traffic volume of the road surface to the database for storage at a frequency of once every 30 minutes.

[0029] In a preferred embodiment, the raw data also includes weather information, and the data processing unit obtains historical weather data and current weather data by parsing the weather information.

[0030] The method for predicting the remaining life of a road surface according to the present invention has the following beneficial effects:

[0031] (1) The present invention uses the cumulative mass of passing vehicles, the cumulative temperature difference, the cumulative sunshine time, the cumulative snow and ice coverage time and the cumulative rain time to predict the remaining life of the road surface. The above data are easy to collect and can achieve accurate prediction of road surface life.

[0032] (2) This invention quickly determines the vehicle type by using image recognition based on convolutional neural networks and estimates the vehicle weight based on the vehicle type. This invention does not require the installation of load sensors and will not affect the normal use of the road. The data collection method is quick and convenient.

[0033] In addition, the present invention also discloses a data-driven pavement remaining life prediction system, which is used to implement the pavement remaining life prediction method as described above. The system includes a data acquisition unit, a data transmission unit, a data processing unit, and a database.

[0034] The data acquisition unit is used to acquire raw data;

[0035] The data transmission unit is used to transmit the raw data to the data processing unit;

[0036] The data processing unit is used to parse the raw data, obtain historical road surface data and current road surface data, train a road surface remaining life prediction model based on historical road surface data, and input the current road surface data and historical road surface data into the road surface remaining life prediction model to obtain the road surface remaining life.

[0037] The database is used to store historical road surface data and current road surface data. The current road surface data is incorporated into the historical road surface data on the second day, and the historical road surface data is updated on the second day.

[0038] In a preferred embodiment, the raw data includes road surface images, and the data processing unit obtains the cumulative mass of passing vehicles by parsing the road surface images.

[0039] In a preferred embodiment, the road surface image is acquired via a camera.

[0040] In a preferred embodiment, the data transmission unit transmits the cumulative traffic volume of the road surface to the database for storage at a frequency of once every 30 minutes.

[0041] In a preferred embodiment, the raw data also includes weather information, and the data processing unit obtains historical weather data and current weather data by parsing the weather information.

[0042] The pavement remaining life prediction system of the present invention has the following beneficial effects:

[0043] This invention uses cameras to collect road surface images and analyzes them to obtain historical and current road surface data. The cameras deployed on the road will not affect the normal passage of vehicles, and the cameras can collect road surface images in real time for a long period of time, enabling real-time prediction of the remaining lifespan of the road surface.

[0044] In addition to the objectives, features, and advantages described above, the present invention has other objectives, features, and advantages. The invention will now be described in further detail with reference to the figures. Attached Figure Description

[0045] The accompanying drawings, which form part of this application, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings:

[0046] Figure 1 This is a flowchart of the steps in the method for predicting the remaining life of a road surface in a preferred embodiment of the present invention. Detailed Implementation

[0047] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings. However, the present invention can be implemented in many different ways as defined and covered by the claims.

[0048] Example:

[0049] See Figure 1 This embodiment discloses a data-driven method for predicting the remaining life of a road surface, the steps of which are as follows:

[0050] Step S1: Collect historical pavement data D of the pavement to be predicted R. o (Tr o W o The historical road surface data D o (Tr o W o )Including historical traffic data Tr o (∑m o ) and historical weather data

[0051] Specifically, the historical traffic data Tr o (∑m o This includes the cumulative mass of passing vehicles ∑m o ;

[0052] Specifically, the historical weather data Including the historical cumulative temperature difference change ∑ΔT o Historical cumulative sunshine hours Historical cumulative snow cover time Historical cumulative rainfall time and historical cumulative precipitation ∑P o .

[0053] Step S2: Train the pavement remaining life prediction model. Specifically, based on data-driven methods, establish the historical pavement data D0(Tr) from step S1. o W o The mapping relationship between the current remaining life L of the road surface R to be predicted can be expressed as F:=[R, D0(Tr)] o W oThe mapping relationship is M0, which is the original pavement remaining life prediction model.

[0054] It should be noted that the current remaining lifespan L of the road surface in this embodiment can be obtained through standard testing methods. The current remaining lifespan L is a parameter used for training. The standard testing method used to test the current remaining lifespan of the road surface is existing technology and will not be described in detail here.

[0055] Step S3: Real-time prediction of remaining pavement life, specifically, collecting daily pavement data D of the pavement R to be predicted. n (Tr n W n The road surface data D for that day n (Tr n W n (Including daily traffic data Tr) n (∑m n ) and the weather data for the day D Road surface data for the day n (Tr n W n ) Merged into historical road surface data D o (Tr o W o The updated historical road surface data is obtained from this process. Updated historical road data The original pavement remaining life prediction model M0 is input into step S2 to obtain the pavement remaining life L of the pavement to be predicted. p .

[0056] The traffic data for that day D n (Tr n W n This includes the cumulative mass of vehicles passing through on that day, ∑m. n The weather data for that day Including the cumulative temperature difference change ∑ΔT on the same day n Daily cumulative sunshine hours Accumulated snow and ice cover time of the day Total rainfall for the day and the cumulative precipitation of the day ∑P n .

[0057] The road surface remaining life prediction method in this embodiment uses the cumulative passing vehicle mass ∑m o Historical cumulative temperature difference change ∑ΔT o Historical cumulative sunshine hours Historical cumulative snow cover time Historical cumulative rainfall time and historical cumulative precipitation ∑Po A pavement remaining life prediction model can be constructed. The above data collection is convenient and can achieve accurate pavement life prediction.

[0058] It should be noted that due to differences in the number and type of vehicles traveling on roads at different locations, roads of the same grade and service time may have significant differences in their remaining lifespan. This embodiment introduces the cumulative mass of passing vehicles ∑m. o A remaining life prediction model is constructed to make the remaining life prediction results more accurate. Specifically, in step S1, historical pavement data D for at least the past three months of the pavement R to be predicted is collected. o (Tr o W o For newly opened roads, three months of historical road surface data can be collected. o (Tr o W o () is used to train the model; for roads that have been in use for a long time, six months of historical road surface data can be collected. o (Tr o W o Based on six months of historical road surface data, D o (Tr o W o Estimate the cumulative mass of vehicles that have traveled on the road since it was put into use, ∑m o .

[0059] Specifically, in step S1, the cumulative passing vehicle mass ∑m o The specific method for obtaining it is as follows:

[0060] Vehicle model V m D is used as an index to build the vehicle type database. b (V m You can query each vehicle model V. m The corresponding vehicle characteristics include vehicle type (vehicle types include non-van trucks, van trucks, buses, cars, and other vehicles, where other vehicles are vehicle types other than non-van trucks, van trucks, buses, and cars, such as engineering vehicles and special vehicles), and vehicle curb weight. m Vehicle full load weight Information such as the number of passengers allowed on board is collected and updated regularly.

[0061] Images of vehicles traveling on the road surface to be predicted are captured, and image recognition is performed using a convolutional neural network to obtain the vehicle model V. m Through the vehicle type database D b (V m Extract vehicle feature information to identify vehicle type;

[0062] Non-box trucks are divided into two categories based on whether they are carrying cargo: empty and fully loaded. If there is cargo in the cargo box, it is considered fully loaded; if there is no cargo, it is considered empty. Empty loads are calculated based on the curb weight, while fully loaded loads are calculated based on the maximum weight. Box trucks are always calculated as fully loaded.

[0063] For passenger cars and sedans, passenger mass is calculated as half of their rated passenger capacity, with passenger mass measured in meters. p Calculation; calculating the weight of buses and cars based on their curb weight;

[0064] For other vehicles, the calculation is based on their full load weight.

[0065] For a period of time [t0, t] f The sequence of vehicles {m1, ..., m} that have traveled along the road surface to be predicted i , ..., m j The total mass of vehicles passing through can be obtained by adding up the masses of the vehicles.

[0066] It should be noted that this embodiment uses image recognition based on convolutional neural networks to quickly determine the vehicle type and estimate the vehicle weight based on the vehicle type. Compared to setting up load sensors to measure vehicle weight, this embodiment uses a camera to collect data, which will not affect the normal use of the road, and the data collection method is quick and convenient.

[0067] Specifically, in step S2, the road surface type includes asphalt pavement R. a and cement concrete pavement R c Furthermore, this embodiment can further classify road surface types by highway grade, specifically including asphalt pavement for Class I highways, cement concrete pavement for Class II highways, etc., and obtain subdivided highway pavement types by combining highway grade and road surface type.

[0068] Specifically, in step S2, the mapping relationship is established based on a convolutional neural network.

[0069] In addition, this embodiment also discloses a data-driven pavement remaining life prediction system, which is used to implement the pavement remaining life prediction method as described above. The system includes a data acquisition unit, a data transmission unit, a data processing unit, and a database.

[0070] The data acquisition unit is used to acquire raw data;

[0071] The data transmission unit is used to transmit the raw data to the data processing unit;

[0072] The data processing unit is used to parse the raw data and obtain historical road surface data D. o (Tro W o ) and the road surface data of the day D n (Tr n W n Based on historical road surface data D o (Tr o W o Training the road surface remaining life prediction model M0, using the road surface data D for the day. n (Tr n W n ) and historical road surface data D o (Tr o W o The merged data yields updated historical road surface data. Inputting it into the pavement remaining life prediction model M0 yields the pavement remaining life L. p ;

[0073] The database is used to store vehicle type data D b (V m Historical road surface data D o (Tr o W o ) and the road surface data of the day D n (Tr n W n The road surface data D for that day n (Tr n W n The historical road surface data D was incorporated on the second day. o (Tr o W o In the historical road surface data D, o (Tr o W o The update was completed on the second day.

[0074] Furthermore, the raw data includes road surface images and weather information, and the data processing unit obtains the cumulative mass of passing vehicles ∑m by parsing the road surface images. o The data processing unit obtains historical weather data by parsing weather information. (Historical weather data here does not include current day's weather data) and current day's weather data. It should be noted that weather information can be obtained from the local meteorological bureau.

[0075] Furthermore, in this embodiment, road surface images are captured by a camera and analyzed to obtain historical road surface data D. o (Tr o W o ) and the road surface data of the day Dn (Tr n W n The cameras deployed on the road will not affect the normal passage of vehicles, and the cameras can collect road images in real time for a long period of time, so as to predict the remaining life of the road surface in real time.

[0076] Furthermore, this embodiment uses a camera to acquire road surface images and a data processing unit to analyze the road surface images to obtain real-time road surface data, thereby predicting the remaining lifespan of the road surface in real time.

[0077] In a preferred embodiment, the data transmission unit transmits the cumulative traffic volume ∑m of the road surface at a frequency of once every 30 minutes. o Add to the database.

[0078] In a preferred embodiment, the data processing unit acquires the daily weather data once a day. And then transfer it to the database for storage.

[0079] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A data-driven pavement remaining life prediction method, characterized by, The steps are as follows: Step S1: Collect historical road surface data of the road surface to be predicted. The historical road surface data includes historical traffic data and historical weather data. The historical traffic data includes the cumulative mass of vehicles passing through, and the historical weather data includes the historical cumulative temperature difference, historical cumulative sunshine duration, historical cumulative snow and ice coverage duration, historical cumulative rainfall duration, and historical cumulative precipitation. The method for obtaining the cumulative mass of passing vehicles is as follows: A vehicle type database is established using vehicle model as an index. The characteristics of each vehicle model are queried, including vehicle type, vehicle curb weight, vehicle load capacity, and passenger capacity. The database is updated regularly. Images of vehicles traveling on the road surface to be predicted are captured, and image recognition is performed through a convolutional neural network to obtain the vehicle model. Vehicle feature information is extracted from the vehicle type database to identify the vehicle type, which includes non-van trucks, van trucks, buses, cars, and other vehicles. For non-box trucks, they are divided into two categories according to whether they are carrying cargo: empty and fully loaded. If there is cargo in the cargo box, it is considered fully loaded; if there is no cargo, it is considered empty. Empty load is calculated based on the curb weight, and fully loaded load is calculated based on the maximum weight. For box trucks, they are all calculated as fully loaded. For passenger cars and passenger buses, the passenger mass is calculated as half of the number of passengers for which the vehicle is certified, where the passenger mass is calculated as half of the number of passengers for which the vehicle is certified, and the passenger car and passenger bus mass is calculated as the curb mass. For other vehicles, the calculation is based on their full load weight; The sum of the masses of vehicles that travel on the predicted road surface is the cumulative mass of passing vehicles. Step S2: Train the pavement remaining life prediction model. Specifically, establish a mapping relationship between the historical pavement data in Step S1 and the current remaining life of the pavement based on data-driven methods. This mapping relationship is the original pavement remaining life prediction model. Step S3: Real-time prediction of road surface remaining life. Specifically, collect the road surface data of the road surface to be predicted for the day. The road surface data of the day includes the traffic data and the weather data of the day. Merge the road surface data of the day into the historical road surface data to obtain the updated historical road surface data. Input the updated historical road surface data into the road surface remaining life prediction model in step S2 to obtain the road surface remaining life of the road surface to be predicted. The daily traffic data includes the cumulative vehicle mass that passed through that day, and the daily weather data includes the cumulative temperature difference, cumulative sunshine hours, cumulative snow and ice coverage time, cumulative rain time, and cumulative precipitation that day.

2. The method for predicting the remaining life of a road surface according to claim 1, characterized in that, In step S1, at least three months of historical road surface data for the road surface to be predicted are collected.

3. The method for predicting the remaining life of a road surface according to claim 1, characterized in that, In step S2, the road surface type includes asphalt road surface and cement concrete road surface.

4. The method for predicting the remaining life of a road surface according to claim 1, characterized in that, In step S2, the mapping relationship is established based on a convolutional neural network.

5. A data-driven pavement remaining life prediction system, characterized in that, The system is used to implement the pavement remaining life prediction method as described in any one of claims 1-4, wherein the system includes a data acquisition unit, a data transmission unit, a data processing unit, and a database; The data acquisition unit is used to acquire raw data; The data transmission unit is used to transmit the raw data to the data processing unit; The data processing unit is used to parse the raw data, obtain historical road surface data and current road surface data, train a road surface remaining life prediction model based on historical road surface data, merge the current road surface data and historical road surface data to obtain updated historical road surface data, and input it into the road surface remaining life prediction model to obtain the road surface remaining life. The database is used to store vehicle type data, historical road surface data, and current road surface data. The current road surface data is merged into the historical road surface data on the second day, and the historical road surface data is updated on the second day. The raw data includes road surface images, and the data processing unit obtains the cumulative mass of passing vehicles by parsing the road surface images. The road surface image was acquired via a camera; The raw data also includes weather information, and the data processing unit obtains historical weather data and current weather data by parsing the weather information.

6. The pavement remaining life prediction system according to claim 5, characterized in that, The data transmission unit transmits the cumulative traffic volume of the road surface to the database for storage once every 30 minutes.

Images