A dynamic prediction method for road surface performance based on high-frequency intelligent inspection data

Abstract

The application relates to a kind of dynamic prediction method of road surface performance based on high-frequency inspection data, which comprises the following steps: step 1: according to road surface performance condition data, respectively construct decay model, and determine the initial model parameters of decay model;Step 2: the current time is set as T0, and the high-frequency inspection time section t1 and the prediction time section t2 for accumulating prediction data are determined;Step 3: the decay model parameters of various road surface performance condition data are calculated, and the decay model parameters are updated according to the regression calculation result;Step 4: verify the reliability of the decay model parameters of various road surface performance condition data;Step 5: based on the decay model after verification, the road surface performance condition data of each type is predicted and weighted calculation is carried out, and the road surface performance comprehensive evaluation index is obtained to represent the road surface performance.Compared with the prior art, the application has the advantages of improving the scientificity and reliability of the prediction model and improving the accuracy of road surface performance prediction.

Description

Technical Field

[0001] This invention relates to the field of road engineering, and in particular to a method for dynamic prediction of road surface performance based on high-frequency intelligent inspection data. Background Technology

[0002] Road pavement performance prediction is an important part of proposing road maintenance strategies, determining maintenance funding, and formulating road network maintenance plans. Accurate and rapid prediction of road pavement performance allows for the development of scientific and reliable pre-maintenance measures based on the trend of road pavement damage, preventing further deterioration of pavement performance, effectively ensuring road service capacity, and extending road service life.

[0003] Existing methods for assessing pavement performance status are based on long-term, periodic inspection data and use statistical regression or mechanical-empirical analysis methods to directly predict the PCI or PQI indicators. However, they do not separately consider factors such as smoothness, road surface bounce, road cracks, potholes, rutting, skid resistance coefficient, road wear, and pavement structural strength involved in the calculation of PCI and PQI. Furthermore, due to the large granularity of periodic inspection data (usually once a year and unable to cover all lanes), the predictive accuracy of statistical regression methods is insufficient, and mechanical-empirical analysis methods involve multiple external parameters (environment, traffic volume, road structure, etc.), making calculations complex and difficult to obtain, thus leading to computational difficulties.

[0004] High-frequency inspection data is an effective supplement to periodic inspection data. It can usually cover a road section every 1 to 5 days. Existing intelligent inspection equipment has the ability to inspect various types of road damage, such as road surface smoothness and road surface damage, providing complete and high-frequency road performance data for road performance prediction. However, there is currently no prediction method for dynamic road performance based on high-frequency inspection data. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the existing technology and provide a method for dynamic prediction of road performance based on high-frequency intelligent inspection data.

[0006] The objective of this invention can be achieved through the following technical solutions:

[0007] A method for dynamic prediction of road surface performance based on high-frequency inspection data, the method comprising the following steps:

[0008] Step 1: Grid the road segment to be predicted according to a fixed length and lanes, and construct decay models based on the obtained road surface performance data, and determine the initial model parameters of the decay models;

[0009] Step 2: Set the current time as T0, and determine the high-frequency inspection time interval t1 and the prediction time interval t2 for the accumulation of prediction data;

[0010] Step 3: Based on the high-frequency inspection data within the prediction data accumulation period, perform regression calculations on the decay model parameters of various road surface performance data, and update the decay model parameters according to the regression calculation results;

[0011] Step 4: Verify the reliability of the decay model parameters for the updated pavement performance data;

[0012] Step 5: Based on the validated decay model, predict various types of pavement performance data, perform weighted calculations on various types of data, and obtain a comprehensive pavement performance evaluation index to characterize pavement performance.

[0013] In step 1, the grid length of the road segment to be predicted ranges from 50 to 200 meters, and the grid is defined as R1, R2, ... R i ,……R n Calculate various road surface performance data within each grid;

[0014] The pavement performance data includes smoothness within the grid, pavement bounce index, pavement cracks, pavement potholes, pavement rutting, pavement skid resistance coefficient, pavement wear, and pavement structural strength. Among them, the smoothness within the grid is characterized by the International Roughness Index (IRI), the pavement bounce index is characterized by the root mean square value of short-time acceleration (rmsa), pavement cracks and potholes are characterized by the total crack length (Lc) and the total pothole area (Sp) within the grid, respectively, pavement rutting is characterized by rutting depth (RD), pavement skid resistance coefficient is characterized by the lateral force coefficient (SFC), pavement wear is characterized by the wear rate (WR), and pavement structural strength is characterized by the pavement structural strength coefficient (SSR).

[0015] Different types of pavement performance data correspond to different decay models, including linear models, exponential models, negative exponential models, and S-models.

[0016] In step 2, the high-frequency inspection time interval t1 for the prediction data accumulation is not less than 30 days, and the prediction time interval t2 is not greater than 100 days, so as to ensure the reliability of the parameters used for decay model regression.

[0017] In step 3, the process of regressing the decay model parameters of various road surface performance data based on high-frequency inspection data within the prediction data accumulation period is as follows:

[0018] Step 301: Establish time-series data of pavement performance based on the high-frequency inspection time interval t1 and the prediction time interval t2 used for data accumulation. and Where n corresponds to different road surface performance data categories, r i and p j These represent the i-th road segment grid and the j-th inspection time, respectively.

[0019] Step 302: Based on the time series data of road surface performance during the high-frequency inspection time segment t1 For the decay models of smoothness, road surface bumps, road surface cracks, road surface potholes, road surface ruts, road surface skid resistance coefficient, road surface wear and road surface structural strength, linear regression or nonlinear regression methods are used for regression analysis to obtain the decay model parameters of various decay models.

[0020] Step 3, which involves updating the parameters of each decay model based on the regression calculation results of the predicted time period, specifically includes the following steps:

[0021] Step 303: Calculate the performance data of various road surfaces within the predicted time interval t2 based on the regression calculation results.

[0022] Step 304: Combine the prediction results obtained in step 301 with the measured pavement performance data within the prediction time interval t2. Compare and construct the objective function;

[0023] Step 305: Determine the parameters to be updated and the boundary conditions for various decay models;

[0024] Step 306: Perform iterative calculations based on the objective function, the parameters to be updated, and the boundary conditions. Stop iterating when the objective function is less than the allowable error or when the maximum number of iterations is reached, and use the optimal solution among the parameters to be updated as the final calibration parameters.

[0025] Step 307: Repeat steps 301 to 306 to obtain the decay model parameters of various pavement performance data for different grid road segments.

[0026] In step 304, the objective function is used to represent the time-series data of the predicted road surface performance conditions. Time series data of various road surface performance conditions measured in actual tests The error can be calculated using methods such as Euclidean distance, Minkowski distance, Manhattan distance, or cosine similarity.

[0027] When using Euclidean distance for calculation, the time-series data of various pavement performance conditions are normalized, and the Euclidean distance between the predicted results and the measured results is calculated, thus obtaining the objective function. The expression of the objective function is as follows:

[0028]

[0029] Where n represents the types of road surface performance conditions, N represents the total number of categories, j represents the inspection time, t represents the total number of inspections, and Norm(·) is the normalization function.

[0030] In step 305, the parameters to be updated are the parameters in the decay model of various road performance data. When the decay model adopts a linear model, the parameter to be updated is the linear decay rate α. linear When the decay model is an exponential or negative exponential model, the parameter to be updated is the exponential decay rate α. exp or -α exp When the decay model is an S-model, the parameter to be updated is the correlation coefficient α. s and β s The initial values ​​of the parameters to be updated are set to 20 to 50 sets and are randomly generated according to the boundary conditions;

[0031] The boundary conditions are the range of values ​​for the parameters to be updated.

[0032] In step 306, the algorithms used for iterative calculation include multi-objective particle swarm optimization, multi-objective genetic algorithm, and nonlinear least squares method, and the maximum number of iterations in the iterative calculation process does not exceed 200.

[0033] In step 4, the process of verifying the reliability of the decay model parameters of the updated pavement performance data is as follows:

[0034] Determine the verification time period t3, and collect high-frequency inspection data within the verification time period t3 to obtain time-series data of the measured pavement performance. Simultaneously, the decay model parameters from step 3 are substituted into each grid segment to calculate the time-series data of the predicted pavement performance within the verification time interval t3. The time series data of predicted road surface performance within the verification time interval t3 were calculated. Time series data of measured pavement performance The error judgment determines whether the decay model parameters are reliable, that is, whether the error of the road performance data is less than the error threshold. If so, the decay model parameters are reliable, and the decay model is used to continue to perform prediction calculations until the error exceeds the set error threshold. If the error exceeds the error threshold, return to step 2. If not, return to step 2 until the decay model parameters are reliable.

[0035] Compared with the prior art, the present invention has the following beneficial effects:

[0036] 1. Based on high-frequency collected data on various road surface performance conditions, this invention uses optimization and iterative calculation methods to dynamically correct and periodically adjust the relevant parameters of the road surface performance prediction model, which can greatly improve the accuracy of road surface performance prediction.

[0037] 2. This invention performs decay model regression and prediction separately for different categories of pavement performance indicators, overcoming the limitations of existing pavement performance prediction methods that mostly predict directly based on comprehensive indicators such as PCI and PQI without considering the inconsistency in decay rates of indicators such as smoothness, road surface bumps, road surface cracks, road surface potholes, road surface rutting, road surface skid resistance coefficient, road surface wear and road surface structural strength. This can further improve the scientificity and reliability of the prediction model. Attached Figure Description

[0038] Figure 1 This is a flowchart of the method of the present invention.

[0039] Figure 2 This is a flowchart illustrating a specific method in an embodiment of the present invention.

[0040] Figure 3 This is a schematic diagram illustrating the representation of various road surface performance data after the road segment to be predicted is gridded in an embodiment of the present invention. Detailed Implementation

[0041] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0042] like Figures 1-2 As shown, this invention provides a method for dynamic prediction of road surface performance based on high-frequency inspection data. The method includes the following steps:

[0043] Step 1: As Figure 3 As shown, the road segment to be predicted is gridded according to a fixed length and lanes. A decay model is constructed for each road surface performance data (road surface damage), and the initial model parameters of the decay model are determined.

[0044] Step 2: Set the current time as T0, and determine the high-frequency inspection time period t1 and the prediction time interval t2 for the accumulation of prediction data;

[0045] Step 3: Based on the high-frequency inspection data within the prediction data accumulation period, perform regression calculations on the decay model parameters of various road surface performance data, and update the decay model parameters according to the regression calculation results;

[0046] Step 4: Verify the reliability of the decay model parameters for the updated pavement performance data;

[0047] Step 5: Based on the validated decay model, predict various types of pavement performance data, perform weighted calculations on various types of data, and obtain a comprehensive pavement performance evaluation index to characterize pavement performance.

[0048] In step 1, the grid length of the road segment to be predicted ranges from 50 to 200 m, and the road grid is defined as R1, R2, ... R i ,……R n Within each grid, various pavement performance data are calculated, including pavement smoothness, pavement bounce index, pavement cracks, pavement potholes, pavement rutting, pavement skid resistance coefficient, pavement wear, and pavement structural strength. Among them, pavement smoothness is characterized by the International Roughness Index (IRI), pavement bounce index is characterized by the root mean square value of short-time acceleration (rmsa), pavement cracks and potholes are characterized by the total crack length (Lc) and the total pothole area (Sp) within the grid, respectively, pavement rutting is characterized by rutting depth (RD), pavement skid resistance coefficient is characterized by the lateral force coefficient (SFC), pavement wear is characterized by the wear rate (WR), and pavement structural strength is characterized by the pavement structural strength coefficient (SSR).

[0049] Different decay models are selected for different types of road damage. Road damage types such as road surface roughness, road surface cracks, and road surface potholes usually show a continuous deterioration trend under external loads, so linear or exponential decay models can be used. Road damage types such as road rutting, skid resistance coefficient, road surface wear, road surface structural strength, and road surface bumps are affected by external factors and deteriorate rapidly, but usually converge after deterioration to a certain extent. Therefore, based on their damage trend, linear, negative exponential, or S-model decay models can be considered. The initial model parameters are set based on experience.

[0050] In step 2, the high-frequency inspection period t1 of the prediction data accumulation is not less than 30 days to ensure the reliability of the model parameter regression. The prediction period t2 is not greater than 100 days. Too long a prediction period will lead to insufficient prediction accuracy. The inspection period of the high-frequency inspection data is 1 to 7 days. The inspection period of the high-frequency inspection data refers to the time difference between two adjacent inspections.

[0051] In step 3, the process of performing regression calculations on the decay model parameters of various road surface performance data based on high-frequency inspection data within the prediction data accumulation period, and updating the decay model parameters according to the regression calculation results, specifically includes the following steps:

[0052] Time series data on pavement performance were established based on the high-frequency inspection time interval t1 and the prediction time interval t2 used for data accumulation for prediction. and Where n corresponds to different road surface performance data categories, r i and p j These represent the i-th road segment grid and the j-th inspection time, respectively. The inspection time is related to the high-frequency inspection cycle, and the interval between adjacent inspection times is usually 1 to 7 days. The field format of the time series data of pavement performance conditions for each category is shown in Table 1.

[0053] Table 1. Field format table for time-series data on pavement performance conditions of various categories

[0054] No Detection time Distress_Type Value Section_ID 1 2022 / 06 / 3011:10:43 Lc 5.5 R1 2 2022 / 06 / 30 11:12:13 Sp 1.8 R2 3 2022 / 06 / 30 11:14:02 RI 3.2 R3 4 2022 / 06 / 3011:15:49 IRI 3.8 R4 5 2022 / 06 / 3011:17:32 rmsa 2.3 R5

[0055] This includes the number, inspection time, pavement performance data category, pavement performance data characterization indicators and quantitative parameters, and the grid number of the road segment to which it belongs;

[0056] Step 302: Based on the time series data of road surface performance during the high-frequency inspection time segment t1 Regression analysis was performed on decay models for roughness, road surface bumps, road surface cracks, road surface potholes, road surface ruts, road surface skid resistance coefficient, road surface wear, and road surface structural strength. The decay model parameters of each decay model were calculated, and linear or nonlinear regression methods were used for regression analysis.

[0057] Step 303: Calculate the performance data of various road surfaces within the prediction time period based on the regression calculation results.

[0058] Step 304: Compare the prediction results obtained in Step 301 with the measured results of various road surface performance conditions within the prediction time period, and construct the objective function;

[0059] Step 305: Determine the parameters to be updated and the boundary conditions for various decay models;

[0060] Step 306: Perform iterative calculations based on the objective function, the parameters to be updated, and the boundary conditions. Stop iterating when the objective function is less than the allowable error or when the maximum number of iterations is reached, and use the optimal solution among the parameters to be updated as the final calibration parameters.

[0061] Step 307: Repeat steps 301 to 306 to obtain the decay model parameters of various pavement performance data for different grid road segments.

[0062] In step 304, the objective function is used to represent the time series data of the predicted pavement performance conditions. Time series data of various road surface performance conditions measured in actual tests The error is calculated using Euclidean distance, Minkowski distance, Manhattan distance, or cosine similarity. In this embodiment, after normalizing the time-series data of various road performance conditions, Euclidean distance is used for calculation. The expression of the objective function is:

[0063]

[0064] Where n represents the types of road surface performance conditions, N represents the total number of categories, j represents the inspection time, and t represents the total number of inspections.

[0065] In step 305, the parameter to be updated is one or more decay model parameters of various pavement performance data. When the decay model adopts a linear model, the parameter to be updated is the linear decay rate α. linear When the decay model is an exponential or negative exponential model, the parameter to be updated is the exponential decay rate α. exp or -α exp When the decay model is an S-model, the parameter to be updated is the correlation coefficient α. s and β s The boundary conditions are the range of values ​​for the parameters to be updated.

[0066] In step 306, iterative calculations are performed using a multi-objective particle swarm optimization algorithm, a multi-objective genetic algorithm, or a nonlinear least squares method. The initial values ​​of the parameters to be updated are set to 20 to 50 sets and are randomly generated according to the boundary conditions. During the iterative calculation, the maximum number of iterations does not exceed 200. Iteration stops when the objective function is less than the allowable error or the maximum number of iterations is reached, and the optimal solution among the parameters to be updated is taken as the final calibration parameter.

[0067] In step 4, the process of verifying the reliability of the decay model parameters of the updated pavement performance data is as follows:

[0068] After determining the verification time interval t3 after time T0, high-frequency inspection data within the t3 time interval are collected to obtain time-series data of the measured pavement performance. Simultaneously, the decay model parameters from step 7 are substituted into each grid segment to calculate the time series data of the predicted pavement performance within the t3 time interval. And calculate the time series data of the measured pavement performance. Time series data for predicting pavement performance conditions The error is used to determine whether the error of the road performance data is less than the error threshold. If it is, the decay model parameters are considered reliable, and the decay model is used to continue to perform prediction calculations until the error exceeds the set error threshold. If the error exceeds the error threshold, the process returns to step 2. If not, the process returns to step 2 until the decay model parameters are reliable. The error is calculated using Euclidean distance, Minkowski distance, Manhattan distance, or cosine similarity.

[0069] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. A method for dynamic prediction of road surface performance based on high-frequency inspection data, characterized in that, The method includes the following steps: Step 1: Grid the road segment to be predicted according to a fixed length and lanes, and construct decay models based on the obtained road surface performance data, and determine the initial model parameters of the decay models; Step 2: Set the current time as T 0, and determine the high-frequency inspection time segment used for predictive data accumulation. t 1 and the predicted time range t 2; Step 3: Based on the high-frequency inspection data within the prediction data accumulation period, perform regression calculations on the decay model parameters of various road surface performance data, and update the decay model parameters according to the regression calculation results; Step 4: Verify the reliability of the decay model parameters for the updated pavement performance data; Step 5: Based on the validated decay model, predict various types of pavement performance data, perform weighted calculations on various types of data, and obtain a comprehensive pavement performance evaluation index to characterize pavement performance; In step 1, the grid length of the road segment to be predicted ranges from 50 to 200 meters, and the grid is defined as... R 1, R 2,…… R i ,…… R n Calculate various road surface performance data within each grid; Road surface performance data includes smoothness, bump index, cracks, potholes, rutting, skid resistance coefficient, wear, and structural strength within the grid. Smoothness within the grid is characterized by the International Roughness Index (IRI), the bump index by the root mean square (RMSA) of short-time acceleration, and cracks and potholes by the total crack length within the grid. Total area of ​​pits and trenches The road surface rutting was characterized by rutting depth RD, the road surface skid resistance coefficient was characterized by lateral force coefficient SFC, the road surface wear was characterized by wear rate WR, and the road surface structural strength was characterized by road surface structural strength coefficient SSR. Different types of pavement performance data correspond to different decay models, including linear models, exponential models, negative exponential models, and S-models. In step 2, the high-frequency inspection time segment for predicted data accumulation is... t 1. The predicted time period is no less than 30 days. t 2. No more than 100 days, to ensure the reliability of the parameters used for decay model regression.

2. The method for dynamic prediction of road performance based on high-frequency inspection data according to claim 1, characterized in that, In step 3, the process of regressing the decay model parameters of various road surface performance data based on high-frequency inspection data within the prediction data accumulation period is as follows: Step 301: Based on the high-frequency inspection time intervals used for predictive data accumulation and predicted time range Establish time series data of pavement performance conditions respectively and ,in, Corresponding to different road surface performance data categories, and They represent the first The first road segment grid and the first Each inspection moment; Step 302: Based on the high-frequency inspection time segment Time series data of road surface performance We used linear or nonlinear regression methods to perform regression analysis on decay models of road surface roughness, road surface bumps, road surface cracks, road surface potholes, road surface ruts, road surface skid resistance coefficient, road surface wear and road surface structural strength, and then obtained the decay model parameters of various decay models.

3. The method for dynamic prediction of road surface performance based on high-frequency inspection data according to claim 2, characterized in that, Step 3, which involves updating the parameters of each decay model based on the regression calculation results, specifically includes the following steps: Step 303: Calculate the prediction time interval based on the regression calculation results. Data on various road surface performance conditions ; Step 304: The predicted time segment obtained in step 303... Data on various road surface performance conditions and prediction time periods Data on the measured performance of various road surfaces within the area Compare and construct the objective function; Step 305: Determine the parameters to be updated and the boundary conditions for various decay models; Step 306: Perform iterative calculations based on the objective function, the parameters to be updated, and the boundary conditions. Stop iterating when the objective function is less than the allowable error or when the maximum number of iterations is reached, and use the optimal solution among the parameters to be updated as the final calibration parameters. Step 307: Repeat steps 301 to 306 to obtain the decay model parameters of various pavement performance data for different grid road segments.

4. The method for dynamic prediction of road surface performance based on high-frequency inspection data according to claim 3, characterized in that, In step 304, the objective function is used to represent the time-series data of the predicted road surface performance conditions. Time series data of various road surface performance conditions measured in actual tests The error can be calculated using methods such as Euclidean distance, Minkowski distance, Manhattan distance, or cosine similarity.

5. The method for dynamic prediction of road surface performance based on high-frequency inspection data according to claim 4, characterized in that, When using Euclidean distance for calculation, the time-series data of various pavement performance conditions are normalized, and the Euclidean distance between the predicted results and the measured results is calculated, thus obtaining the objective function. The expression of the objective function is as follows: in, These are the types of road surface performance conditions. This represents the total number of categories. For inspection time, Total number of inspections This is the normalization function.

6. The method for dynamic prediction of road surface performance based on high-frequency inspection data according to claim 3, characterized in that, In step 305, the parameters to be updated are the parameters in the decay model of various road surface performance data. When the decay model adopts a linear model, the parameter to be updated is the linear decay rate. When the decay model is an exponential or negative exponential model, the parameter to be updated is the exponential decay rate. or When the decay model is an S-model, the parameter to be updated is the correlation coefficient. and The initial values ​​of the parameters to be updated are set to 20-50 sets and are randomly generated according to the boundary conditions; The boundary conditions are the range of values ​​for the parameters to be updated.

7. The method for dynamic prediction of road surface performance based on high-frequency inspection data according to claim 3, characterized in that, In step 306, the algorithms used for iterative calculation include multi-objective particle swarm optimization, multi-objective genetic algorithm, and nonlinear least squares method, and the maximum number of iterations in the iterative calculation process does not exceed 200.

8. The method for dynamic prediction of road surface performance based on high-frequency inspection data according to claim 3, characterized in that, In step 4, the process of verifying the reliability of the decay model parameters of the updated pavement performance data is as follows: Determine the verification time range Data collection and verification time period High-frequency inspection data within the system yields time-series data of measured pavement performance. Simultaneously, the decay model parameters from step 3 are substituted into each grid segment to calculate the verification time interval. Time series data of predicted pavement performance within By calculating the verification time period Time series data of predicted pavement performance within Time series data of measured pavement performance The error judgment determines whether the decay model parameters are reliable, that is, whether the error of the road performance data is less than the error threshold. If so, the decay model parameters are reliable, and the decay model is used to continue to perform prediction calculations until the error exceeds the set error threshold. If the error exceeds the error threshold, return to step 2. If not, return to step 2 until the decay model parameters are reliable.

Images