Asphalt pavement performance and service life prediction method

Abstract

The invention discloses an asphalt pavement performance and service life prediction method, and relates to the technical field of road engineering, and the method comprises the steps: carrying out the preprocessing of multi-source data, and constructing an influence factor comprehensive quantification model based on the preprocessed multi-source data; using the preprocessed multi-source data to train a prediction model; inputting real-time multi-source data of a to-be-predicted asphalt pavement into the trained prediction model to obtain pavement performance indexes, and using a curve fitting technology to obtain a function expression of the pavement performance indexes changing along with time; and calculating the time required by the pavement performance index to reach the corresponding critical value, and selecting the minimum time value as the pavement residual life. According to the method, by comprehensively considering the influence factors of multiple dimensions such as traffic load, environmental factors and pavement structure characteristics, the specific influence of the actual situation on the use performance and the residual life of the asphalt pavement can be disclosed more comprehensively and accurately.

Description

Technical Field

[0001] The invention relates to the technical field of road engineering, and in particular to a method for predicting the performance and life of an asphalt pavement. Background Art

[0002] Asphalt pavement is a core component of modern transportation infrastructure, and its sustained and stable performance is crucial to ensure traffic safety and efficiency. However, due to the influence of various factors such as environmental changes, traffic load, and material aging, the performance of asphalt pavement will gradually decline over time. In order to effectively manage and extend the service life of roads, it is necessary to accurately predict the performance and remaining life of asphalt pavement through scientific methods, so as to provide a basis for road maintenance and repair decisions and achieve the optimal allocation of resources.

[0003] However, existing prediction methods have several limitations, which are mainly reflected in the following two aspects: on the one hand, many existing technologies only consider some factors that affect pavement performance and ignore other variables that may be equally important, resulting in a lack of comprehensiveness and accuracy in the prediction results; on the other hand, some prediction models are built based on overly simplified assumptions, such as using linear relationships to describe complex and changeable actual conditions, which makes it difficult for the models to adapt to the actual performance of asphalt pavements under different environments and conditions.

[0004] Currently, no effective solution has been proposed for the problems in the related technologies. Summary of the invention

[0005] In view of the problems in the related art, the present invention proposes a method for predicting the performance and life of an asphalt pavement to overcome the above-mentioned technical problems existing in the existing related art.

[0006] To this end, the specific technical solution adopted by the present invention is as follows:

[0007] A method for predicting asphalt pavement performance and lifespan, the method comprising the following steps:

[0008] S1. Obtain multi-source data of the asphalt pavement to be predicted, pre-process the multi-source data, and build a comprehensive quantitative model of influencing factors based on the pre-processed multi-source data;

[0009] S2. Based on the influencing factors in the comprehensive quantitative model of influencing factors, a prediction model based on deep learning is constructed, and the prediction model is trained using preprocessed multi-source data;

[0010] S3, inputting the real-time multi-source data of the asphalt pavement to be predicted into the trained prediction model to obtain the pavement performance index, and using the curve fitting technology to obtain the functional expression of the pavement performance index changing with time;

[0011] S4. According to the critical value and function expression of the pavement performance index, the time required for the pavement performance index to reach the corresponding critical value is calculated, and the minimum time value is selected as the remaining life of the pavement.

[0012] Furthermore, obtaining multi-source data of the asphalt pavement to be predicted and preprocessing the multi-source data, and constructing a comprehensive quantitative model of influencing factors based on the preprocessed multi-source data includes the following steps:

[0013] S11, obtaining multi-source data of the asphalt pavement to be predicted, the multi-source data including pavement structure data, traffic load data, environmental factor data and pavement performance historical monitoring data;

[0014] S12, preprocessing the acquired multi-source data;

[0015] S13, using the pre-processed multi-source data, respectively calculating the traffic load influencing factor, the comprehensive environmental influencing factor and the pavement structure characteristic influencing factor;

[0016] S14. Combine traffic load influencing factors, comprehensive environmental influencing factors and pavement structure characteristics influencing factors to construct a comprehensive quantitative model of influencing factors.

[0017] Furthermore, preprocessing the acquired multi-source data includes the following steps:

[0018] S121, using a standard score conversion algorithm to standardize the acquired multi-source data;

[0019] S122, using a mean filtering algorithm to perform denoising on the standardized multi-source data.

[0020] Furthermore, using the pre-processed multi-source data, respectively calculating the traffic load influencing factor, the environmental comprehensive influencing factor and the pavement structure characteristic influencing factor includes the following steps:

[0021] S131, extracting traffic load data from the preprocessed multi-source data, and applying the traffic load impact factor formula to calculate the traffic load impact factor;

[0022] S132, extracting environmental factor data from the preprocessed multi-source data, and applying the environmental comprehensive impact factor formula to calculate the environmental comprehensive impact factor;

[0023] S133. Extract pavement structure data from the preprocessed multi-source data, and apply the pavement structure characteristic influencing factor formula to calculate the pavement structure characteristic influencing factor.

[0024] Furthermore, the traffic load influencing factor formula is:

[0025]

[0026] Where, TLF represents the traffic load influencing factor;

[0027] n represents the number of vehicle types;

[0028] i represents the index value;

[0029] q i Indicates the traffic volume of different types of vehicles;

[0030] P zi Indicates axle weight;

[0031] v i Indicates vehicle speed.

[0032] Furthermore, the formula for the influencing factor of pavement structure characteristics is:

[0033]

[0034] Where, PSF represents the influencing factor of pavement structure characteristics;

[0035] l represents the number of pavement structure layers;

[0036] k represents the number of vehicle types;

[0037] h k Indicates the thickness of the pavement structure layer;

[0038] E k Represents the modulus of the pavement structure layer.

[0039] Furthermore, based on the influencing factors in the comprehensive quantitative model of influencing factors, a prediction model based on deep learning is constructed, and the prediction model is trained using the preprocessed multi-source data, including the following steps:

[0040] S21. Determine the input dimension and output dimension of the prediction model according to the influencing factors of the comprehensive quantitative model of influencing factors;

[0041] S22. Based on the determined input dimension and output dimension, a prediction model including a long short-term memory network layer and a fully connected layer is constructed using a deep learning framework;

[0042] S23. Use the preprocessed multi-source data to train the constructed prediction model.

[0043] Furthermore, training the constructed prediction model using the preprocessed multi-source data includes the following steps:

[0044] S231, dividing the preprocessed multi-source data into a training set and a validation set;

[0045] S232, setting parameter configuration in the training process and setting initial parameters of the prediction model;

[0046] S233, training the prediction model batch by batch using the training set, calculating the mean square error between the predicted value and the true value as the loss function, and optimizing the parameters of the prediction model through an adaptive moment estimation optimizer;

[0047] S234, monitoring the loss change on the validation set to evaluate the performance of the prediction model, and optimizing the prediction model based on the evaluation results;

[0048] S235. Repeat steps S233 to S234 until a preset number of training iterations is reached.

[0049] Furthermore, the real-time multi-source data of the asphalt pavement to be predicted is input into the prediction model after training to obtain the pavement performance index, and the function expression of the pavement performance index changing with time is obtained by using the curve fitting technology, including the following steps:

[0050] S31. Acquire real-time multi-source data of the asphalt pavement to be predicted according to a preset time period, where the real-time multi-source data includes real-time traffic load data, real-time environmental factor data, and real-time pavement structure data;

[0051] S32, inputting the real-time multi-source data into the trained prediction model to obtain the road performance index value of each time period;

[0052] S33. Use curve fitting technology to fit the pavement performance index values ​​in each time period to obtain a functional expression of the pavement performance index changing with time.

[0053] Furthermore, the pavement performance index values ​​include: smoothness index, rutting depth, and pavement strength.

[0054] The beneficial effects of the present invention are:

[0055] 1. The present invention can more comprehensively and accurately reveal the specific impact of actual conditions on the performance of asphalt pavement and its remaining life by comprehensively considering the influencing factors of multiple dimensions such as traffic load, environmental factors and pavement structure characteristics.

[0056] 2. The present invention constructs a prediction model by using the long short-term memory network architecture of deep learning, which can efficiently capture the time series characteristics and complex nonlinear relationships in the data, thereby breaking through the limitations faced when dealing with complex changes in road performance.

[0057] 3. The present invention determines the critical value of each pavement performance index and combines the curve fitting technology to predict the remaining life, so that the prediction result can be closer to the needs of actual engineering applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0059] Figure 1 is a flow chart of a method for predicting performance and life of an asphalt pavement according to an embodiment of the present invention;

[0060] Figure 2 It is an architecture diagram of a method for predicting asphalt pavement performance and life according to an embodiment of the present invention. DETAILED DESCRIPTION

[0061] To further illustrate each embodiment, the present invention provides drawings, which are part of the disclosure of the present invention and are mainly used to illustrate the embodiments. They can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these contents, ordinary technicians in this field should be able to understand other possible implementation methods and advantages of the present invention.

[0062] According to an embodiment of the present invention, a method for predicting the performance and life of an asphalt pavement is provided.

[0063] The present invention will now be further described with reference to the accompanying drawings and specific embodiments. Figure 1-Figure 2 As shown, according to the asphalt pavement performance and life prediction method of the embodiment of the present invention, the method comprises the following steps:

[0064] S1. Obtain multi-source data of the asphalt pavement to be predicted, pre-process the multi-source data, and build a comprehensive quantitative model of influencing factors based on the pre-processed multi-source data.

[0065] Specifically, obtaining multi-source data of the asphalt pavement to be predicted, preprocessing the multi-source data, and constructing a comprehensive quantitative model of influencing factors based on the preprocessed multi-source data includes the following steps:

[0066] S11. Acquire multi-source data of the asphalt pavement to be predicted, where the multi-source data includes pavement structure data, traffic load data, environmental factor data, and pavement performance historical monitoring data.

[0067] It should be noted that the pavement structure data: detailed collection of structural information of each layer of the asphalt pavement section to be predicted, including but not limited to the thickness of each layer (such as the thickness of the upper layer, middle layer, lower layer, etc., respectively recorded as h1, h2, h3, etc., unit: centimeter), material type (such as asphalt concrete of different grades, etc.) and its corresponding material characteristic parameters (such as elastic modulus E1, E2, E3, etc., unit: MPa). These data can be obtained from road engineering design drawings and related construction materials.

[0068] Traffic load data: Set up high-precision traffic flow monitoring equipment on the road section to be predicted, and continuously collect traffic load data for at least [X] years, covering the traffic volume of various types of vehicles (recorded as q1, q2, q3, etc., in units of vehicles / day, corresponding to different vehicle types), axle weight distribution (single axle weight of each vehicle type is recorded as P Z1 , P Z2 , P Z3 etc., unit: kN) and vehicle speed distribution (the average speed of each vehicle type is recorded as v1, v2, v3, etc., unit: km / h).

[0069] Environmental factor data: Obtain meteorological data from the local meteorological department during the same period as the road surface is used, including daily average temperature (recorded as T, unit: Celsius), precipitation (recorded as R, unit: millimeter), sunshine duration (recorded as S, unit: hour) and frequency of extreme weather events (such as heavy rain, heavy snow, high temperature heat waves, etc., respectively recorded as f1, f2, f3, etc., unit: times / year). At the same time, collect geological environmental data around the road section, such as groundwater depth (recorded as H, unit: meter), soil type and its corresponding bearing capacity, etc.

[0070] Historical pavement performance monitoring data: Collect existing pavement performance monitoring data for the road section, such as flatness at different times (characterized by the International Roughness Index IRI, unit: meter / kilometer), rutting depth (denoted as RD, unit: millimeter), pavement strength (converted from the deflection value obtained by testing with a Beckman beam deflectometer or a drop weight deflectometer, denoted as LD, unit: millimeter), etc.

[0071] S12. Preprocess the acquired multi-source data.

[0072] Specifically, preprocessing the acquired multi-source data includes the following steps:

[0073] S121. Use a standard score conversion algorithm to standardize the acquired multi-source data.

[0074] It should be noted that the standard score conversion algorithm (Z-score standardization) is used to transform each eigenvalue of the data so that it has a distribution with a mean of 0 and a standard deviation of 1. For the data of a certain feature column, its standardized value xnew The calculation formula is:

[0075]

[0076] In the formula, x new represents the data after standardization; x represents the original data; μ represents the mean of the feature column; σ represents the standard deviation of the feature column.

[0077] S122, using a mean filtering algorithm to perform denoising on the standardized multi-source data.

[0078] It should be noted that denoising based on statistical methods (mean filtering) is as follows: for time series data or image data, etc., take a certain data point as the center, take the data points within a certain range around it (such as data of several time steps before and after in the time series, pixels around a certain pixel in the image), calculate the average value of these data points, and use the average value to replace the value of the central data point. This can smooth out some local fluctuations and achieve a denoising effect.

[0079] S13. Using the pre-processed multi-source data, calculate the traffic load influencing factors, the comprehensive environmental influencing factors and the pavement structure characteristic influencing factors respectively.

[0080] Specifically, using the pre-processed multi-source data, respectively calculating the traffic load influencing factor, the comprehensive environmental influencing factor and the pavement structure characteristic influencing factor includes the following steps:

[0081] S131. Extracting traffic load data from the preprocessed multi-source data, and applying the traffic load impact factor formula to calculate the traffic load impact factor.

[0082] Specifically, the traffic load influencing factor formula is:

[0083]

[0084] In the formula, TLF represents the traffic load influencing factor; n represents the number of vehicle types; i represents the index value; q i represents the traffic volume of different types of vehicles; P zi Indicates axle weight; v i Indicates vehicle speed.

[0085] It should be noted that the traffic load factor (TLF) takes into account the complexity of traffic loads, including not only axle weight and traffic volume, but also vehicle speed factors to comprehensively quantify its impact on the road surface. The traffic load factor formula comprehensively reflects the cumulative damage effect of traffic loads on the road surface by weighted summing the traffic volume, axle weight cube and vehicle speed square of different vehicle types and dividing it by a normalized constant.

[0086] S132. Extract environmental factor data from the preprocessed multi-source data, and apply the environmental comprehensive impact factor formula to calculate the environmental comprehensive impact factor.

[0087] It should be noted that the environmental comprehensive impact factor (ELF) is a comprehensive environmental impact factor formula that considers the impact of environmental factors on asphalt pavement. This factor integrates multiple factors such as temperature, precipitation, sunshine duration, and extreme weather events. The formula is as follows:

[0088]

[0089] In the formula, EIF represents the comprehensive environmental impact factor; a1 represents the empirical coefficient; a2 represents the empirical coefficient; a3 represents the empirical coefficient; a4 represents the empirical coefficient; T represents temperature; R represents precipitation; S represents the length of the day and week; m represents the number of extreme weather event types; j represents the index value; f i A numerical value indicating extreme weather.

[0090] Among them, a1, a2, a3, and a4 are empirical coefficients determined through a large number of experiments and data analysis based on local environmental characteristics and pavement material properties.

[0091] S133. Extract pavement structure data from the preprocessed multi-source data, and apply the pavement structure characteristic influencing factor formula to calculate the pavement structure characteristic influencing factor.

[0092] Specifically, the formula for the influencing factor of pavement structure characteristics is:

[0093]

[0094] In the formula, PSF represents the influencing factor of pavement structure characteristics; l represents the number of pavement structure layers; k represents the number of vehicle types; h k Indicates the thickness of the pavement structure layer; E k Represents the modulus of the pavement structure layer.

[0095] It should be noted that the pavement structure characteristic influence factor (PSF) is based on the pavement structure data, taking into account the influence of each layer thickness and material elastic modulus on pavement performance. The pavement structure characteristic influence factor formula sums the product of each layer thickness and elastic modulus and divides it by a normalized constant to reflect the influence of the pavement structure's own characteristics on its performance.

[0096] S14. Combine traffic load influencing factors, comprehensive environmental influencing factors and pavement structure characteristics influencing factors to construct a comprehensive quantitative model of influencing factors.

[0097] S2. Based on the influencing factors in the comprehensive quantitative model of influencing factors, a prediction model based on deep learning is constructed, and the prediction model is trained using the preprocessed multi-source data.

[0098] Specifically, based on the influencing factors in the comprehensive quantitative model of influencing factors, a prediction model based on deep learning is constructed, and the prediction model is trained using the preprocessed multi-source data, including the following steps:

[0099] S21. Determine the input dimension and output dimension of the prediction model based on the influencing factors of the comprehensive quantitative model.

[0100] S22. Based on the determined input dimension and output dimension, a prediction model including a long short-term memory network layer and a fully connected layer is constructed using a deep learning framework.

[0101] S23. Use the preprocessed multi-source data to train the constructed prediction model.

[0102] Specifically, training the constructed prediction model using the preprocessed multi-source data includes the following steps:

[0103] S231, dividing the preprocessed multi-source data into a training set and a validation set;

[0104] S232, setting parameter configuration in the training process and setting initial parameters of the prediction model;

[0105] S233, training the prediction model batch by batch using the training set, calculating the mean square error between the predicted value and the true value as the loss function, and optimizing the parameters of the prediction model through an adaptive moment estimation optimizer;

[0106] S234, monitoring the loss change on the validation set to evaluate the performance of the prediction model, and optimizing the prediction model based on the evaluation results;

[0107] S235. Repeat steps S233 to S234 until a preset number of training iterations is reached.

[0108] It should be noted that the long short-term memory network (LSTM network) in the prediction model is trained (initializing model parameters, training batch by batch, monitoring validation set loss), and the training goal is to minimize the mean square error (MSE) between the predicted road performance indicators (such as predicted IRI, RD, LD, etc.) output by the network and the actual monitored historical road performance indicators. Through repeated training with a large amount of training data (historical data from multiple similar asphalt road sections), the weight parameters of the network are adjusted until a satisfactory training effect is achieved.

[0109] S3. Input the real-time multi-source data of the asphalt pavement to be predicted into the prediction model after training to obtain the pavement performance index, and use the curve fitting technology to obtain the functional expression of the pavement performance index changing with time.

[0110] Specifically, the real-time multi-source data of the asphalt pavement to be predicted is input into the prediction model after training to obtain the pavement performance index, and the function expression of the pavement performance index changing with time is obtained by using the curve fitting technology, including the following steps:

[0111] S31. Acquire real-time multi-source data of the asphalt pavement to be predicted according to a preset time period, wherein the real-time multi-source data includes real-time traffic load data, real-time environmental factor data and real-time pavement structure data.

[0112] S32, inputting the real-time multi-source data into the trained prediction model to obtain the road performance index value for each time period.

[0113] Specifically, the pavement performance index values ​​include: flatness index, rutting depth, and pavement strength.

[0114] It should be noted that the network will output the predicted road performance index value, such as the predicted roughness index IRI crit , rutting depth RD crit , road surface strength LD crit Etc., so as to realize the prediction of the performance of asphalt pavement in the current and future period.

[0115] S33. Use curve fitting technology to fit the pavement performance index values ​​in each time period to obtain a functional expression of the pavement performance index changing with time.

[0116] It should be noted that based on the predicted trend of the pavement performance indicators changing over time (by regularly repeating the above performance prediction steps to obtain the predicted values ​​at different time points), curve fitting technology (such as polynomial fitting, exponential fitting, etc.) is used to fit the predicted values ​​to obtain the functional expressions of the pavement performance indicators changing over time, such as IRI(t), RD(t), LD(t), etc.

[0117] S4. According to the critical value and function expression of the pavement performance index, the time required for the pavement performance index to reach the corresponding critical value is calculated, and the minimum time value is selected as the remaining life of the pavement.

[0118] It should be noted that the critical values ​​of various pavement performance indicators, such as IRI, are determined based on relevant industry standards and actual engineering experience. crit , RD crit , LD critWhen the pavement performance index reaches or exceeds its corresponding critical value, the pavement is considered to have reached the end of its service life.

[0119] Solve the time t when each pavement performance index function expression is equal to its corresponding critical value IRI ,t RD ,t LD etc., and take the minimum value as the remaining life of the asphalt pavement, that is:

[0120] Lremain=min(tIRI,tRD,tLD);

[0121] Where, L remain Indicates the remaining life of the asphalt pavement; t IRI Indicates the time to reach the critical value of the flatness index; t RD Indicates the time to reach the critical value of rutting depth index; t LD Indicates the time to reach the critical value of the pavement strength index.

[0122] In a specific embodiment, it is assumed that the performance and remaining life of a certain section of asphalt pavement need to be predicted:

[0123] 1. Basic information about the road section

[0124] Pavement structure data:

[0125] Upper layer: 5 cm thick, made of AC-13 asphalt concrete, elastic modulus 1200 MPa;

[0126] Middle surface layer: 6 cm thick, made of AC-20 asphalt concrete, elastic modulus 1000 MPa;

[0127] Lower layer: thickness 8 cm, material is AC-25 asphalt concrete, elastic modulus 800 MPa.

[0128] Traffic load data (average data collected in the past three years):

[0129] Small car: traffic volume 5,000 vehicles / day, single axle weight 10 kN, average speed 60 km / h;

[0130] Medium-sized vehicles: traffic volume 2,000 vehicles / day, single axle weight 15 kN, average speed 50 km / h;

[0131] Large vehicles: Traffic volume 1,000 vehicles / day, single axle weight 20 kN, average speed 40 km / h.

[0132] Environmental factor data (average data for the past three years):

[0133] The average daily temperature is 18 degrees Celsius;

[0134] Precipitation: 800 mm;

[0135] The sunshine duration is 6 hours.

[0136] Extreme weather events: Heavy rains occur 10 times per year.

[0137] Historical monitoring data of pavement performance (average data for the past three years):

[0138] International roughness index 2m / km;

[0139] Rutting depth 4 mm;

[0140] Road surface strength (converted into deflection value) is 0.5 mm.

[0141] At the same time, based on the local environmental characteristics and pavement material properties, the empirical coefficients in the comprehensive environmental impact factor formula were determined through a large number of experiments and data analysis: a1=0.05, a2=0.03, a3=0.02, a4=0.1.

[0142] 2. Calculation of various influencing factors

[0143] Traffic Load Factor (TLF) calculation:

[0144] Assume n = 3 (number of vehicle types);

[0145] TLF=47675000.

[0146] Calculation of Environmental Impact Factor (ELF):

[0147] Assume m = 1 (number of extreme weather event types);

[0148] ELF=26.02.

[0149] Calculation of pavement structural characteristics factor (PSF):

[0150] PSF=184.

[0151] 3. Prediction model training based on deep learning

[0152] In practical applications, it is necessary to collect historical data of multiple similar asphalt pavement sections to train the long short-term memory network (LSTM) to minimize the mean square error (MSE) between the predicted pavement performance indicators output by the network and the actual monitored historical pavement performance indicators.

[0153] 4. Prediction of Asphalt Pavement Performance

[0154] The current traffic load impact factor (TLF=47675000), comprehensive environmental impact factor (ELF=26.02) and pavement structure characteristic impact factor (PSF=184) of the road section, together with other real-time collected or newly acquired relevant data (such as historical pavement performance monitoring data, etc.), are input into the trained LSTM network.

[0155] After network prediction, the predicted values ​​of pavement performance indicators in the next year are as follows:

[0156] The predicted international roughness index is 2.5 m / km;

[0157] Predicted rutting depth 5 mm;

[0158] Predicted pavement strength 0.6 mm.

[0159] 5. Prediction of Remaining Life of Asphalt Pavement

[0160] According to relevant industry standards and actual engineering experience, the critical values ​​of various pavement performance indicators are determined:

[0161] The critical value of the international roughness index is 3.5 meters per kilometer;

[0162] The critical value of rutting depth is 6 mm;

[0163] The critical value of road surface strength is 0.8 mm.

[0164] Based on the predicted trend of the pavement performance indicators over time (by regularly repeating the above performance prediction steps, the predicted values ​​at different time points in the next few years are obtained), the polynomial fitting technology is used to fit the predicted values.

[0165] After fitting, the functional expression of the international roughness index changing with time is obtained as follows:

[0166] IRI(t)=0.05t+2 (wherein is time, unit: year);

[0167] The functional expression of rutting depth changing with time is: RD(t)=0.1t+4;

[0168] The functional expression of pavement strength changing with time is: LD(t)=0.01t+0.5.

[0169] Solve the time when each pavement performance index function expression is equal to its corresponding critical value:

[0170] For the International Roughness Index: tIRI = 30;

[0171] For rutting depth: tRD = 20;

[0172] For road surface strength: tLD = 30;

[0173] Take the minimum value as the remaining life of the asphalt pavement: 20 (unit: year)

[0174] Therefore, according to this forecast, the remaining life of the asphalt pavement is about 20 years.

[0175] In summary, with the help of the above technical solution of the present invention, by comprehensively considering the influencing factors of multiple dimensions such as traffic load, environmental factors and pavement structure characteristics, the specific impact of actual conditions on the performance of asphalt pavement and its remaining life can be revealed more comprehensively and accurately. By using the long short-term memory network architecture of deep learning to build a prediction model, the time series characteristics and complex nonlinear relationships in the data can be efficiently captured, thus breaking through the limitations faced when dealing with complex changes in pavement performance. By determining the critical values ​​of each pavement performance indicator and combining curve fitting technology to predict the remaining life, the prediction results can be closer to the needs of actual engineering applications.

[0176] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for predicting asphalt pavement performance and life, characterized in that: The method comprises the following steps: S1. Obtain multi-source data of the asphalt pavement to be predicted, pre-process the multi-source data, and build a comprehensive quantitative model of influencing factors based on the pre-processed multi-source data; S2. Based on the influencing factors in the comprehensive quantitative model of influencing factors, a prediction model based on deep learning is constructed, and the prediction model is trained using preprocessed multi-source data; S3, inputting the real-time multi-source data of the asphalt pavement to be predicted into the trained prediction model to obtain the pavement performance index, and using the curve fitting technology to obtain the functional expression of the pavement performance index changing with time; S4. According to the critical value and function expression of the pavement performance index, the time required for the pavement performance index to reach the corresponding critical value is calculated, and the minimum time value is selected as the remaining life of the pavement.

2. The asphalt pavement performance and life prediction method according to claim 1, characterized in that: The method of obtaining multi-source data of the asphalt pavement to be predicted, preprocessing the multi-source data, and constructing a comprehensive quantitative model of influencing factors based on the preprocessed multi-source data includes the following steps: S11, obtaining multi-source data of the asphalt pavement to be predicted, wherein the multi-source data includes pavement structure data, traffic load data, environmental factor data and pavement performance historical monitoring data; S12, preprocessing the acquired multi-source data; S13, using the pre-processed multi-source data, respectively calculating the traffic load influencing factor, the comprehensive environmental influencing factor and the pavement structure characteristic influencing factor; S14. Combine traffic load influencing factors, comprehensive environmental influencing factors and pavement structure characteristics influencing factors to construct a comprehensive quantitative model of influencing factors.

3. The method for predicting asphalt pavement performance and life according to claim 2, characterized in that: The preprocessing of the acquired multi-source data comprises the following steps: S121, using a standard score conversion algorithm to standardize the acquired multi-source data; S122, using a mean filtering algorithm to perform denoising on the standardized multi-source data.

4. The asphalt pavement performance and life prediction method according to claim 3, characterized in that: The method of using the pre-processed multi-source data to respectively calculate the traffic load influencing factor, the environmental comprehensive influencing factor and the pavement structure characteristic influencing factor comprises the following steps: S131, extracting traffic load data from the preprocessed multi-source data, and applying the traffic load impact factor formula to calculate the traffic load impact factor; S132, extracting environmental factor data from the preprocessed multi-source data, and applying the environmental comprehensive impact factor formula to calculate the environmental comprehensive impact factor; S133. Extract pavement structure data from the preprocessed multi-source data, and apply the pavement structure characteristic influencing factor formula to calculate the pavement structure characteristic influencing factor.

5. The method for predicting asphalt pavement performance and life according to claim 4, characterized in that: The traffic load influencing factor formula is: Where, TLF represents the traffic load influencing factor; n represents the number of vehicle types; i represents the index value; q i Indicates the traffic volume of different types of vehicles; P zi Indicates axle weight; v i Indicates vehicle speed.

6. A method for predicting asphalt pavement performance and life according to claim 5, characterized in that: The formula for the influencing factor of the pavement structure characteristics is: Where, PSF represents the influencing factor of pavement structure characteristics; l represents the number of pavement structure layers; k represents the number of vehicle types; h k Indicates the thickness of the pavement structure layer; E k Represents the modulus of the pavement structure layer.

7. The method for predicting asphalt pavement performance and life according to claim 1, characterized in that: The method of constructing a prediction model based on deep learning based on the influencing factors in the comprehensive quantitative model of influencing factors, and training the prediction model using the preprocessed multi-source data includes the following steps: S21. Determine the input dimension and output dimension of the prediction model according to the influencing factors of the comprehensive quantitative model of influencing factors; S22. Based on the determined input dimension and output dimension, a prediction model including a long short-term memory network layer and a fully connected layer is constructed using a deep learning framework; S23. Use the preprocessed multi-source data to train the constructed prediction model.

8. The method for predicting asphalt pavement performance and life according to claim 7, characterized in that: The method of training the constructed prediction model using the preprocessed multi-source data includes the following steps: S231, dividing the preprocessed multi-source data into a training set and a validation set; S232, setting parameter configuration in the training process and setting initial parameters of the prediction model; S233, training the prediction model batch by batch using the training set, calculating the mean square error between the predicted value and the true value as the loss function, and optimizing the parameters of the prediction model through an adaptive moment estimation optimizer; S234, monitoring the loss change on the validation set to evaluate the performance of the prediction model, and optimizing the prediction model based on the evaluation results; S235. Repeat steps S233 to S234 until a preset number of training iterations is reached.

9. The method for predicting asphalt pavement performance and life according to claim 1, characterized in that: The method of inputting the real-time multi-source data of the asphalt pavement to be predicted into the trained prediction model to obtain the pavement performance index, and using the curve fitting technology to obtain the functional expression of the pavement performance index changing with time includes the following steps: S31, acquiring real-time multi-source data of the asphalt pavement to be predicted according to a preset time period, wherein the real-time multi-source data includes real-time traffic load data, real-time environmental factor data and real-time pavement structure data; S32, inputting the real-time multi-source data into the trained prediction model to obtain the road performance index value of each time period; S33. Use curve fitting technology to fit the pavement performance index values ​​in each time period to obtain a functional expression of the pavement performance index changing with time.

10. The method for predicting asphalt pavement performance and life according to claim 9, characterized in that: The road performance index values ​​include: flatness index, rutting depth, and road strength.

Images