A comprehensive performance evaluation method for asphalt pavement based on unascertained measure theory

Abstract

The present invention proposes a comprehensive performance evaluation method for asphalt pavement based on unascertained measurement theory, which relates to the technical field of road engineering, including: establishing a performance index system for asphalt pavement, determining a road section to be evaluated; determining a grading standard matrix based on performance index data; establishing a measurement function parameter optimization framework, establishing a measurement function set, and optimizing the parameters of the measurement function; calculating the single index measurement value of the road section to be evaluated in the evaluation grade space based on the measurement function, and obtaining a single index measurement matrix; calculating the combined weight vector of each index; combining the single index measurement matrix and the combined weight vector to calculate the comprehensive measurement evaluation matrix of the road section to be evaluated; and determining the final evaluation grade of the road section to be evaluated based on the confidence recognition criterion. The present invention can realize a comprehensive evaluation of the surface and internal performance of asphalt pavement, and overcomes the technical defects of insufficient consideration of internal diseases, single evaluation index, and unreasonable weight distribution in traditional evaluation methods.

Description

Technical Field

[0001] The invention relates to the technical field of road engineering, and in particular to a method for evaluating the comprehensive performance of asphalt pavement based on unascertained measurement theory. Background Art

[0002] With the continuous expansion of the road network and the continuous growth of traffic load, asphalt pavement damage has intensified and performance degradation has become increasingly obvious. Road maintenance work has gradually shifted from the construction period to the maintenance period. In the face of the growing maintenance demand, how to accurately evaluate the performance of the pavement and achieve scientific maintenance decisions has become a key technical issue that needs to be solved in the field of road engineering.

[0003] Unplanned pavement renovation and large-scale repairs will not only lead to excessive consumption of maintenance resources, but also cause a series of problems such as traffic congestion, safety hazards and environmental pollution. Therefore, establishing a scientific asphalt pavement comprehensive performance evaluation system has important technical guidance significance for optimizing maintenance strategies, extending pavement service life and reducing maintenance costs.

[0004] At present, the performance evaluation of asphalt pavement is usually based on a series of surface performance indicators, such as pavement flatness, rutting depth, skid resistance, pavement damage rate and structural strength. These indicators mainly reflect the disease condition and structural bearing capacity of the pavement surface. However, in actual engineering, the impact of hidden diseases inside the pavement (such as voids, crack expansion and internal damage) on the overall performance of the pavement cannot be ignored. Many surface diseases are often caused by the gradual extension of internal diseases to the surface. It is difficult to fully reflect the actual condition of the pavement by relying solely on surface performance indicators for evaluation. This limitation may lead to a lack of pertinence in the formulation of maintenance plans, resulting in waste of resources and poor maintenance results. Summary of the invention

[0005] In view of this, the present invention proposes a comprehensive performance evaluation method for asphalt pavement based on the unascertained measurement theory. By establishing a complete evaluation system, combining the measurement function parameters optimized by neural network and the combined weighting model based on the analytic hierarchy process and the entropy weight method, a comprehensive evaluation of the surface and internal performance of the asphalt pavement is achieved, which overcomes the technical defects of the traditional evaluation methods such as insufficient consideration of internal diseases, single evaluation index, and unreasonable weight distribution, thereby providing more accurate technical support for making scientific maintenance decisions.

[0006] The technical solution of the present invention is implemented as follows: The present invention provides a method for evaluating the comprehensive performance of asphalt pavement based on the unascertained measurement theory, comprising:

[0007] S1. Establish a performance index system for asphalt pavement, determine the road section to be evaluated, and obtain performance index data for the road section to be evaluated; when establishing the performance index system, select relevant indicators of the road section to be evaluated in the performance index set according to the screening criteria to form a performance index system; the performance index set includes basic performance indicators and extended performance indicators, and the basic performance indicators include flatness index IRI, rutting depth RD, lateral force coefficient SFC, structural strength coefficient SSI, pavement crack rate PCR, pavement repair rate PPR and internal damage rate IDR; the extended performance indicators include temperature sensitivity index TEI, noise attenuation index NAI, water damage resistance index WRI, aging index AGI and traffic loss index TDI; the screening criteria are: based on the service life of the pavement, climate regional characteristics, and traffic volume, calculate the importance of the indicators, and include the indicators with an importance greater than the preset threshold into the performance index system;

[0008] S2. Based on the performance indicator data, establish an evaluation level space and determine the grading standard matrix of each performance indicator at different evaluation levels;

[0009] S3. Establish a neural network-based measurement function parameter optimization framework, establish a measurement function set including basic measurement functions, extreme value measurement functions and composite measurement functions based on the unascertained measurement theory, select the measurement function type according to the performance indicator characteristics, and optimize the measurement function parameters through the optimization framework; wherein, the optimization framework includes constructing a parameter optimization objective function, designing parameter update rules and determining the optimization termination conditions;

[0010] S4. Calculate the single index measurement value of the road section to be evaluated in the evaluation level space based on the optimized measurement function to obtain a single index measurement matrix;

[0011] S5. Establish a combined weighting model based on the hierarchical analysis method and entropy weight method, and calculate the combined weight vector of each indicator by combining the indicator correlation analysis;

[0012] S6. Calculate the comprehensive measurement evaluation matrix of the road section to be evaluated by combining the single index measurement matrix and the combined weight vector;

[0013] S7. Based on the confidence recognition criterion, the final evaluation level of the road section to be evaluated is determined according to the comprehensive measurement evaluation matrix.

[0014] Based on the above solution, preferably, the set of n road sections to be evaluated is defined as the evaluation space, denoted as X={X 1 ,X 2 ,…,X n}; The set of m performance indicators contained in each road section to be evaluated is defined as the indicator space, denoted as Y={Y 1 ,Y 2 ,...,Y m}, x ij Indicates the road segment X to be evaluated i About indicator Y j The measured value, for each x ij There are p evaluation levels C 1 ,C 2 ,...,C p , that is, the level space C={C 1 ,C 2 ,...,C p}, let μ ijk Represents the measured value x ij Belongs to the kth evaluation level C k The degree of jk For a single indicator Y j The interval scale of the grading criterion, and μ ijk Satisfy 0≤μ ijk ≤1, normalization and additivity requirements.

[0015] Based on the above scheme, preferably, the indicator importance is calculated as follows:

[0016] ,

[0017] In the formula, I(Y j ) represents the indicator Y j The importance of f 1 (t) represents the influence of the pavement service life t; f 2 (c) Characterize the periodic impact of climate regional characteristics c; f 3 (v) represents the impact of traffic volume v; f 4 (e j ) Characterize environmental factors j For indicator Y j The impact of 1 , b 2 , b 3 , b 4 is the weight coefficient.

[0018] Based on the above solution, preferably, step S3 includes:

[0019] S31. The measurement function types include basic measurement function, extreme value measurement function and composite measurement function. The measurement function type is selected according to the performance indicator characteristics.

[0020] S32, establishing a neural network optimization framework, including constructing a parameter optimization objective function, designing a parameter update rule, and determining an optimization termination condition;

[0021] S33. Initialize the measurement function parameters, and use the neural network optimization framework to iteratively optimize the parameter values ​​to obtain the optimized measurement function.

[0022] Based on the above solution, preferably, the basic measurement function is divided into the following according to positive indicators and negative indicators:

[0023] For positive indicators, that is, the larger the better indicators:

[0024] ,

[0025] In the formula, x j Y j The measured value of jk is the measured value x j Belongs to the kth evaluation level C k degree; jk Y j In the evaluation level C k The eigenvalue on a jk+1 Y j In the evaluation level C k+1 The eigenvalue on α 1 is the adjustment parameter of the positive indicator, α 1 >0; β 1 is the correction coefficient of the positive indicator, 0≤β 1 ≤1;

[0026] For negative indicators, that is, the smaller the better the indicator:

[0027] ,

[0028] In the formula, α 2 is the adjustment parameter of the negative indicator, α 2 >0; β 2 is the correction coefficient of negative indicators, 0≤β 2 ≤1;

[0029] The extreme value measure function is:

[0030] ,

[0031] In the formula, x* represents the optimal value, σ is the standard deviation parameter, and γ j is the adjustment coefficient;

[0032] The composite measure function is:

[0033] ,

[0034] In the formula, μ base is the basic measurement function, λ is the periodic influence coefficient, κ is the cumulative effect coefficient, and ω is the frequency parameter;

[0035] Among them, α 1 , β1 , α 2 , β 2 , σ, γ j , λ, κ, ω are the parameters to be optimized.

[0036] Based on the above solution, preferably, the parameter optimization objective function is:

[0037] ,

[0038] Where L(θ) is the loss function to be minimized; θ represents the parameter to be optimized of the measurement function; Indicates indicator Y j In the evaluation level C k The observed measurement value under jk (x j ;θ) represents the value of the measurement function calculated using the parameter θ; ε represents the regularization parameter, which is used to prevent overfitting; Represents the square norm of the vectorized parameter θ;

[0039] Among them, the observed measurement value It is the real measurement value used to guide parameter optimization, which can be obtained as follows:

[0040] Collect sample data of road sections with known evaluation levels, including all performance indicators Y j The measured value x j And the corresponding evaluation level C k ;

[0041] According to expert evaluation, determine each indicator Y j In different evaluation levels C k The observed measurement value under , and normalize it to the interval [0,1].

[0042] Based on the above solution, preferably, in step S5, the indicator correlation analysis uses the Pearson correlation coefficient to calculate the correlation between the indicators:

[0043] ,

[0044] In the formula, r sj Indicates indicator Y s With indicator Y j The correlation between si 、x ji are the index Y of the i-th road section to be evaluated s , Y j The measured value of; n represents the number of road sections to be evaluated; , The indicators Y s , Yj The average value of the measurements;

[0045] The formula for calculating the combined weight is:

[0046] ,

[0047] Where W j Y j The combined weight of is the subjective weight obtained through AHP; is the objective weight obtained by entropy weight method; δ j Y j The average correlation of ; Based on indicator Y j adjustment factors for variability;

[0048] The combined weights of m indicators form a combined weight vector W = {W 1 ,W 2 ,…,W m}.

[0049] Based on the above solution, preferably, in step S6, the calculation process of the comprehensive measurement evaluation matrix is:

[0050] Calculate the road segment X to be evaluated i Rating level C k The comprehensive measure value μ ik :

[0051] ,

[0052] In the formula, x ij The road segment X to be evaluated i In the indicator Y j The measured value on jk (x ij ) is the single indicator measurement value calculated based on the measurement function;

[0053] The comprehensive measurement values ​​of all the sections to be evaluated are combined into a comprehensive measurement evaluation matrix μ of the evaluation space, which is in the form of:

[0054] ,

[0055] Where n is the number of road sections to be evaluated, and p is the number of evaluation levels.

[0056] Based on the above solution, preferably, step S7 includes:

[0057] According to the confidence calculation formula, we can get the confidence value of each road section to be evaluated. iThe confidence level at all evaluation levels is calculated as follows:

[0058] ,

[0059] Where P ik The road segment X to be evaluated i Rating level C k Confidence level; is the dynamic adjustment coefficient; η k Rating C k The dynamic adjustment factor is defined as:

[0060] ,

[0061] In the formula, R jk Indicates indicator Y j With evaluation grade C k The correlation is calculated based on the correlation and variability of the indicators; m is the number of performance indicators;

[0062] Adopt the maximum confidence principle to determine the road section X to be evaluated. i Initial evaluation grade C k* ,in:

[0063] ,

[0064] In order to avoid excessive jumps in evaluation levels, adjustment rules are defined in combination with the adjacency constraints of evaluation levels:

[0065] ,

[0066] Where P ik’ is the confidence level of the evaluation level corresponding to the second highest confidence level; is the confidence difference threshold;

[0067] According to the adjusted evaluation level C k* , output the final evaluation level of each road section to be evaluated.

[0068] Compared with the prior art, the present invention has the following beneficial effects:

[0069] (1) The asphalt pavement comprehensive performance evaluation method based on the unascertained measurement theory proposed in this invention achieves a comprehensive evaluation of the pavement surface and internal performance by establishing a complete evaluation index system and measurement function set, combining a neural network optimization framework and a combined weighting model, thereby improving the accuracy and reliability of the evaluation results and providing a more scientific technical basis for making maintenance decisions;

[0070] (2) The performance index system established by the present invention includes basic performance indexes and extended performance indexes, and introduces an index importance calculation method based on the pavement service life, climate region characteristics and traffic volume, so that the selection of performance indexes is more targeted and can more comprehensively reflect the actual condition of the pavement;

[0071] (3) The measurement function parameter optimization framework designed by the present invention optimizes the parameters of the basic measurement function, the extreme value measurement function and the composite measurement function through neural network iteration, thereby improving the adaptability of the measurement function to different types of indicator characteristics and enhancing the generalization ability of the evaluation model;

[0072] (4) The combined weighting model adopted by the present invention effectively balances expert experience and data statistical information by combining the subjective weight of the hierarchical analysis method and the objective weight of the entropy weight method, and introduces indicator correlation analysis, making the weight distribution more reasonable;

[0073] (5) The rating method based on the confidence identification criterion proposed in the present invention effectively avoids the jump phenomenon of evaluation level by introducing dynamic adjustment factors and rating adjustment rules, thereby improving the stability and credibility of the evaluation results. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] 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 or the description of the prior art 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.

[0075] Figure 1 is a flow chart of a method according to an embodiment of the present invention;

[0076] Figure 2 It is the IRI measurement function diagram in the embodiment of the present invention;

[0077] Figure 3 It is a graph of RD measurement function in an embodiment of the present invention;

[0078] Figure 4 It is a SFC measurement function diagram in an embodiment of the present invention;

[0079] Figure 5 It is a graph of the SSI measurement function in an embodiment of the present invention;

[0080] Figure 6 It is a PCR measurement function diagram in an embodiment of the present invention;

[0081] Figure 7 It is a PPR measurement function diagram in an embodiment of the present invention;

[0082] Figure 8 It is a diagram of the IDR measurement function in an embodiment of the present invention. DETAILED DESCRIPTION

[0083] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0084] like Figure 1 As shown, the present invention provides a method for evaluating the comprehensive performance of asphalt pavement based on the unascertained measurement theory, comprising:

[0085] S1. Establish a performance index system for asphalt pavement, determine the road section to be evaluated, and obtain performance index data for the road section to be evaluated; when establishing the performance index system, select relevant indicators of the road section to be evaluated in the performance index set according to the screening criteria to form a performance index system; the performance index set includes basic performance indicators and extended performance indicators, and the basic performance indicators include flatness index IRI, rutting depth RD, lateral force coefficient SFC, structural strength coefficient SSI, pavement crack rate PCR, pavement repair rate PPR and internal damage rate IDR; the extended performance indicators include temperature sensitivity index TEI, noise attenuation index NAI, water damage resistance index WRI, aging index AGI and traffic loss index TDI; the screening criteria are: based on the service life of the pavement, climate regional characteristics, and traffic volume, calculate the importance of the indicators, and include the indicators with an importance greater than the preset threshold into the performance index system;

[0086] S2. Based on the performance indicator data, establish an evaluation level space and determine the grading standard matrix of each performance indicator at different evaluation levels;

[0087] S3. Establish a neural network-based measurement function parameter optimization framework, establish a measurement function set including basic measurement functions, extreme value measurement functions and composite measurement functions based on the unascertained measurement theory, select the measurement function type according to the performance indicator characteristics, and optimize the measurement function parameters through the optimization framework; wherein, the optimization framework includes constructing a parameter optimization objective function, designing parameter update rules and determining the optimization termination conditions;

[0088] S4. Calculate the single index measurement value of the road section to be evaluated in the evaluation level space based on the optimized measurement function to obtain a single index measurement matrix;

[0089] S5. Establish a combined weighting model based on the hierarchical analysis method and entropy weight method, and calculate the combined weight vector of each indicator by combining the indicator correlation analysis;

[0090] S6. Calculate the comprehensive measurement evaluation matrix of the road section to be evaluated by combining the single index measurement matrix and the combined weight vector;

[0091] S7. Based on the confidence recognition criterion, the final evaluation level of the road section to be evaluated is determined according to the comprehensive measurement evaluation matrix.

[0092] Specifically, in one embodiment of the present invention, before the asphalt pavement performance evaluation is performed, the following assumptions are made about the evaluation object and the evaluation level:

[0093] The set of n road sections to be evaluated is defined as the evaluation space, denoted as X={X 1 ,X 2 ,…,X n}; The set of m performance indicators contained in each road section to be evaluated is defined as the indicator space, denoted as Y={Y 1 ,Y 2 ,...,Y m}, x ij Indicates the road segment X to be evaluated i About indicator Y j The measured value, for each x ij There are p evaluation levels C 1 ,C 2 ,...,C p , that is, the level space C={C 1 ,C 2 ,...,C p}, let μ ijk Represents the measured value x ij Belongs to the kth evaluation level C k The degree of jk For a single indicator Y j The interval scale of the grading criterion, and μ ijk Satisfy 0≤μ ijk ≤1, the requirements of normalization and additivity are as follows:

[0094] ,

[0095] ,

[0096] ,

[0097] Then μ is called an unascertained measure. The above three formulas respectively indicate that μ satisfies "non-negative boundedness", "normalization" and "additivity" for the rank space C.

[0098] Specifically, in one embodiment of the present invention, when establishing the performance indicator system, the relevant indicators of the road section to be evaluated are screened from the performance indicator set according to the screening criteria to form the performance indicator system;

[0099] The performance index set includes basic performance index and extended performance index. The basic performance index includes roughness index IRI, rutting depth RD, lateral force coefficient SFC, structural strength coefficient SSI, pavement crack rate PCR, pavement repair rate PPR and internal damage rate IDR; the extended performance index includes temperature sensitivity index TEI, noise attenuation index NAI, water damage resistance index WRI, aging index AGI and traffic loss index TDI.

[0100] The screening criteria are as follows: based on the age of the pavement, climate zone characteristics, and traffic volume calculation indicator importance, indicators with an importance greater than a preset threshold are included in the performance indicator system.

[0101] In a specific example, IRI, RD and SFC are directly obtained from a multifunctional road detection vehicle, and SSI, PCR, PPR and IDR are calculated as follows:

[0102] Structural Strength Index (SSI):

[0103] ,

[0104] Where: Design deflection for the road surface; It represents the measured deflection.

[0105] Pavement crack rate (PCR):

[0106] ,

[0107] Where: L i is the length of the i-th crack (m); ω i is the weight of the ith crack, as shown in Table 1; A is the total pavement area of ​​the investigated section (m 2 ).

[0108] When calculating the pavement crack rate, ω i The determination is shown in Table 1.

[0109] Table 1 Asphalt pavement crack weights

[0110]

[0111] Pavement Repair Rate (PPR):

[0112] ,

[0113] Where: A i is the road repair area at the ith location (m 2 ); A is the total road surface area of ​​the surveyed section (m 2 ).

[0114] Internal damage rate (IDR):

[0115] ,

[0116] Where: A i is the internal damaged area of ​​the road surface at the i-th location (m 2 );ω i is the internal damage weight of the road surface at the ith location, as shown in Table 2; A is the total road surface area of ​​the investigated section (m 2 ).

[0117] When calculating the damage rate of the road surface, ω i The determination is shown in Table 2.

[0118] Table 2 Internal damage weights of asphalt pavement

[0119]

[0120] Temperature Sensitivity Index (TEI):

[0121] The temperature sensors buried in the road surface are used to record the temperature changes of the surface and bottom layers of the road surface at different time periods (temperature difference). T). Substitute the obtained temperature difference into the formula:

[0122] ,

[0123] Where L is the temperature gradient influence range.

[0124] Noise Attenuation Index (NAI):

[0125] Use noise sensing equipment to directly test the noise level N generated when a vehicle passes through different types of roads before and the attenuated noise level N after :

[0126] NAI=N before -N after .

[0127] Water Resistance Index (WRI):

[0128] Use water seepage equipment to measure the water seepage time t of the road surface w , flow rate q and water pressure conditions; at the same time, the anti-adhesion ability of the surface material is analyzed through relevant experiments. Combining the various parameters obtained, the anti-water damage index formula is introduced:

[0129]

[0130] Where A represents the seepage area.

[0131] The aging index (AGI) is calculated based on the material strength, elongation, etc., and the change ratio of the data collected twice before and after is obtained to obtain the aging index AGI.

[0132] Traffic loss index (TDI):

[0133] Through sensors or traffic monitoring equipment to obtain traffic volume (ADT), heavy vehicle ratio (P h ), durability parameters of pavement materials. These data are related to the pavement usage time to reflect the cumulative damage of traffic load to the pavement:

[0134] ,

[0135] Among them, k is the material durability coefficient and t is the service time.

[0136] The indicator importance is calculated as follows:

[0137] ,

[0138] In the formula, I(Y j ) represents the indicator Y j The importance of f 1 (t) represents the influence of the pavement service life t; f 2 (c) Characterize the periodic impact of climate regional characteristics c; f 3 (v) represents the impact of traffic volume v; f 4 (e j ) Characterize environmental factors j For indicator Y j The impact of 1 , b 2 , b 3 , b 4 is the weight coefficient.

[0139] By calculating the importance value of each indicator, comparing the indicator importance with the preset threshold, the indicators with importance greater than the threshold are included in the final evaluation system.

[0140] Specifically, in one embodiment of the present invention, the evaluation levels are divided into five levels: excellent, good, medium, poor, and bad, that is, the level space C={C 1 ,C 2 ,C 3 ,C 4 ,C 5 Each performance indicator Y j The grading standard matrix a jk It is used to define the grading interval of the indicator at different evaluation levels. The establishment of the grading standard matrix is ​​based on the following principles:

[0141] Data source: Determine the grading range for each indicator by combining historical test data, expert experience and relevant technical specifications.

[0142] Interval division: The grading interval of each indicator [a jk ,a jk+1 ]Meet the positive or negative requirements of indicator characteristics.

[0143] In a specific example, taking the basic performance indicators as an example, the grading standards are shown in Table 3.

[0144] Table 3 Classification Standards

[0145]

[0146] The above classification standards are organized into a matrix form, denoted as a jk , which is used to calculate the single indicator measurement value in the subsequent steps. Each column of the matrix corresponds to an evaluation level, and each row corresponds to a grading interval of a performance indicator.

[0147] Specifically, in one embodiment of the present invention, step S3 includes:

[0148] S31. The measurement function types include basic measurement function, extreme value measurement function and composite measurement function. The measurement function type is selected according to the performance indicator characteristics.

[0149] S32, establishing a neural network optimization framework, including constructing a parameter optimization objective function, designing a parameter update rule, and determining an optimization termination condition;

[0150] S33. Initialize the measurement function parameters, and use the neural network optimization framework to iteratively optimize the parameter values ​​to obtain the optimized measurement function.

[0151] In this embodiment, the basic measurement function is divided into the following according to positive indicators and negative indicators:

[0152] For positive indicators, that is, the larger the better indicators:

[0153] ,

[0154] In the formula, x j Y j The measured value of jk is the measured value x j Belongs to the kth evaluation level C k degree; jk Y j In the evaluation level C k The eigenvalue on jk+1 Y j In the evaluation level C k+1 The eigenvalue on α1 is the adjustment parameter of the positive indicator, α 1 >0; β 1 is the correction coefficient of the positive indicator, 0≤β 1 ≤1;

[0155] For negative indicators, that is, the smaller the better the indicator:

[0156] ,

[0157] In the formula, α 2 is the adjustment parameter of the negative indicator, α 2 >0; β 2 is the correction coefficient of negative indicators, 0≤β 2 ≤1;

[0158] The extreme value measure function is:

[0159] ,

[0160] In the formula, x* represents the optimal value, σ is the standard deviation parameter, and γ j is the adjustment coefficient;

[0161] The composite measure function is:

[0162] ,

[0163] In the formula, μ base is the basic measurement function, λ is the periodic influence coefficient, κ is the cumulative effect coefficient, and ω is the frequency parameter;

[0164] Among them, α 1 , β 1 , α 2 , β 2 , σ, γ j , λ, κ, ω are the parameters to be optimized.

[0165] Specifically, the mapping relationship between each performance indicator and the measurement function is as follows:

[0166] IRI (Roughness Index): Negative basic measurement function, because the smaller the value, the smoother the road surface; RD (Rutting Depth): Negative basic measurement function, the smaller the value, the better; SFC (Lateral Force Coefficient): Positive basic measurement function, the larger the value, the better the anti-skid performance; SSI (Structural Strength Coefficient): Positive basic measurement function, the larger the value, the better the structural strength; PCR (Pavement Crack Rate): Negative basic measurement function, the smaller the value, the better; PPR (Pavement Repair Rate): Negative basic measurement function, the smaller the value, the better; IDR (Internal Damage Rate): Negative basic measurement function, the smaller the value, the better.

[0167] TEI (temperature sensitivity index): an extreme value measurement function with an optimal temperature sensitivity range; NAI (noise attenuation index): a composite measurement function that takes seasonal effects into account; WRI (water damage resistance index): a composite measurement function that takes the impact of rainfall cycles into account; AGI (aging index): a composite measurement function that takes cumulative effects into account; TDI (traffic loss index): a composite measurement function that takes cumulative effects into account.

[0168] This embodiment uses the unascertained measure theory to construct a measure function, taking the basic measure function as an example:

[0169] In order to obtain the unascertained measure value of the performance index, it is necessary to construct the unascertained measure function according to the definition of the unascertained measure. The unascertained measure function is a piecewise function, which is constructed based on the evaluation level space. For an ordered partition class {C 1 ,C 2 ,···,C p}, let a k (k=1,2,···,p) represents the eigenvalue of the corresponding evaluation level. The actual meaning of the measure function is that the eigenvalue is from a k Increase to a k+1 In the process of k The degree of the level, the corresponding function value is the measure of the indicator, and a linear method is adopted to construct the unascertained measure function based on comprehensive considerations.

[0170] The level space established in the embodiment of the present invention is C={C 1 ,C 2 ,C 3 ,C 4 ,C 5}={excellent, good, medium, inferior, poor}. According to the grading standard, the unascertained measurement function of the performance index is constructed, and the characteristic value of the index is obtained using the interval number, C 1 The grading standard takes the lower limit of the interval, C 5 The grading standard takes the upper limit of the interval, C 2 , C 3 , C 4 The grading standard is the median of the interval numbers.

[0171] Specifically, in this embodiment, the constructed parameter optimization objective function is:

[0172] ,

[0173] Where L(θ) is the loss function to be minimized; θ represents the parameter to be optimized of the measurement function; Indicates indicator Y j In the evaluation level C k The observed measurement value under jk(x j ;θ) represents the value of the measurement function calculated using the parameter θ; ε represents the regularization parameter, which is used to prevent overfitting; Represents the square norm of the vectorized parameter θ;

[0174] Among them, the observed measurement value It is the real measurement value used to guide parameter optimization, which can be obtained as follows:

[0175] Collect sample data of road sections with known evaluation levels, including all performance indicators Y j The measured value x j And the corresponding evaluation level C k ;

[0176] According to expert evaluation, determine each indicator Y j In different evaluation levels C k The observed measurement value under , and normalize it to the interval [0,1].

[0177] The parameter optimization steps are as follows:

[0178] Collect sample data of road sections with known evaluation levels; obtain observation measurement values ​​evaluated by experts ; Initialize the parameter θ to be optimized; Use the neural network to iteratively optimize the parameter value; Stop the optimization when the loss function converges or reaches the maximum number of iterations.

[0179] In this embodiment, the parameter update adopts the gradient descent method, and the formula is as follows:

[0180] ,

[0181] In the formula, is the learning rate, Represents the gradient of the loss function. When calculating the gradient, the gradient of the error term is:

[0182] ,

[0183] The gradient of the regularization term is:

[0184] ,

[0185] Through the above rules, the parameter θ of the measurement function can be continuously adjusted during the optimization process to improve the evaluation accuracy and reliability of the model.

[0186] Specifically, in one embodiment of the present invention, step S4 includes:

[0187] According to the measurement function of each performance index, the single index unascertained measurement evaluation matrix of different road sections is solved. In this step, the single index unascertained measurement matrix is:

[0188] .

[0189] Specifically, in one embodiment of the present invention, step S5 includes:

[0190] S51. Calculate the subjective weight. According to the basic principle of hierarchical analysis method, use the "1-9" scale method to judge the relative importance of each performance indicator and give the relative importance value a. ij , the importance scale is shown in Table 4; thus establishing the judgment matrix A=(a ij ) n×n .

[0191] Table 4 Meaning of the “1-9” scale

[0192]

[0193] To ensure the credibility of the judgment matrix, calculate the consistency ratio:

[0194] ,

[0195] Among them, CR is the consistency ratio, which is generally less than 0.1, which is considered to have passed the consistency test; n is the order of the judgment matrix; RI is the average random consistency index, which can be found in Table 5; λ max is the maximum eigenvalue.

[0196] Table 5 Random consistency index

[0197]

[0198] Calculate the geometric mean of each row of the judgment matrix:

[0199] ,

[0200] After normalization, we get the index Y j Subjective weight of:

[0201] .

[0202] S52. Calculate the objective weight and the contribution of each evaluation index:

[0203] ,

[0204] Among them, n is the total number of road sections to be evaluated, x ij The road segment X to be evaluated i Index Y j The data, p ij The road segment X to be evaluated i For indicator Y jdegree of contribution.

[0205] Calculate the entropy value:

[0206] ,

[0207] Among them, e j Indicates indicator Y j Its value is positively correlated with the dispersion degree of the evaluation index.

[0208] Calculate the entropy redundancy of the indicator:

[0209] ,

[0210] Among them, d j Y j The entropy redundancy of , its value directly affects the value of the weight.

[0211] Calculate the objective weight of each indicator:

[0212] ,

[0213] Among them, m is the number of indicators, Y j objective weight.

[0214] S53. Use Pearson correlation coefficient to calculate the correlation between indicators:

[0215] ,

[0216] In the formula, r sj Indicates indicator Y s With indicator Y j The correlation between si 、x ji are the index Y of the i-th road section to be evaluated s , Y j The measured value of; n represents the number of road sections to be evaluated; , The indicators Y s , Y j The average value of the measurements.

[0217] S54. Calculate the combined weight of each indicator:

[0218] ,

[0219] Where W j Y j The combined weight of is the subjective weight obtained through AHP; is the objective weight obtained by entropy weight method; δj Y j The average correlation of ; Based on indicator Y j The adjustment factor of variability; the combined weights of m indicators form a combined weight vector W = {W 1 ,W 2 ,…,W m}.

[0220] Specifically, in one embodiment of the present invention, step S6 includes:

[0221] According to the combined weight vector W={W 1 ,W 2 ,…,W m},w j Represents the evaluation object X i The weight value of the jth indicator, then w j satisfy:

[0222] ,

[0223] Let μ ik =μ(X i C k ) represents the evaluation object X i Rating level C k The weight of each index is introduced to form the road section to be evaluated X i Rating level C k The comprehensive measure value μ ik :

[0224] ,

[0225] In the formula, x ij The road segment X to be evaluated i In the indicator Y j The measured value on jk (x ij ) is the single index measurement value calculated based on the measurement function; the above formula obviously satisfies μ ik [0,1] and , then μ ik satisfies the definition of an unascertained measure, and μ ik Represents the evaluation object X i Rating level C k The multi-index comprehensive measure of all evaluation objects can form the multi-index comprehensive measure evaluation matrix μ of the evaluation space, which is in the form of:

[0226] ,

[0227] Where n is the number of road sections to be evaluated, and p is the number of evaluation levels.

[0228] Specifically, in one embodiment of the present invention, step S7 includes:

[0229] According to the confidence calculation formula, we can get the confidence value of each road section to be evaluated. i The confidence level at all evaluation levels is calculated as follows:

[0230] ,

[0231] Where P ik The road segment X to be evaluated i Belongs to the evaluation level C k confidence level; is the dynamic adjustment coefficient; η k Rating level C k The dynamic adjustment factor is defined as:

[0232] ,

[0233] In the formula, R jk Indicates indicator Y j With evaluation grade C k The correlation is calculated based on the correlation and variability of the indicators; m is the number of performance indicators;

[0234] Adopt the maximum confidence principle to determine the road section X to be evaluated. i Initial evaluation grade C k* ,in:

[0235] ,

[0236] In order to avoid excessive jumps in evaluation levels, adjustment rules are defined in combination with the adjacency constraints of evaluation levels:

[0237] ,

[0238] Where P ik’ is the confidence level of the evaluation level corresponding to the second highest confidence level; is the confidence difference threshold;

[0239] According to the adjusted evaluation level C k* , output the final evaluation level of each road section to be evaluated.

[0240] The scheme of the present invention is described in detail with a specific embodiment below:

[0241] The section from Shaoguan Gantang to Guangzhou Taihe of Beijing-Hong Kong-Macao Expressway in Guangdong Province was taken as the evaluation object. According to the calculation and screening of the index importance, the performance index system of this embodiment is the basic performance index, which includes the flatness index IRI, rutting depth RD, lateral force coefficient SFC, structural strength coefficient SSI, pavement crack rate PCR, pavement repair rate PPR and internal damage rate IDR. The pavement disease detection data of 10 sections were collected by using a comprehensive road inspection vehicle, and the performance index was calculated, as shown in Table 6.

[0242] Table 6 Pavement performance indicators of the test section

[0243]

[0244] According to the grading standards of each performance index in Table 3, the established index measurement functions are as follows: Figure 2-Figure 8 shown.

[0245] According to the measurement function of the single indicator, the specific value of the performance indicator can be substituted to obtain the measurement value of each level. The single indicator unascertained measurement evaluation matrix of the 10 sections to be evaluated is as follows:

[0246] ,

[0247] ,

[0248] ,

[0249] ,

[0250] ,

[0251] ,

[0252] ,

[0253] ,

[0254] ,

[0255] ,

[0256] Use the analytic hierarchy process to find the subjective weight:

[0257] Invite relevant experts to compare the performance indicators in pairs and score them according to the "1-9" scale method to obtain the judgment matrix as shown below:

[0258] ,

[0259] When the consistency test is satisfied, the eigenvector corresponding to the maximum eigenvalue of the judgment matrix is ​​obtained and normalized to obtain the subjective weight vector:

[0260] ,

[0261] According to the index values ​​in Table 6, the entropy weight method is used to find the objective weight of each index. The results are shown in Table 7:

[0262] Table 7 Performance index objective weight calculation table

[0263]

[0264] In this embodiment, the combined weight vector can be obtained using the formula:

[0265] ,

[0266] Combining the combined weights of each performance index and the single index unascertained measurement matrix of the 10 sections to be evaluated, the multi-index unascertained measurement evaluation matrix for comprehensive evaluation can be obtained:

[0267] ,

[0268] The confidence identification method was used to rate the comprehensive performance of asphalt pavement, and the results are shown in Table 8.

[0269] Table 8 Comprehensive performance rating results of asphalt pavement in each road section

[0270]

[0271] In order to compare with the standard evaluation method, this embodiment collects the road surface comprehensive condition index PQI of each road section. The evaluation results of the two methods are shown in Table 9:

[0272] Table 9 Evaluation results and comparison

[0273]

[0274] It can be found that except for Section 6, Section 7, and Section 9, the evaluation results of the combined weighted unascertained measurement method proposed by the present invention for the remaining sections are one level lower than those of the standard evaluation method, indicating that the comprehensive evaluation considering the hidden diseases inside the asphalt pavement can more accurately reflect the actual situation of the asphalt pavement. In addition, the combined weighted method fully considers the changes in the conditions of the asphalt pavement in different sections, thereby realizing the organic combination of subjective and objective factors in the evaluation process, and effectively avoiding the adverse effects of using fixed weights. More importantly, the unascertained measurement theory uses the confidence recognition criterion to determine the evaluation level, avoiding the defect of ignoring certain evaluation index information when identifying the maximum membership criterion, making the comprehensive performance evaluation of the asphalt pavement more reasonable. Therefore, compared with the current specifications, the present invention proposes a more comprehensive and reasonable comprehensive evaluation method, which can more accurately describe the performance of the asphalt pavement, and solves the problem that the existing evaluation method cannot simultaneously evaluate from the three aspects of surface diseases, internal diseases, and structural strength, thereby more scientifically guiding the subsequent maintenance plan decision.

[0275] 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 principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A comprehensive performance evaluation method for asphalt pavement based on unascertained measure theory, characterized in that: include: S1. Establish a performance index system for asphalt pavement, determine the road section to be evaluated, and obtain performance index data for the road section to be evaluated; when establishing the performance index system, select relevant indicators of the road section to be evaluated in the performance index set according to the screening criteria to form a performance index system; the performance index set includes basic performance indicators and extended performance indicators, and the basic performance indicators include flatness index IRI, rutting depth RD, lateral force coefficient SFC, structural strength coefficient SSI, pavement crack rate PCR, pavement repair rate PPR and internal damage rate IDR; the extended performance indicators include temperature sensitivity index TEI, noise attenuation index NAI, water damage resistance index WRI, aging index AGI and traffic loss index TDI; the screening criteria are: based on the service life of the pavement, climate regional characteristics, and traffic volume, calculate the importance of the indicators, and include the indicators with an importance greater than the preset threshold into the performance index system; S2. Based on the performance indicator data, establish an evaluation level space and determine the grading standard matrix of each performance indicator at different evaluation levels; S3. Establish a neural network-based measurement function parameter optimization framework, establish a measurement function set including basic measurement functions, extreme value measurement functions and composite measurement functions based on the unascertained measurement theory, select the measurement function type according to the performance indicator characteristics, and optimize the measurement function parameters through the optimization framework; wherein, the optimization framework includes constructing a parameter optimization objective function, designing parameter update rules and determining the optimization termination conditions; Step S3 includes: S31. The measurement function types include basic measurement function, extreme value measurement function and composite measurement function. The measurement function type is selected according to the performance indicator characteristics. S32, establishing a neural network optimization framework, including constructing a parameter optimization objective function, designing a parameter update rule, and determining an optimization termination condition; S33, initializing measurement function parameters, and iteratively optimizing parameter values ​​using a neural network optimization framework to obtain an optimized measurement function; The basic measurement function is divided into positive and negative indicators: For positive indicators, that is, the larger the better indicators: In the formula, x j Y j The measured value of jk is the measured value x j Belongs to the kth evaluation level C k degree; jk Y j In the evaluation level C k The eigenvalue on jk+1 Y j In the evaluation level C k+1 The characteristic value on; α1 is the adjustment parameter of the positive indicator, α1>0; β1 is the correction coefficient of the positive indicator, 0≤β1≤1; For negative indicators, that is, the smaller the better the indicator: In the formula, α2 is the adjustment parameter of the negative indicator, α2>0; β2 is the correction coefficient of the negative indicator, 0≤β2≤1; The extreme value measure function is: In the formula, x* represents the optimal value, σ is the standard deviation parameter, and γ j is the adjustment coefficient; The composite measure function is: In the formula, μ base is the basic measurement function, λ is the periodic influence coefficient, κ is the cumulative effect coefficient, and ω is the frequency parameter; Among them, α1, β1, α2, β2, σ, γ j 、λ、k、ω are parameters to be optimized; S4. Calculate the single index measurement value of the road section to be evaluated in the evaluation level space based on the optimized measurement function to obtain a single index measurement matrix; S5. Establish a combined weighting model based on the hierarchical analysis method and entropy weight method, and calculate the combined weight vector of each indicator by combining the indicator correlation analysis; S6. Calculate the comprehensive measurement evaluation matrix of the road section to be evaluated by combining the single index measurement matrix and the combined weight vector; S7. Based on the confidence recognition criterion, the final evaluation level of the road section to be evaluated is determined according to the comprehensive measurement evaluation matrix.

2. The method for evaluating comprehensive performance of asphalt pavement based on unascertained measure theory as claimed in claim 1, characterized in that: The set of n road sections to be evaluated is defined as the evaluation space, denoted as X={X1,X2,…,X n }; The set of m performance indicators contained in each road section to be evaluated is defined as the indicator space, denoted as Y={Y1,Y2,...,Y m }, x ij Indicates the road segment X to be evaluated i About indicator Y j The measured value, for each x ij There are p evaluation levels C1, C2, ..., C p , that is, the level space C={C1,C2,...,C p }, let μ ijk Represents the measured value x ij Belongs to the kth evaluation level C k The degree of jk For a single indicator Y j The interval scale of the grading criterion, and μ ijk Satisfy 0≤μ ijk ≤1, normalization and additivity requirements.

3. The method for evaluating comprehensive performance of asphalt pavement based on unascertained measure theory as claimed in claim 1, characterized in that: The indicator importance is calculated as follows: In the formula, I(Y j ) represents the indicator Y j f1(t) represents the influence of the road surface age t; f2(c) represents the periodic influence of climate zone characteristics c; f3(v) represents the influence of traffic volume v; f4(e j ) Characterize environmental factors j For indicator Y j The influence of; b1, b2, b3, b4 are weight coefficients.

4. The method for evaluating comprehensive performance of asphalt pavement based on unascertained measure theory according to claim 1, characterized in that: The parameter optimization objective function is: Where L(θ) is the loss function to be minimized; θ represents the parameter to be optimized of the measurement function; Indicates indicator Y j In the evaluation level C k The observed measurement value under jk (x j ;θ) represents the value of the measurement function calculated using the parameter θ; ε represents the regularization parameter, which is used to prevent overfitting; Represents the square norm of the vectorized parameter θ; Among them, the observed measurement value It is the real measurement value used to guide parameter optimization, which can be obtained as follows: Collect sample data of road sections with known evaluation levels, including all performance indicators Y j The measured value x j And the corresponding evaluation level C k ; According to expert evaluation, determine each indicator Y j In different evaluation levels C k The observed measurement value under , and normalize it to the interval [0,1].

5. The method for evaluating comprehensive performance of asphalt pavement based on unascertained measure theory as claimed in claim 2, characterized in that: In step S5, the indicator correlation analysis uses the Pearson correlation coefficient to calculate the correlation between indicators: In the formula, r sj Indicates indicator Y s With indicator Y j The correlation between si 、x ji are the index Y of the i-th road section to be evaluated s , Y j The measured value of; n represents the number of road sections to be evaluated; , The indicators Y s , Y j The average value of the measurements; The formula for calculating the combined weight is: Where W j Y j The combined weight of is the subjective weight obtained through AHP; is the objective weight obtained by entropy weight method; δ j Y j The average correlation of ; Based on indicator Y j adjustment factors for variability; The combined weights of m indicators form a combined weight vector W={W1,W2,…,W m }.

6. The method for evaluating comprehensive performance of asphalt pavement based on unascertained measure theory as claimed in claim 2, characterized in that: In step S6, the calculation process of the comprehensive measurement evaluation matrix is: Calculate the road segment X to be evaluated i Belongs to the evaluation level C k The comprehensive measure value μ ik : In the formula, x ij The road segment X to be evaluated i In the indicator Y j The measured value on jk (x ij ) is the single indicator measurement value calculated based on the measurement function; The comprehensive measurement values ​​of all the sections to be evaluated are combined into a comprehensive measurement evaluation matrix μ of the evaluation space, which is in the form of: Where n is the number of road sections to be evaluated, and p is the number of evaluation levels.

7. The method for evaluating comprehensive performance of asphalt pavement based on unascertained measure theory as claimed in claim 6, characterized in that: Step S7 includes: According to the confidence calculation formula, we can get the confidence value of each road section to be evaluated. i The confidence level at all evaluation levels is calculated as follows: Where P ik The road segment X to be evaluated i Belongs to the evaluation level C k Confidence level; is the dynamic adjustment coefficient; η k Rating C k The dynamic adjustment factor is defined as: In the formula, R jk Indicates indicator Y j With evaluation grade C k The correlation is calculated based on the correlation and variability of the indicators; m is the number of performance indicators; Adopt the maximum confidence principle to determine the road section X to be evaluated. i Initial evaluation grade C k* ,in: In order to avoid excessive jumps in evaluation levels, adjustment rules are defined in combination with the adjacency constraints of evaluation levels: Where P ik’ is the confidence level of the evaluation level corresponding to the second highest confidence level; is the confidence difference threshold; according to the adjusted evaluation level C k* , output the final evaluation level of each road section to be evaluated.

Images