A nondestructive testing method and system for the aging degree of asphalt

Abstract

The application discloses a kind of asphalt aging degree nondestructive testing method and system, belong to road material detection technical field.The method is first prepared by different aging degree standard asphalt sample, and the real aging degree is calibrated in combination with destructive test, and the broadband dielectric spectrum under multiple temperature points is measured, and the multi-temperature field standard dielectric spectrum-aging degree database is constructed;Further, the comprehensive dielectric characteristic parameters containing the fixed-point dielectric characteristics under specific temperature point and the evolution dielectric characteristics changing with temperature are extracted from the database;Using these characteristic parameters, the artificial neural network model is trained, the nonlinear mapping relationship between aging degree is established, and the aging degree prediction model is formed;Finally, the nondestructive multi-temperature field dielectric detection is carried out to the asphalt sample to be measured, and the obtained dielectric data is extracted after the same feature is input into the prediction model, and the aging degree is calculated.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of road material detection, and particularly relates to a method and system for nondestructive detection of the aging degree of asphalt. BACKGROUND

[0002] Asphalt materials are exposed to environmental factors such as oxygen, ultraviolet light and temperature cycles during service, which will cause different degrees of aging, leading to the gradual degradation of their physical and mechanical properties, and thus affecting the durability and service life of asphalt pavements. Accurate assessment of the aging state of asphalt is of great significance for pavement maintenance decision-making and residual life prediction.

[0003] Currently, the widely used aging degree evaluation methods in the industry are mainly divided into destructive tests and nondestructive detection. Destructive tests such as Fourier transform infrared spectroscopy (FTIR) and dynamic shear rheological (DSR) testing can provide accurate aging indicators (such as carbonyl index, complex modulus, etc.), but they require sampling and laboratory analysis, which is time-consuming and not suitable for on-site rapid evaluation and in-situ monitoring. Therefore, researchers have gradually turned to the development of nondestructive detection techniques, among which methods based on dielectric properties have attracted attention due to their fast response speed and sensitivity to material polarity changes.

[0004] Existing studies have attempted to indirectly reflect the aging state of asphalt by measuring its dielectric parameters (such as dielectric constant, loss factor) at a single frequency or a single temperature. However, asphalt is a temperature-sensitive material, and its dielectric behavior is strongly dependent on temperature changes. Measurements at a single temperature are easily disturbed by environmental fluctuations, leading to insufficient stability of the evaluation results. In addition, existing methods rely on simple empirical formulas or linear regression models, which are difficult to capture the complex nonlinear relationship between dielectric response and microstructure changes during the aging process, limiting their applicability and accuracy in wide aging ranges and actual temperature variation scenarios. SUMMARY

[0005] The present application overcomes the shortcomings of the prior art and provides a method and system for nondestructive detection of the aging degree of asphalt.

[0006] To achieve the above-mentioned purpose, the technical solution adopted by the present application is as follows: a method for nondestructive detection of the aging degree of asphalt, comprising the following steps:

[0007] S1: Constructing a multi-temperature domain standard dielectric spectrum-aging degree database; the multi-temperature domain standard dielectric spectrum-aging degree database is obtained by measuring the dielectric spectrum of standard asphalt samples at multiple temperature points under gradient aging degrees, and correlating the dielectric spectrum with the aging degree indicators quantified by destructive tests;

[0008] S2: Extracting the comprehensive dielectric characteristic parameters sensitive to aging based on the multi-temperature zone standard dielectric spectrum-aging degree database; the comprehensive dielectric characteristic parameters include fixed-point dielectric characteristic parameters and evolution dielectric characteristic parameters; the fixed-point dielectric characteristic parameters are extracted from the dielectric spectrum, and the evolution dielectric characteristic parameters are regularity indexes of the fixed-point dielectric characteristic parameters changing with temperature;

[0009] S3: Constructing an aging degree prediction model based on the comprehensive dielectric characteristic parameters;

[0010] S4: Performing non-destructive multi-temperature zone dielectric detection on the asphalt sample to be tested to obtain dielectric data of the asphalt sample to be tested;

[0011] S5: Inputting the dielectric data of the asphalt sample to be tested into the aging degree prediction model to calculate the aging degree of the asphalt sample to be tested.

[0012] Further, step S1 includes:

[0013] S11: Preparing standard asphalt samples;

[0014] S12: Performing artificial accelerated aging treatment on the standard asphalt samples to obtain samples with gradient aging degrees;

[0015] S13: Quantitatively characterizing the real aging degrees of each aging asphalt sample through destructive tests;

[0016] S14: Performing multi-temperature zone dielectric spectrum measurement on each aging asphalt sample;

[0017] S15: Associating the dielectric spectrum data of each aging asphalt sample with the corresponding real aging degree to construct a multi-temperature zone standard dielectric spectrum-aging degree database.

[0018] Further, the fixed-point dielectric characteristic parameters include dielectric constant real part values and dielectric loss factor imaginary part values, peak frequency and peak amplitude in the dielectric loss factor imaginary part spectrum, and average relaxation time and relaxation time distribution broadening parameters obtained through Cole-Cole diagram analysis.

[0019] Further, the evolution dielectric characteristic parameters include apparent activation energy obtained based on temperature dependence analysis of the dielectric loss peak frequency, slope of the dielectric constant real part or the dielectric loss factor imaginary part changing with temperature, and Cole-Cole parameter trend changing with temperature.

[0020] Further, the aging degree prediction model in step S3 adopts an artificial neural network model.

[0021] Further, the artificial neural network model comprises an input layer, a plurality of hidden layers and an output layer, wherein the number of neurons of the input layer corresponds to the number of comprehensive dielectric characteristic parameters, the hidden layers comprise at least two layers, each layer contains a plurality of neurons, the hidden layers adopt a ReLU activation function, and the output layer contains one neuron.

[0022] Further, the step S4 comprises:

[0023] S41: preparing a to-be-tested asphalt sample and cleaning a measurement surface;

[0024] S42: setting a multi-temperature-range dielectric spectrum measurement condition, including a temperature range, a temperature step, a stabilization time, a frequency range and an applied voltage;

[0025] S43: performing a multi-temperature-range dielectric spectrum scanning measurement to obtain complex impedance data;

[0026] S44: calculating a real part of a dielectric constant and an imaginary part of a dielectric loss factor according to the complex impedance data to form a multi-temperature-range dielectric spectrum sequence.

[0027] Further, the step S5 comprises:

[0028] S51: extracting comprehensive dielectric characteristic parameters from the dielectric data of the to-be-tested asphalt sample;

[0029] S52: inputting the comprehensive dielectric characteristic parameters into the aging degree prediction model to obtain an aging degree estimation value;

[0030] S53: calculating a confidence interval of the prediction result;

[0031] S54: generating an aging degree evaluation report.

[0032] The present application provides another technical solution: an asphalt aging degree nondestructive testing system for realizing the above method, comprising:

[0033] A database construction module is configured to construct a multi-temperature-range standard dielectric spectrum-aging degree database;

[0034] A feature extraction module is connected to the database construction module and configured to extract comprehensive dielectric characteristic parameters sensitive to aging based on the database;

[0035] A model construction module is connected to the feature extraction module and configured to construct an aging degree prediction model based on the comprehensive dielectric characteristic parameters;

[0036] A nondestructive testing module is configured to perform nondestructive multi-temperature-range dielectric testing on a to-be-tested asphalt sample to obtain dielectric data of the to-be-tested asphalt sample;

[0037] The aging degree calculation module is connected to the nondestructive testing module and the model construction module, and is configured to input the dielectric data into the aging degree prediction model to calculate the aging degree of the asphalt sample to be tested.

[0038] The present application solves the defects in the background art and has the following advantages:

[0039] By using multi-temperature domain standard dielectric spectrum measurement and combining with the extraction of fixed-point and evolution dielectric characteristic parameters, the present application can comprehensively capture the complete dielectric response behavior of the asphalt material under different aging states with temperature changes, wherein the evolution parameters such as apparent activation energy directly quantify the change of the potential barrier to be overcome by the thermal motion of molecular segments or dipoles, which is closely related to the microscopic mechanism of the increase of intermolecular force and the change of the proportion of polar components caused by aging. This multi-dimensional feature extraction makes the evaluation of the aging degree no longer rely on the instantaneous measurement at a single temperature point, thereby significantly reducing the measurement error caused by temperature fluctuations and improving the stability of the detection. Compared with the method in the prior art which is based on only the single-temperature dielectric characteristic, the present application realizes more accurate and robust characterization of the entire life cycle of the asphalt from the initial aging to the near failure by wide-temperature domain scanning and feature evolution analysis, especially under complex environmental temperature, which can still maintain high precision and avoid the prediction deviation caused by temperature sensitivity of the traditional method.

[0040] By introducing a data-driven model such as artificial neural network to construct the aging degree prediction model, the present application can automatically learn the complex nonlinear mapping relationship between the comprehensive dielectric characteristic parameters and the aging degree, and the model uses multiple hidden layers and ReLU activation function to effectively capture the high-order patterns in the features, avoiding the simplification assumptions and limitations of the traditional physical model in describing the aging behavior of asphalt. This modeling method not only improves the prediction accuracy, but also enhances the generalization ability of the model to unknown data, so that the prediction result is closer to the benchmark value of the destructive test. Compared with the traditional method based on empirical formula or simple regression, the machine learning model of the present application can mine the internal rules from a large amount of multi-temperature domain data, thereby providing more consistent and reliable aging degree quantification output in a wide aging gradient range, reducing the uncertainty of human intervention and model parameter adjustment.

[0041] The present application further realizes effective decoupling of temperature effect and aging effect by combining multi-temperature domain dielectric characteristics with a machine learning prediction model. The multi-temperature domain characteristics provide a dynamic map of the changes of dielectric parameters with temperature, and the machine learning model learns the correlation between these changes and the aging degree, thereby automatically separating temperature interference in the prediction process. This combination enables the model to adapt to a wide temperature range of environmental conditions in practical applications, maintains the robustness of the prediction results, and avoids the problem of a sharp decline in performance of a single-temperature model outside the calibration point. Therefore, the present application not only improves the single-index precision of aging detection, but also ensures the consistency and practicality of evaluation in the actual variable-temperature scene, providing more reliable technical support for asphalt pavement maintenance decisions. BRIEF DESCRIPTION OF DRAWINGS

[0042] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0043] Figure 1 It is a flowchart of a nondestructive testing method for the aging degree of asphalt.

[0044] Figure 2 It is an architectural diagram of a nondestructive testing system for the aging degree of asphalt. DETAILED DESCRIPTION

[0045] The technical solutions in the embodiments of the present application will be described clearly and completely in the following with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0046] In the following description, many specific details are set forth in order to provide a thorough understanding of the present application, but the present application can also be implemented in other ways different from those described herein, therefore, the scope of protection of the present application is not limited by the specific embodiments disclosed below.

[0047] In the following description, all raw materials can come from commercial or be prepared by conventional methods in the art, unless otherwise specified. Among them, the base asphalt is Sinopec Donghai AH-70 base asphalt.

[0048] Exemplary method:

[0049] As Figure 1As shown, a non-destructive detection method for the aging degree of asphalt includes the following steps:

[0050] S1: Construct a multi-temperature field standard dielectric spectrum-aging degree database;

[0051] S2: Based on the multi-temperature field standard dielectric spectrum-aging degree database, extract the comprehensive dielectric characteristic parameters sensitive to aging;

[0052] S3: Based on the comprehensive dielectric characteristic parameters, construct an aging degree prediction model;

[0053] S4: Non-destructive multi-temperature field dielectric detection is performed on the asphalt sample to be tested to obtain dielectric data of the asphalt sample to be tested;

[0054] S5: Input the dielectric data of the asphalt sample to be tested into the aging degree prediction model to calculate the aging degree of the asphalt sample to be tested.

[0055] In the following, each step will be described in detail.

[0056] Step S1 captures the dielectric behavior of asphalt materials under different aging conditions and different temperature conditions, constructs a multi-temperature field standard dielectric spectrum-aging degree database, and includes the following steps:

[0057] S11: Prepare standard asphalt samples.

[0058] The standard asphalt samples are selected from commonly used asphalt base materials in engineering, such as road petroleum asphalt with a penetration grade of 70-100, or SBS, SBR, EVA polymer modified asphalt. When selecting the base asphalt, record the initial chemical components including the proportions of saturates, aromatics, resins, and asphaltenes, as well as the physical properties including penetration, softening point, and ductility. These initial properties constitute the baseline of dielectric performance.

[0059] S12: Perform artificial accelerated aging treatment on the standard asphalt samples to obtain samples with gradient aging degrees.

[0060] The artificial accelerated aging treatment simulates the oxidation and degradation of asphalt during actual service. First, short-term thermal oxidation treatment is performed using a thin film oven test at a temperature of 163°C for 85 minutes, and 4L / min of air is introduced to cause preliminary aging of the asphalt film. Subsequently, long-term oxidation hardening treatment is performed using a pressure aging vessel test at a temperature range of 90°C to 110°C with an air pressure of 2.1MPa for 20 hours to 80 hours on the short-term aged samples. By controlling the pressure aging vessel test time, discrete aging levels from trace aging to severe aging are obtained, covering the complete life cycle of asphalt from initial aging to near failure. The standard asphalt samples after aging treatment are aged asphalt samples.

[0061] S13: Quantitatively characterize the real aging degree of each aged asphalt sample by destructive tests.

[0062] Carbonyl index is measured by Fourier transform infrared spectroscopy. Carbonyl index is calculated as the ratio of the carbonyl peak area at wave number 1700 cm -1 to the methylene peak area at wave number 1460 cm -1 . The absorption peak at 1700 cm -1 is originated from carbonyl stretching vibration, and its intensity increases with the aging degree. The absorption peak at 1460 cm -1 is originated from methylene bending vibration, and it is relatively stable during aging. Carbonyl index sensitively reflects the oxidation aging degree of asphalt.

[0063] Or measure the complex shear modulus and phase angle of asphalt samples at a specific temperature and frequency. Dynamic shear rheometer test is conducted at 25 °C temperature and 10 rad / s frequency, and the value is calculated. The value represents the contribution of viscous component of asphalt. Aging leads to the increase of asphalt hardness and elasticity, the increase of the value, the decrease of the value, and the increase of the value with the deepening of aging degree. These quantitative indexes provided by destructive tests serve as real aging labels.

[0064] S14: Perform multi-temperature domain dielectric spectroscopy on each aged asphalt sample.

[0065] Aged asphalt samples are prepared into disc-shaped or rectangular thin slices with thickness of 2-5 mm and diameter of 20-30 mm by melting and pouring followed by cooling or hot-pressing molding. The surface is flat and the thickness is uniform.

[0066] Dielectric spectroscopy is measured at multiple gradient temperature points. The temperature range is set to -20 °C to 60 °C, and the temperature step is 5 °C, with a total of 17 discrete temperature points. At each temperature point, the sample is stabilized for 15 minutes. At each stabilized temperature, a sinusoidal alternating current electric field with a frequency range of 10 Hz to 10 MHz and a voltage of 1 VRMS is applied. The complex impedance is measured, and the complex dielectric constant is calculated through the sample geometry parameters and the unloaded capacitance. The complex dielectric constant contains the real part of the dielectric constant and the imaginary part of the dielectric loss factor. The real part of the dielectric constant reflects the energy storage capacity of the material under the action of an electric field, and is related to the degree of polarization. The imaginary part of the dielectric loss factor Reflects energy dissipation, associated with dipole relaxation loss and direct current conduction loss. Wide frequency range covers multiple typical dielectric relaxation processes in asphalt materials, including dipole orientation relaxation and ion conduction and interfacial polarization effects.

[0067] In dielectric spectrum measurement, the complex dielectric constant calculation formula is: ; wherein, is the complex dielectric constant (dimensionless); is the imaginary unit; is the angular frequency, with the unit of rad / s; is the empty capacitance, with the unit of F; is the complex impedance, with the unit of Ω; through the relationship between dielectric constant and impedance, the measured electrical signal is converted into material dielectric properties according to the principle of electromagnetism, which is used to quantify the polarization response of asphalt.

[0068] S15: Constructing multi-temperature field standard dielectric spectrum-aging degree database.

[0069] The dielectric constant and dielectric loss factor spectrum data of each aged asphalt sample at each gradient temperature point are associated with the aging degree index quantified by destructive test of the corresponding sample, forming a structured database. Each record of the database contains the original type of asphalt, initial aging state, specific aging degree quantification index after artificial accelerated aging, and complex dielectric constant spectrum data at multiple discrete temperature points and wide frequency range.

[0070] The aging process of asphalt leads to significant changes in the content of internal polar components, molecular size distribution, and intermolecular interaction force, which affect the dielectric properties at a single temperature and change the response sensitivity of dielectric relaxation process to temperature changes. Multi-temperature field dielectric spectrum measurement comprehensively captures the change map of dielectric relaxation activation energy and relaxation time distribution caused by aging.

[0071] Step S2 identifies and quantifies the key dielectric characteristics that can stably represent the aging degree of asphalt from the multi-temperature field standard dielectric spectrum-aging degree database constructed in step S1. The comprehensive dielectric characteristic parameters include fixed-point dielectric characteristic parameters and evolving dielectric characteristic parameters, which together represent the multi-dimensional information of the aging state of asphalt.

[0072] The method for extracting aging-sensitive comprehensive dielectric characteristic parameters in step S2 includes the following steps:

[0073] S21: Extracting fixed-point dielectric characteristic parameters.

[0074] Fixed-point dielectric characteristic parameters refer to numerical values ​​characterizing dielectric properties obtained under specific temperature and frequency conditions or through analysis of dielectric spectrum diagrams at a single temperature. The dielectric spectrum data of each aged asphalt sample in the multi-temperature domain standard dielectric spectrum-aging degree database were processed to extract the following fixed-point dielectric characteristic parameters.

[0075] First, the real part of the dielectric constant and the imaginary part of the dielectric loss factor are extracted at characteristic frequency points under preset characteristic temperature points. The characteristic temperature points are selected from a temperature range of -10℃ to 60℃, such as 0℃, 25℃, and 50℃, and the characteristic frequency points are selected from a frequency range of 100Hz to 1MHz, such as 1kHz, 10kHz, and 100kHz. These frequency points are associated with specific dielectric relaxation modes in the asphalt material, specifically α-relaxation, β-relaxation, or interfacial polarization effects at different temperatures. These relaxation modes are sensitive to the molecular structure and motion state of the asphalt.

[0076] Secondly, extract the imaginary part of the dielectric loss factor at each gradient temperature point. Peak frequency in the spectrum and peak amplitude Peak frequency It is the imaginary part of the dielectric loss factor. Logarithmic coordinate graph of frequency The frequency at which the maximum value is reached. Peak frequency. The peak frequency is closely related to the average relaxation time of the asphalt molecular chain segments. Shifts towards lower frequencies indicate impaired molecular motion, reduced free volume, or increased system viscosity—typical signs of asphalt aging. Peak amplitude The intensity of this relaxation process is also related to the polar component content and dipole density.

[0077] Third, Cole-Cole parameters are extracted by performing Cole-Cole plot analysis on the dielectric spectrum data. The Cole-Cole plot is derived by representing the imaginary part of the dielectric loss factor. The vertical axis represents the real part of the dielectric constant. The curve plotted as the x-axis is an arc. The average relaxation time is extracted from the curve using a nonlinear circular arc fitting method, such as fitting the Cole-Cole equation based on least squares. , scaling parameter of relaxation time distribution High-frequency dielectric constant and low-frequency dielectric constant Mean relaxation time Characterizing the time required for a molecule or chain segment to reach equilibrium, broadening parameters reflecting the non-uniformity of the relaxation process or the distribution width of the relaxation time, ranges from 0 to 1. Aging of bitumen leads to the increase of intermolecular forces and the increase of molecular size, which affect these parameters, for example, increase the average relaxation time , decrease the spread parameter or present non-monotonic changes.

[0078] S22: Extracting evolution dielectric characteristic parameters.

[0079] Evolution dielectric characteristic parameters are indicators of regular changes in dielectric characteristic parameters at a specified point as temperature changes. These parameters can capture the dynamic response characteristics of bitumen materials at different temperatures, reflecting changes in the activation energy of molecular thermal motion. Further processing of the fixed-point dielectric characteristic parameters extracted in S21 extracts the following evolution dielectric characteristic parameters.

[0080] First, analyze the temperature dependence of the dielectric loss peak frequency . The values of measured at different temperature points for each aged bitumen sample are plotted as an Arrhenius plot, i.e., ln( ) vs. 1 / T, where T is the absolute temperature unit K. By linearly fitting the Arrhenius plot, the apparent activation energy is calculated. The apparent activation energy quantifies the energy required for a molecular segment or dipole in bitumen to overcome the potential barrier for thermal motion, with units of kJ / mol. Aging of bitumen leads to the increase of intermolecular forces, the increase of molecular size, and the decrease of free volume, which increases the potential barrier of molecular motion and increases the apparent activation energy value.

[0081] In analyzing the temperature dependence of , the relationship between relaxation rate and temperature is described according to the principle of thermal activation process, which is used to calculate the apparent activation energy from the temperature dependence of as an evolution characteristic parameter representing the degree of aging. The Arrhenius equation is expressed as: or in linear form: ; where is the peak frequency, with units of Hz; is the pre-exponential factor, with units of Hz; is the apparent activation energy, with units of J / mol; is the gas constant, approximately 8.314 J / (mol·K); is the absolute temperature, with units of K.

[0082] As an alternative, a Vogel-Fulcher-Tammann plot can be drawn, i.e., Plot 1 / (T-T0) vs. frequency, where T0 is the Vogel temperature, to obtain the VFT parameters A, B and T0 by non-linear fitting. The VFT model provides a more accurate description of the molecular dynamics behavior in the glass transition region.

[0083] Secondly, calculate the slope of the real part of the dielectric constant or the imaginary part of the dielectric loss factor vs. temperature at a specific frequency. For example, calculate the linear regression slope of the real part of the dielectric constant vs. temperature at a frequency of 10 kHz. The slope reflects the sensitivity of the polarization ability of the material at a specific frequency to the change in temperature, and aging will change this sensitivity.

[0084] Thirdly, analyze the trend of Cole-Cole parameters vs. temperature. For example, calculate the average relaxation time and the spread parameter vs. temperature. The derivative of these parameters vs. temperature (the slope of the change with temperature) quantifies the temperature sensitivity of these parameters. These rates of change reflect the impact of aging on the thermal responsiveness of the microstructure of asphalt.

[0085] The mechanism of extracting evolving dielectric characteristic parameters lies in the fact that the impact of asphalt aging on different dielectric polarization mechanisms exhibits different sensitivities in different temperature intervals and frequency ranges. Combining fixed-point dielectric characteristic parameters and evolving dielectric characteristic parameters can more comprehensively quantify the changes in the internal microstructure and chemical composition of asphalt materials due to aging, highlighting the physical indicator of molecular thermal motion activation energy. These parameters together form a feature vector that is highly sensitive to the aging state and robust to temperature fluctuations.

[0086] Step S3 establishes a quantitative mapping relationship from the comprehensive dielectric characteristic parameters to the aging degree of asphalt, realizing the prediction of the aging degree based on dielectric data. Step S3 takes the comprehensive dielectric characteristic parameters extracted in step S2 and the aging degree index quantified in step S1 as inputs, and constructs a prediction model through a machine learning algorithm, including the following steps:

[0087] S31: Prepare the model training data set.

[0088] Take the fixed-point dielectric characteristic parameters and the evolving dielectric characteristic parameters extracted in step S2 as independent variables, i.e. input features. Take the aging degree index quantified through destructive testing in step S1 as the dependent variable, i.e. the output label. The aging degree index includes the carbonyl index obtained by Fourier transform infrared spectroscopy analysis or the value measured by a dynamic shear rheometer. The comprehensive dielectric characteristic parameters of each aged asphalt sample and the corresponding aging degree index form a data sample, and all data samples constitute the model training data set.

[0089] S32: Configure the aging degree prediction model structure.

[0090] The aging degree prediction model adopts a data-driven machine learning algorithm, specifically an artificial neural network model. The artificial neural network model architecture includes an input layer, multiple hidden layers, and an output layer. The number of neurons in the input layer corresponds to the total number of comprehensive dielectric feature parameters, for example, if 20 feature parameters are extracted, 20 neurons are set in the input layer. The hidden layer is set to 2 to 4 layers, and the number of neurons in each layer is adjusted between 64 and 256, specifically configured as 128 neurons in the first hidden layer, 64 neurons in the second hidden layer, and 32 neurons in the third hidden layer. The hidden layer uses a ReLU activation function, with a function expression of to enhance the non-linear mapping ability of the model. The output layer contains one neuron for predicting the numerical value of the aging degree, and the output layer does not use an activation function to adapt to the continuous output of the regression task. Through this multi-layer non-linear transformation structure, the artificial neural network model can learn the complex mapping relationship between the comprehensive dielectric feature parameters and the aging degree.

[0091] S33: Divide the data set and set the training parameters.

[0092] The model training data set is divided into a training set, a validation set, and a test set. Random sampling is used for division, with a proportion of 70% training set, 15% validation set, and 15% test set. The training set is used for parameter optimization and learning of the model, the validation set is used to evaluate the model performance during training and optimize the hyperparameters, and the test set is used to evaluate the independent generalization ability of the trained model. The model training uses the Adam optimizer, which can adaptively adjust the learning rate, accelerate the convergence speed, and improve the training efficiency. The training process uses mean squared error as the loss function, which calculates the average of the square of the difference between the predicted value and the true value. The number of training iterations is set to 200 epochs, and the batch size is set to 32 samples.

[0093] S34: Train and optimize the aging degree prediction model.

[0094] The artificial neural network model is optimized through the training process. The training process uses the backpropagation algorithm to adjust the weights and biases of the neural network, so that the model can accurately predict the aging degree of asphalt from the input comprehensive dielectric feature parameters by minimizing the loss function. During the training process, after completing one epoch of training, the model performance is evaluated using the validation set, the loss function value on the validation set is monitored to prevent overfitting of the model. When the validation set loss function value does not decrease for multiple epochs in a row, the training process is terminated early, and the best model parameters are saved.

[0095] S35: Evaluate the model performance.

[0096] After the training, the performance of the artificial neural network model is comprehensively evaluated using the test set. The evaluation indicators include the coefficient of determination, the root mean square error, and the mean absolute error. The coefficient of determination measures the degree of explanation of the model to the variability of the data, and the closer the value is to 1, the better the fitting effect of the model. The root mean square error and the mean absolute error measure the deviation between the predicted value and the true value, and the smaller the value, the higher the prediction accuracy. The coefficient of determination value of a high-performance model is close to 1, and the root mean square error and the mean absolute error value are as small as possible, indicating that the prediction result has accuracy and stability. Through the test set evaluation, it is ensured that the model can still maintain high precision and stability when facing unseen data, providing reliable performance basis for the practical application of the model.

[0097] The artificial neural network model is based on machine learning algorithm, has strong non-linear mapping ability and pattern recognition ability, and can deeply mine the internal relationship between the complex law of dielectric response evolution with temperature in multiple temperature domains and the change of asphalt microstructure and chemical composition. Through training on multiple temperature domain data, the model learns the combined effect of aging and temperature on dielectric parameters, effectively decouples the temperature effect in practical application, and realizes the robust prediction of aging degree. This data-driven modeling method overcomes the limitations of traditional physical models in capturing complex material behavior, and provides more accurate and universal prediction ability.

[0098] Step S4 acquires the dielectric spectrum data of the asphalt sample to be tested at different temperatures as the input of the aging degree prediction model, and the whole process does not produce any damage to the sample. Step S4 takes the asphalt sample to be tested in the actual road surface or laboratory as input, obtains the dielectric spectrum sequence through multi-temperature domain sweep frequency measurement, including the following steps:

[0099] S41: Prepare the asphalt sample to be tested and clean the measurement surface.

[0100] The asphalt sample to be tested is a core sample directly obtained from the field asphalt pavement structure or a plate specimen, or a laboratory-prepared sample to be evaluated. The surface of the asphalt sample to be tested is cleaned to remove dust, oil stains or loose particles. Isopropyl alcohol or special cleaning agent is used to wipe the surface, ensuring thorough drying, so that the measurement electrode and the surface of the asphalt sample to be tested form a stable capacitive coupling, avoiding the introduction of measurement errors by pollutants or air gaps.

[0101] S42: Set the multi-temperature domain dielectric spectrum measurement conditions.

[0102] The measurement conditions are consistent with those of the multi-temperature domain standard dielectric spectrum-aging degree database established in step S1. The temperature control range is set to -20°C to 60°C, with a temperature step of 5°C, for a total of 17 discrete temperature points. At each temperature point, the stabilization time is set to 15 minutes. The frequency scan range is set to 10 Hz to 10 MHz, with a scan point number of 50-100 points / octave, and the applied voltage is 1 VRMS of sinusoidal alternating current field.

[0103] S43: Perform multi-temperature domain dielectric spectrum scan measurement.

[0104] The measurement electrode is tightly attached to the surface of the asphalt sample to be measured, ensuring that there is no air gap between the electrode and the sample. According to the preset temperature sequence, the temperature is gradually increased from -20°C to 60°C with a step of 5°C. After reaching stability at each temperature point, a frequency scan measurement is performed, continuously scanning from 10 Hz to 10 MHz, and recording the complex impedance data at each frequency point. The dielectric constant real part and dielectric loss factor imaginary part spectrum data are calculated from the complex impedance data and electrode geometry parameters.

[0105] S44: Record dielectric spectrum data.

[0106] The dielectric spectrum data obtained at each temperature point is recorded in real time, including frequency points, dielectric constant real part values, and dielectric loss factor imaginary part values, forming a multi-temperature domain dielectric spectrum sequence of the asphalt sample to be measured. This data sequence contains complete dielectric response information of the asphalt sample to be measured at 17 temperature points and in the frequency range of 10 Hz to 10 MHz.

[0107] Step S4 obtains a complete information map of the polarization response as a function of temperature by applying an external electric field excitation to the asphalt material without damaging the sample at different temperatures. This map includes the static dielectric properties of asphalt at a specific temperature, as well as the dynamic sensitivity of dielectric relaxation processes to temperature changes, thereby comprehensively and robustly reflecting the internal aging state.

[0108] Step S5 uses the aging degree prediction model constructed in step S3 to analyze the dielectric data of the asphalt sample to be measured obtained in step S4, and outputs a quantitative evaluation result of the aging degree. Step S5 takes the multi-temperature domain dielectric spectrum sequence obtained in step S4 as input, and obtains the final aging degree evaluation through feature extraction and model prediction, including the following steps:

[0109] S51: Extract the comprehensive dielectric characteristic parameters of the sample to be measured.

[0110] From the multi-temperature domain dielectric spectrum sequence of the to-be-tested asphalt sample measured in step S4, the same fixed-point dielectric characteristic parameters and evolution dielectric characteristic parameters are extracted according to the same rules and algorithms defined in step S2. The feature extraction process is strictly consistent with the database construction phase to ensure the consistency of the input features. The fixed-point dielectric characteristic parameters include the real part values and the imaginary part values of the dielectric constant at 1 kHz, 10 kHz and 100 kHz frequencies at 0°C, 25°C and 50°C temperatures, the peak frequency and the peak amplitude in the imaginary part spectrum of the dielectric loss factor at each temperature point, and the average relaxation time and the broadening parameter obtained by Cole-Cole diagram analysis. The evolution dielectric characteristic parameters include the apparent activation energy obtained by temperature dependence analysis of the dielectric loss peak frequency, the linear regression slope of the real part of the dielectric constant with respect to temperature at 10 kHz frequency, and the trend of the Cole-Cole parameters with respect to temperature.

[0111] S52: Calculate the aging degree estimate value using the prediction model.

[0112] The comprehensive dielectric characteristic parameters of the to-be-tested sample extracted in S51 are taken as inputs and substituted into the aging degree prediction model trained and verified in step S3. The prediction model processes the input features and outputs the aging degree estimate value of the to-be-tested asphalt sample. The aging degree estimate value can be a quantitative carbonyl index, a value, or an index corresponding to a specific aging level such as mild aging, moderate aging, and severe aging. The model output represents the current true aging state of the to-be-tested asphalt sample.

[0113] S53: Calculate the confidence interval of the prediction result.

[0114] Based on the performance of the prediction model on the test set, the confidence interval of the aging degree estimate value is calculated. A 95% confidence level is adopted to determine the fluctuation range of the estimate value through the distribution characteristics of the model prediction error. For example, when the predicted carbonyl index is 0.32, the confidence interval is [0.30, 0.34], indicating that there is a 95% chance that the true carbonyl index falls within this range. The confidence interval quantifies the uncertainty of the prediction result and provides a reliability reference for decision-making.

[0115] The confidence interval calculation formula is: ; wherein, is the confidence interval; is the point estimate of the prediction value; is the quantile of the standard normal distribution (1.96 corresponds to a 95% confidence level); is the standard error of the prediction value.

[0116] S54: Generate an aging degree assessment report.

[0117] The estimated aging degree and its confidence interval are arranged into a structured report. The report includes basic information of the sample to be tested, measurement conditions, a summary of the original dielectric data, extracted characteristic parameter values, predicted aging degree indicators and confidence intervals. The report format uses a standardized template to ensure the standardization and readability of the results.

[0118] Step S5 converts the multi-temperature domain dielectric response of the asphalt sample to be tested into a unified feature vector, which is input into the aging degree prediction model trained by big data. The aging degree prediction model automatically decouples the interference of temperature on dielectric parameters, and outputs the real aging degree based on the essential changes of dielectric relaxation behavior of asphalt materials at different temperatures. This makes the method realize quantitative, robust and non-destructive evaluation of the aging state of asphalt, effectively overcoming the limitations of traditional non-destructive testing methods in complex variable temperature environments.

[0119] Exemplary system:

[0120] As shown in Figure 2 An asphalt aging degree non-destructive testing system for performing and implementing the above asphalt aging degree non-destructive testing method, comprising a database construction module, a feature extraction module, a model construction module, a non-destructive testing module and an aging degree calculation module connected in sequence.

[0121] The database construction module is used to construct a multi-temperature domain standard dielectric spectrum-aging degree database, comprising:

[0122] The sample preparation unit is configured to prepare selected standard asphalt samples;

[0123] The aging treatment unit is configured to perform gradient time artificial accelerated aging treatment on the standard asphalt samples to obtain a series of aged asphalt samples;

[0124] The aging degree calibration unit is configured to quantitatively measure the real aging degree of each aged asphalt sample by destructive test;

[0125] The dielectric spectrum measurement unit is configured to measure the wideband dielectric spectrum of each aged asphalt sample at multiple gradient temperature points;

[0126] The data integration unit is configured to associate and integrate the dielectric spectrum data of each aged asphalt sample with the corresponding real aging degree to form a structured database.

[0127] The output of the sample preparation unit is connected to the input of the aging treatment unit; the output of the aging treatment unit is connected to the input of the aging degree calibration unit and the dielectric spectrum measurement unit; the outputs of the aging degree calibration unit and the dielectric spectrum measurement unit are jointly connected to the input of the data integration unit

[0128] The feature extraction module is connected to the data integration unit of the database construction module, and is configured to extract aging-sensitive comprehensive dielectric feature parameters based on the database, including:

[0129] The fixed-point feature extraction unit is configured to extract dielectric parameter values, dielectric loss peak frequency and amplitude, and relaxation parameters obtained through Cole-Cole diagram analysis at a specific temperature and frequency point from the database;

[0130] The evolution feature extraction unit is configured to analyze the variation law of the fixed-point feature parameters with temperature, and calculate evolution feature parameters such as apparent activation energy and temperature slope.

[0131] The input of the fixed-point feature extraction unit is connected to the output of the data integration unit of the database construction module, and the output of the fixed-point feature extraction unit is connected to the input of the evolution feature extraction unit.

[0132] The model construction module is connected to the output of the feature extraction module, and is configured to construct an aging degree prediction model based on the comprehensive dielectric feature parameters, including:

[0133] The data set preparation unit is configured to use the comprehensive dielectric feature parameters output by the feature extraction module as independent variables, and use the real aging degree provided by the database construction module as dependent variables, to form a model training data set;

[0134] The model training unit is configured to use an artificial neural network algorithm to train the model training data set, and optimize network parameters to establish a mapping relationship between feature parameters and aging degree;

[0135] The model verification unit is configured to use a verification set to monitor the training process to prevent overfitting, and use a test set to evaluate the prediction performance indicators of the trained model.

[0136] The input of the data set preparation unit is connected to the output of the feature extraction module and the output of the aging degree calibration unit in the database construction module, respectively; the output of the data set preparation unit is connected to the input of the model training unit; the model training unit and the model verification unit interact during the training process.

[0137] The nondestructive testing module is configured to perform nondestructive multi-temperature domain dielectric testing on the asphalt sample to be tested, and obtain dielectric data, including:

[0138] The sample processing unit is configured to prepare and clean the measurement surface of the asphalt sample to be tested;

[0139] The measurement control unit is configured to set multi-temperature domain sweep measurement conditions consistent with the database construction stage;

[0140] a signal acquisition unit configured to control the measurement electrode to perform frequency scanning on the sample to be measured under a preset temperature sequence and to acquire a complex impedance signal;

[0141] a data calculation unit configured to calculate a dielectric spectrum sequence of the sample to be measured according to the acquired complex impedance signal and electrode parameters.

[0142] The output of the sample processing unit is connected to the input of the measurement control unit; the measurement control unit controls the operation of the signal acquisition unit; the output of the signal acquisition unit is connected to the input of the data calculation unit.

[0143] The aging degree calculation module is connected to the data calculation unit of the nondestructive testing module and the model training unit of the model construction module, and is used to calculate and output the aging degree evaluation result of the sample to be measured, including:

[0144] a feature parameter calculation unit configured to extract the same comprehensive dielectric feature parameters from the dielectric spectrum sequence of the sample to be measured according to the same rules as the feature extraction module;

[0145] an aging degree prediction unit configured to input the extracted feature parameters into the trained aging degree prediction model to calculate an aging degree estimate value;

[0146] a confidence assessment unit configured to calculate the confidence interval of the prediction result based on the model performance;

[0147] a report generation unit configured to integrate the prediction result, the confidence interval and related information to generate a standardized evaluation report.

[0148] The input of the feature parameter calculation unit is connected to the output of the data calculation unit of the nondestructive testing module; the output of the feature parameter calculation unit is connected to the input of the aging degree prediction unit; the output of the aging degree prediction unit is connected to the input of the confidence assessment unit; the outputs of the aging degree prediction unit and the confidence assessment unit are jointly connected to the input of the report generation unit. The aging degree prediction unit calls the prediction model parameters trained and saved in the model construction module.

[0149] Embodiment 1:

[0150] The embodiment provides a nondestructive testing method for the aging degree of asphalt, including the following steps:

[0151] S1: Construct a multi-temperature-domain standard dielectric spectrum-aging degree database:

[0152] S11: Prepare standard asphalt samples, and select road petroleum asphalt with a penetration grade of 70 as the matrix asphalt.

[0153] S12: Perform artificial accelerated aging treatment on the standard asphalt samples;

[0154] Firstly, thin film oven test (163°C, 85 minutes, air flow 4 L / min) was performed on all base asphalt samples;

[0155] Subsequently, pressure aging vessel test (temperature 100°C, air pressure 2.1 MPa) was performed, with aging time gradient set as 20h, 30h, 40h, 50h, 60h, 70h, 80h, 90h, 100h, totally 9 gradients;

[0156] 25 samples were prepared independently for each aging gradient, totally 225 aging asphalt samples.

[0157] S13: Quantitatively characterize the real aging degree of each aging asphalt sample through destructive test;

[0158] Fourier transform infrared spectrometer was used to measure each sample, and carbonyl index was calculated;

[0159] Dynamic shear rheometer was used to measure complex shear modulus and phase angle at 25°C, 10 rad / s frequency, and value was calculated.

[0160] S14: Multi-temperature domain dielectric spectrum measurement was performed on each aging asphalt sample;

[0161] The sample was made into a disc-shaped sheet with a thickness of 3mm and a diameter of 25mm;

[0162] Dielectric spectrum measurement was performed at -20°C, -15°C, -10°C, -5°C, 0°C, 5°C, 10°C, 15°C, 20°C, 25°C, 30°C, 35°C, 40°C, 45°C, 50°C, 55°C, 60°C, totally 17 temperature points, with 15 minutes of stabilization at each temperature point;

[0163] At each stable temperature, impedance analyzer was used to apply a sinusoidal alternating current field with a frequency range of 10Hz to 10MHz and a voltage of 1VRMS to perform scanning, measure complex impedance, and calculate complex dielectric constant (real part of dielectric constant and imaginary part of dielectric loss factor).

[0164] S15: Construct multi-temperature domain standard dielectric spectrum-aging degree database;

[0165] All dielectric spectrum data (frequency, temperature, real part of dielectric constant, imaginary part of dielectric loss factor) of each sample were associated with corresponding carbonyl index and value, forming a structured database.

[0166] S2: Based on the multi-temperature domain standard dielectric spectrum-aging degree database, extract comprehensive dielectric feature parameters sensitive to aging.

[0167] S21: Extracting the fixed-point dielectric characteristic parameters;

[0168] From the dielectric spectrum data of each sample in the database, the following parameters are extracted:

[0169] The real part of dielectric constant and the imaginary part of dielectric loss factor at 1 kHz, 10 kHz, and 100 kHz frequencies at 0°C, 25°C, and 50°C temperatures;

[0170] In the imaginary part of the dielectric loss factor spectrum at each temperature point, identify the peak frequency and peak amplitude;

[0171] Draw Cole-Cole plots (imaginary part of dielectric loss factor vs. real part of dielectric constant) for the dielectric data at each temperature point, and extract the average relaxation time and the broadening parameter of the relaxation time distribution by nonlinear arc fitting method.

[0172] S22: Extracting evolution dielectric characteristic parameters.

[0173] Further process the fixed-point dielectric characteristic parameters extracted in S21, and extract the following evolution dielectric characteristic parameters:

[0174] For each sample, draw an Arrhenius plot (ln(peak frequency) vs. 1 / T) of the peak frequency at different temperatures, and perform linear fitting to calculate the apparent activation energy;

[0175] Calculate the linear regression slope of the real part of the dielectric constant with respect to temperature at 10 kHz frequency;

[0176] Analyze the trend of the average relaxation time and the broadening parameter with respect to temperature.

[0177] S3: Constructing an aging degree prediction model based on comprehensive dielectric characteristic parameters.

[0178] S31: Preparing the model training data set;

[0179] Taking the fixed-point dielectric characteristic parameters and the evolution dielectric characteristic parameters as independent variables, and taking the carbonyl index and the value as dependent variables, the data of each sample constitutes a data sample.

[0180] S32: Configuring the structure of the aging degree prediction model;

[0181] The aging degree prediction model adopts an artificial neural network model; the number of input layer neurons corresponds to the total number of comprehensive dielectric characteristic parameters; the hidden layer is set to three layers: 128 neurons in the first hidden layer, 64 neurons in the second hidden layer, and 32 neurons in the third hidden layer; the hidden layer uses a linear rectifier function as the activation function; the output layer contains one neuron and does not use an activation function.

[0182] S33: Divide the data set and set the training parameters;

[0183] The model training data set is randomly divided into 70% training set, 15% validation set, and 15% test set. The adaptive moment estimation optimizer is used for model training. The mean square error is used as the loss function in the training process. The number of training iterations is set to 200 epochs, and the batch size is set to 32 samples.

[0184] S34: Train and optimize the aging degree prediction model;

[0185] The artificial neural network model is optimized through the training process. The back propagation algorithm is used in the training process. During the training process, the model performance is evaluated using the validation set after completing one epoch of training. When the validation set loss function value does not decrease for 10 consecutive epochs, the training process is terminated in advance, and the best model parameters are saved.

[0186] S35: Evaluate the model performance.

[0187] After training, the performance of the artificial neural network model is evaluated using the test set. The evaluation indicators include the coefficient of determination, the root mean square error, and the mean absolute error.

[0188] S4: Perform non-destructive multi-temperature domain dielectric detection on the asphalt sample to be tested to obtain the dielectric data of the asphalt sample to be tested.

[0189] S41: Prepare the asphalt sample to be tested and clean the measurement surface;

[0190] The asphalt sample to be tested is a laboratory-prepared sample to be evaluated. The matrix asphalt is the same as the standard sample and has also undergone 20h to 100h pressure aging vessel test processing. Three independent samples are prepared for each aging gradient. Isopropyl alcohol is used to wipe the surface of the asphalt sample to be tested and thoroughly dry.

[0191] S42: Set the multi-temperature domain dielectric spectrum measurement conditions;

[0192] The measurement temperature range is set to -20°C to 60°C, with a temperature step of 5°C, a total of 17 discrete temperature points. At each temperature point, the stabilization time is set to 15 minutes. The frequency scanning range is set to 10Hz to 10MHz, and the applied voltage is 1VRMS sinusoidal alternating current field.

[0193] S43: Perform multi-temperature domain dielectric spectrum scanning measurement;

[0194] The measurement electrode is closely attached to the surface of the asphalt sample to be measured, and the temperature sequence is preset to start from -20°C and gradually increase to 60°C at a step of 5°C; after reaching stability at each temperature point, frequency scanning measurement is performed, and the complex impedance data at each frequency point are recorded to calculate the frequency spectrum data of the real part of the dielectric constant and the imaginary part of the dielectric loss factor.

[0195] S44: Record the dielectric spectrum data;

[0196] The dielectric spectrum data obtained at each temperature point is recorded in real time to form a multi-temperature domain dielectric spectrum sequence of the asphalt sample to be measured; the measurement results of 3 independent samples of the same aging gradient are averaged to obtain the final dielectric data of the sample to be measured.

[0197] S5: Input the dielectric data of the asphalt sample to be measured into the aging degree prediction model to calculate the aging degree of the asphalt sample to be measured.

[0198] S51: Extract the comprehensive dielectric characteristic parameters of the sample to be measured;

[0199] From the average multi-temperature domain dielectric spectrum sequence of the asphalt sample to be measured obtained in S44, the same fixed-point dielectric characteristic parameters and evolution dielectric characteristic parameters are extracted according to the same rules and algorithms defined in S2.

[0200] S52: Calculate the aging degree estimate value using the prediction model;

[0201] The comprehensive dielectric characteristic parameters of the sample to be measured extracted in S51 are input into the aging degree prediction model trained in S3. The model outputs the aging degree estimate value of the asphalt sample to be measured.

[0202] S53: Calculate the confidence interval of the prediction result.

[0203] Based on the performance of the prediction model on the test set, the 95% confidence interval of the aging degree estimate value is calculated.

[0204] S54: Generate an aging degree evaluation report.

[0205] The predicted aging degree estimate value and its confidence interval are arranged into a structured report.

[0206] Comparative Example 1:

[0207] This comparative example provides a method for non-destructive detection of the aging degree of asphalt different from Example 1, comprising the following steps:

[0208] S1: Construct a multi-temperature domain standard dielectric spectrum-aging degree database:

[0209] S11: Prepare standard asphalt samples, and select road petroleum asphalt with a penetration grade of 70 as the base asphalt.

[0210] S12: Artificially accelerated aging treatment is performed on the standard asphalt sample;

[0211] First, thin film oven test (163°C, 85 minutes, air flow 4 L / min) is performed on all matrix asphalt samples;

[0212] Subsequently, pressure aging vessel test (temperature 100°C, air pressure 2.1 MPa) is performed, and aging time gradient is set to 20h, 30h, 40h, 50h, 60h, 70h, 80h, 90h, 100h, a total of 9 gradients;

[0213] 25 samples are independently prepared for each aging gradient, and a total of 225 aged asphalt samples are obtained.

[0214] S13: The real aging degree of each aged asphalt sample is quantitatively characterized by destructive test;

[0215] Fourier transform infrared spectrometer is used to measure each sample, and carbonyl index is calculated;

[0216] Dynamic shear rheometer is used to measure complex shear modulus and phase angle at 25°C and 10 rad / s frequency, and value is calculated.

[0217] S14: Multi-temperature domain dielectric spectrum measurement is performed on each aged asphalt sample;

[0218] The sample is made into a disc-shaped sheet with a thickness of 3mm and a diameter of 25mm;

[0219] Dielectric spectrum measurement is performed at 17 temperature points of -20°C, -15°C, -10°C, -5°C, 0°C, 5°C, 10°C, 15°C, 20°C, 25°C, 30°C, 35°C, 40°C, 45°C, 50°C, 55°C, 60°C, and each temperature point is stable for 15 minutes; after the road petroleum asphalt with penetration grade of 70 is sequentially subjected to thin film oven test (163°C, 85 minutes, air flow 4 L / min) and pressure aging vessel test (temperature 100°C, air pressure 2.1 MPa, aging time gradient is set to 20h, 30h, 40h, 50h, 60h, 70h, 80h, 90h, 100h, a total of 9 gradients), carbonyl index and

[0220] At each stable temperature, impedance analyzer is used to apply a sinusoidal alternating current field with a frequency range of 10Hz to 10MHz and a voltage of 1VRMS to scan, measure complex impedance, and calculate complex dielectric constant (real part of dielectric constant and imaginary part of dielectric loss factor).

[0221] S15: Constructing multi-temperature domain standard dielectric spectrum-aging degree database;

[0222] Correlating all dielectric spectrum data (frequency, temperature, real part of dielectric constant, imaginary part of dielectric loss factor) of each sample with corresponding carbonyl index and value, forming a structured database.

[0223] S2: Extracting fixed-point dielectric characteristic parameters based on multi-temperature domain standard dielectric spectrum-aging degree database.

[0224] From the dielectric spectrum data of each sample in the database, the following parameters are extracted:

[0225] Only at 25℃ temperature, the real part of dielectric constant and the imaginary part of dielectric loss factor at 1kHz, 10kHz, 100kHz frequencies;

[0226] Only at 25℃ temperature, the peak frequency and peak amplitude in the imaginary part of dielectric loss factor spectrum are identified;

[0227] At 25℃ temperature, Cole-Cole plot (imaginary part of dielectric loss factor vs. real part of dielectric constant) is drawn for the dielectric data, and the average relaxation time and the broadening parameter of relaxation time distribution are extracted by nonlinear circular arc fitting method.

[0228] S3: Constructing aging degree prediction model based on comprehensive dielectric characteristic parameters.

[0229] S31: Preparing model training data set;

[0230] Taking fixed-point dielectric characteristic parameters as independent variables and carbonyl index and value as dependent variables, the data of each sample constitutes a data sample.

[0231] S32: Configuring aging degree prediction model structure;

[0232] The aging degree prediction model adopts artificial neural network model; the number of input layer neurons corresponds to the total number of fixed-point dielectric characteristic parameters; the hidden layer is set to three layers: 128 neurons in the first hidden layer, 64 neurons in the second hidden layer, and 32 neurons in the third hidden layer; the hidden layer uses linear rectification function as the activation function; the output layer contains one neuron and does not use activation function.

[0233] S33: Dividing data set and setting training parameters;

[0234] The model training dataset is randomly divided into 70% training set, 15% validation set and 15% test set; the adaptive matrix estimation optimizer is used for model training; the mean square error is used as the loss function in the training process; the number of training iterations is set to 200 epochs, and the batch size is set to 32 samples.

[0235] S34: training and optimizing the aging degree prediction model;

[0236] The artificial neural network model is optimized through the training process; the back propagation algorithm is used in the training process; during the training process, after completing one epoch of training, the model performance is evaluated using the validation set; when the validation set loss function value does not decrease for 10 consecutive epochs, the training process is terminated in advance, and the best model parameters are saved.

[0237] S35: evaluating the model performance.

[0238] After training, the performance of the artificial neural network model is evaluated using the test set; the evaluation indicators include the coefficient of determination, the root mean square error and the mean absolute error.

[0239] S4: performing non-destructive multi-temperature domain dielectric detection on the asphalt sample to be tested to obtain dielectric data of the asphalt sample to be tested.

[0240] S41: preparing the asphalt sample to be tested and cleaning the measurement surface;

[0241] The asphalt sample to be tested is a laboratory-prepared sample to be evaluated, the matrix asphalt is the same as the standard sample, and both are subjected to 20h to 100h pressure aging vessel test processing; 3 independent samples are prepared for each aging gradient; isopropyl alcohol is used to wipe the surface of the asphalt sample to be tested and completely dry.

[0242] S42: setting the multi-temperature domain dielectric spectrum measurement conditions;

[0243] The measurement temperature range is set to -20℃ to 60℃, the temperature step is 5℃, and there are 17 discrete temperature points; at each temperature point, the stabilization time is set to 15 minutes; the frequency scanning range is set to 10Hz to 10MHz, and the applied voltage is 1VRMS of sinusoidal alternating current field.

[0244] S43: performing multi-temperature domain dielectric spectrum scanning measurement;

[0245] The measurement electrode is tightly attached to the surface of the asphalt sample to be tested, and the temperature sequence is preset to start from -20℃ and gradually increase to 60℃ at a step of 5℃; after reaching stability at each temperature point, frequency scanning measurement is performed, and the complex impedance data at each frequency point are recorded to calculate the dielectric constant real part and dielectric loss factor imaginary part spectrum data.

[0246] S44: record dielectric spectrum data;

[0247] The dielectric spectrum data obtained at each temperature point is recorded in real time to form a multi-temperature domain dielectric spectrum sequence of the asphalt sample to be tested. The measurement results of three independent samples of the same aging gradient are averaged to obtain the final dielectric data of the sample to be tested.

[0248] S5: input the dielectric data of the asphalt sample to be tested into the aging degree prediction model to calculate the aging degree of the asphalt sample to be tested.

[0249] S51: extract the comprehensive dielectric characteristic parameters of the sample to be tested;

[0250] From the average multi-temperature domain dielectric spectrum sequence of the asphalt sample to be tested obtained in S44, only the fixed-point dielectric characteristic parameters at 25℃ are extracted, and the extraction rule is the same as that in S21.

[0251] S52: calculate the aging degree estimate value using the prediction model;

[0252] The fixed-point dielectric characteristic parameters of the sample to be tested at 25℃ are input into the aging degree prediction model trained in S3. The model outputs the aging degree estimate value of the asphalt sample to be tested.

[0253] S53: calculate the confidence interval of the prediction result.

[0254] Based on the performance of the prediction model on the test set, the 95% confidence interval of the aging degree estimate value is calculated.

[0255] S54: generate an aging degree evaluation report.

[0256] The predicted aging degree estimate value and its confidence interval are arranged into a structured report.

[0257] Based on the measured and predicted results of Example 1 and Comparative Example 1, the following table is obtained:

[0258] Table 1 Comparison of aging degree prediction results based on carbonyl index

[0259]

[0260] Model performance evaluation (carbonyl index):

[0261] Example 1: the coefficient of determination is 0.997, the root mean square error is 0.0018, and the average absolute error is 0.0017;

[0262] Comparative Example 1: the coefficient of determination is 0.935, the root mean square error is 0.0085, and the average absolute error is 0.0076.

[0263] Table 2 Aging degree prediction results based on Comparison of aging degree prediction results of values

[0264]

[0265] Model performance evaluation (R2, RMSE, MAE) Values, unit kPa·°:

[0266] Example 1: Determination coefficient is 0.998, root mean square error is 2.41, and mean absolute error is 2.25;

[0267] Comparative Example 1: Determination coefficient is 0.976, root mean square error is 10.58, and mean absolute error is 9.28.

[0268] In Table 1 and Table 2, the true value is the benchmark value measured based on destructive test, and the predicted value is the aging degree estimated value calculated by the model of Example 1 and Comparative Example 1 respectively.

[0269] The multi-temperature domain detection method adopted in Example 1 can capture the complete law of dielectric response evolution with temperature, while the single-temperature detection method adopted in Comparative Example 1 only extracts dielectric characteristic parameters at 25℃, resulting in significant differences between the prediction results of Example 1 and Comparative Example 1.

[0270] Specifically, the fixed-point dielectric characteristic parameters relied on by the single-temperature detection method can only reflect the instantaneous polarization state of asphalt at a certain specific temperature point. This state is strongly affected by the temperature at the time of measurement, and cannot effectively separate the temperature fluctuation and aging effect. The multi-temperature domain detection method directly quantifies the potential barrier that needs to be overcome by the thermal motion of molecular segments or dipoles in asphalt by extracting evolution dielectric characteristic parameters including apparent activation energy.

[0271] The aging process of asphalt leads to the increase of intermolecular force, the change of the proportion of polar components, and the decrease of free volume. These changes in microstructure will significantly increase the potential barrier of molecular motion, that is, the apparent activation energy of dielectric relaxation process increases. Therefore, the apparent activation energy characterizing the temperature dependence of relaxation rate becomes a sensitive indicator connecting the micro aging mechanism and the macro dielectric response.

[0272] The prediction result of Comparative Example 1 is closest to the true value when the aging time is 40h, corresponding to its single measurement temperature 25℃, and the error is comparable to Example 1. This phenomenon exactly exposes the inherent defect of the single-temperature model, that is, the model performs best near the specific temperature point used to establish the calibration relationship. Once the actual state of the sample to be measured or the environmental temperature deviates from this calibration point, the performance of the model will decrease sharply.

[0273] The multi-temperature zone model of embodiment 1 maintains high precision and stability in the whole aging gradient range and wide temperature range, proving that the model successfully learns and decouples the combined effects of temperature and aging on dielectric parameters. The model builds a robust prediction relationship resistant to temperature interference by integrating the fixed-point characteristics at multiple temperature points and the trend of the characteristics changing with temperature.

[0274] In summary, embodiment 1 verifies the effectiveness of the multi-temperature zone dielectric detection method. The method realizes a deep capture of the changes in the micro-molecular motion during the aging process of asphalt by constructing a quantitative mapping relationship between the features of the dielectric spectrum at multiple temperature points and the evolution law and the aging degree of asphalt. Compared with the single-temperature method represented by comparative example 1, the multi-temperature zone dielectric detection method provided by embodiment 1 significantly improves the prediction accuracy and robustness of the aging degree evaluation in a complex variable temperature environment.

[0275] Based on the ideal embodiments of the present application, related personnel can make various changes and modifications without deviating from the technical idea of the present application. The technical scope of the present application is not limited to the contents of the specification, and must be determined by the scope of the claims.

Claims

1. A non-destructive method for determining the degree of aging of bitumen, characterized in that, The method comprises the following steps: S1: constructing a multi-temperature zone standard dielectric spectrum-aging degree database; the multi-temperature zone standard dielectric spectrum-aging degree database is obtained by measuring the dielectric spectrum of a standard asphalt sample at multiple temperature points under gradient aging degrees and associating the dielectric spectrum with an aging degree index quantified by a destructive test; S2: extracting comprehensive dielectric characteristic parameters sensitive to aging based on the multi-temperature zone standard dielectric spectrum-aging degree database; the comprehensive dielectric characteristic parameters include fixed-point dielectric characteristic parameters and evolution dielectric characteristic parameters; the fixed-point dielectric characteristic parameters are extracted from the dielectric spectrum and include the real part value of the dielectric constant and the imaginary part value of the dielectric loss factor, the peak frequency and peak amplitude in the imaginary part spectrum of the dielectric loss factor, and the average relaxation time and relaxation time distribution broadening parameter obtained by Cole-Cole diagram analysis; the evolution dielectric characteristic parameters are regularity indexes of the fixed-point dielectric characteristic parameters changing with temperature and include the apparent activation energy obtained by temperature dependence analysis of the dielectric loss peak frequency, the slope of the real part of the dielectric constant or the imaginary part of the dielectric loss factor with respect to temperature change, and the trend of the Cole-Cole parameters changing with temperature; S3: constructing an aging degree prediction model based on the comprehensive dielectric characteristic parameters; S4: performing non-destructive multi-temperature zone dielectric detection on the asphalt sample to be tested to obtain dielectric data of the asphalt sample to be tested; S5: inputting the dielectric data of the asphalt sample to be tested into the aging degree prediction model to calculate the aging degree of the asphalt sample to be tested; In step S3, the aging degree prediction model adopts an artificial neural network model; the artificial neural network model includes an input layer, multiple hidden layers, and an output layer, wherein the number of neurons of the input layer corresponds to the number of the comprehensive dielectric characteristic parameters, the hidden layers include at least two layers, each layer contains multiple neurons, the hidden layers adopt a ReLU activation function, and the output layer contains one neuron.

2. The method of claim 1, wherein, Step S1 comprises: S11: preparing a standard asphalt sample; S12: performing artificial accelerated aging treatment on the standard asphalt sample to obtain samples with gradient aging degrees; S13: quantitatively characterizing the true aging degree of each aged asphalt sample by a destructive test; S14: performing multi-temperature zone dielectric spectrum measurement on each aged asphalt sample; S15: associating the dielectric spectrum data of each aged asphalt sample with the corresponding true aging degree to construct a multi-temperature zone standard dielectric spectrum-aging degree database.

3. The method of claim 1, wherein, Step S4 comprises: S41: preparing the asphalt sample to be tested and cleaning the measurement surface; S42: setting multi-temperature zone dielectric spectrum measurement conditions, including temperature range, temperature step, stabilization time, frequency range, and applied voltage; S43: performing multi-temperature zone dielectric spectrum scanning measurement to obtain complex impedance data; S44: calculating the real part of the dielectric constant and the imaginary part of the dielectric loss factor from the complex impedance data to form a multi-temperature zone dielectric spectrum sequence.

4. The method of claim 1, wherein, Step S5 comprises: S51: extracting the comprehensive dielectric characteristic parameters from the dielectric data of the asphalt sample to be tested; S52: inputting the comprehensive dielectric characteristic parameters into the aging degree prediction model to obtain an aging degree estimate value; S53: calculating a confidence interval of the prediction result; S54: generating an aging degree evaluation report.

5. A non-destructive testing system for the degree of aging of bitumen, for implementing the method according to any one of claims 1 to 4, characterized in that, The system comprises: a database construction module for constructing a multi-temperature field standard dielectric spectrum-aging degree database; a feature extraction module connected to the database construction module, for extracting aging-sensitive comprehensive dielectric feature parameters based on the database; a model construction module connected to the feature extraction module, for constructing an aging degree prediction model based on the comprehensive dielectric feature parameters; a non-destructive testing module for performing non-destructive multi-temperature field dielectric testing on the asphalt sample to be tested to obtain dielectric data of the asphalt sample to be tested; an aging degree calculation module connected to the non-destructive testing module and the model construction module, for inputting the dielectric data into the aging degree prediction model to calculate the aging degree of the asphalt sample to be tested.

Images