A regenerant aging failure behavior recognition method and system based on differential spectrum extraction

Abstract

The present invention relates to the technical field of chemistry and materials science, and particularly relates to a method and system for identifying the aging failure behavior of rejuvenators based on differential spectrum extraction. The method includes the following steps: obtaining the infrared spectral data of the recycled asphalt sample and the aged asphalt sample at the same stage; preprocessing the infrared spectral data; constructing a characteristic differential spectrum library of the rejuvenator; extracting the change amount of the characteristic peak area; and obtaining the identification result of the aging failure behavior of the rejuvenator by establishing a correlation model between the change amount of the characteristic peak area and the macroscopic performance index. By combining the adaptive morphology - quantile regression baseline correction, peak entropy weighted normalization, and differential spectrum data enhancement algorithms, the present invention realizes the high - sensitivity extraction of the signals of 3% low - dosage rejuvenators; adopts the Gaussian - Lorentz mixed peak - fitting model and the aging response index model to correlate the changes of microscopic functional groups with macroscopic properties, accurately tracking the aging process of rejuvenators, and providing a universal technical framework for the failure determination and performance optimization of rejuvenators.

Description

Technical Field

[0001] The present invention relates to the technical field of chemistry and materials science, and particularly relates to a method and system for identifying the aging failure behavior of regenerants based on differential spectrum extraction. Background Art

[0002] As road infrastructure enters a large-scale maintenance cycle, asphalt recycling technology has become a key means to extend the pavement life and promote resource recycling. Due to its green environmental protection and high component activity, bio-oil regenerant has shown significant advantages in restoring the performance of aged asphalt. However, during long-term service, the regenerant is affected by environmental factors such as heat, oxygen, and ultraviolet rays, and its chemical structure is prone to oxidation and cracking reactions, resulting in a gradual decline in performance, which directly affects the durability of recycled asphalt. Therefore, accurately identifying the aging failure behavior of the regenerant and revealing the correlation mechanism between its microstructure and macroscopic properties are of great significance for improving the design and application level of the regenerant.

[0003] Currently, in the prior art, the aging state of the regenerant is mainly indirectly evaluated by monitoring the macroscopic performance indicators of recycled asphalt, such as penetration and softening point. However, this method cannot distinguish the respective aging contributions of the asphalt matrix and the regenerant. Especially under the condition of a low dosage of 3%, the weak signal of the regenerant is easily interfered and masked by the matrix, resulting in insufficient identification accuracy. In addition, the existing infrared spectroscopy analysis is difficult to separately extract the characteristic functional group information of the regenerant from the composite system, and lacks standardized processing and dynamic enhancement methods for differential spectrum data, making it difficult to extract the characteristic peaks of trace components and unable to establish a quantitative correlation model between the evolution of functional groups and the attenuation of performance. These defects limit the in-depth study and accurate prediction of the aging failure mechanism of the regenerant.

[0004] To solve the above problems, the present invention proposes a method for identifying the aging failure behavior of regenerants based on differential spectrum extraction. By combining adaptive morphology-quantile regression baseline correction, peak entropy weighted normalization, and differential spectrum data enhancement algorithms, high-sensitivity extraction of the characteristic signals of low-dosage regenerants is achieved. Further, a Gaussian-Lorentz mixed peak deconvolution model and an aging response index calculation model are constructed to dynamically correlate the microscopic functional group changes with the macroscopic performance indicators, solving the problem that the existing methods cannot accurately track the aging process of the regenerant. This method not only provides a scientific basis for the failure determination of the regenerant, but also establishes a universal technical framework for the performance optimization and screening of different regenerants. Summary of the Invention

[0005] Aiming at the defects in the prior art, the present invention provides a method and system for identifying the aging failure behavior of regenerants based on differential spectrum extraction.

[0006] In a first aspect, a method for identifying the aging failure behavior of a rejuvenator based on differential spectrum extraction provided by the present invention includes the following steps: obtaining infrared spectral data of a regenerated asphalt sample and an aged asphalt sample at the same stage; preprocessing the infrared spectral data to obtain a preprocessing result, where the preprocessing includes baseline correction and normalization; constructing a characteristic differential spectrum library of the rejuvenator with the aged asphalt spectrum as the background spectrum according to the preprocessing result; extracting the change amount of the characteristic peak area of typical functional groups in the rejuvenator based on the characteristic differential spectrum library; and obtaining the identification result of the aging failure behavior of the rejuvenator by establishing a correlation model between the change amount of the characteristic peak area and the macroscopic performance index. By obtaining the infrared spectral data of the regenerated asphalt and the aged asphalt at the same stage and performing baseline correction and normalization, the present invention realizes high-quality standardization processing of the spectral data, providing a reliable basis for subsequent differential spectrum analysis; by constructing a characteristic differential spectrum library of the rejuvenator with the aged asphalt spectrum as the background spectrum, the functional group signal differences between the rejuvenator and the aged asphalt are effectively separated, significantly improving the accuracy and sensitivity of rejuvenator characteristic identification; by extracting the change amount of the characteristic peak area of typical functional groups in the rejuvenator, the dynamic tracking of the chemical structure evolution process of the rejuvenator at the molecular level is realized, revealing the microscopic mechanism of its aging failure; and by establishing a correlation model between the change amount of the characteristic peak area and the macroscopic performance index, the quantitative correlation between the microscopic functional group changes and the macroscopic performance attenuation is established, providing an important method system for the identification and prediction of the aging failure behavior of the rejuvenator.

[0007] Optionally, the preprocessing of the infrared spectral data to obtain a preprocessing result includes: using a baseline correction method based on adaptive morphology - quantile regression to perform baseline correction processing on the infrared spectral data to obtain the infrared spectral data after baseline correction processing; using the peak entropy weighting method to perform normalization processing on the infrared spectral data after baseline correction processing to obtain the infrared spectral data after normalization processing, where the normalization processing includes using an improved Burg entropy criterion to locate characteristic peaks; and using an improved adaptive iterative reweighted penalty least squares method to perform secondary baseline correction processing on the infrared spectral data after normalization processing to obtain the infrared spectral data after secondary baseline correction processing. By using a baseline correction method based on adaptive morphology - quantile regression, the present invention effectively suppresses baseline drift and background interference in the infrared spectrum, significantly improving the signal - to - noise ratio and characteristic peak identification accuracy of the spectral data; by introducing the peak entropy weighting method for normalization processing and accurately locating characteristic peaks based on the improved Burg entropy criterion, the adaptive weighted normalization of the spectral data is realized, avoiding the suppression problem of weak peaks by existing normalization methods and ensuring the complete retention of key functional group information; and by using an improved adaptive iterative reweighted penalty least squares method for secondary baseline correction processing, dynamically adjusting the weight function and penalty parameters, the problem of complex spectral baseline fluctuations is solved, further improving the comparability and analysis reliability of the spectral data.

[0008] Optionally, constructing a characteristic differential spectrum library of the regenerant with the aged asphalt spectrum as the background spectrum according to the preprocessing result includes: establishing a differential spectrum intensity calculation model based on the infrared spectral data after baseline quadratic correction, with the aged asphalt spectrum as the background spectrum; obtaining differential spectrum data according to the normalized differential spectrum intensity calculation model; constructing a differential spectrum data enhancement algorithm model based on the differential spectrum data; obtaining enhanced differential spectrum data through the differential spectrum data enhancement algorithm model; constructing a fingerprint matching degree scoring model using the enhanced differential spectrum data; and constructing a characteristic differential spectrum library of the regenerant in combination with the fingerprint matching degree scoring model. By establishing a differential spectrum intensity calculation model with the aged asphalt spectrum as the background spectrum, the present invention effectively separates the superimposed spectral signals of the regenerant and the aged asphalt, accurately extracts the characteristic functional group responses of the regenerant, and provides high-purity differential spectrum data for subsequent analysis; by constructing a differential spectrum data enhancement algorithm model, the weak characteristic peaks of the regenerant are amplified using signal enhancement technology, significantly improving the detection sensitivity of low-content functional groups; by introducing a fingerprint matching degree scoring model, intelligent screening and classification of differential spectrum data are realized. In combination with the construction of the characteristic differential spectrum library, a rapid comparison system for the chemical fingerprints of the regenerant is established, providing a quantifiable discrimination basis for the failure behavior of regenerants at different aging stages.

[0009] Optionally, the differential spectrum intensity calculation model satisfies the following expression:

[0010]

[0011] where is the differential spectrum intensity, is the spectral intensity of the regenerant at the wavenumber , is the spectral intensity of the aged asphalt at the wavenumber , is the dynamic background deduction factor, is the frequency domain weight function; the differential spectrum data enhancement algorithm model satisfies the following expression:

[0012]

[0013] where is the enhanced differential spectrum signal, is the differential spectrum intensity, is the convolution operator, is the spectral intensity of the regenerant at the wavenumber , is the spectral intensity of the aged asphalt at the wavenumber , is the central wavenumber of the target characteristic peak, is the standard deviation of the Gaussian function; the fingerprint matching degree scoring model satisfies the following expression:

[0014]

[0015] wherein, is the comprehensive score of the regenerant matching degree, is the enhanced difference spectrum signal, is the spectral feature vector of the standard regenerant in the reference spectral library, is the common characteristic peak area between the difference spectrum and the reference spectrum, is the total characteristic peak area between the difference spectrum and the reference spectrum, 、 are the integral intervals. By constructing a difference spectrum intensity calculation model, the present invention realizes the accurate decoupling of the regenerant and the aged asphalt spectrum, effectively eliminates matrix interference and highlights the characteristic peaks of the regenerant, and significantly improves the accuracy of difference spectrum analysis; by designing a difference spectrum data enhancement algorithm model including derivative operation and Gaussian enhancement, the differential change rate detection and local signal focusing techniques are skillfully combined, greatly enhancing the recognition ability of weak characteristic peaks and solving the bottleneck problem of insufficient response of existing methods to trace components; by establishing a fingerprint matching degree scoring model integrating spectral graph convolution integral and set similarity metric, the multi-dimensional intelligent comparison between the enhanced difference spectrum and the standard spectral library is realized, providing an important mathematical criterion for the quantitative evaluation of the aging state of the regenerant.

[0016] Optionally, the extraction of the characteristic peak area change amount of the typical functional groups in the regenerant based on the characteristic difference spectral library includes: establishing a Gaussian-Lorentz mixed peak separation model based on the characteristic difference spectral library; using the Gaussian-Lorentz mixed peak separation model to fit the characteristic peaks of the typical functional groups in the regenerant to obtain a fitting result; establishing a characteristic peak area change amount calculation model according to the fitting result; and extracting the characteristic peak area change amount of the typical functional groups in the regenerant through the characteristic peak area change amount calculation model. By establishing a Gaussian-Lorentz mixed peak separation model to analyze the characteristic difference spectrum, the present invention significantly improves the characteristic peak separation accuracy and quantitative reliability; by using a mixed peak type to fit the characteristic peaks of the typical functional groups in the regenerant, the accurate analysis of complex spectral shapes is realized, providing a high-quality fitting result for subsequent quantitative analysis; by constructing a characteristic peak area change amount calculation model, the peak separation fitting result is converted into a quantifiable functional group evolution index, establishing a bridge from molecular structure change to aging degree evaluation.

[0017] Optionally, the Gaussian-Lorentz mixed peak separation model satisfies the following relational expression:

[0018]

[0019] wherein, is the mathematical expression of the differential spectrum of the regenerant characteristics, is the dynamic baseline, is the number of Gaussian peak functions, is the peak height, is the peak center wave number, is the Gaussian full width at half maximum, is the Lorentzian full width at half maximum, is the mixing coefficient, is the wave number; the calculation model of the characteristic peak area change satisfies the following relationship:

[0020]

[0021] where, is the change amount of the characteristic peak area, , are the fitting areas of the regenerant and the aged asphalt at the th characteristic peak, is the reference value of the corresponding peak area of the standard regenerant, is the weight factor of the wave number interval, is the wave number interval. By constructing a Gaussian-Lorentzian mixed peak separation model and introducing dynamic baseline correction, the present invention realizes the accurate analysis of overlapping peaks and asymmetric peaks in the differential spectrum of regenerant characteristics, significantly improves the separation degree and fitting accuracy of characteristic peaks of complex functional groups; by designing the mixing coefficient to adjust the contribution ratio of Gaussian and Lorentzian functions, the problem of insufficient adaptability of a single peak type function is solved, providing a flexible mathematical expression for characteristic peaks of different forms; by establishing a calculation model of the change amount of the characteristic peak area and introducing a weight factor and wave number interval correction, the standardized quantitative characterization of the change of the functional groups of the regenerant is realized, effectively eliminating the systematic error caused by instrument fluctuation and test condition difference.

[0022] Optionally, the obtaining of the identification result of the aging failure behavior of the regenerant by establishing the correlation model between the change amount of the characteristic peak area and the macroscopic performance index includes: by the change amount of the characteristic peak area, combined with the macroscopic performance index, establishing an aging response index calculation model, and the macroscopic performance index includes penetration, softening point, complex shear modulus and creep stiffness; the aging response index calculation model satisfies the following relationship:

[0023]

[0024] where, is the aging response index, is the change amount of the characteristic peak area, is the initial peak area, , , , respectively represent the initial penetration, softening point, complex shear modulus, and creep stiffness, , , , respectively represent the measured performance values of the initial penetration, softening point, complex shear modulus, and creep stiffness after aging, , , , , , ,

[0025] are weight coefficients; according to the aging response index calculation model, the aging response index is obtained; based on the aging response index, the identification result of the aging failure behavior of the rejuvenator is obtained. By establishing an aging response index calculation model that integrates the change amount of microscopic characteristic peak area and macroscopic performance indicators, the present invention realizes the cross-scale correlation analysis of the evolution of functional groups at the molecular level and macroscopic pavement performance, providing a multi-dimensional comprehensive evaluation system for the aging failure behavior of rejuvenators; by designing a composite calculation model including weight coefficient adjustment, the change laws of infrared spectrum characteristic parameters and key performance indicators are effectively integrated; by constructing an aging response index calculation model, the data is converted into a unified evaluation index, realizing the rapid diagnosis of the aging state of rejuvenators.

[0025] Optionally, the obtaining the identification result of the aging failure behavior of the rejuvenator based on the aging response index includes: establishing a coupling correlation coefficient calculation model based on the aging response index; obtaining a coupling correlation coefficient through the coupling correlation coefficient calculation model; evaluating the correlation between the aging of the rejuvenator and the performance attenuation according to the coupling correlation coefficient to obtain the performance change trend; establishing a rejuvenator failure determination model through the aging response index and the performance change trend; obtaining the identification result of the aging failure behavior of the rejuvenator according to the rejuvenator failure determination model. By establishing a coupling correlation coefficient calculation model, the present invention realizes the quantitative correlation analysis between the aging response index and the performance attenuation trend, revealing the internal connection mechanism between the chemical structure and macroscopic performance of the rejuvenator; by adopting a multi-parameter coupling analysis method, the aging response index is dynamically correlated with the performance change trend, solving the bottleneck problem that it is difficult to accurately evaluate the aging stage of rejuvenators by existing methods; by constructing a rejuvenator failure determination model, the chemical aging index and engineering performance threshold are integrated, realizing the accurate judgment from molecular evolution to functional failure.

[0026] Optionally, the coupling correlation coefficient calculation model satisfies the following relationship:

[0027]

[0028] wherein, is the coupling correlation coefficient, is the aging response index of the th sample, is the sample mean of the aging response index, is the performance degradation index of the th sample, is the sample mean of the performance degradation index, is the number of samples; the regenerant failure determination model satisfies the following expression:

[0029]

[0030] wherein, is the state of the regenerant, is the critical failure threshold, is the performance retention rate, is the initial performance benchmark, is the material degradation coefficient, is the warning buffer interval, is the time. By establishing a coupling correlation coefficient calculation model and adopting the covariance and standard deviation processing methods, the present invention realizes the quantification of the dynamic correlation degree between the aging response index and the performance degradation index, providing a mathematical tool for revealing the synchronous evolution law of the chemical aging and performance decline of the regenerant; by constructing a three-dimensional failure determination model including the critical failure threshold, the performance retention rate and the degradation coefficient, combining the static threshold criterion with the dynamic change rate analysis, the present invention realizes the multi-level intelligent identification of the failure state of the regenerant; by setting the warning buffer interval and the hierarchical determination conditions, the present invention establishes a progressive warning mechanism from the safe state to the failure state, significantly improving the forward-looking and reliability of the evaluation of the failure state of the regenerant.

[0031] Second aspect, the present invention provides a regenerant aging failure behavior recognition system based on differential spectrum extraction. The system includes an input device, a processor, an output device, and a memory. The input device, the processor, the output device, and the memory are interconnected. Among them, the memory is used to store computer programs, and the computer programs include program instructions. The processor is configured to call the program instructions. The system uses the regenerant aging failure behavior recognition method based on differential spectrum extraction described above. The system provided by the present invention has a high degree of integration, and the information transmission between components is smooth. By integrating the infrared spectrum differential spectrum extraction and enhancement algorithm, high-sensitivity detection of the characteristic functional group signals of the regenerant is realized, overcoming the defect of insufficient recognition of trace components in the existing methods, and significantly improving the ability to capture weak changes in the initial stage of aging; by integrating the Gaussian-Lorentz mixed peak fitting model and the dynamic baseline correction technology, the problem of complex peak overlap resolution is solved, providing an accurate quantitative analysis method for the molecular structure evolution of the regenerant; by coupling the change amount of the characteristic peak area with the macroscopic performance index, a cross-scale aging response index model is established, realizing the full-chain correlation analysis from chemical structure changes to engineering performance decline; by developing a failure determination algorithm based on dynamic threshold and change rate, a hierarchical recognition system including safety, warning, and failure states is established, greatly improving the accuracy and forward-looking of the regenerant aging state assessment. Description of the Drawings

[0032] Figure 1 It is a flowchart of the regenerant aging failure behavior recognition method based on differential spectrum extraction according to an embodiment of the present invention;

[0033] Figure 2 It is a correction result diagram of different baseline correction methods according to an embodiment of the present invention;

[0034] Figure 3 It is a relationship curve diagram of the characteristic peak area and penetration according to an embodiment of the present invention;

[0035] Figure 4 It is a structural schematic diagram of the regenerant aging failure behavior recognition system based on differential spectrum extraction according to an embodiment of the present invention. Detailed Embodiments

[0036] The specific embodiments of the present invention will be described in detail below. It should be noted that the embodiments described here are only for illustrative purposes and are not used to limit the present invention. In the following description, in order to provide a thorough understanding of the present invention, a large number of specific details are set forth. However, it is obvious to those of ordinary skill in the art that the present invention does not have to be implemented with these specific details. In other instances, well-known circuits, software, or methods are not specifically described to avoid obscuring the present invention.

[0037] Throughout the specification, references to "one embodiment", "an embodiment", "one example", or "an example" mean that a particular feature, structure, or characteristic described in connection with the embodiment or example is included in at least one embodiment of the present invention. Thus, the phrases "in one embodiment", "in an embodiment", "one example", or "an example" appearing throughout the specification do not necessarily all refer to the same embodiment or example. In addition, the specific features, structures, or characteristics may be combined in any suitable combination and / or sub-combination in one or more embodiments or examples. Further, those of ordinary skill in the art should understand that the diagrams provided herein are for illustrative purposes only and are not necessarily drawn to scale.

[0038] Please refer to Figure 1 , an embodiment of the present invention provides a method for identifying the aging failure behavior of a regenerant based on differential spectrum extraction, and the method includes the following steps:

[0039] S1. Obtain the infrared spectral data of the regenerated asphalt sample and the co-stage aged asphalt sample.

[0040] In one embodiment, sample preparation and stage division are first carried out to obtain the regenerated asphalt sample and the co-stage aged asphalt sample for infrared spectral testing.

[0041] Specifically, 70# base asphalt is selected as the raw material, and the rolling thin film oven test (RTFOT) and pressure aging vessel accelerated aging test (PAV) are carried out in sequence according to the standard process. RTFOT simulates the performance changes of asphalt during short-term aging, and PAV further simulates the influence of long-term aging on asphalt. Through the treatment of these two steps, the asphalt samples corresponding to the aging stage of the regenerated asphalt are successfully prepared.

[0042] Further, the prepared aged asphalt sample is taken out, and the regenerant is accurately weighed according to a mass fraction of 3%, and it is fully mixed with the aged asphalt. During the mixing process, it is necessary to ensure uniform stirring so that the regenerant can be fully dispersed in the aged asphalt, and finally the regenerated asphalt sample is prepared.

[0043] Further, prepare the samples and test conditions required for infrared spectral testing.

[0044] Specifically, appropriate amounts of the prepared aged asphalt sample and regenerated asphalt sample are respectively placed on a clean sample rack. For viscous samples such as asphalt, a suitable sample preparation method is required, such as heating the sample to a flowing state and then evenly smearing it on the sample rack, ensuring that the thickness of the smeared layer is uniform and the surface is flat, and avoiding the occurrence of bubbles or uneven thickness to ensure the accuracy of the test results. At the same time, attention should be paid to preventing the sample from being contaminated during the smearing process.

[0045] Further, a Fourier Transform Infrared Spectrometer (FTIR) is used for testing. According to the characteristics of the asphalt sample, appropriate test parameters are set. For example, the scanning range is set to , and this range can cover the characteristic absorption peaks of various functional groups in the asphalt; the resolution is set to to obtain a relatively clear spectrogram; the number of scans is set to 32 times. By taking the average of multiple scans, the signal-to-noise ratio of the spectral signal is improved, and the influence of random errors on the test results is reduced.

[0046] Further, infrared spectroscopy tests are carried out and data are obtained.

[0047] Specifically, before formally testing the sample, a background scan is first performed on the blank sample holder. The sample holder without the sample is placed in the optical path of the FTIR, and the scanning program is started to collect background spectral data. The purpose of the background scan is to eliminate the interference of the instrument itself, environmental factors, and the sample holder on the test results, and ensure the accuracy and reliability of the sample spectral data obtained in the subsequent tests.

[0048] Further, the processed aged asphalt sample holder is placed in the optical path of the FTIR, ensuring that the sample completely covers the optical path and the surface of the sample is perpendicular to the optical path. The test program is started, and the instrument begins to scan the aged asphalt sample to collect its infrared spectral data. During the scanning process, the instrument records the absorption intensity of the sample for infrared light at different wavenumbers, and finally generates an infrared spectrogram of the aged asphalt sample. This spectrogram contains the characteristic absorption information of various chemical bonds and functional groups in the aged asphalt sample.

[0049] Further, after completing the test of the aged asphalt sample, the aged asphalt sample holder is taken out, and the sample holder containing the regenerated asphalt sample is replaced, also ensuring that the sample is placed correctly. The test program is started again to scan the regenerated asphalt sample and collect its infrared spectral data, generating an infrared spectrogram of the regenerated asphalt sample. Through the above steps, the infrared spectral data of the regenerated asphalt sample and the aged asphalt sample at the same stage can be obtained respectively.

[0050] S2. Preprocess the infrared spectral data to obtain a preprocessing result. The preprocessing includes baseline correction and normalization processing.

[0051] Among them, S2 further includes the following steps:

[0052] S21. Perform automatic baseline correction processing on the infrared spectral data to obtain the infrared spectral data after baseline correction processing.

[0053] In one embodiment, a baseline correction method based on adaptive morphology - quantile regression is first proposed for performing automatic baseline correction processing on the infrared spectral data.

[0054] Specifically, a top-hat transform is performed using a wavelength-adaptive elliptical structuring element, and the length of its major axis varies non-linearly with the wavelength. The expression for the length of the major axis is as follows:

[0055]

[0056] Wherein, is the length of the major axis, is the maximum window width, is the slope factor, is the wavelength, is the spectral center wavelength.

[0057] Furthermore, for the residual signal after the top-hat transform, the quantile is calculated within a sliding window to construct an initial baseline, and it is optimized by Tikhonov regularization. The expression of the Tikhonov regularization model is as follows:

[0058]

[0059] Wherein, is the second-order difference operator, is the smoothing factor, is the initial baseline value, is the target baseline value.

[0060] The advantage of this method is that it combines morphological filtering and quantile regression to solve the problem of baseline drift in wide spectral bands through a dynamic structuring element.

[0061] S22. Normalize the infrared spectral data after baseline correction to obtain the normalized infrared spectral data.

[0062] In one embodiment, peak entropy weighted normalization is used to process the data after baseline correction.

[0063] Specifically, an improved Burg entropy criterion is used to locate the characteristic peaks. The location formula is as follows:

[0064]

[0065] Wherein, is the intensity value of the spectrum at the th wavelength, is the total energy of the spectrum, is the number of spectral data points, is the improved Burg entropy value, reflecting the degree of dispersion of the spectral energy distribution.

[0066] Furthermore, the entropy weight is calculated for each characteristic peak region, and its calculation expression is as follows:

[0067]

[0068] Among them, is the th characteristic peak region, is the th entropy weight factor, is the maximum entropy value of the full spectrum, which is used for weight normalization, is the local entropy value of region .

[0069] It should be noted that the improved Burg entropy criterion performs normalization by introducing the total spectral energy , converts the absolute energy calculation of the traditional Burg entropy into the measurement of the discreteness of the relative energy distribution, and enhances the comparability between spectra of different intensities. At the same time, using to replace the original amplitude highlights the contribution weight of the characteristic peaks, making the entropy value more sensitive to reflect the degree of energy concentration in the peak region. In addition, by calculating the weight through the ratio of the local entropy to the global maximum entropy, adaptive region normalization is achieved, avoiding the distortion of peak features caused by global uniform scaling.

[0070] Furthermore, the normalized spectral data is obtained, and its normalization expression is as follows:

[0071]

[0072] Among them, is the normalized spectrum, is the spectrum after baseline correction but before normalization, is the th entropy weight factor, is the peak area.

[0073] The advantage of this method is that it automatically balances the contributions of strong / weak peaks through entropy weights, solving the deviation problem of existing normalization methods in heterogeneous samples.

[0074] S23. Using the adaptive iteratively reweighted penalized least squares method, perform baseline secondary correction processing on the normalized infrared spectral data to obtain the infrared spectral data after baseline secondary correction processing.

[0075] In one embodiment, an improved adaptive iteratively reweighted penalized least squares method (airPLS) is constructed to perform baseline secondary correction processing, also known as baseline flattening processing.

[0076] Specifically, the improved airPLS solves for the baseline value by minimizing the following objective function, thereby achieving baseline secondary correction processing:

[0077]

[0078] Among them, is the baseline value to be solved, the weight of the th iteration, is the original spectral data, is the second-order difference operator, is the L1 penalty coefficient, represents point-by-point multiplication, forcing the baseline to be close to the original spectrum in the non-peak region.

[0079] Furthermore, a dynamic weight update model is established to update the weights. In the th iteration, the weight vector is updated according to the dynamic weight update model. The dynamic weight update model is as follows:

[0080]

[0081] Among them, is the weight of the th data point in the th iteration, is the second derivative of the spectral data at the th data point, is the derivative threshold multiplier, represents the median of the absolute value of the second derivative of the full spectrum, represents any data point, represents the intensity of the th point of the original spectrum, represents the baseline estimate of the th iteration, is the residual scale factor, represents other cases.

[0082] Please refer to Figure 2 , Figure 2 which is the calibration result graph of different baseline correction methods of the present invention. Figure 2 shows the relationship between absorbance and wavenumber after untreated, OMNIC program processed, and improved airPLS method processed. The absorbance is a dimensionless parameter, and the OMNIC is software for infrared spectral data acquisition and analysis.

[0083] From Figure 2 it can be seen that the improved airPLS method can correct baseline drift more thoroughly compared with untreated and conventional software program processing. For example, in the wavenumber region of , it can dynamically identify background noise and retain true signals, especially optimizing weak peak detection. For example, in the wavenumber region of , it improves data reliability.

[0084] It should be noted that the improvement of the airPLS of the present invention lies in: changing the existing L2 penalty term to the L1 norm, enhancing the sparse constraint on the sudden change of the baseline curvature, and improving the fitting ability for complex baselines; introducing a dynamic derivative threshold mechanism to adaptively identify the peak region and non-peak region through the median of the spectral second derivative, avoiding manual setting of a fixed threshold.

[0085] Furthermore, through the objective function and the dynamic weight update model, the infrared spectral data after baseline quadratic correction processing is obtained.

[0086] The advantage of this method is that it can effectively retain important true signals by using the derivative constraint mechanism. Compared with the existing airPLS method, it achieves a significant reduction in error and better performance.

[0087] S3. According to the preprocessing result, using the aged asphalt spectrum as the background spectrum, a characteristic difference spectrum library of the rejuvenator is constructed.

[0088] In one embodiment, according to the preprocessing result, using the aged asphalt spectrum as the background spectrum, a difference spectrum intensity calculation model is established, and the difference spectrum intensity calculation model satisfies the following expression:

[0089]

[0090] where is the difference spectrum intensity, is the spectral intensity of the rejuvenator at the wavenumber , is the spectral intensity of the aged asphalt at the wavenumber , is the dynamic background subtraction factor, an adjustment coefficient varying with the wavenumber, dynamically calculated through the intensity gradient of the aging characteristic peak, and used to accurately subtract the background interference of the aged asphalt, is the frequency domain weight function, which assigns higher weights in the characteristic peak region of the rejuvenator to enhance the contribution of the target functional groups.

[0091] Furthermore, a difference spectrum data enhancement algorithm model is constructed, and the original difference spectrum is subjected to feature extraction through the convolution kernel , and the difference spectrum data enhancement algorithm model satisfies the following expression:

[0092]

[0093] where is the enhanced difference spectrum signal, and features are extracted through convolution operation, is the convolution operator, representing the integral overlap calculation between functions, is a composite convolution kernel, composed of a derivative term and a Gaussian term, and satisfies the following expression:

[0094]

[0095] Among them, is the spectral intensity of the regenerant at the wavenumber ; is the spectral intensity of the aged asphalt at the wavenumber ; is the central wavenumber of the target characteristic peak, is the standard deviation of the Gaussian function, which determines the width of the wavenumber range for extracting the characteristic peak.

[0096] The advantage of this method is that it uses the derivative term to capture the difference in the spectral line slopes between the regenerant and the aged asphalt, and the Gaussian term focuses on the characteristic peak region. Compared with conventional wavelet denoising, it can more effectively retain the fingerprint information of functional groups and achieve the directional enhancement of weak characteristic signals.

[0097] Furthermore, a fingerprint matching degree scoring model is constructed for establishing a fingerprint difference spectrum library of regenerants. The fingerprint matching degree scoring model realizes the identification of regenerants through the matching degree scoring of typical functional group characteristic vectors, and the fingerprint matching degree scoring model satisfies the following expression:

[0098]

[0099] Among them, is the comprehensive scoring of the regenerant matching degree, is the enhanced difference spectrum signal, is the spectral characteristic vector of the standard regenerant in the reference spectrum library, is the common characteristic peak area between the difference spectrum and the reference spectrum, is the total characteristic peak area of the difference spectrum and the reference spectrum, , are the integration intervals, covering the characteristic wavenumber range of the target functional group.

[0100] The advantage of this method is that it uses the fingerprint matching degree scoring model to integrate the two parameters of spectral intensity integration and spectral shape topological similarity. Compared with the existing single peak height ratio method, it can significantly improve the specificity of regenerant identification, especially suitable for distinguishing regenerants with similar chemical compositions.

[0101] S4. Based on the characteristic difference spectrum library, extract the change amount of the characteristic peak area of the typical functional groups in the regenerant.

[0102] In one embodiment, an adaptive Gaussian-Lorentz mixed peak separation model is first constructed, and the expression of this model is as follows:

[0103]

[0104] Among them, is the mathematical expression of the characteristic difference spectrum of the regenerant, used to fit the preprocessed infrared spectral data, is the dynamic baseline, and piecewise polynomial fitting is adopted. is the number of Gaussian peak functions. is the th Gaussian peak function, which satisfies the following expression:

[0105]

[0106] where is the peak height, is the peak center wavenumber, is the Gaussian full width at half maximum, is the th Lorentzian peak function, which satisfies the following expression:

[0107]

[0108] where is the Lorentzian full width at half maximum, is the mixing coefficient, which dynamically adjusts the ratio of the Gaussian and Lorentzian components and is calculated through the peak shape asymmetry. The relevant expressions are as follows:

[0109]

[0110]

[0111] where and respectively represent the absolute values of the slopes on the left and right sides of the peak center.

[0112] The advantage of this method is that it adaptively adjusts the Gaussian / Lorentz ratio according to the peak shape asymmetry, overcoming the defect of the existing method using fixed empirical values; for the complex system of regenerant-aged asphalt, the baseline fitting interval is automatically divided according to the distribution of characteristic peaks, overcoming the defect of the existing method using a global baseline that leads to fitting deviation.

[0113] Furthermore, a calculation model for the change in characteristic peak area is established, and the calculation model for the change in characteristic peak area satisfies the following expression:

[0114]

[0115] where is the change in characteristic peak area, and are the fitting areas of the regenerant and aged asphalt at the th characteristic peak, is the reference value of the corresponding peak area of the standard regenerant (unaged), is the wavenumber interval weight factor, which is determined by the signal-to-noise ratio of the characteristic peak.

[0116] The advantage of this method is that it uses the signal-to-noise ratio of characteristic peaks for dynamic weighting to improve the reliability of weak peaks; by introducing the reference value of the corresponding peak area of the standard regenerant, the difference between regenerant batches is eliminated.

[0117] S5. By establishing a correlation model between the change amount of the characteristic peak area and the macroscopic performance index, the identification result of the aging failure behavior of the regenerant is obtained.

[0118] In one embodiment, first, an aging response index calculation model based on the change amount of the characteristic peak area and the macroscopic performance index is established. The macroscopic performance index includes penetration, softening point, complex shear modulus, and creep stiffness. The aging response index calculation model satisfies the following expression:

[0119]

[0120] where is the aging response index, is the change amount of the characteristic peak area, is the initial peak area, , , , respectively represent the initial penetration, softening point, complex shear modulus, and creep stiffness, , , , respectively represent the measured performance values of the initial penetration, softening point, complex shear modulus, and creep stiffness after aging, , , , , are the weight coefficients. The penetration is the depth that a standard needle vertically penetrates an asphalt specimen within 5 seconds under specified temperature and load, expressed in units of 0.1 ; the softening point is the temperature at which the asphalt material begins to soften and reaches a certain sag amount under specific conditions; the complex shear modulus is a complex modulus that describes the stress-strain relationship of asphalt materials under dynamic shear action, including a real part and an imaginary part; the creep stiffness is a parameter that describes the change of strain with time when the asphalt material is under a constant stress at low temperature, reflecting the anti-deformation ability and anti-cracking performance of the asphalt material at low temperature.

[0121] It should be noted that if the characteristic peak area changes significantly and has a good linear relationship with all macroscopic performance indicators, it can be used to judge the oxidative failure of the regenerant. That is to say, if there is no good linear relationship with one of the macroscopic performance indicators, the regenerant has undergone oxidative failure.

[0122] See also Figure 3 In a specific embodiment, a curve showing the relationship between characteristic peak area and needle penetration is given. Figure 3 It can be seen that the linear expression is as follows:

[0123]

[0124] in, Indicates needle penetration. represents the characteristic peak area, the correlation coefficient between the needle penetration and the characteristic peak area , which means that the relationship curve has a good linear relationship.

[0125] It should be noted that when the actual measured penetration deviates significantly from the predicted value of the regression equation, it indicates that the regenerant has been oxidized and failed.

[0126] The advantage of this method is that it combines chemical characteristic peaks with multiple macroscopic performance indicators, overcoming the limitations of existing indicators.

[0127] Furthermore, based on the aging response index calculation model, a coupling correlation coefficient calculation model is established to evaluate the nonlinear correlation between the aging of the regenerant and the performance attenuation. The coupling correlation coefficient calculation model satisfies the following expression:

[0128]

[0129] in, is the coupling correlation coefficient, measuring The statistical correlation between performance degradation and the range of , the closer The stronger the correlation, For the The aging response index of the samples, is the sample mean of the aging response index, For the The performance degradation index of each sample can be the penetration loss rate, softening point growth rate or other macro performance degradation. is the sample mean of the performance decay index, indicating the average level of performance decay. is the sample size used to calculate statistical correlation.

[0130] The advantage of this method is that it quantifies the degree of coupling between aging and performance through the coupling correlation coefficient, which is better than the existing Pearson correlation coefficient and can capture nonlinear relationships.

[0131] Further, through The threshold value and the performance change trend are combined to establish a regeneration agent failure judgment model, which satisfies the following expression:

[0132]

[0133] Among them, is the state of the regenerant, is the critical failure threshold, is the performance retention rate, is the initial performance benchmark, is the material degradation coefficient, is the early warning buffer interval, is the time.

[0134] Furthermore, using the regenerant failure determination model, the aging failure behavior of the regenerant is identified, that is, when is in the failure state, the regenerant ages and fails.

[0135] The advantage of this method is that by combining the threshold and the performance change rate, a multi-level failure early warning mechanism is constructed, realizing the accurate quantitative evaluation of the material degradation process, significantly improving the sensitivity and reliability of the regenerant state identification, and providing a theoretical basis for predictive maintenance.

[0136] Please refer to Figure 4 , Figure 4 which is the structural schematic diagram of a regenerant aging failure behavior identification system based on differential spectrum extraction in an embodiment of the present invention. The system includes an input device, a processor, an output device, and a memory. The input device, the processor, the output device, and the memory are interconnected. Among them, the memory is used to store computer programs, the computer programs include program instructions, the processor is configured to call the program instructions, and the system uses the above-mentioned regenerant aging failure behavior identification method based on differential spectrum extraction.

[0137] In this embodiment, the input device includes an infrared spectrometer, a mechanical property tester, and a data interface module; the infrared spectrometer is used to collect infrared spectrum data of the regenerant and aged asphalt; the mechanical property tester is used to obtain macroscopic property test data, and use the data interface module to receive and store these data to provide input for subsequent correlation analysis; the input device supports data format standardization to ensure that the collected and received data is compatible with the processor.

[0138] The processor includes a preprocessing module, a differential spectrum analysis module, a peak fitting module, a correlation modeling module, and a failure determination module; the preprocessing module is used to denoise, correct, and enhance the original spectral data to improve the signal-to-noise ratio; the differential spectrum analysis module is used to extract the differential spectrum signals of the characteristic functional groups of the regenerant and construct a differential spectrum database; the peak fitting module is used to quantify the correlation between the peak area change and the macroscopic performance; the correlation modeling module is used to calculate the aging response index in real time; the failure determination module is used to perform failure early warning in combination with the performance decay trend.

[0139] The output device includes a display terminal, a report generation module, and a warning prompt module; the display terminal is used to intuitively display the aging degree of the regenerant, such as the safe / warning / failure status; the report generation module is used to generate a complete report containing differential spectrum analysis, peak area change, and aging response index data; the warning prompt module is used to give a warning prompt when necessary.

[0140] The memory uses a high-speed solid-state drive, which has the characteristics of fast read and write speed, large capacity, and high reliability. It is mainly used to store the data input by the input device and the result data processed by the processor, and can meet the needs of storing a large amount of data.

[0141] In summary, the present invention realizes the highly sensitive extraction of the characteristic signals of the low-dosage regenerant by combining the adaptive morphology-quantile regression baseline correction, peak entropy weighted normalization, and differential spectrum data enhancement algorithms; further constructs a Gaussian-Lorentz mixed peak fitting model and an aging response index calculation model to dynamically correlate the microscopic functional group changes with the macroscopic performance indicators, solving the problem that the existing methods cannot accurately track the aging process of the regenerant. This method not only provides a scientific basis for the failure determination of the regenerant, but also establishes a universal technical framework for the performance optimization and screening of different regenerants.

[0142] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered by the scope of the claims and the description of the present invention.

Claims

1. A method for identifying the aging failure behavior of a regenerant based on differential spectrum extraction, characterized in that, The method includes the following steps: Obtain the infrared spectral data of the recycled asphalt sample and the aged asphalt sample at the same stage; Preprocess the infrared spectral data to obtain a preprocessing result, where the preprocessing includes baseline correction and normalization; Based on the preprocessing result, with the aged asphalt spectrum as the background spectrum, construct a characteristic difference spectrum library of the rejuvenator; Based on the characteristic difference spectrum library, extract the change amount of the characteristic peak area of the typical functional groups in the rejuvenator; By establishing a correlation model between the change amount of the characteristic peak area and the macroscopic performance index, obtain the recognition result of the aging failure behavior of the rejuvenator.

2. The identification method for the aging failure behavior of the regenerant based on differential spectrum extraction according to claim 1, wherein The preprocessing of the infrared spectral data to obtain a preprocessing result includes: Use a baseline correction method based on adaptive morphology - quantile regression to perform baseline correction processing on the infrared spectral data to obtain the infrared spectral data after baseline correction processing; Use the peak entropy weighted method to perform normalization processing on the infrared spectral data after baseline correction processing to obtain the infrared spectral data after normalization processing, where the normalization processing includes using an improved Burg entropy criterion to locate the characteristic peaks; Use an improved adaptive iterative reweighted penalty least squares method to perform baseline correction processing on the infrared spectral data after normalization processing to obtain the infrared spectral data after baseline secondary correction processing.

3. The method for identifying the aging failure behavior of a regenerant based on differential spectrum extraction according to claim 1, wherein The constructing of the characteristic difference spectrum library of the rejuvenator with the aged asphalt spectrum as the background spectrum based on the preprocessing result includes: Based on the infrared spectral data after baseline secondary correction processing, with the aged asphalt spectrum as the background spectrum, establish a differential spectrum intensity calculation model; According to the differential spectrum intensity calculation model, obtain differential spectrum data; Based on the differential spectrum data, construct a differential spectrum data enhancement algorithm model; Through the differential spectrum data enhancement algorithm model, obtain the differential spectrum data with enhanced signals; Use the differential spectrum data with enhanced signals to construct a fingerprint matching degree scoring model; Combined with the fingerprint matching degree scoring model, construct the characteristic difference spectrum library of the rejuvenator.

4. The method for identifying the aging failure behavior of a regenerant based on differential spectrum extraction according to claim 3, characterized in that, The differential spectrum intensity calculation model satisfies the following expression: ， Among them, is the differential spectrum intensity, is the spectral intensity of the regenerant at the wavenumber ; is the spectral intensity of the aged asphalt at the wavenumber ; is the dynamic background subtraction factor, is the frequency domain weight function; the differential spectrum data enhancement algorithm model satisfies the following expression: ， Among them, is the enhanced differential spectrum signal, is the differential spectrum intensity, is the convolution operator, is the spectral intensity of the regenerant at the wavenumber ; is the spectral intensity of the aged asphalt at the wavenumber ; is the central wavenumber of the target characteristic peak, is the standard deviation of the Gaussian function; the fingerprint matching degree scoring model satisfies the following expression: ， Among them, is the comprehensive score of the regenerant matching degree, is the enhanced differential spectrum signal, is the spectral feature vector of the standard regenerant in the reference spectral library, is the common characteristic peak area of the differential spectrum and the reference spectrum, is the total characteristic peak area of the differential spectrum and the reference spectrum, and are the integration intervals.

5. A method for identifying the aging failure behavior of a regenerant based on differential spectrum extraction according to claim 1, characterized in that, The extracting of the change amount of the characteristic peak area of the typical functional groups in the rejuvenator based on the characteristic difference spectrum library includes: Based on the characteristic difference spectrum library, establish a Gaussian - Lorentz mixed deconvolution model; Use the Gaussian - Lorentz mixed deconvolution model to fit the characteristic peaks of the typical functional groups in the rejuvenator to obtain a fitting result; Based on the fitting result, establish a characteristic peak area change amount calculation model; Through the characteristic peak area change amount calculation model, extract the change amount of the characteristic peak area of the typical functional groups in the rejuvenator.

6. The method for identifying the aging failure behavior of a regenerant based on differential spectrum extraction according to claim 5, characterized in that, The Gaussian - Lorentz mixed deconvolution model satisfies the following relationship: ， Among them, is the mathematical expression of the differential spectrum of the regenerant characteristics, is the dynamic baseline, is the number of Gaussian peak functions, is the peak height, is the peak center wavenumber, is the Gaussian full width at half maximum, is the Lorentzian full width at half maximum, is the mixing coefficient, is the wavenumber; the calculation model of the change amount of the characteristic peak area satisfies the following relational expression: ， Among them, is the change amount of the characteristic peak area, and are the fitting areas of the regenerant and the aged asphalt at the th characteristic peak, is the reference value of the corresponding peak area of the standard regenerant, is the weight factor of the wave number interval, is the wave number interval.

7. A method for identifying the aging failure behavior of a regenerant based on differential spectrum extraction according to claim 1, wherein The obtaining of the recognition result of the aging failure behavior of the rejuvenator by establishing a correlation model between the change amount of the characteristic peak area and the macroscopic performance index includes: Through the change amount of the characteristic peak area, combined with the macroscopic performance index, establish an aging response index calculation model, where the macroscopic performance index includes penetration, softening point, complex shear modulus, and creep stiffness; the aging response index calculation model satisfies the following relationship: ， Among them, is the aging response index, is the change in the characteristic peak area, is the initial peak area, , , , respectively represent the initial penetration, softening point, complex shear modulus, and creep stiffness, , , , respectively represent the measured performance values of the initial penetration, softening point, complex shear modulus, and creep stiffness after aging, , , , , are the weight coefficients; According to the aging response index calculation model, obtain the aging response index; Based on the aging response index, an identification result of the aging failure behavior of the regenerant is obtained.

8. The identification method for the aging failure behavior of the regenerant based on differential spectrum extraction according to claim 7, wherein The obtaining of the identification result of the aging failure behavior of the regenerant based on the aging response index includes: Based on the aging response index, a coupling correlation coefficient calculation model is established; Through the coupling correlation coefficient calculation model, a coupling correlation coefficient is obtained; Based on the coupling correlation coefficient, the correlation between the aging of the regenerant and the performance degradation is evaluated, and a performance change trend is obtained; Through the aging response index and the performance change trend, a regenerant failure determination model is established; According to the regenerant failure determination model, an identification result of the aging failure behavior of the regenerant is obtained.

9. The method for identifying the aging failure behavior of a regenerant based on differential spectrum extraction according to claim 8, wherein The coupling correlation coefficient calculation model satisfies the following relationship: ， Among them, is the coupling correlation coefficient, is the aging response index of the th sample, is the sample mean of the aging response index, is the performance decay index of the th sample, is the sample mean of the performance decay index, is the number of samples; the regenerant failure determination model satisfies the following expression: ， Among them, is the state of the regenerant, is the critical failure threshold, is the performance retention rate, is the initial performance benchmark, is the material degradation coefficient, is the warning buffer interval, is the time.

10. A regenerant aging failure behavior recognition system based on differential spectrum extraction, the system uses a regenerant aging failure behavior recognition method according to any one of claims 1 to 9, characterized in that, The system includes an input device, a processor, an output device, and a memory. The input device, the processor, the output device, and the memory are interconnected. Among them, the memory is used to store a computer program, and the computer program includes program instructions. The processor is configured to call the program instructions.

Images