A method for evaluating the performance of cold-mixed and cold-laid modified emulsified asphalt mixtures

Abstract

This invention relates to the field of road engineering materials technology and discloses a method for evaluating the performance of cold-mixed modified emulsified asphalt mixtures. The method includes: collecting the preparation and construction parameters of the mixture; classifying the parameters according to preset performance evaluation intervals, extracting material characteristic indicators and construction characteristic indicators within each interval, and generating a set of mixture characteristics; detecting the frequency of evaluation interval switching, and initiating a dynamic correction strategy to mark abnormal evaluation intervals when the frequency exceeds a preset threshold; decomposing the characteristic set of non-abnormal intervals based on the matching degree between historical performance data and current material characteristic indicators, and generating a subset of performance degradation features; correcting the feature subset by combining the coupling relationship between material mechanical parameters and environmental parameters; tracing potential defect factors along the performance degradation path, and generating performance optimization suggestions. This method can improve the comprehensiveness and accuracy of mixture performance evaluation and provide effective guidance for engineering optimization.

Description

Technical Field

[0001] This invention relates to the field of road engineering materials technology, specifically to a method for evaluating the performance of cold-mixed and cold-laid modified emulsified asphalt mixtures. Background Technology

[0002] In the field of road construction and maintenance, cold-mixed modified emulsified asphalt mixtures are widely used in temporary repairs, low-traffic road paving, and emergency projects due to their advantages such as no need for high-temperature heating, low energy consumption, and convenient construction. However, the performance of this type of mixture is significantly affected by multiple parameters in the preparation and construction stages. Fluctuations in these parameters can easily lead to deviations in key performance indicators such as the mixture's strength and durability, thereby affecting the quality and service life of road projects.

[0003] Currently, the industry's evaluation methods for the performance of cold-mixed and cold-laid modified emulsified asphalt mixtures largely rely on static parameter testing and comparison with fixed standards. This involves collecting single preparation or construction parameters and comparing them to preset fixed thresholds to determine whether the mixture's performance meets the standards. This evaluation method has significant limitations: it fails to consider the correlation between preparation and construction parameters, as well as the impact of dynamic factors such as environmental temperature and humidity on the parameters, making it difficult to comprehensively reflect the true performance of the mixture in actual engineering environments. Furthermore, existing evaluation methods lack monitoring and analysis of the switching patterns of performance evaluation intervals. When parameter fluctuations cause frequent switching of evaluation intervals, potential performance anomalies cannot be identified in a timely manner, easily leading to delayed evaluation results.

[0004] In the performance degradation analysis stage, existing technologies often simply compare historical data with current test data without refining the matching degree between material property indicators and historical data. This makes it difficult to accurately extract the subset of performance degradation features. Furthermore, they neglect the coupling relationship between material mechanical parameters and environmental parameters, failing to perform targeted parameter correction for the subset of performance degradation features. This leads to biases when tracing potential defect factors later, and fails to provide accurate guidance for mixture performance optimization. These problems result in insufficient reliability and practicality of current performance evaluation methods for cold-mixed and cold-paved modified emulsified asphalt mixtures, hindering the further promotion and application of this type of mixture in road engineering. Summary of the Invention

[0005] The purpose of this invention is to provide a method for evaluating the performance of cold-mixed and cold-laid modified emulsified asphalt mixtures, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides a method for evaluating the performance of cold-mixed and cold-laid modified emulsified asphalt mixtures, the method comprising:

[0007] The preparation parameters and construction parameters of the cold-mixed and cold-laid modified emulsified asphalt mixture are collected. The preparation parameters include emulsified asphalt content, aggregate gradation and mixing temperature. The construction parameters include paving thickness and ambient temperature and humidity.

[0008] The preparation parameters and construction parameters are classified according to the preset performance evaluation intervals, and the material characteristic indicators and construction characteristic indicators within each performance evaluation interval are extracted to generate a set of mixture characteristics.

[0009] The switching frequency of the performance evaluation interval is detected. When the switching frequency exceeds a preset switching threshold, a dynamic correction strategy is activated to mark the abnormal evaluation interval.

[0010] Based on the matching degree between historical performance data and current material property indicators, the set of mixture characteristics in the non-abnormal evaluation interval is decomposed to generate a subset of performance degradation characteristics.

[0011] The performance degradation feature subset is modified according to the coupling relationship between material mechanical parameters and environmental parameters;

[0012] Based on the modified subset of performance degradation features, potential defect factors are traced back along the performance degradation path to generate performance optimization suggestions.

[0013] Preferably, the collection of the preparation parameters and construction parameters includes:

[0014] Simultaneously record the viscosity of emulsified asphalt, the moisture content of aggregates, and the uniformity of mixing, and align the emulsified asphalt viscosity, the moisture content of aggregates, and the uniformity of mixing with time series.

[0015] The moving average method was used to remove noise data in the viscosity of the emulsified asphalt, and abnormal data points were removed based on the correlation between aggregate moisture content and mixing uniformity.

[0016] The processed data is resampled at preset sampling intervals to generate a standardized parameter dataset.

[0017] Preferably, the extraction of the material property indicators and construction property indicators includes:

[0018] The standardized parameter dataset is divided into low-content, medium-content, and high-content ranges according to the emulsified asphalt content.

[0019] Time-domain and frequency-domain features were extracted for emulsified asphalt viscosity, aggregate gradation, and paving thickness within each content range.

[0020] The time-domain and frequency-domain characteristics within the same content range are integrated into a multi-dimensional set of mixture characteristics according to time windows.

[0021] Preferably, the time-domain characteristics of the emulsified asphalt viscosity include viscosity change rate and fluctuation amplitude, the frequency-domain characteristics of the aggregate gradation include gradation dispersion and main peak frequency, and the time-domain characteristics of the paving thickness include thickness uniformity and settlement rate.

[0022] Preferably, marking the abnormal evaluation interval includes:

[0023] Count the number of switching times within the performance evaluation interval within the preset time window, and calculate the switching frequency per unit time.

[0024] When the switching frequency exceeds the preset switching threshold, the material property indicators of adjacent performance evaluation intervals are matched by a dynamic alignment method, and cross-interval consistency indicators are extracted.

[0025] If the cross-interval consistency index is lower than the preset consistency threshold, the performance evaluation interval corresponding to the current time window will be marked as an abnormal evaluation interval.

[0026] Preferably, extracting the cross-interval consistency index includes:

[0027] Align the emulsified asphalt viscosity change rate and aggregate gradation dispersion of adjacent performance evaluation intervals;

[0028] Calculate the correlation coefficient between the viscosity change rate and the gradation dispersion after alignment, and use the mean of the correlation coefficient as the consistency index across intervals.

[0029] Preferably, generating the subset of performance degradation features includes:

[0030] Historical degradation stages are defined based on crack development rate and strength decay trend in historical performance data.

[0031] Match the current material property indicators with the similarity to the historical degradation stage;

[0032] If the similarity is greater than or equal to a preset matching threshold, then a subset of low-frequency performance degradation features is extracted from the mixture characteristic set.

[0033] If the similarity is less than the preset matching threshold, then the feature enhancement decomposition interval is selected based on the correlation between the harmonic components of aggregate gradation and mechanical parameters.

[0034] Preferably, parameter correction of the subset of performance degradation features includes:

[0035] Based on the correlation between the elastic modulus of asphalt binder and the viscosity change rate, the viscosity fluctuation amplitude in the performance degradation feature subset is corrected.

[0036] Based on the matching relationship between ambient temperature and humidity and the settlement rate of paving thickness, the thickness uniformity index is compensated and corrected.

[0037] The corrected viscosity fluctuation amplitude and thickness uniformity index are combined into a parameter coupling correction result.

[0038] Preferably, tracing the potential defect factors includes:

[0039] Construct a performance degradation map that includes material defect propagation paths and construction defect propagation paths;

[0040] Identify anomalous indicators from the revised subset of performance degradation features and match them with corresponding nodes in the performance degradation map;

[0041] Traverse backwards along the defect propagation direction to locate the root defect node of the abnormal indicator.

[0042] Performance optimization suggestions are generated based on the hierarchical depth of the root cause defect node and the number of abnormal indicators.

[0043] Preferably, generating the performance optimization suggestions includes:

[0044] For nodes with abnormal emulsified asphalt content in the root cause defect nodes, it is recommended to adjust the emulsified asphalt blending ratio.

[0045] For nodes with abnormal paving thickness among the root cause defect nodes, it is recommended to optimize the paver travel speed and vibration frequency parameters;

[0046] The adjustment suggestions are sorted by node level and output as a structured optimization scheme.

[0047] Compared with the prior art, the beneficial effects of the present invention are:

[0048] This performance evaluation method for cold-mixed and cold-laid modified emulsified asphalt mixtures overcomes the limitations of traditional evaluation methods that rely on single-parameter detection by systematically collecting preparation and construction parameters. It comprehensively captures key factors affecting mixture performance. Integrating preparation and construction parameters into a unified evaluation system fully considers the interactions between parameters, avoiding biased evaluations caused by isolated parameter analysis, and making the evaluation results more closely aligned with the actual application scenarios of the mixture in engineering projects.

[0049] In the parameter classification and characteristic set generation stage, parameters are classified according to preset performance evaluation intervals, and material characteristic indicators and construction characteristic indicators are extracted to form a mixture characteristic set, which enables stratified and refined characterization of mixture performance. This classification method can clearly present the performance status of mixtures under different parameter combinations, providing a structured data foundation for subsequent performance analysis, and facilitating a more intuitive identification of the differences and patterns in mixture performance within different parameter intervals;

[0050] The activation of the monitoring and dynamic correction strategy for the switching frequency of performance evaluation intervals effectively solves the problem of the inability to identify performance anomalies in a timely manner in traditional evaluation methods. When the switching frequency exceeds a preset threshold, the abnormal evaluation interval can be marked to quickly identify performance risk points caused by parameter fluctuations, prevent abnormal states from continuously affecting the performance of the mixture, improve the timeliness of the evaluation process and the ability to warn of risks, and reduce potential engineering quality hazards caused by performance anomalies.

[0051] Based on a decomposition method that matches historical performance data with current material property indicators, a subset of performance degradation features is generated within the non-abnormal evaluation interval, enabling precise correlation between historical and current data. By filtering historical data with high correlation to the current mixture performance through matching degree analysis, the accuracy of performance degradation feature extraction can be improved, avoiding misjudgments of degradation features due to data mismatch, and providing reliable feature basis for subsequent defect factor tracing.

[0052] By combining the coupling relationship between material mechanical parameters and environmental parameters, parameter correction is applied to the performance degradation feature subset, fully considering the objective law that the performance of the mixture is affected by both mechanical properties and environmental factors. This correction method can eliminate the interference of environmental parameter fluctuations on the mechanical parameter test results, making the performance degradation feature subset more realistically reflect the performance degradation trend of the mixture itself, reducing the impact of external factors on the evaluation results, and improving the reliability of performance analysis.

[0053] By tracing the performance degradation path backward to identify potential defect factors and generating performance optimization suggestions, performance evaluation is closely integrated with actual engineering optimization. By accurately locating the defect factors leading to performance degradation, targeted suggestions for adjusting preparation or construction parameters can be proposed, avoiding the blindness of traditional optimization processes and making optimization measures more operable. This helps improve the performance stability of cold-mixed and cold-laid modified emulsified asphalt mixtures, extend the service life of road projects, reduce engineering maintenance costs, and promote the efficient application of this type of mixture in the road engineering field. Attached Figure Description

[0054] Figure 1 This is a schematic diagram illustrating the working principle of the method for evaluating the performance of cold-mixed and cold-laid modified emulsified asphalt mixtures as described in this invention.

[0055] Figure 2 A flowchart illustrating the method for preparing and collecting construction parameters of cold-mixed modified emulsified asphalt mixtures;

[0056] Figure 3 A flowchart illustrating the method for extracting characteristic indicators of cold-mixed and cold-laid modified emulsified asphalt mixtures. Detailed Implementation

[0057] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0058] Please see Figure 1 This invention provides a method for evaluating the performance of cold-mixed, cold-laid modified emulsified asphalt mixtures. The method includes systematically collecting and processing the preparation and construction parameters of the mixture to achieve accurate evaluation and optimization of its performance. Preparation and construction parameters of the cold-mixed, cold-laid modified emulsified asphalt mixture are collected. Preparation parameters include emulsified asphalt content, aggregate gradation, and mixing temperature; construction parameters include paving thickness and ambient temperature and humidity. These parameters are standardized and categorized according to preset performance evaluation intervals, thereby extracting material and construction characteristic indicators within each interval and generating a mixture characteristic set. The switching frequency of the performance evaluation intervals is detected. When the switching frequency exceeds a preset switching threshold, a dynamic correction strategy is initiated to mark abnormal evaluation intervals. For non-abnormal evaluation intervals, the mixture characteristic set is decomposed based on the matching degree between historical performance data and current material characteristic indicators to generate a performance degradation feature subset. Next, based on the coupling relationship between material mechanical parameters and environmental parameters, the performance degradation feature subset is parameter-corrected to eliminate data bias. Based on the modified subset of performance degradation features, potential defect factors are traced back along the performance degradation path, and targeted performance optimization suggestions are generated, thereby improving the overall performance reliability of the mixture.

[0059] Example 1: See Figure 2 When collecting preparation and construction parameters for cold-mixed modified emulsified asphalt mixtures, the process begins with the simultaneous recording of several key data points. These data specifically include the viscosity of the emulsified asphalt, the moisture content of the aggregates, and the mixing uniformity of the mixture. All parameters are accompanied by high-precision timestamps during collection. Time series alignment is the first step in achieving effective data fusion. It establishes a unified timeline to accurately match data points from different sensors or manual records. For example, it correlates the viscosity value recorded at a certain moment with the moisture content value measured at the same moment. This process needs to address potential time drift or sampling delays between devices. Linear interpolation is typically used to align non-uniformly sampled data points onto a unified time grid.

[0060] The moving average method is used to process emulsified asphalt viscosity data to suppress unavoidable random noise interference during the measurement process. The size of the moving average window needs to be reasonably selected based on the data acquisition frequency and the physical characteristics of viscosity changes. A larger window can more effectively smooth high-frequency fluctuations but may also mask the true rapid changing trend, while a smaller window retains more details but has limited noise reduction effect. When removing outlier data points, it is necessary to deeply analyze the intrinsic relationship between aggregate moisture content and mixing uniformity. For example, when the moisture content is abnormally high, the mixing uniformity index may decrease significantly due to aggregate agglomeration. By calculating the statistical correlation between these two parameter sequences, data points that deviate from the main trend can be identified and treated as invalid values ​​for removal. After noise removal and outlier removal, the data resampling step converts the non-equal interval or high-frequency raw data into a sequence of standard time intervals. This preset sampling interval should comprehensively consider the real-time requirements of the evaluation and the burden of data processing. The resampling process usually uses an anti-aliasing filter to prevent frequency aliasing and generate the final standardized parameter dataset.

[0061] The process of extracting material and construction characteristic indicators is based on a standardized parameter dataset. The first step is to divide the entire dataset into three subsets according to the numerical range of emulsified asphalt content: low-content, medium-content, and high-content ranges. The threshold values ​​for these ranges are not fixed but are determined by the material design specifications and historical experience data for specific mixtures. For example, in some engineering applications, the low-content range might be defined as below 4%, while the high-content range might be defined as above 8%. The purpose of this division is to categorize data with similar material properties for more targeted analysis. For each defined content range, the time-domain and frequency-domain characteristics of the included emulsified asphalt viscosity, aggregate gradation, and paving thickness data need to be extracted. Time-domain characteristic analysis focuses on the direct statistical properties of parameters changing over time, such as calculating the mean and variance of viscosity to understand its central tendency and dispersion, and calculating the rate of change to capture its dynamic behavior. Frequency-domain characteristic analysis uses mathematical transformations to convert the signal from the time dimension to the frequency dimension, thereby revealing its periodic fluctuation patterns. For example, spectral analysis of the aggregate gradation data sequence can identify the dominant fluctuation frequency components. After feature extraction is completed, the time domain and frequency domain features of all parameters within the same content range need to be integrated according to a preset time window. The selection of this time window should reflect a complete construction operation cycle or a meaningful performance evolution stage. The result after integration is a multi-dimensional set of mixture characteristics. Each data point in this set represents the comprehensive state of the mixture within a specific time window and content range.

[0062] In the low asphalt content range, the viscosity behavior of emulsified asphalt may exhibit different characteristics compared to the medium-to-high content range. Its viscosity change rate may be more sensitive to small fluctuations in aggregate moisture content. Therefore, when extracting time-domain features in this range, special attention needs to be paid to the calculation accuracy of the change rate, potentially requiring a finer differential step size. The frequency domain characteristics of aggregate gradation may show higher dispersion in the low-content range because insufficient asphalt content can lead to uneven aggregate coating, resulting in richer frequency components in the gradation detection signal. Identifying the dominant frequency in this case helps determine whether there is a periodic pattern in the gradation unevenness. The time-domain characteristics of paving thickness, such as thickness uniformity, need to be evaluated in the low-content range using a longer observation window. This is because lower asphalt content may result in poorer initial stability of the mixture, and thickness settlement is a relatively slow process. Calculating the settlement rate requires a sufficiently long time series to obtain a reliable trend.

[0063] In the medium content range, the viscosity of emulsified asphalt is generally within a relatively stable range with small fluctuations. Therefore, the extracted time-domain characteristics are more likely to reflect fluctuations in the construction process than the inherent instability of the material itself. The frequency domain characteristics of aggregate gradation typically exhibit low dispersion in this range, with a more prominent peak frequency, indicating optimal gradation control. The focus of frequency domain analysis is monitoring for shifts in the peak frequency, which could be an early signal of changes in aggregate supply or mixing processes. The uniformity of paving thickness is crucial for evaluating construction quality in this range. Due to the relatively balanced material properties, any significant non-uniformity in thickness is more likely to be directly attributed to the paver's operating parameters or the smoothness of the base layer. Therefore, the calculation of the settlement rate needs to be correlated with the paver's real-time operating parameters, such as travel speed and vibration frequency.

[0064] In the high asphalt content range, extracting the temporal characteristics of emulsified asphalt viscosity faces new challenges. High asphalt content may cause viscosity measurements to tend towards their upper limit, and dynamic range compression necessitates higher precision in calculating the rate of change and fluctuation amplitude. Methods such as logarithmic transformation may be needed to enhance sensitivity to changes. The frequency domain characteristics of aggregate gradation may exhibit unique patterns in this range. High asphalt content may mask some gradation defects, artificially reducing frequency dispersion, but it may also reveal abnormal peaks in certain high-frequency bands, indicating localized asphalt enrichment or fiber agglomeration. The settlement rate of paving thickness is typically slower in the high asphalt content range because abundant asphalt provides better lubrication and adhesion. Evaluation of thickness uniformity should focus on lateral uniformity rather than longitudinal uniformity to prevent excessively thick asphalt layers from causing later-stage rutting and other defects.

[0065] When integrating features from different content ranges into a multidimensional feature set, a unified data structure needs to be established. This structure must be able to accommodate all feature variables extracted from the time and frequency domains, and label each variable with a time window and content range. This multidimensional set constitutes the foundational data model for subsequent performance evaluation and analysis, and its quality directly affects the accuracy and reliability of a series of subsequent steps, such as anomaly interval identification and performance degradation feature extraction. The entire implementation process emphasizes the synchronicity of data acquisition, the standardization of processing, and the targeted nature of feature extraction, ensuring a scientific and effective transformation from raw parameters to evaluation indicators through a systematic process.

[0066] Example 2: See Figure 3 Regarding the extraction of the time-domain characteristics of emulsified asphalt viscosity, the viscosity change rate is a dynamic indicator. It is calculated by analyzing the viscosity measurements at consecutive time points to determine the relative magnitude of the change. In practice, the ratio of the viscosity difference between adjacent time points to the time interval is used to characterize the severity of the change. The fluctuation amplitude reflects the overall oscillation range of the viscosity data over a certain period. It is necessary to identify local maxima and minima from the preprocessed viscosity time series and use their difference to quantify the degree of fluctuation. Obtaining the frequency domain characteristics of aggregate gradation requires treating the gradation data as a discrete signal. The gradation dispersion is calculated by analyzing the uniformity of the distribution of different sieve apertures, reflecting the concentration or dispersion trend of aggregate particle distribution. The main peak frequency needs to be determined by spectral analysis of the gradation data sequence. The calculation process involves converting the time-domain gradation percentage sequence to the frequency domain using a fast Fourier transform, and then finding the frequency components with the most concentrated energy in the spectrum. The thickness uniformity index in the time-domain characteristics of paving thickness is obtained by calculating the statistical coefficient of variation of thickness data from multiple measuring points. This coefficient is the ratio of the standard deviation to the mean, which can eliminate the influence of absolute thickness on uniformity evaluation. The calculation of settlement rate requires continuous monitoring of the thickness value at the same location. The settlement amount per unit time is obtained by linearly fitting the slope of the curve of thickness changing with time.

[0067] When extracting viscosity change rate, the choice of time interval is crucial. Too short an interval will result in the change rate being overly affected by random measurement noise, while too long an interval may mask the true dynamic change process. A reasonable time step is usually determined by comprehensively considering the rhythm of the mixing process and the frequency of data acquisition. When calculating the fluctuation amplitude, a complete fluctuation period needs to be clearly defined. This can be achieved by identifying inflection points in the time series or using peak detection algorithms. For non-periodic irregular fluctuations, statistical methods such as calculating the root mean square value are sometimes used to characterize the equivalent fluctuation amplitude. The calculation of gradation dispersion relies on complete gradation composition data. Typically, it is necessary to obtain the passing rates of key sieve openings such as 0.075mm, 2.36mm, 4.75mm, and 9.5mm, and then analyze the deviation of these passing rate values ​​from the design gradation median. The dispersion index can be expressed in the form of standard deviation or coefficient of variation. Identifying the main peak frequency requires spectral analysis of the gradation data sequence. This requires the data to have sampling characteristics at equal time intervals. Therefore, it may be necessary to resample the original data before analysis. After spectral analysis, it is necessary to find the frequency component with the largest amplitude within the effective frequency range. This component often corresponds to a certain periodic gradation fluctuation during mixing or transportation.

[0068] Assessing thickness uniformity requires setting up multiple measuring points on the paved surface. The number and layout of these points should be representative of the entire paved surface, typically arranged in a longitudinal and transverse grid pattern. When calculating uniformity, not only the overall coefficient of variation must be considered, but sometimes it's also necessary to analyze the differences in uniformity across different directions. Measuring the settlement rate requires thickness monitoring equipment with sufficient accuracy and stability, as the settlement may be very small initially. During measurement, interference from environmental factors such as paver vibration and temperature changes needs to be eliminated. This is usually achieved by setting benchmark points and using differential measurement methods to improve accuracy. For calculating viscosity change rate, data gaps or anomalies may occur in actual engineering projects. In such cases, data interpolation methods are needed to avoid significant deviations in feature extraction. Calculating fluctuation amplitude sometimes requires distinguishing between short-term fluctuations and long-term trends. High-pass filtering can be used to remove trend terms before calculating fluctuation characteristics, resulting in purer amplitude information representing short-term instability. The analysis of gradation dispersion needs to be considered in conjunction with the specific material type. The normal range of gradation dispersion may differ for mixtures with different maximum nominal particle sizes. Therefore, dispersion indices usually need to be compared with empirical thresholds or historical data to be meaningful. The physical meaning of the main peak frequency needs to be interpreted in conjunction with the actual process parameters. For example, the main peak frequency may be related to the vibration frequency of the mixing plant screening equipment and the feeding cycle of the cold material silo, which helps to trace the cause of gradation fluctuations.

[0069] Modern thickness uniformity measurement often employs non-contact laser or ultrasonic thickness measurement technologies, enabling continuous acquisition of thickness data over large areas. The monitoring time for settling rate needs to be determined based on the type of mixture and environmental conditions. For emulsified asphalt mixtures, the initial settling rate is faster, gradually decreasing later; therefore, the rate calculation needs to explicitly specify the corresponding time period. The implementation of these feature extraction processes requires corresponding hardware measurement equipment and software algorithm support. Viscosity measurement typically uses an online rotational viscometer system, gradation detection can use image-recognition-based particle analyzers or laser particle size analyzers, and thickness measurement commonly uses laser rangefinders or ultrasonic thickness gauges. The extracted time-domain and frequency-domain features are ultimately stored in a structured data format, with each feature carrying a timestamp and corresponding process parameter information, providing input data for subsequent performance evaluation and anomaly diagnosis. The accuracy and reliability of feature extraction directly affect the effectiveness of the entire performance evaluation system; therefore, in practical applications, it is necessary to periodically calibrate the measuring instruments and verify the feature extraction algorithm.

[0070] Example 3: The process of marking abnormal evaluation intervals begins with the statistical analysis of the number of performance evaluation interval switching times within a preset time window. The length of this time window needs to be reasonably set according to the data acquisition density and evaluation cycle of the specific project. For example, in continuous paving operations, half an hour may be selected as a statistical window. The switching frequency per unit time is obtained by calculating the ratio of the number of interval switching times occurring within the window to the window duration. Its mathematical expression can be written as:

[0071]

[0072] Where: symbol The calculated switching frequency represents the average number of times the performance evaluation interval changes per unit time; the symbol... Indicates the selected time window The total number of interval switching events obtained from internal statistics. Time window. The value of directly affects the sensitivity of frequency calculation. If the window is too short, the frequency value will be too affected by occasional and short-term interval fluctuations. If the window is too long, it will smooth out meaningful frequent switching signals. Therefore, calibration is required based on historical data analysis and engineering experience.

[0073] When the calculated switching frequency When the preset switching threshold is exceeded, the system will trigger the dynamic correction strategy. This threshold is an empirical critical value used to distinguish between normal interval evolution and abnormal frequent jumps. The implementation of the dynamic alignment method is a key step in matching material property indicators within two adjacent performance evaluation intervals. This method needs to handle the time scaling and phase shift issues that may exist between data in different intervals. The dynamic time warping algorithm aligns two time series by finding an optimal bending path to minimize the cumulative distance between corresponding points in the series. The alignment operation mainly targets indicators with time series characteristics, such as continuous monitoring values ​​of emulsified asphalt viscosity and periodic detection values ​​of aggregate gradation. The aligned data series have the same number of time points, and the corresponding points represent the same process time. The core work of extracting cross-interval consistency indicators is to calculate the correlation between the key parameters after alignment. Specifically, it involves using the emulsified asphalt viscosity change rate series and aggregate gradation dispersion series within adjacent intervals that have already undergone time alignment as the analysis objects. The correlation coefficient is calculated using statistical methods to measure the degree of linear association between two sequences. The coefficient value ranges from negative to positive one; the closer the absolute value is to one, the higher the consistency. The calculation is performed on each pair of adjacent intervals, yielding multiple correlation coefficient values. The final cross-interval consistency indicator used for judgment is the arithmetic mean of these correlation coefficients. This mean comprehensively reflects the synchronicity and stability of indicator changes across multiple interval boundaries.

[0074] If the calculated cross-interval consistency index is lower than the preset consistency threshold, the performance evaluation interval corresponding to the current time window is determined to be an abnormal evaluation interval. The consistency threshold setting needs to consider the material characteristics and measurement uncertainties. A standard that is too low will fail to effectively capture anomalies, while a standard that is too high may lead to misjudgment. Marking abnormal intervals typically involves adding anomaly tags or metadata to the data for that time period in the data management system, so that it can be excluded from the normal performance degradation model construction process in subsequent analysis steps, or given special attention and processing. In the specific operation of counting the number of switches, it is necessary to clearly define what constitutes a "switch." This usually refers to a change in the combination of characteristic indicators of the mixture from one preset evaluation interval range to another different interval range, such as a sudden change in the emulsified asphalt content from the medium content range to the high content range. The statistical process needs to traverse all data records within the entire selected time window, accurately identifying the start and end points of each interval change, thereby accurately counting. When executing the dynamic alignment method, it is necessary to select appropriate local path constraints for the two indicator sequences to be matched. These conditions limit the curvature of the sequences on the time axis to avoid producing unrealistic alignment results. Alignment quality can be assessed by calculating the residual distance or correlation between aligned sequences to ensure the effectiveness of the match.

[0075] When calculating the correlation coefficient between viscosity change rate and gradation dispersion, it is necessary to check whether the data sequence satisfies the assumption of linear correlation. If necessary, data transformation (such as taking the logarithm) may be required to improve the linear relationship. For cases with obvious nonlinear relationships, nonlinear correlation measures such as mutual information may be used as supplementary judgment criteria. The determination of the preset consistency threshold often relies on the post-hoc analysis of a large amount of normal construction historical data. By statistically analyzing the distribution range of consistency indicators under normal operating conditions, a specific quantile (such as the 5th percentile) is taken as the threshold, so that situations below this threshold can be considered low-probability events, i.e., abnormal situations. After marking abnormal intervals, the system usually records detailed information about the anomaly, including the time of occurrence, the type of performance evaluation interval involved, the calculated consistency index value, and related process parameters. This information is crucial for subsequent analysis of the root cause of the anomaly and optimization of process control. The entire implementation process emphasizes capturing the dynamic characteristics of the data and judging its statistical significance. Abnormal states are objectively identified through quantified frequency and consistency indicators, reducing the bias that may be caused by subjective experience judgment.

[0076] Example 4: The process of generating a subset of performance degradation features relies on a long-term accumulated historical performance database. This database typically contains sequences of performance indicators such as crack width, flexural strength, or resilient modulus of the mixture over time, recorded under various working conditions. The primary task in dividing historical degradation stages is to perform morphological analysis on these historical performance data curves to identify key inflection points in performance degradation. For example, a typical crack development curve may exhibit an initial slow growth stage, a subsequent linear stable expansion stage, and a final accelerated failure stage. The stage division is based on the significant change in the performance degradation rate. When matching the similarity between current material property indicators and historical degradation stages, it is necessary to construct a feature vector from indicators such as emulsified asphalt content, aggregate gradation, and viscosity obtained within the current monitoring period. This feature vector is compared with the standard feature template of each degradation stage in the historical database. Similarity calculation can be performed by measuring the distance between vectors; the closer the distance, the higher the similarity. Assuming the key indicators currently monitored, refer to Table 1.

[0077] Table 1: Material property indicators for the current monitoring period

[0078] Monitoring time points Emulsified asphalt content (%) 2.36mm sieve aperture passing rate (%) Average viscosity Viscosity fluctuation amplitude Day 1 6.5 45.2 1.25 0.08 Day 2 6.3 46.1 1.18 0.12 Day 3 6.7 44.8 1.30 0.15 Day 4 6.2 47.5 1.15 0.20 Day 5 6.4 45.9 1.22 0.18

[0079] If the calculated similarity is greater than or equal to the preset matching threshold, it means that the current behavior of the mixture highly matches a known historical degradation pattern. The system will then extract low-frequency features that change slowly and dominate the long-term trend from the current complete set of mixture characteristics. The extraction of low-frequency features relies on signal processing techniques. Short-term fluctuation noise is removed through wavelet transform or high-pass filters, retaining components that can reflect the essential trend of performance degradation. These components constitute a subset of performance degradation features.

[0080] If the similarity between the current indicator and historical stages is below a preset threshold, it indicates a possible new or atypical degradation pattern, requiring the activation of alternative strategies. This strategy focuses on analyzing the correlation between the periodic harmonic components inherent in the aggregate gradation sequence and mechanical parameters (such as compressive strength and modulus). By analyzing the strength of this correlation, the characteristic bands most sensitive to performance evolution are selected as the intervals for enhanced decomposition. Parameter correction of the performance degradation characteristic subset is to compensate for measurement errors and environmental interference. The correction process is based on established physical empirical relationships. The first step is to calibrate the viscosity fluctuation amplitude contained in the subset based on the correlation curve between the elastic modulus of asphalt binder determined by indoor tests and the viscosity change rate measured in the field. For example, when the measured value of the elastic modulus is too high, it may indicate that the asphalt aging degree is accelerated, and the corresponding viscosity fluctuation amplitude may need to be corrected upwards according to the correlation.

[0081] The next step is to compensate based on the matching relationship between ambient temperature and humidity and the settlement rate of paving thickness. This relationship is usually obtained through regression analysis of a large amount of field data, and the current actual ambient temperature and humidity data must be input during correction. When compensating for thickness uniformity, for example, in high-temperature and low-humidity environments, rapid evaporation of moisture in the mixture may lead to premature formation of a hard crust on the surface, causing distortion of the thickness uniformity measurement. In this case, it is necessary to perform reverse compensation on the thickness uniformity index according to the degree of deviation of temperature and humidity from standard conditions. Combining the corrected viscosity fluctuation amplitude with the thickness uniformity index into a parameter-coupled correction result requires a data fusion method. During merging, the weights of the influence of different parameters on the overall performance must be considered. The weight coefficients can be determined based on parameter sensitivity analysis or expert experience. The final correction result is a feature subset that more accurately reflects the material state.

[0082] When delineating historical degradation stages, massive amounts of historical data need to be processed. This necessitates effective feature extraction and dimensionality reduction of historical performance curves. For example, principal component analysis can be used to transform lengthy performance time series into a few representative feature indicators, thus simplifying the stage delineation process. The process of matching similarity often faces computational efficiency issues due to the large historical database. In practical applications, cluster analysis may be used to first classify historical degradation stages, establishing a limited library of typical degradation patterns. Then, the current data is matched against these typical patterns to improve computational speed. When extracting low-frequency dominant features, the choice of wavelet basis functions and the setting of the decomposition level significantly affect the results. Different basis functions are suitable for capturing signals with different characteristics, and their selection needs to be based on the characteristics of the mixture performance data. Selecting feature-enhanced decomposition intervals based on the correlation between harmonic components and mechanical parameters is an experimental method. Its effectiveness depends on the accuracy of the correlation model; therefore, continuous validation and updating of the correlation model with new data is necessary.

[0083] The physical empirical relationships relied upon in the parameter correction process are not static. The properties of asphalt binder may vary depending on the oil source and production process, and the relationship between environmental temperature and humidity and thickness settlement also varies depending on regional climate and base course conditions. Therefore, the correlation model used for correction needs to have a mechanism for regular updates. The determination of the correction amount involves a certain degree of uncertainty. In practice, interval estimation or fuzzy logic methods are sometimes used to handle this uncertainty, rather than providing a fixed correction value. When merging the corrected parameters, the allocation of weights is a problem that requires careful consideration. Fixed weights may not be suitable for all working conditions. Sometimes, it is necessary to dynamically adjust the weights according to the current degradation stage or environmental conditions, which makes the merging algorithm more complex but more realistic. The entire implementation process embodies the idea of ​​learning from historical experience and adaptively adjusting to the current situation. Through dynamic feature extraction and parameter correction, the performance evaluation is made more closely reflect the true state of the mixture. The data table in the example shows the monitoring values ​​for five consecutive days. The monitoring period and the number of indicators may be extended and increased according to project needs. The core is to transform the raw data into reliable degradation characteristic information through a systematic process.

[0084] Example 5: The construction of the map relies on the integrated analysis of numerous historical engineering cases, systematic laboratory test results, and materials science theoretical knowledge. Nodes in the map represent specific defect states, such as "low emulsified asphalt content," "aggregate gradation segregation," or "uneven paving thickness." The directed edges connecting the nodes represent the possible initiation or aggravation relationships between defects, and the direction of the edges indicates the transmission path of the defect's influence. A completed map typically contains two main paths: the material defect transmission path and the construction defect transmission path. The material path may begin with "unstable emulsified asphalt formulation," leading to "excessive asphalt viscosity fluctuations," ultimately resulting in "insufficient mixture adhesion." The construction path may start with "paver baseline calibration error," leading to "uneven initial paving thickness," and subsequently causing "density differences after compaction." The level of detail in the map determines the accuracy of tracing. A complete map defines the hierarchy of nodes, with root causes located in the upper layers and manifested end defects in the lower layers. Identifying anomalous indicators from a subset of performance degradation features after parameter correction is the starting point for tracing. The identification process relies on preset thresholds or statistical control limits. For example, when the corrected viscosity fluctuation amplitude exceeds the upper limit of its historical normal fluctuation range for multiple consecutive monitoring periods, the indicator is marked as anomalous. Matching the corresponding node in the performance degradation map is achieved through a search algorithm. The system compares the name or code of the anomalous indicator with the attributes of all nodes in the map to locate the node representing the anomalous phenomenon. For example, high viscosity fluctuation amplitude may match the "excessive asphalt viscosity fluctuation" node in the map.

[0085] The core logic for locating the root cause of a problem is to traverse backwards along the defect propagation direction. The traversal starts from the matched end node and explores upstream nodes layer by layer along the opposite direction of the directed edges (i.e., the opposite direction indicated by the arrows), checking whether the state of each parent node also shows signs of anomaly. During the traversal, the activation probability or anomaly contribution of nodes on the path is calculated. If a node has multiple paths pointing to it, or multiple downstream child nodes are simultaneously anomalous, the probability that the node is the root cause defect increases significantly. After locating the root cause defect node, it is necessary to record the node's type, its level depth in the graph, and the number of all its connected child nodes in an anomalous state. Generating performance optimization suggestions is a process of transforming diagnostic conclusions into specific measures. For the type of root cause defect node located, the system will call preset adjustment schemes from the knowledge base. For the "abnormal emulsified asphalt content" node identified as the root cause, the knowledge base may have pre-stored suggestions for different abnormal situations. For example, when the content is consistently low, the suggestion will clearly indicate that the specific range of increase in the emulsified asphalt blending ratio needs to be calculated based on the deviation between the current mix proportion and the design requirements, and may be related to the adjustment settings of the mixing plant control parameters.

[0086] For construction-related root cause issues like "abnormal paving thickness," the generated recommendations will be more specific, focusing on the operation of mechanical equipment. For example, it might suggest optimizing the paver's travel speed to make it more uniform and stable, and adjusting the screed's vibration frequency and amplitude parameters accordingly to improve the initial density and thickness uniformity of the mixture. The generation of recommendations needs to consider the node level. For defects at deeper levels (i.e., closer to the root cause), the optimization measures given are usually more fundamental and important; for defects at shallower levels, the measures may focus more on temporary adjustments or process control. Multiple adjustment recommendations are sorted by node level and output as a structured optimization scheme, ensuring the logic and operability of the recommendations. The sorting rule usually prioritizes recommendations for root cause defects, while also considering the severity of the defect and the urgency of repair. The structured scheme can be presented in the form of lists, tree diagrams, or flowcharts. Each recommendation includes a defect description, root cause analysis, specific optimization measures, and expected adjustment goals. After the scheme is output, it can be directly used by field engineers to guide process adjustments.

[0087] In the practical application of graph construction, it is necessary to continuously validate and update the graph with new case data to enhance its accuracy and coverage. For example, when a new abnormal correlation pattern between aggregate moisture content and mixing uniformity is discovered, corresponding nodes and connecting edges need to be added to the graph. When identifying abnormal indicators, the scientific nature of threshold setting is crucial. Static thresholds may not adapt to changes in working conditions; sometimes dynamic thresholds or adaptive control chart methods are required to make anomaly judgments more reflective of the current process's true state. The reverse traversal algorithm needs to handle complex situations such as loops and multiple connections to avoid getting stuck in cycles or missing important paths during the tracing process. The algorithm design must ensure that all possible upstream nodes can be traversed. The knowledge base for generating optimization suggestions needs to be synchronized with the latest construction specifications, material research results, and equipment operation manuals to ensure the technical rationality and advancement of the suggestions. The presentation of structured optimization solutions needs to consider user habits; clear formats and explicit priority markings help on-site personnel quickly understand and implement them.

[0088] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0089] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended related limitations and their equivalents.

Claims

1. A method for evaluating the performance of cold-mixed and cold-laid modified emulsified asphalt mixture, characterized in that, The method comprises the following steps: Collecting preparation parameters and construction parameters of cold-mixed and cold-laid modified emulsified asphalt mixture, the preparation parameters including emulsified asphalt content, aggregate gradation and mixing temperature, and the construction parameters including paving thickness and environmental temperature and humidity; Classifying the preparation parameters and construction parameters according to preset performance evaluation intervals, extracting material characteristic indexes and construction characteristic indexes in each performance evaluation interval, and generating a mixture characteristic set; Detecting the switching frequency of the performance evaluation intervals, and starting a dynamic correction strategy to mark an abnormal evaluation interval when the switching frequency exceeds a preset switching threshold; Decomposing the mixture characteristic set of a non-abnormal evaluation interval based on the matching degree of historical performance data and current material characteristic indexes, and generating a performance degradation feature subset; Parameter correcting the performance degradation feature subset according to the coupling relationship between material mechanical parameters and environmental parameters; Generating performance optimization suggestions by reversely tracing potential defect factors along a performance degradation path based on the corrected performance degradation feature subset.

2. The method for evaluating the performance of cold-recycled modified emulsified asphalt mixture according to claim 1, characterized in that, The collecting of the preparation parameters and construction parameters comprises: Synchronously recording emulsified asphalt viscosity, aggregate moisture content and mixing uniformity, and time series aligning the emulsified asphalt viscosity, aggregate moisture content and mixing uniformity; Removing noise data in the emulsified asphalt viscosity by using a sliding average method, and eliminating abnormal data points based on the correlation between the aggregate moisture content and the mixing uniformity; Resampling the processed data at a preset sampling interval to generate a standardized parameter data set.

3. The method for evaluating the performance of cold-recycled modified emulsified asphalt mixture according to claim 1, characterized in that, The extracting of the material characteristic indexes and the construction characteristic indexes comprises: Dividing the standardized parameter data set into a low content interval, a medium content interval and a high content interval according to the emulsified asphalt content; Extracting time domain features and frequency domain features of the emulsified asphalt viscosity, aggregate gradation and paving thickness in each content interval, respectively; Integrating the time domain features and the frequency domain features in the same content interval into a multi-dimensional mixture characteristic set according to a time window.

4. The method for evaluating the performance of cold-recycled modified emulsified asphalt mixture according to claim 3, characterized in that, The time domain features of the emulsified asphalt viscosity include a viscosity change rate and a fluctuation amplitude, the frequency domain features of the aggregate gradation include a gradation dispersion and a main peak frequency, and the time domain features of the paving thickness include a thickness uniformity and a settlement rate.

5. The method for evaluating the performance of cold-recycled modified emulsified asphalt mixture according to claim 1, characterized in that, The marking of the abnormal evaluation interval comprises: Counting the switching frequency in a preset time window, and calculating the switching frequency per unit time; When the switching frequency exceeds a preset switching threshold, matching the material characteristic indexes of adjacent performance evaluation intervals by a dynamic alignment method, and extracting an inter-interval consistency index; If the inter-interval consistency index is lower than a preset consistency threshold, marking the performance evaluation interval corresponding to the current time window as an abnormal evaluation interval.

6. The method for evaluating the performance of cold-recycled modified emulsified asphalt mixture according to claim 5, characterized in that, The extracting of the inter-interval consistency index comprises: Aligning the emulsified asphalt viscosity change rate and the aggregate gradation dispersion of adjacent performance evaluation intervals; Calculating the correlation coefficient of the aligned viscosity change rate and the gradation dispersion, and taking the average of the correlation coefficient as the inter-interval consistency index.

7. The method for evaluating the performance of cold-recycled modified emulsified asphalt mixture according to claim 1, characterized in that, The generating of the performance degradation feature subset comprises: Dividing historical degradation stages according to the crack development rate and the strength attenuation trend in historical performance data; Matching the similarity of the current material characteristic indexes and the historical degradation stages; If the similarity is greater than or equal to a preset matching threshold, a low-frequency dominant performance degradation feature subset is extracted from the mixture property set; If the similarity is less than the preset matching threshold, a feature enhancement decomposition interval is selected based on the correlation between the aggregate gradation harmonic component and the mechanical parameters.

8. The method for evaluating the performance of cold-recycled modified emulsified asphalt mixture according to claim 1, characterized in that, The parameter correction on the performance degradation feature subset includes: According to the correlation between the asphalt binder elastic modulus and the viscosity change rate, the viscosity fluctuation amplitude in the performance degradation feature subset is corrected; Based on the matching relationship between the environmental temperature and humidity and the paving thickness settlement rate, the thickness uniformity index is compensated and corrected; The corrected viscosity fluctuation amplitude and the thickness uniformity index are combined as a parameter coupling correction result.

9. The method for evaluating the performance of cold-recycled modified emulsified asphalt mixture according to claim 1, characterized in that, The potential defect factors are traced back, including: A performance degradation map including material defect transmission paths and construction defect transmission paths is constructed; An abnormal index is identified from the corrected performance degradation feature subset, and a corresponding node in the performance degradation map is matched; In the reverse direction along the defect transmission direction, the root defect node of the abnormal index is located; According to the hierarchical depth of the root defect node and the number of abnormal indexes, a performance optimization suggestion is generated.

10. The method for evaluating the performance of cold-recycled modified emulsified asphalt mixture according to claim 9, characterized in that, The performance optimization suggestion includes: For the emulsified asphalt content abnormal node in the root defect node, it is suggested to adjust the emulsified asphalt mixing ratio; For the paving thickness abnormal node in the root defect node, it is suggested to optimize the paving machine travel speed and vibration frequency parameters; The adjustment suggestions are sorted according to the node level and output as a structured optimization scheme.

Images