Rubber modified asphalt mixture aging test method under multi-factor coupling conditions

Abstract

The present application relates to the technical field of road engineering material testing, in particular to a rubber modified asphalt mixture aging test method under multi-factor coupling conditions, comprising: taking preset environment data as a test set to perform aging test to obtain aging difference and divide environment data groups, predicting new environment data based on each group containing maximum and minimum aging difference environment data P1 and P2, recording the adjustment vector of P1, and integrating the new environment data into the test set; in the repeating process, P1 used for prediction is recorded as reference data, in a plurality of environment data most similar to the reference data, a plurality of adjustment vectors with the greatest consistency in the direction of the adjustment vector of the reference data are obtained, and the test duration of the new predicted environment data is prolonged based on the credibility of these adjustment vectors. The present application efficiently explores key aging environment parameters and intelligently optimizes the test duration, thereby improving the pertinence and efficiency of the aging test.

Description

Technical Field

[0001] This invention relates to the field of road engineering material testing technology, specifically to an aging test method for rubber-modified asphalt mixtures under multi-factor coupling conditions. Background Technology

[0002] To assess the durability of rubber-modified asphalt mixtures or optimize their processes and raw material proportions, accelerated aging tests are required in a laboratory to simulate the service environment of the road surface. Conventional laboratory aging methods include oven aging, ultraviolet aging, pressure aging, humidity aging, and load impact aging. These methods typically employ a single environmental factor or apply multiple environmental factors sequentially. To more realistically simulate the complex aging scenarios of road surfaces under the combined effects of sunlight, high temperature, rain, oxidation, and traffic loads, existing technologies have developed test methods capable of simultaneously or in a highly synergistic manner applying multiple aging factors (such as temperature, humidity, ultraviolet radiation, and mechanical loads). This method aims to simulate the coupled effects of multiple factors and, theoretically, reflects the actual service condition of the material better than single or sequential aging.

[0003] However, when multiple aging factors are applied simultaneously, these factors are not independent and exhibit complex interactions, known as coupling effects. The physicochemical mechanisms of these coupling effects are not yet clear, and a clear theoretical model is lacking. Because it is unclear how these factors interact and collectively influence material aging, it is impossible to scientifically and reliably determine a representative combination of multi-factor environmental parameters (e.g., how to combine specific values ​​for parameters such as temperature, humidity, and load). Furthermore, even after selecting a set of environmental parameters, it is impossible to predict how long the test will need to last before significant differences in material aging become apparent. Too short a test time may fail to observe effective performance differentiation signals; too long a test time leads to low experimental efficiency.

[0004] Due to the aforementioned shortcomings, current multi-factor coupled aging tests lack theoretical guidance in the selection of environmental parameters and the setting of test duration, resulting in significant arbitrariness and hindering the accurate and efficient evaluation of the durability of rubber-modified asphalt mixtures under complex environments. Summary of the Invention

[0005] To address the lack of theoretical guidance and significant uncertainty in the selection of environmental parameters and setting of test duration in current multi-factor coupled aging tests, this invention provides an aging test method for rubber-modified asphalt mixtures under multi-factor coupled conditions.

[0006] The aging test method for rubber-modified asphalt mixtures under multi-factor coupling conditions of the present invention adopts the following technical solution: One embodiment of the present invention provides an aging test method for rubber-modified asphalt mixtures under multi-factor coupling conditions, the method comprising the following steps: S1: Several preset environmental data are used as a test set. Under each environmental data, an aging test of preset test duration is conducted on the test group and the control group to obtain the aging difference under each environmental data. The test group is rubber modified asphalt mixture, and each environmental data includes: temperature, humidity, and load. S2: Based on the aging difference, all environmental data in the test set are divided into several groups of data. Each group of data contains two environmental data. The aging difference between the two environmental data in each group of data has the largest difference among all groups of data. Based on the environmental data P1 with the largest aging difference and the environmental data P2 with the smallest aging difference in each data set, new environmental data is predicted, and the difference between the new environmental data and P1 is recorded as the adjustment vector of P1; under the new environmental data, aging tests are carried out simultaneously on the test group and the control group, and the aging difference of the new environmental data is recorded as the confidence of the adjustment vector; the new environmental data is incorporated into the test set. S3: Each time S2 is repeated, P1 used to predict new environmental data is recorded as reference data. Among several environmental data that are most similar to the reference data, several adjustment vectors with the greatest directional consistency with the adjustment vector of the reference data are obtained. The test duration of the new environmental data predicted by the reference data is extended. The extended test duration is negatively correlated with the reliability of the several adjustment vectors.

[0007] Preferably, the specific steps for dividing all environmental data in the test set into several groups based on the aging differences are as follows: For the aging differences of all environmental data in the test set, sort all environmental data in the test set in ascending order of aging differences. In the resulting sorted order, select n environmental data from left to right as Class I environmental data and select n environmental data from right to left as Class II environmental data. Sort all Class I environmental data in ascending order and all Class II environmental data in ascending order. Class I environmental data and Class II environmental data with the same sorted order are matched together.

[0008] Preferably, the specific steps for predicting new environmental data are as follows: Obtain the difference vector between environmental data P1 and environmental data P2, and record the sum of environmental data P1 and the difference vector as the new environmental data.

[0009] Preferably, the specific steps for obtaining several adjustment vectors that have the greatest directional consistency with the adjustment vectors of the reference data are as follows: For any set of data obtained each time S2 is executed, after predicting a new environmental data p using environmental data P1 and environmental data P2 in the set of data, environmental data P1 in the set of data is recorded as reference data, wherein the adjustment vector Q corresponding to the reference data is equal to the difference between environmental data p and reference data; several environmental data in the test set that are most similar to the reference data are recorded as similar data. For all adjusted vectors of all similar data, the directional consistency between each adjusted vector and the adjusted vector Q and the cosine similarity between each adjusted vector and the adjusted vector Q are positively correlated. The aforementioned adjustment vectors with the greatest directional consistency refer to the adjustment vectors with the greatest directional consistency with adjustment vector Q among all adjustment vectors of all similar data.

[0010] Preferably, the specific steps for obtaining the directional consistency between each adjustment vector and the adjustment vector Q are as follows: For any one adjustment vector q among all adjustment vectors of all similar data; when the similar data corresponding to adjustment vector q has more than one adjustment vector, all adjustment vectors of the similar data corresponding to adjustment vector q are called the first vector; according to the order of the cosine similarity between the first vector and adjustment vector Q from large to small, the numbers of all first vectors are sequentially increased; the directional consistency between adjustment vector q and adjustment vector Q and the cosine similarity between adjustment vector q and adjustment vector Q are positively correlated, and negatively correlated with the number of adjustment vector q.

[0011] Preferably, the specific steps for extending the testing duration of the new environmental data predicted by the reference data are as follows: The adjustment vectors with the greatest directional consistency are denoted as reference vectors. The mean of the confidence of all reference vectors is obtained. The extended test duration is negatively correlated with the mean, representing the duration of the aging test under environmental data p.

[0012] Preferably, the specific steps for obtaining the aging differences under each environmental data are as follows: The absolute value of the difference in aging assessment indicators obtained from aging tests on the control group and the test group under each environmental data is denoted as the aging difference. The method for obtaining the aging assessment indicators is as follows: Material strength tests were conducted on the control group or test group to obtain the strength limit; the difference between the strength limit of the control group or test group before the start of the aging test and the strength limit after the end of the aging test was obtained, and the aging assessment index of the control group or test group was positively correlated with this difference.

[0013] Preferably, when the value of any dimension in the new environmental data exceeds the preset value range of that dimension, the new environmental data is redefined, including the following specific steps: Dimensions in the new environmental data that exceed the preset value range are denoted as out-of-limit dimensions. The environmental data P1 is summed with w times the difference vector to obtain temporary environmental data, where w represents an undetermined coefficient. The excess amount of each out-of-limit dimension in the temporary environmental data is calculated, and the value of w that makes the excess amount of each out-of-limit dimension equal to 0 is solved and denoted as the preferred coefficient of each out-of-limit dimension. The minimum value of the preferred coefficients of all out-of-limit dimensions is denoted as w0, and the temporary environmental data obtained under w0 is used as the newly determined environmental data.

[0014] Preferably, when the magnitude of the newly determined environmental data is less than a preset percentage of the magnitude of the difference vector, the difference between P1 and w1 times the difference vector is calculated and used as the new environmental data, where w1 is a scaling factor less than 1.

[0015] Preferably, the specific steps for calculating the excess amount of each out-of-limit dimension in the temporary environment data that exceeds the value range are as follows: If the value of any out-of-limit dimension exceeds the upper limit of the preset value range of that out-of-limit dimension, the difference between the value of the out-of-limit dimension and the upper limit of the value range is recorded as the excess amount; if the value of any out-of-limit dimension exceeds the lower limit of the preset value range of that out-of-limit dimension, the difference between the lower limit of the value range and the value of the out-of-limit dimension is recorded as the excess amount.

[0016] The beneficial effects of the technical solution of the present invention are: The present invention provides a method for aging tests of rubber-modified asphalt mixtures under multi-factor coupled conditions. By constructing a grouping-prediction-verification loop guided by aging differences, and combining it with a mechanism for dynamically optimizing test duration based on historical adjustment effects, this method transforms multi-factor coupled aging tests from a trial-and-error process relying on prior knowledge and experience into an intelligent process that autonomously and efficiently explores key aging conditions. This method can quickly identify the environmental combinations that best highlight the relative advantages and disadvantages of the material's aging resistance even when the coupling mechanism is unclear, and intelligently suggest the required test time.

[0017] Specifically, firstly, by dividing the environmental data in the test set into several groups that maximize aging differences, and predicting new environmental data based on the environmental data P1 with the largest aging difference and the environmental data P2 with the smallest aging difference in each group, targeted exploration of environmentally sensitive areas was achieved. This effectively increased the likelihood of significant aging differences in the newly predicted environmental data, thereby enhancing the targeting and effectiveness of environmental parameter exploration. Secondly, by recording the difference between the new environmental data and P1 as an adjustment vector, and using the aging difference under the new environmental data as the confidence level of this adjustment vector, a historical record of the environmental data adjustment effect was established, providing a data foundation for subsequent optimization. Finally, by obtaining several adjustment vectors with the greatest directional consistency with the adjustment vector of the reference data from the environmental data most similar to the reference data, and dynamically extending the test duration of the newly predicted environmental data based on the confidence level of these adjustment vectors, adaptive optimization of the test duration setting was achieved, avoiding the problems of insufficient or excessive test time.

[0018] Overall, compared with existing or conventional methods that blindly select environmental parameters and set test durations, this invention significantly improves the efficiency and reliability of multi-factor coupled aging tests while ensuring test results, providing a scientific and efficient test method for durability assessment and process optimization of rubber-modified asphalt mixtures. Attached Figure Description

[0019] Figure 1 The flowchart illustrates the steps of an aging test method for rubber-modified asphalt mixtures under multi-factor coupling conditions provided in one embodiment of the present invention. Figure 2 A flowchart of a first novel environmental data generation and aging test provided in one embodiment of the present invention; Figure 3 This is a second novel environmental data generation and aging test flowchart provided in one embodiment of the present invention. Detailed Implementation

[0020] The following description, in conjunction with the accompanying drawings, details the specific scheme of the aging test method for rubber-modified asphalt mixtures under multi-factor coupling conditions provided by the present invention.

[0021] Please see Figure 1 The diagram illustrates a flowchart of a method for aging tests of rubber-modified asphalt mixtures under multi-factor coupling conditions, according to an embodiment of the present invention. The method includes the following steps: Step S101: Take several preset environmental data as a test set, and conduct aging tests on the test group and control group for a preset test duration under each environmental data to obtain the aging difference under each environmental data; the test group is rubber modified asphalt mixture, and each environmental data includes: temperature, humidity, and load.

[0022] The aging test in this embodiment begins with a set of preset environmental data (i.e., a test set). Each environmental data point refers to a multidimensional parameter vector of a set of aging conditions simultaneously applied to the asphalt specimen, which includes at least temperature, humidity, and static load. In other embodiments, parameters such as ultraviolet radiation intensity and oxygen concentration may also be included. Each environmental data point represents a specific, coupled, multi-factor aging scenario.

[0023] The test group consists of the novel rubber-modified asphalt mixture to be evaluated (referred to as the new material). In this embodiment, the new material uses 70# road petroleum asphalt as the base asphalt and desulfurized waste rubber powder as the modifier. The rubber powder content can be 25%, the gradation type is AC-13, the optimal asphalt-aggregate ratio is 5.3%, and the preparation process is the shear-stir-stabilization development method. The control group consists of existing asphalt mixtures with known performance and mature technology. In this embodiment, the control group uses SBS modified asphalt, which uses 70# road petroleum asphalt as the base asphalt (softening point 46.5℃), and the SBS modifier content can be 4%. The core purpose of setting up the control group is to establish a stable and relative evaluation benchmark. By comparing it with the control group under the same accelerated aging conditions, the systematic error caused by the uncertainty of equivalence between laboratory accelerated aging and actual long-term aging can be offset, thereby more reliably judging whether the new material is more resistant to aging or more prone to aging than the existing benchmark material, that is, highlighting its relative performance advantages and disadvantages.

[0024] Under any given environmental data, after conducting aging tests on the test group and the control group for the same duration (i.e., the predicted test duration), their respective aging assessment results will be obtained. The difference between the two assessment results is the aging difference under that environmental data. The greater the aging difference, the more significant the performance change (whether for better or worse) of the new material compared to the reference material under that environmental data, and the more sensitive that environmental data is to distinguish material performance; the smaller the aging difference, the more similar the aging behavior of the two materials is under that environment, and the less easily distinguishable they are.

[0025] In one embodiment of the present invention, a method for acquiring several preset environmental data includes: Set a value range for each dimension of the environmental data. Randomly sample one data point from each dimension's value range. The vector formed by the sampled values ​​from all dimensions constitutes one set of environmental data. Randomly sample multiple (e.g., 10) environmental data points using this method as the preset environmental data. Specifically, if a sampled environmental data point is identical to an already sampled environmental data point, resampling is required.

[0026] In other embodiments, environmental data can also be manually selected or sampled from the above-mentioned value range based on human experience as preset environmental data. The preferred number of preset environmental data is greater than 6 (e.g., 8). The advantage of this embodiment is that it effectively combines human experience (e.g., historical production experience of rubber-modified asphalt mixtures) to provide a relatively reliable cold start condition for this embodiment.

[0027] As an example, the typical range for temperature is -20℃ to 85℃, where the low temperature simulates cold climates and the high temperature simulates summer road surface temperatures. In this example, to accelerate the aging test process, a more extreme range is used: -20℃ to 10℃ and 52℃ to 76℃. For humidity, this embodiment simulates humidity conditions under precipitation by controlling the spray intensity to facilitate experimentation and variable control. That is, this embodiment uses spray intensity to equivalently characterize humidity, with a spray intensity range of 0 to 180 ml / min. In this example, to accelerate the aging test process, the spray intensity is set to approximately 150 ml / min, specifically 120 to 180 ml / min. The typical range for load is 0.1 kPa to 2.5 MPa, used to simulate traffic load levels (such as rut ​​load levels for light vehicles and heavy trucks). The load uses a strain control mode. In this example, to accelerate the aging test process, a more extreme range is used: 0.6 to 2.5 MPa.

[0028] In this embodiment, 60mW / cm² was used to further accelerate the aging test. 2 The high-pressure mercury lamp is used for ultraviolet irradiation. In other embodiments, the irradiation intensity of the high-pressure mercury lamp can also be used as a dimension of environmental data to simulate ultraviolet irradiation conditions, for example, its range can be set to 30~60mW / cm. 2 .

[0029] In other embodiments, the value range can be set according to the specific application scenario of the new material. For example, when applied to highways in areas with high average temperature or low air humidity, the lower limit of the temperature value range can be increased or the upper limit of the humidity value range can be decreased. This embodiment does not specifically limit the above value range, nor does it elaborate on the value ranges of other dimensions.

[0030] In another embodiment of the present invention, a method for acquiring several preset environmental data includes: For the asphalt material in the control group, since it is an existing material and has already undergone a large number of aging tests under multiple coupled factors, the environmental data of it under different aging conditions can be manually selected as the environmental data for testing new materials.

[0031] In summary, the preset results of the environmental data are only to provide a cold start condition for the implementation of this embodiment. The aging test in this embodiment serves to analyze the aging condition, which is based on the test results of a large amount of environmental data. The analysis of material aging condition is not sensitive to the preset results of a small amount of environmental data. Therefore, this embodiment does not specifically limit the preset environmental data settings.

[0032] To facilitate subsequent analysis and calculation, all embodiments of this invention require dimension removal for each environmental data point. As an example, the dimension removal method is: divide each dimension of each environmental data point by the maximum value within the corresponding range of values ​​for each dimension in the above example. It should be noted that when conducting aging tests using the environmental data later, the dimensions in the environmental data need to be restored. Specifically, this is done by multiplying each dimension of the environmental data by the maximum value within the corresponding range of values ​​for each dimension in the above example.

[0033] In one embodiment of the present invention, the aging test process includes: The specimens made of the material to be tested (i.e., rubber-modified asphalt mixture or SBS-modified asphalt) (i.e., the specimens prepared for the test group or control group) are placed in a sealed test chamber, where the temperature, humidity, and load are all controllable. The preferred test duration in this embodiment is 8 to 24 hours; this embodiment uses 12 hours as an example. In other embodiments, without using the more extreme value ranges described above, the preset test duration can be extended to 24 to 48 hours.

[0034] In some embodiments, multiple (e.g., eight) test chambers can be used for simultaneous testing, with each pair of test chambers forming a group. Within the same group, the same environmental data is used, with one chamber containing specimens from the control group and the other from the test group. Different groups are tested for different environmental data, and all groups are tested in parallel to greatly improve testing efficiency.

[0035] It should be noted that the test process for each specimen (control group or test group) is considered a destructive test. Therefore, a large number of parallel specimens need to be prepared. That is, the control group or test group corresponds to multiple specimens. These specimens are all made of the same batch of asphalt material (i.e., produced using the same process and the same raw material ratio) and are of the same specifications (e.g., all are round disc-shaped specimens of the same volume and shape).

[0036] In an optional embodiment of the present invention, the method for obtaining the aging assessment indicators is as follows: After each aging test, a crack distribution map of the corresponding specimen in the control group or test group is collected and denoted as F1. The crack distribution map of the corresponding specimen in the control group or test group before the aging test is denoted as F2. The crack distribution map represents the distribution of microcracks inside the specimen. All cracked areas in crack distribution maps F1 or F2 are extracted, and the ratio of the sum of the areas of all cracked areas to the total area of ​​the crack distribution map is denoted as the crack density of F1 or F2. The difference between the crack density of F1 and the crack density of F2 is obtained, and the ratio of this difference to the test duration of each aging test (i.e., the actual test time) is denoted as the aging evaluation index of the control group or test group (when the aging evaluation index is less than 0, it is set to 0). The larger the aging evaluation index, the more cracks appear after the aging test, and the weaker the aging resistance.

[0037] The absolute value of the difference in aging assessment indicators between the control group and the test group is recorded as the aging difference of the environmental data used in each aging test.

[0038] The crack distribution map in this embodiment is better able to describe the local and internal aging of the specimen (i.e., crack distribution). The aging assessment index is more accurate, but it requires high-precision and expensive crack distribution map detection equipment (such as ultrasonic equipment or CT equipment).

[0039] The crack distribution map is obtained using ultrasonic detection, that is, by scanning the surface of the test piece with an ultrasonic probe, and the resulting ultrasonic image is used as the crack distribution map. In other embodiments, CT scanning technology can also be used to obtain the crack distribution map. The method of obtaining the crack distribution map is not limited in all embodiments of the present invention.

[0040] In this embodiment, a semantic segmentation network (e.g., DeepLabV3) is used to extract the semantic regions (i.e., all crack regions) of all cracks in the crack distribution map F1 or F2. DeepLabV3 is an existing mature technology. Compared with other technologies (e.g., Canny edge detection technology or threshold segmentation technology), DeepLabV3 can avoid interference from the inherent texture edges in the specimen (e.g., particle edges in asphalt mixture).

[0041] In a preferred embodiment of the present invention, the method for obtaining the aging assessment index is as follows: Before and after the aging test, material strength tests (e.g., strength limit tests) are performed on the corresponding specimens of the control group or test group. The difference between the strength limit before the test and the strength limit after the test is obtained. The ratio of this difference to the test duration (i.e., the actual test time) is recorded as the aging evaluation index for the control group or test group. In this embodiment, the material strength testing technology is mature and low-cost, but the efficiency of obtaining the aging evaluation index is low. A larger aging evaluation index indicates a significant reduction in the strength limit after the aging test, and a weaker aging resistance.

[0042] Step S102: Based on the aging difference, divide all environmental data in the test set into several groups of data, so that the aging difference between the two environmental data contained in each group of data has the maximum difference.

[0043] Based on the aging differences corresponding to all environmental data in the test set, the environmental data are paired to form several groups. The goal of grouping is to make the aging differences between the two environmental data within the same group as large as possible (i.e., with the largest difference), that is, one has a large aging difference (labeled P1), and the other has a small aging difference (labeled P2). This means that from environmental P2 to P1, the new material changes from having no significant aging difference compared to the existing reference material to having a significant aging difference, that is, the change in the environmental data of this group causes a significant change in the material's aging response (from insensitive to sensitive). Here, "no significant aging difference" and "significant aging difference" are relative terms.

[0044] Step S103: Based on the environmental data P1 with the largest aging difference and the environmental data P2 with the smallest aging difference in each group of data, predict new environmental data, and record the difference between the new environmental data and P1 as the adjustment vector of P1; under the new environmental data, conduct aging tests on the test group and the control group at the same time, and record the aging difference of the new environmental data as the confidence of the adjustment vector; incorporate the new environmental data into the test set.

[0045] First, it should be noted that one important basis for setting new environmental data in this embodiment is that the new material must still exhibit a significant difference in aging compared to the existing benchmark material under the new environmental data (a large difference in aging). For example, if the new material has stronger aging resistance, it serves as a positive sample, helping to acquire or analyze the characteristics of the new material (the environmental conditions required for the new material to exhibit aging resistance). Conversely, if the new material has very weak aging resistance, it can serve as a negative sample, providing a reference for optimizing and improving the preparation process of the new material. If the new material still does not exhibit a significant difference in aging compared to the existing benchmark material under the new environmental data (a small difference in aging), it means that the new material cannot serve as a positive sample (i.e., it has no advantage over the existing benchmark material) nor as a negative sample (it cannot provide a reference for improving the new material). For the rapid iteration and update of new materials or process optimization in the early stages, the production and testing of the new material is a waste of time and efficiency.

[0046] Based on this, and in conjunction with the above description: from environment P2 to P1, the new material changes from having no significant aging difference compared to the existing reference material to having a significant aging difference. This embodiment uses P1 (sensitive point) and P2 (insensitive point) to predict the new environmental data.

[0047] Although the aging process of asphalt mixtures and environmental data have a complex nonlinear relationship, the macroscopic trend of changing from insensitive to sensitive provides directional clues for predicting new potential sensitive points. For example, new environmental data can be inferred in the direction of the line connecting P1 and P2 or in its vicinity. Existing direct prediction methods based on machine learning or neural network models rely on datasets and data annotations. However, the physicochemical mechanism of the multi-factor coupling effect in this embodiment is not yet clear, and a clear theoretical model is lacking, which further increases the difficulty of data annotation. Therefore, compared with existing prediction methods, the method of setting new environmental data through prediction in this embodiment is simpler, more efficient, and more effective.

[0048] Furthermore, the difference between the new environmental data and P1 is defined as an adjustment vector starting from P1, representing an environmental data adjustment method or adjustment record based on P1.

[0049] Furthermore, aging tests were conducted on the control and test groups under the new environmental data to obtain the aging differences corresponding to the new environmental data. This aging difference was used as the confidence level of the adjustment vector. High confidence level means that adjusting from P1 in this direction (adjustment vector) can reach another sensitive point that also produces significant aging differences (i.e., large aging differences), indicating that this adjustment direction may be effective and meet expectations. Low confidence level means that this adjustment may have caused the material to move out of the sensitive area, and the aging difference is not significant (i.e., small aging differences).

[0050] Finally, the new environmental data was incorporated into the test set, thus expanding the test set.

[0051] Preferred methods for predicting new environmental data include: Obtain the difference vector between P1 and P2 (that is, the displacement vector from P2 to P1), and record the sum of P1 and the difference vector as the new environmental data. In other words, extend a certain distance from P1 along the direction from P2 to P1 (that is, the distance from P2 to P1) to obtain the new environmental data.

[0052] Specifically, if the value of any dimension in the new environmental data exceeds the value range of that dimension (the value range corresponding to each dimension as described in step S101), then the new environmental data is redefined, specifically including: In the new environmental data, dimensions that exceed the range of values ​​are denoted as out-of-limit dimensions. P1 is summed with w times the difference vector to obtain temporary environmental data, where w represents an undetermined coefficient (a scalar). The excess amount of each out-of-limit dimension in the temporary environmental data is calculated, and the value of w that makes the excess amount of each out-of-limit dimension equal to 0 is determined and denoted as the optimal coefficient for each out-of-limit dimension. In other words, the optimal coefficient for any out-of-limit dimension ensures that the value of the temporary environmental data in that out-of-limit dimension does not exceed the range of values ​​for that out-of-limit dimension.

[0053] The minimum value of the optimization coefficients for all out-of-limit dimensions is denoted as w0, and the temporary environmental data obtained under w0 is used as the newly determined environmental data.

[0054] The solution involves finding the value of w that makes the excess of each out-of-limit dimension equal to 0. This is a well-known mathematical solution technique, which will not be described in detail in this embodiment.

[0055] During this process, the temporary environmental data is kept in the extension direction from P2 to P1. w0 is used to ensure that the temporary environmental data obtained by the extension (that is, the newly determined environmental data) does not exceed the value range in all out-of-limit dimensions.

[0056] Specifically, if a dimension does not have a set range of values, then it is determined that there are no cases where the value of that dimension exceeds the range.

[0057] Specifically, when the magnitude of the newly determined environmental data is less than a preset percentage of the magnitude of the difference vector (in this embodiment, the preset percentage is preferably 20% to 50%, and 30% is used as an example), it indicates that the extension space from P1 to P2 in the direction from P2 to P1 is relatively small. In this case, the difference between P1 and the difference vector multiplied by w1 is taken as the new environmental data, where w1 is a scaling factor less than 1. In this example, w1 is equal to the preset percentage. This process means that when the extension space from P2 to P1 is relatively small, the extension is carried out from P1 in the opposite direction from P2 to P1 (i.e., from P1 to P2) to obtain new environmental data. This avoids the aging test being under excessively extreme test conditions when there are other suitable environmental data exploration spaces (i.e., when the magnitude of the new environmental data is less than the preset percentage of the magnitude of the difference vector). (Excessively extreme test conditions refer to the situation where the newly determined environmental data does not exceed the value range of some dimensions.)

[0058] Preferably, the method for calculating the excess amount of each out-of-limit dimension in the temporary environmental data, which exceeds the value range, includes: If the value of any out-of-limit dimension exceeds the upper limit of the range of values ​​for that out-of-limit dimension, the difference between the value of the out-of-limit dimension and the upper limit of the range is recorded as the excess amount; if the value of any out-of-limit dimension exceeds the lower limit of the range of values ​​for that out-of-limit dimension, the difference between the lower limit of the range and the value of the out-of-limit dimension is recorded as the excess amount.

[0059] Specifically, if the range of values ​​for a certain over-limit dimension consists of multiple non-overlapping value intervals, the minimum distance (i.e., the absolute value of the difference) between the value of the over-limit dimension and the two endpoints of each value interval is obtained and denoted as the proximity distance between the over-limit dimension and each value interval. In this case, the value range mentioned above refers to the value interval with the smallest proximity distance (i.e., the value interval closest to the value of the over-limit dimension).

[0060] Step S104: When new environmental data is generated, P1 used to predict the new environmental data is recorded as reference data. Among several environmental data that are most similar to the reference data, several adjustment vectors with the greatest directional consistency with the adjustment vector of the reference data are obtained. The test duration of the new environmental data predicted by the reference data is extended. The extended test duration is negatively correlated with the reliability of the several adjustment vectors.

[0061] In this embodiment, steps S102 and S103 are repeatedly executed to continuously generate new environmental data. During the repetition process, regardless of whether the aging of the new environmental data is significant, it is incorporated into the test set. If the new environmental data produces a significant aging difference (high confidence), this new environmental data may become the new P1 in subsequent iterations, continuing to guide the exploration of sensitive areas; if the new environmental data does not produce a significant aging difference (low confidence), it may become the new P2, helping to define the boundary between sensitive and non-sensitive areas. By continuously repeating this process, the test set is continuously expanded, enabling intelligent expansion and exploration of environmental regions sensitive to material aging performance.

[0062] In summary, by repeatedly executing S102 and S103, ensuring that the aging difference between the two environmental data points within each data set is maximized, and predicting new environmental data based on P1 and P2 in each data set, the system not only efficiently and effectively predicts new environmental data (where "efficient" means without relying on unknown physicochemical mechanisms of coupling effects, and "effective" means that it can be expected before the actual experiment that the new material will exhibit a significant aging difference compared to the existing benchmark material), but also, regardless of whether the predicted new environmental data is effective (i.e., whether or not a significant aging difference is observed after the actual experiment), it can serve as a basis for exploring diverse and sensitive environmental data. Each exploration process is based on environmental data when the new material changes from having no significant aging difference to having a significant aging difference, ensuring that the expansion of environmental data during the aging experiment is efficient and evidence-based.

[0063] Furthermore, this embodiment takes into account that the magnitude of aging differences is also related to the test duration. Under some environmental data, a test time is not required for the test group and the control group to show obvious aging differences. If the test time is set too long, the efficiency will be low. Under some environmental data, a shorter test time cannot clearly reflect the aging differences, which will make the above prediction process more unreliable.

[0064] Based on this, in each repeated iteration, when a certain environmental data in the test set is used as a sensitive point (i.e., marked as P1 with the largest aging difference in step S102) and used to predict new environmental data, it is marked as reference data; as can be seen from step S103, when the reference data is used as a sensitive point, an adjustment vector is also obtained (which represents an adjustment method or adjustment record of environmental data under the reference data); the following description will take the reference data and the adjustment vector as an example.

[0065] This embodiment retrieves other environmental data in the test set that are most similar to the current reference data, and examines historical adjustment records from which the adjustment vectors are closest in direction to the reference data (i.e., several adjustment vectors with the greatest directional consistency with the reference data's adjustment vectors). If the reliability of these historically similar adjustment vectors is generally high, it indicates that the new environmental data obtained by this environmental data adjustment method based on the reference data's adjustment vectors (i.e., the new environmental data predicted based on the reference data) is more likely to have significant aging differences. In this case, aging tests can continue to be performed on the new environmental data according to the test duration. If the reliability of these historically similar adjustment vectors is generally low, it suggests that the aging differences after testing of the new environmental data obtained by this environmental data adjustment method based on the reference data's adjustment vectors may not be significant. One reason for this insignificance may be that the preset test duration is insufficient. Therefore, this embodiment will extend the preset test duration for this newly predicted environmental data. Conversely, if the reliability of historically similar adjustments is high, a shorter preset duration will be maintained or used.

[0066] Based on this, the test duration of the new environmental data predicted by the reference data is extended, and the extended test duration is negatively correlated with the reliability of the aforementioned adjustment vectors.

[0067] It should be noted that the update or extension of the test duration is a heuristic optimization based on data experience. In this embodiment, low reliability may have multiple causes (such as actually entering a non-sensitive area), and is not necessarily due to insufficient test duration. However, using low reliability of historical similar adjustments as a strong clue to consider extending the test duration is a significant improvement in the overall efficiency of discovering effective test conditions during multiple trial explorations (i.e., during repeated execution of S102 and S103), compared to conventional methods that use a fixed, long time consumption or blindly set the duration. This embodiment achieves synergistic optimization of environmental parameter exploration and test duration setting.

[0068] In summary, steps S101-S104 of this embodiment transform multi-factor coupled aging tests from a trial-and-error process relying on prior knowledge and experience into an intelligent process that autonomously and efficiently explores key aging conditions by constructing a grouping-prediction-verification loop guided by aging differences and combining it with a mechanism for dynamically optimizing test duration based on historical adjustment effects. This allows for the rapid identification of harsh environmental combinations that best highlight the relative superiority or inferiority of material aging resistance even when the coupling mechanism is unclear, and intelligently prompts for the required test time. Compared to existing or conventional methods that blindly set environmental data, this embodiment significantly improves the targeting and efficiency of the test while ensuring test effectiveness.

[0069] In a preferred embodiment of the present invention, the method for dividing all environmental data in the test set into several groups of data based on the aging differences includes: For all environmental data in the test set, each environmental data corresponds to an aging difference. All environmental data in the test set are sorted in ascending order of aging difference. From the resulting sorted order, n environmental data are selected from left to right and designated as Class I environmental data; n environmental data are selected from right to left and designated as Class II environmental data. The aging differences of Class I environmental data are relatively small, while the aging differences of Class II environmental data are relatively large.

[0070] The KM algorithm is used to match Class I and Class II environmental data to obtain all matching results. Each matching result contains two environmental data from Class I and Class II environmental data, respectively. The absolute value of the difference in aging difference between these two environmental data is denoted as the matching difference of each matching result. In this embodiment, the matching method is as follows: Class I environmental data are arranged from smallest to largest, and Class II environmental data are also arranged from smallest to largest. Data with the same arrangement order in Class I and Class II environmental data are matched together.

[0071] Each of the above matching results serves as the data set for that set, and the aging differences in the environmental data contained across all data sets have the greatest variation.

[0072] Where n is half the total number of environmental data in the test set (rounded down). In particular, when the total number of environmental data is odd, there is one remaining environmental data in the test set that has not participated in the matching. In this case, the environmental data with the smallest aging difference in the test set and the remaining environmental data are used as a set of data.

[0073] Although this process locally satisfies the condition that the aging difference between the two environmental data contained in each data set is the greatest, the calculation process is simple and efficient, applicable to all embodiments of the present invention, and is therefore implemented as a preferred example.

[0074] In an optional embodiment of the present invention, the method for dividing all environmental data in the test set into several groups of data based on the aging differences includes: The environmental data in the test set are grouped into pairs, with each pair containing two environmental data points. The two environmental data points in different pairs are not allowed to be completely identical (at most one identical environmental data point is allowed). The absolute value of the difference in aging difference between the two environmental data points in each pair is denoted as the matching difference for each pair, and the mean of the matching differences for all pairs is denoted as the pair index.

[0075] At this point, each grouping method yields multiple groups and a grouping metric. By iterating through all grouping methods, the grouping method with the highest grouping metric is identified; each group obtained by this method constitutes the data for that group.

[0076] The grouped data obtained by this process globally satisfies the condition that the aging difference between the two environmental data in each group has the maximum difference. However, when there is a large amount of environmental data in the test set, the computation is large. Therefore, this process is implemented as an optional example.

[0077] In an optional embodiment of the present invention, obtaining a plurality of adjustment vectors that have the greatest directional consistency with the adjustment vectors of the reference data includes the following method: First, it should be noted that each time steps S102 and S103 are repeated, for any set of data obtained, after predicting a new environmental data (denoted as p) using P1 and P2 from that set of data, P1 from that set of data is recorded as the reference data. The reference data also corresponds to an adjustment vector, denoted as Q (i.e., the difference between the environmental data p and the reference data). The environmental data sets in the test set that are most similar to the reference data are recorded as similar data.

[0078] It should be further noted that after repeatedly executing S102 and S103, the same similar data may not have a corresponding adjustment vector in the historical repeated iterations, may have a corresponding adjustment vector, or may have a corresponding multiple adjustment vectors.

[0079] For all adjustment vectors (equivalent to historical adjustment records) of all similar data, the cosine similarity between each adjustment vector and Q is denoted as the directional consistency between Q and each adjustment vector. The greater the directional consistency, the more consistent the orientation of Q with each adjustment vector, indicating that the adjustment direction of the reference data is closest to one of the historical adjustment records.

[0080] Among all adjustment vectors of all similar data, select the few adjustment vectors that have the greatest consistency with the Q direction, and denote them as reference vectors.

[0081] In a preferred embodiment of the present invention, obtaining a plurality of adjustment vectors that have the greatest directional consistency with the adjustment vectors of the reference data includes the following method: The difference between this preferred embodiment and the optional embodiment is as follows: For all adjustment vectors of all similar data, and for any one of these adjustment vectors q. As described in the optional embodiment, one similar data point may correspond to multiple adjustment vectors; If the similar data corresponding to the adjustment vector q has multiple (two or more) adjustment vectors, then all the adjustment vectors (including q) of the similar data corresponding to the adjustment vector q are called the first vector. All the first vectors are numbered (e.g., numbered 0, 1, 2, ...) in descending order of the cosine similarity between the first vector and Q. The directional consistency between the adjustment vector q and Q is positively correlated with the cosine similarity between q and Q, and negatively correlated with the number of q.

[0082] As a specific example, the specific formula for calculating directional consistency is: directional consistency equals the cosine similarity between q and Q multiplied by (1-h), where h represents the attenuation factor, which is used to attenuate the cosine similarity. The attenuation factor h is positively correlated with the number of q.

[0083] Among all adjustment vectors of all similar data, select the few adjustment vectors that have the greatest consistency with the Q direction, and denote them as reference vectors.

[0084] In this embodiment, when the same similar data corresponds to multiple adjustment vectors, the smaller the number of the adjustment vector (the greater the cosine similarity with Q), the smaller its corresponding decay factor and the greater its directional consistency; the larger the number of the adjustment vector (the smaller the cosine similarity with Q), the larger its corresponding decay factor and the smaller its directional consistency. This allows for the priority selection of adjustment vectors with a high cosine similarity to Q from different similar data when filtering reference vectors based on directional consistency. Adjustment vectors with a high cosine similarity to Q within the same similar data, but not the largest (i.e., those with higher numbers), have their directional consistency attenuated or reduced compared to the optional embodiments described above. In other words, this embodiment focuses on adjustment vectors with a high cosine similarity to Q and located within different similar data sets. The aim is to ensure that the adjustment directions represented by adjustment vectors from different similar data sets are more comprehensive and referential (i.e., the historical adjustment records referenced or relied upon are diverse), avoiding potential noise interference issues when relying too heavily on historical adjustment records represented by multiple adjustment vectors from the same or a portion of similar data (e.g., abnormal interference where the same or a portion of similar data is similar to the reference data but has a different impact on the aging process), thereby ensuring the reliability of subsequent test duration updates.

[0085] Preferably, the formula for calculating the attenuation factor h includes: h = tanh(r × R), where R represents the number of the adjustment vector q, tanh represents the hyperbolic tangent function, used to limit r × R within [0, 1), where when R = 0, h = 0, and when R is large, h approaches 1, where r represents the reduction factor, which is greater than 0 and less than 1, used to avoid the situation where h approaches 1 when R is small (e.g., when R equals 2 or 3). In this example, r = 0.18.

[0086] Preferably, the environmental data most similar to the reference data in the test set are categorized as similar data, including the following methods: Obtain the Mahalanobis distance between any environmental data point in the test set (excluding reference data) and the reference data. Select several environmental data points (excluding reference data) with the smallest Mahalanobis distance from the reference data as similar data. These similar data points have the smallest difference from the reference data (i.e., the most similar). Considering that the relationship between asphalt aging and environmental data is complex and nonlinear, it still has continuity (i.e., the abrupt changes in the aging of asphalt under environmental data with small differences are negligible or insignificant), the historical adjustment records of similar data can be used as a reference for the reference data.

[0087] Preferably, the 10 environmental data points with the smallest Mahalanobis distance to the reference data are selected as similar data (if there are fewer than 10, all environmental data points are considered similar data). If, among all similar data points, the Mahalanobis distance of a particular similar data point to the reference data is greater than a preset distance threshold, that similar data point is discarded so that it no longer participates in the calculation of maximum directional consistency. The purpose is to avoid a situation where the environmental data point with the smallest difference from the reference data still has significant differences, rendering its historical adjustment records unusable as a valid reference.

[0088] The preset distance threshold is set to 10% to 40% of the average Mahalanobis distance between all environmental data in the test set, preferably 30%. The purpose is to avoid the preset distance threshold being too large (corresponding to 10% to 40%), and to always use the environmental data in the test set as a reference (corresponding to the average Mahalanobis distance between all environmental data in the test set), which is suitable for the specific distribution of environmental data in the continuously expanding test set.

[0089] Specifically, when no similar data exists, the subsequent test duration update process is not executed (i.e., the test duration is not updated).

[0090] Preferably, when obtaining reference vectors in the above process, only the 10 adjustment vectors with the highest directional consistency are used as reference vectors (if there are fewer than 10, all adjustment vectors are used as reference vectors). Specifically, among all reference vectors, if the directional consistency of a reference vector with Q is less than the consistency threshold, that reference vector is discarded. The purpose is to avoid the introduction and interference of invalid historical adjustment records with excessively large differences in adjustment direction. If the consistency threshold is too small, it will lead to more interference (interference from historical adjustment records with no reference value); if the consistency threshold is too large, it will result in fewer reference vectors and poor coverage of the referenced historical adjustment records. The preferred value range for the consistency threshold is 0.6~0.85. This example uses 0.78 as an example.

[0091] Specifically, when there are no reference vectors, the subsequent test duration update process is not performed (i.e., the test duration is not updated).

[0092] In one embodiment of the present invention, extending the testing duration of new environmental data predicted by reference data includes the following method: obtaining the mean confidence level of all reference vectors, and the ratio of this mean to a preset scaling factor is denoted as an adjustment factor. The purpose of comparing the mean with the preset scaling factor is to remove the dimensions and orders of magnitude of the confidence level.

[0093] The extended test duration (i.e., the duration of the aging test under the new environmental data p) is H = H0 × (1 + b × exp(-h)), where H0 represents the preset test duration described in step S101, h represents the adjustment coefficient, and exp() represents an exponential function with the natural constant as the base. b represents a preset coefficient; for example, in this example, when H0 is set to 12 hours, b is set to 3 so that the adjusted test duration does not exceed 48 hours.

[0094] Optionally, the preset scaling factor can be obtained as follows: The reliability of the reference vector is essentially the aging difference of the new environmental vector predicted by the reference vector (as known from step S103). This aging difference refers to the absolute value of the difference between the aging assessment indicators of the control group and the test group (as known from step S101). Here, the sum of the aging assessment indicators of the control group and the test group is used as the normalization coefficient of this aging difference. At this time, the reliability of the reference vector corresponds to the normalization coefficient mentioned above. The average of the normalization coefficients of the reliability of all reference vectors is used as the preset proportional coefficient. Specifically, when the preset proportional coefficient is equal to 0, it means that the aging assessment indicators of the control group and the test group are both 0, which means that no aging phenomenon occurred after the aging test of the control group and the test group. At this time, the preset proportional coefficient is no longer set, and the adjusted test duration is directly set to 48 hours.

[0095] In an optional embodiment of the present invention, according to as follows Figure 2 The process shown is used to iteratively execute S102 and S103; in a preferred embodiment of the present invention, according to... Figure 3 The process shown is used to iteratively execute S102 and S103.

[0096] Figure 2 and Figure 3 The difference is: Figure 2 In this process, whenever all environmental data in the test set is divided into several groups, predictions are made based on all groups of data to obtain new environmental data corresponding to each group. Then, aging tests are conducted using the new environmental data corresponding to each group. Finally, all the new environmental data obtained are incorporated into the test set.

[0097] Figure 3 In this process, whenever all environmental data in the test set is divided into several groups, only a portion of the data is selected for prediction to obtain new environmental data corresponding to the partial group data. Then, aging tests are conducted using the new environmental data corresponding to the partial group data. Finally, the obtained new environmental data is incorporated into the test set.

[0098] Figure 2In the described embodiment, each iteration executes S102 and S103 (including the first execution of S102 and S103), obtaining a large amount of new environmental data at once. The advantage is that the test set can be quickly expanded and supplemented. The disadvantage is that a large number of aging tests need to be conducted based on a large amount of new environmental data each time the aging test process is slow.

[0099] Figure 3 In the described embodiment, each iteration executes S102 and S103 (including the first execution of S102 and S103), obtaining only a small amount of new environmental data at a time. The disadvantage is that it cannot quickly expand and supplement the test set. The advantage is that each iteration only needs to perform a large number of aging tests based on a small amount of new environmental data, making the overall aging test relatively efficient.

[0100] Preferably, methods for predicting only a subset of data groups include: For all groups of data divided in the test set, each group corresponds to a matching difference (as described above). The groups with the largest matching differences are then selected (e.g., ...). Figure 3 The three sets of data shown indicate that the aging response of the material has changed more significantly. Prediction and aging tests are based on these three sets of data first, while other sets of data are not used in the current iteration.

[0101] Special, regardless Figure 2 still Figure 3 In the corresponding embodiment, if new environmental data predicted based on a certain set of data already exists in the test set, then aging tests will not be conducted based on it, nor will it be repeatedly incorporated into the test set. In this case, for Figure 3 If the new environmental data predicted by some of the selected data sets already exists in the test set, then these data sets are marked as ignorable data, and some data sets need to be re-selected (e.g., three data sets are re-selected). The ignorable data sets do not need to be selected during the re-selection process.

[0102] This invention is based on Figure 3 In the illustrated embodiment, steps S102 and S103 are repeated 20 times before the repetition process ends. This invention follows... Figure 2 When the illustrated embodiment is executed, since the environmental data is amplified significantly after each iteration, steps S102 and S103 are repeated a relatively small number of times (e.g., 10 times) before the repetition process ends.

[0103] The more repetitions, the more complete and reliable the aging test becomes, but the longer it takes. Other embodiments... Figure 3 or Figure 2The maximum number of repetitions should not exceed 30 or 15 (e.g., 20 or 10 as mentioned above).

[0104] Once the repetitive process ends, all the above embodiments of the present invention have completed their operation.

[0105] Furthermore, after the iterative repetition process ends, the present invention provides an additional embodiment for evaluating the performance of rubber-modified asphalt mixtures based on the test results of all environmental data in the test set, and obtaining performance evaluation indicators. These performance evaluation indicators can be used to assess whether the production process of rubber-modified asphalt mixtures has been improved or optimized. The specific process for obtaining the performance evaluation indicators is as follows: At this point, for each environmental data point, the test group and the control group each correspond to an aging assessment index; the larger the aging assessment index, the weaker the aging resistance, that is, the easier it is to age.

[0106] Environmental data for which the aging assessment index of the test group is lower than that of the control group are recorded as the first reference data, and environmental data for which the aging assessment index of the test group is higher than that of the control group are recorded as the second reference data. The mean of all aging differences under the first reference data is recorded as the first mean, and the mean of all aging differences under the second reference data is recorded as the second mean.

[0107] The test group under the first reference data showed relatively strong aging resistance; the test group under the second reference data showed relatively weak aging resistance.

[0108] A higher first mean indicates that the test group not only exhibits relatively strong aging resistance but also shows a significant difference in aging assessment indicators compared to the control group, meaning the test group is significantly more resistant to aging than the control group under the first reference data. A lower first mean indicates that while the test group shows relatively strong aging resistance, the difference in aging assessment indicators compared to the control group is small, meaning the test group's aging resistance under the first reference data does not significantly exceed that of the control group.

[0109] The smaller the second mean, the more likely the test group is to show relatively weak aging resistance and have a small difference from the aging assessment index of the control group. This indicates that the aging resistance of the test group under the second reference data is equal to or close to that of the control group. The larger the second mean, the more likely the test group is to show relatively weak aging resistance and have a large difference from the aging assessment index of the control group. This indicates that the aging resistance of the test group under the second reference data is significantly weaker than that of the control group.

[0110] The ratio of the first mean to the second mean is used as a performance evaluation index for rubber-modified asphalt mixtures. A larger index (i.e., a larger first mean and a smaller second mean) indicates that, after conducting a limited number of aging tests on the test group using limited environmental data, the rubber-modified asphalt mixture in the test group has relatively stronger anti-aging capabilities (compared to the control group). This suggests that the process and raw material ratios used to manufacture the test group contribute to the production of more anti-aging rubber-modified asphalt mixtures, indicating that the production process or raw material ratios for rubber-modified asphalt mixtures have been improved or optimized. Conversely, a smaller index (i.e., a smaller first mean and a larger second mean) indicates that, after conducting a limited number of aging tests on the test group using limited environmental data, the rubber-modified asphalt mixture in the test group has relatively weaker anti-aging capabilities (compared to the control group). This suggests that the process and raw material ratios used to manufacture the test group are not conducive to the production of more anti-aging rubber-modified asphalt mixtures, indicating that the production process or raw material ratios for rubber-modified asphalt mixtures have not been improved or optimized.

[0111] This embodiment describes the improvement or optimization of process parameters such as rubber powder content, gradation type, and optimal asphalt-aggregate ratio in rubber-modified asphalt mixtures.

[0112] Thus, for rubber-modified asphalt mixtures produced by the same process and with the same raw material ratio, after multiple aging tests in steps S101 to S104, their corresponding performance evaluation indicators were obtained.

[0113] For rubber-modified asphalt mixtures produced at different times using different processes or raw material ratios, it is determined whether the performance evaluation index has increased. If an increase occurs, it indicates that the process or raw material ratio has been reliably optimized; if no increase occurs, it indicates that further improvement of the process or raw material ratio is needed. When the performance evaluation index is greater than 5-10 (e.g., greater than 8), it is considered that the stage of process or raw material ratio optimization process has been completed, and the aging performance test is then carried out. The produced rubber-modified asphalt mixture is put into use (e.g., paved as a test road surface). After a certain period of use (e.g., 6-12 months, preferably 8 months), samples are taken from the test road surface and aging tests are conducted to obtain aging evaluation indicators. If the aging evaluation indicators are still greater than those of the control group road surface (after the same period of use), it indicates that the rubber-modified asphalt mixture has reached a better level than the existing benchmark material and can be put into production. If it is not greater, it indicates that the rubber-modified asphalt mixture has no significant advantage over the existing benchmark material. At this time, it is necessary to adjust (manually adjust) the process or raw material ratio, and then continue to carry out aging tests according to all the above embodiments of the present invention, and continue to optimize the process or raw material ratio.

[0114] Specifically, if there is no second reference data when calculating the performance evaluation index, it means that there is no situation where the aging evaluation index of the test group is greater than that of the control group. In this case, the performance evaluation index will not be calculated further, and the aging performance test described above can be carried out directly.

[0115] It should be noted that since the aging test is conducted in a laboratory setting and simulates multi-factor coupling conditions, it requires aging tests based on a large amount of environmental data, and each test has a limited duration. This means that the environmental data corresponding to the first and second reference data mentioned above cannot be equated to actual working conditions, and the performance of the rubber-modified asphalt mixture in the test group cannot be directly identified based on the distribution of the environmental data corresponding to the first and second reference data. However, the performance evaluation index obtained above is based on the comparison between the test group and the control group, and is not sensitive to whether it is equivalent to actual working conditions. This allows the magnitude of the performance evaluation index to effectively assess the performance of the rubber-modified asphalt mixture and provide a reliable basis for optimizing the performance of the rubber-modified asphalt mixture.

[0116] It should also be noted that the control group was always produced using the same process and raw material ratio (e.g., the current best and most mature process).

[0117] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for aging test of rubber-modified asphalt mixture under multi-factor coupled conditions, characterized in that, The method includes the following steps: S1: Several preset environmental data are used as a test set. Under each environmental data, an aging test of preset test duration is conducted on the test group and the control group to obtain the aging difference under each environmental data. The test group is rubber modified asphalt mixture, and each environmental data includes: temperature, humidity, and load. S2: Based on the aging difference, all environmental data in the test set are divided into several groups of data. Each group of data contains two environmental data. The aging difference between the two environmental data in each group of data has the largest difference among all groups of data. Based on the environmental data P1 with the largest aging difference and the environmental data P2 with the smallest aging difference in each data set, new environmental data is predicted, and the difference between the new environmental data and P1 is recorded as the adjustment vector of P1; under the new environmental data, aging tests are carried out simultaneously on the test group and the control group, and the aging difference of the new environmental data is recorded as the confidence of the adjustment vector; the new environmental data is incorporated into the test set. S3: Each time S2 is repeated, P1 used to predict new environmental data is recorded as reference data. Among several environmental data that are most similar to the reference data, several adjustment vectors with the greatest directional consistency with the adjustment vector of the reference data are obtained. The test duration of the new environmental data predicted by the reference data is extended. The extended test duration is negatively correlated with the reliability of the several adjustment vectors.

2. The aging test method for rubber-modified asphalt mixtures under multi-factor coupling conditions according to claim 1, characterized in that, The specific steps involved in dividing all environmental data in the test set into several groups based on the aging differences are as follows: For the aging differences of all environmental data in the test set, sort all environmental data in the test set in ascending order of aging differences. In the resulting sorted order, select n environmental data from left to right as Class I environmental data and select n environmental data from right to left as Class II environmental data. Sort all Class I environmental data in ascending order and all Class II environmental data in ascending order. Class I environmental data and Class II environmental data with the same sorted order are matched together.

3. The aging test method for rubber-modified asphalt mixtures under multi-factor coupling conditions according to claim 1, characterized in that, The specific steps involved in predicting new environmental data are as follows: Obtain the difference vector between environmental data P1 and environmental data P2, and record the sum of environmental data P1 and the difference vector as the new environmental data.

4. The aging test method for rubber-modified asphalt mixtures under multi-factor coupling conditions according to claim 1, characterized in that, The specific steps involved in acquiring several adjustment vectors that have the greatest directional consistency with the adjustment vectors of the reference data are as follows: For any set of data obtained each time S2 is executed, after predicting a new environmental data p using environmental data P1 and environmental data P2 in the set of data, environmental data P1 in the set of data is recorded as reference data, wherein the adjustment vector Q corresponding to the reference data is equal to the difference between environmental data p and reference data; several environmental data in the test set that are most similar to the reference data are recorded as similar data. For all adjusted vectors of all similar data, the directional consistency between each adjusted vector and the adjusted vector Q and the cosine similarity between each adjusted vector and the adjusted vector Q are positively correlated. The aforementioned adjustment vectors with the greatest directional consistency refer to the adjustment vectors with the greatest directional consistency with adjustment vector Q among all adjustment vectors of all similar data.

5. The aging test method for rubber-modified asphalt mixtures under multi-factor coupling conditions according to claim 4, characterized in that, The specific steps for obtaining the directional consistency between each adjustment vector and the adjustment vector Q are as follows: For any one adjustment vector q among all adjustment vectors of all similar data; when the similar data corresponding to adjustment vector q has more than one adjustment vector, all adjustment vectors of the similar data corresponding to adjustment vector q are called the first vector; according to the order of the cosine similarity between the first vector and adjustment vector Q from large to small, the numbers of all first vectors are sequentially increased; the directional consistency between adjustment vector q and adjustment vector Q and the cosine similarity between adjustment vector q and adjustment vector Q are positively correlated, and negatively correlated with the number of adjustment vector q.

6. The aging test method for rubber-modified asphalt mixtures under multi-factor coupling conditions according to claim 4, characterized in that, The specific steps involved in extending the testing duration of the new environmental data predicted from the reference data are as follows: The adjustment vectors with the greatest directional consistency are denoted as reference vectors. The mean of the confidence of all reference vectors is obtained. The extended test duration is negatively correlated with the mean, representing the duration of the aging test under environmental data p.

7. The aging test method for rubber-modified asphalt mixtures under multi-factor coupling conditions according to claim 1, characterized in that, The specific steps for obtaining the aging differences under each environmental data are as follows: The absolute value of the difference in aging assessment indicators obtained from aging tests on the control group and the test group under each environmental data is denoted as the aging difference. The method for obtaining the aging assessment indicators is as follows: Material strength tests were conducted on the control group or test group to obtain the strength limit; the difference between the strength limit of the control group or test group before the start of the aging test and the strength limit after the end of the aging test was obtained, and the aging assessment index of the control group or test group was positively correlated with this difference.

8. The aging test method for rubber-modified asphalt mixtures under multi-factor coupling conditions according to claim 3, characterized in that, When the value of any dimension in the new environmental data exceeds the preset range for that dimension, the new environmental data is redefined, including the following specific steps: Dimensions in the new environmental data that exceed the preset value range are denoted as out-of-limit dimensions. The environmental data P1 is summed with w times the difference vector to obtain temporary environmental data, where w represents an undetermined coefficient. The excess amount of each out-of-limit dimension in the temporary environmental data is calculated, and the value of w that makes the excess amount of each out-of-limit dimension equal to 0 is solved and denoted as the preferred coefficient of each out-of-limit dimension. The minimum value of the preferred coefficients of all out-of-limit dimensions is denoted as w0, and the temporary environmental data obtained under w0 is used as the newly determined environmental data.

9. The aging test method for rubber-modified asphalt mixtures under multi-factor coupling conditions according to claim 8, characterized in that, When the magnitude of the newly determined environmental data is less than a preset percentage of the magnitude of the difference vector, the difference between P1 and w1 times the difference vector is calculated and used as the new environmental data, where w1 is a scaling factor less than 1.

10. The method for aging test of rubber-modified asphalt mixture under multi-factor coupling conditions according to claim 8, characterized in that, The specific steps for calculating the excess amount of each out-of-limit dimension in the temporary environment data that exceeds the value range are as follows: If the value of any out-of-limit dimension exceeds the upper limit of the preset value range of that out-of-limit dimension, the difference between the value of the out-of-limit dimension and the upper limit of the value range is recorded as the excess amount; if the value of any out-of-limit dimension exceeds the lower limit of the preset value range of that out-of-limit dimension, the difference between the lower limit of the value range and the value of the out-of-limit dimension is recorded as the excess amount.

Images