A dynamic modulus test system, method and prediction method for asphalt mixture

Abstract

This invention provides a dynamic modulus testing system, method, and prediction method for asphalt mixtures, belonging to the technical field of asphalt material testing. It includes a test chamber for placing multiple asphalt mixture specimens; a strain acquisition component for acquiring the strain of the specimens placed in the test chamber due to load; a temperature acquisition component for acquiring the temperature of the specimens placed in the test chamber; a circulating temperature control device connected to the test chamber for controlling the temperature of the specimens placed in the test chamber, ensuring the dynamic modulus test is performed at a target temperature; and an accelerated loading device including a tire, which accelerates the loading of the specimens placed in the test chamber by adjusting the tire. This invention can simulate the testing of the dynamic modulus of asphalt mixtures under real-world conditions and achieve accurate prediction of the dynamic modulus of asphalt mixtures.

Description

Technical Field

[0001] This invention relates to the field of asphalt material testing technology, and in particular to a dynamic modulus testing system, method, and prediction method for asphalt mixtures. Background Technology

[0002] Traditional asphalt pavement design typically uses static modulus for pavement mechanics calculations. However, asphalt pavement operates under varying conditions, including dynamic loads at different temperatures, frequencies, and load ranges. Therefore, static modulus cannot objectively reflect the mechanical state of asphalt pavement, and its calculation results often deviate significantly from actual working conditions, leading to a disconnect between asphalt pavement design, construction, and performance. Therefore, it is necessary to test the dynamic modulus of asphalt mixtures.

[0003] The dynamic modulus, as an important material parameter, has the following mathematical expression: The dynamic modulus reflects the relationship between compressive stress and strain changes in a material under strain, exhibiting strong material matching ability and reliability. For asphalt mixtures, which need to withstand changes in vehicle traffic loads and climate temperature during long-term use, the assessment and control of the dynamic modulus is of great significance for ensuring pavement quality and service life. Summary of the Invention

[0004] To address the above problems, this invention provides a dynamic modulus testing system, method, and prediction method for asphalt mixtures, enabling accurate measurement and precise prediction of asphalt mixtures.

[0005] This invention provides a dynamic modulus testing system for asphalt mixtures, comprising a test chamber, a strain acquisition component, a temperature acquisition component, a circulating temperature control device, and an accelerated loading device, wherein: The test trough is used to place multiple specimens prepared from asphalt mixtures; The strain acquisition component is used to acquire the strain of the specimen placed on the test chamber due to the load; The temperature acquisition component is used to acquire the temperature of the specimen placed on the test tank; The circulating temperature control device is connected to the test tank and is used to control the temperature of the specimen placed on the test tank so that the specimen can be subjected to dynamic modulus testing at the target temperature. The accelerated loading device includes a tire, which accelerates the loading of the specimen placed on the test groove by adjusting the tire.

[0006] As a further improvement of the present invention, the strain acquisition component includes a strain gauge and a strain data acquisition device. The strain gauge is used to connect with the specimen to monitor the strain of the specimen, and the strain data acquisition device is connected with the strain gauge to acquire the strain of the specimen monitored by each strain gauge.

[0007] As a further improvement of the present invention, the temperature acquisition component includes a temperature sensor and a temperature data acquisition device. The temperature sensor is used to connect to the specimen to monitor the temperature of the specimen, and the temperature data acquisition device is connected to the temperature sensor to acquire the temperature of the specimen monitored by each of the temperature sensors.

[0008] This invention provides a method for testing the dynamic modulus of asphalt mixtures, comprising: Multiple asphalt mixture specimens were prepared, and strain acquisition components and temperature acquisition components were connected to the specimens respectively. The specimens are placed sequentially in the test chamber, and the specimens are heated using a circulating temperature control device. The temperature of the specimens is collected using the temperature acquisition component, and the specimens are kept warm for a set period of time after the target temperature is reached. The accelerated loading device is activated to load the specimen at a set loading pressure and / or loading frequency, and the strain amplitude of the specimen is acquired using the strain acquisition component. The dynamic modulus of the specimen was calculated based on the loading pressure and strain amplitude.

[0009] As a further improvement of the present invention, the target temperature range is 0-60℃, the heat preservation time range is 0.5h-6h, the loading pressure range is 0.5~1.2MPa, and the loading frequency range is 0~2.5Hz.

[0010] As a further improvement of the present invention, the connection position of the strain acquisition component and the temperature acquisition component with the specimen is located at 1 / 2 height of the specimen.

[0011] This invention provides a method for predicting the dynamic modulus of asphalt mixtures, comprising: Historical data of dynamic modulus testing of asphalt mixtures are obtained, and a dynamic modulus dataset is constructed using the historical data, wherein the historical data includes raw material parameters, working condition parameters and dynamic modulus data of asphalt mixtures; The dynamic modulus dataset is divided into a training set and a test set. The improved neural network model is trained using the training set, and the trained neural network model is tested using the test set to obtain a dynamic modulus prediction model. The raw material parameters and working condition parameters of the asphalt mixture to be predicted are input into the dynamic modulus prediction model to calculate the dynamic modulus prediction result of the asphalt mixture.

[0012] As a further improvement of the present invention, the raw material parameters include asphalt content, asphalt viscosity at a specific temperature, aggregate lithology, aggregate sphericity, and gradation factor, and the working condition parameters include temperature, speed, load level, and number of loading cycles.

[0013] As a further improvement of the present invention, after constructing a dynamic modulus dataset using the historical data, the dynamic modulus dataset is preprocessed to generate a standardized dynamic modulus dataset.

[0014] As a further improvement of the present invention, the improved neural network model includes an input layer, an attention calculation layer, a hidden layer and an output layer, wherein the hidden layer includes a first hidden layer and a second hidden layer, the first hidden layer adopts the LeakyReLU activation function and the second hidden layer adopts the ELU activation function.

[0015] This invention provides a dynamic modulus testing system, method, and prediction method for asphalt mixtures. It addresses the problems of existing uniaxial compression dynamic modulus tests failing to simulate actual operating environments and load conditions, and the difficulty in accurately predicting dynamic modulus when testing conditions are lacking. This invention uses a standardized specimen and a water-circulation temperature control system combined with a small-scale accelerated loading device to simulate actual load and temperature environments to test the dynamic modulus. Simultaneously, based on neural network machine learning, it predicts the dynamic modulus by inputting raw material and test condition parameters. The method of this invention can accurately reflect the actual stress state of asphalt mixtures and can make accurate predictions even when testing conditions are lacking, making it suitable for asphalt pavement design and performance evaluation. Attached Figure Description

[0016] Figure 1 This is a front view of the dynamic modulus testing system for asphalt mixtures according to an embodiment of the present invention.

[0017] Figure 2 This is a top view of the test chamber portion of the asphalt mixture dynamic modulus testing system according to an embodiment of the present invention.

[0018] Figure 3 This is a schematic diagram of the test specimen.

[0019] Figure 4 This is a flowchart illustrating the dynamic modulus testing method for asphalt mixtures according to an embodiment of the present invention.

[0020] Figure 5 This is a flowchart illustrating the dynamic modulus prediction method for asphalt mixtures according to an embodiment of the present invention.

[0021] Explanation of reference numerals in the attached drawings: 1. Specimen; 2. Test chamber; 3. Strain acquisition assembly; 31. Strain gauge; 32. Strain data acquisition device; 4. Temperature acquisition assembly; 41. Temperature sensor; 42. Temperature data acquisition device; 5. Circulating temperature control device; 51. Circulating temperature control chamber; 52. Liquid outlet pipeline; 53. Liquid return pipeline; 6. Accelerated loading device; 61. Accelerated loading device; 611. Tire; 612. Transmission mechanism; 62. Equipment controller. Detailed Implementation

[0022] The following describes specific embodiments and appendices. Figure 1 - Appendix Figure 5 The invention is described in detail so that those skilled in the art can more fully understand its purpose, features and effects.

[0023] Unless otherwise specified, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. In the event of any discrepancy between the definitions of terms in this invention and their commonly understood meaning by one of ordinary skill in the art to which this invention pertains, the definitions provided in this invention shall prevail.

[0024] Currently, domestic and international design standards generally use the uniaxial compression dynamic modulus test to test the dynamic characteristics of asphalt mixtures. This method has the following shortcomings: (1) The stress mode is simple and cannot reflect the confining pressure effect of nearby structures; (2) It is limited by the test equipment. For example, the commonly used SPT (Simple Performance Tester), UMT and MTS test equipment can only control the temperature of asphalt mixtures under dry conditions and cannot reflect the water-temperature coupling environment of asphalt mixtures; (3) The test range of this method is linear viscoelasticity, while asphalt mixtures are viscoelastic-plastic materials. With the increase of the service life, the dynamic modulus of asphalt mixtures is constantly changing. Therefore, it is necessary to realistically simulate the actual environmental conditions and load conditions of asphalt mixtures in order to test their dynamic modulus.

[0025] This invention provides a dynamic modulus testing system and method for asphalt mixtures, which solves the problems of existing dynamic modulus testing methods for asphalt mixtures being unable to simulate actual operating environments (such as temperature changes and real load patterns), having a disconnect between the stress mode and reality, and being unable to accurately predict the dynamic modulus when lacking test conditions. Example 1

[0026] As a specific embodiment of the present invention, this embodiment provides a dynamic modulus testing system for asphalt mixtures, referring to... Figure 1 , Figure 2It includes a test chamber 2 for placing multiple specimens 1, a strain acquisition component 3, a temperature acquisition component 4, a circulating temperature control device 5, and an accelerated loading device 6. The specimens 1 are made of asphalt mixture and are made of the same material as the road surface at the construction site.

[0027] In this embodiment, the test groove 2 provides stable support for the specimen 1, fixing it in place. Specimens 1 of predetermined specifications and shapes are arranged sequentially along the length of the test groove 2, with multiple specimens arranged to allow for testing the dynamic modulus of multiple specimens 1 in a single test. Adjacent specimens 1 are tightly fitted without gaps, simulating a real road surface, thus providing mutual support and mimicking the confining pressure of surrounding specimens on the stressed specimen, reducing deformation perpendicular to the pressure loading direction. To enhance the stability of the specimen 1 within the test groove 2, thermally conductive silicone is filled into the gap between the specimen 1 and the test groove 2, with a 100% filling rate.

[0028] Optionally, the test tank 2 is made of stainless steel.

[0029] The strain acquisition component 3 is used to acquire the strain of each specimen 1 under pressure during the test, and the temperature acquisition component 4 is used to acquire the ambient temperature during the test. The temperature acquisition component 4 can be configured to acquire only the overall ambient temperature of all specimens 1, or it can be configured to acquire the ambient temperature of each individual specimen 1.

[0030] The circulating temperature control device 5 is used to adjust the ambient temperature of the specimen 1 during the test, so that the specimen 1 can be tested for dynamic modulus at the target temperature. The circulating temperature control device 5 allows the specimen 1 to be tested under various ambient temperatures to obtain the influence of ambient temperature on the dynamic modulus test results of asphalt mixtures.

[0031] The accelerated loading device 6 is used to accelerate the loading output of multiple specimens 1 located on the test chamber 2, so that the specimens 1 deform under stress, so that the strain acquisition component 3 can acquire the strain of the specimens 1.

[0032] In this embodiment of the asphalt mixture dynamic modulus testing system, the specimen 1 is first placed sequentially on the test chamber 2 and fixed to prevent movement during the test, ensuring that the upper surfaces of all specimens 1 are at the same level. The circulating temperature control device 5 is then activated to bring the ambient temperature of the specimen 1 to the target temperature. The accelerated loading device 6 is then activated to test the specimen 1 under a set pressure. The strain acquisition component 3 is used to monitor the strain of the specimen 1 during the test, and finally, the dynamic modulus of the asphalt mixture is obtained based on the monitoring results.

[0033] Specifically, the shape of the specimen 1 is as follows: Figure 3 As shown, from left to right, the views are the top view, front view, and left view of the specimen 1. Specifically, the horizontal cross-section of the specimen 1 is waist-shaped, which facilitates molding. Preferably, the specimen 1 has a length of 150mm, a width of 100mm, and a height of 100mm.

[0034] The specimen 1 can be made by the following method: First, make an asphalt mixture cylindrical sample with a diameter of 150mm and a height of not less than 120mm; then cut the top and bottom surfaces of the cylindrical sample in parallel to make its height 100mm; cut the asphalt mixture with a thickness of 25mm in parallel along both sides of the axis of the cylindrical sample to finally form the specimen 1 with a diameter of 150×100×100mm.

[0035] In other feasible embodiments, the shape and size of the specimen 1 may also be different from that of this embodiment. For example, the shape of the specimen 1 may be a cuboid or a cube, and the height may be greater than 100mm or less than 100mm.

[0036] During the processing of specimen 1, a rotary compactor was used to compact and shape an asphalt mixture cylinder with the same design mix as the road surface at the construction site, with a diameter of 150mm. After cooling to room temperature, it was cut into 150×100×100mm size using a cutting machine.

[0037] In this embodiment, nine specimens 1 are required for one test, that is, nine specimens 1 are fixed on the test tank 2 along its length.

[0038] Among them, the test specimen 1 can be a prefabricated test block in the laboratory or a field test block after the road construction is completed. Field sampling can ensure consistency, has stronger representativeness, and reflects the true dynamic modulus of the road surface.

[0039] Specifically, the strain acquisition component 3 includes a strain gauge 31 and a strain data acquisition device 32. For example... Figure 2 The strain gauge 31 is located on the side (arc-shaped surface) of the specimen 1 and is fixedly connected to the specimen 1 by adhesive. The strain gauge 31 may be installed on only one side of the specimen 1. The number of strain gauges 31 is equal to the number of specimens 1; when there are 9 specimens 1, the number of strain gauges 31 is also 9. Multiple strain gauges 31 are uniformly connected to a single strain data acquisition device 32.

[0040] During the test, the strain gauge 31 is fixed to the side of the specimen 1, and the strain gauge 31 is connected to the strain data acquisition device 32.

[0041] Preferably, the strain data acquisition device 32 can acquire and display the strain data of each specimen 1 in real time, so that the test personnel can observe the strain of the specimen 1 at any time.

[0042] Specifically, the temperature acquisition component 4 includes at least one temperature sensor 41 and a temperature data acquisition unit 42. The temperature sensor 41 is fixed on the side of the specimen 1 away from the strain gauge 31. Optionally, the number of temperature sensors 41 is equal to the number of specimens 1, that is, each side of the specimen 1 is provided with a temperature sensor 41 to accurately monitor the ambient temperature of each specimen 1. When there are multiple temperature sensors 41, the multiple temperature sensors 41 are connected to one temperature data acquisition unit 42.

[0043] During the test, the temperature sensor 41 is fixed to the side of the test piece 1, and the temperature sensor 41 is connected to the temperature data acquisition device 42.

[0044] Preferably, the temperature data acquisition device 42 can display the test data of each specimen 1 in real time, which allows the test personnel to keep track of the ambient temperature of the specimen 1 at any time and ensure that the test is carried out at the target temperature. If the real-time temperature is found to deviate from the target temperature, the circulating temperature control device 5 is activated to adjust the temperature so that the ambient temperature reaches the target temperature.

[0045] Preferably, the strain gauge 31 and the temperature sensor 41 are fixed at 1 / 2 height of the specimen 1. The temperature sensor 41 is a PT100 temperature sensor (resistance temperature sensor), which features high accuracy, high stability, excellent linearity, high environmental adaptability, and convenient maintenance.

[0046] Specifically, the circulating temperature control device 5 includes a circulating temperature control chamber 51 (constant temperature water tank) and an outlet pipe 52 and a return pipe 53 connected to the test tank 2. Under the action of a circulating pump in the circulating temperature control chamber 51, water flows out from the outlet pipe 52 and returns to the circulating temperature control chamber 51 via the return pipe 53. The temperature of the specimen 1 is regulated through heat conduction between the tank wall of the test tank 2 and the thermally conductive silicone. The circulating temperature control chamber 51 uses water circulation to control the temperature of the test tank 2. Once the target temperature is reached, it can maintain a constant temperature for a long time. The outlet pipe 52 is used to deliver hot liquid to the test tank 2, and the return pipe 53 is used for liquid recovery. It should be noted that the specimen 1 does not directly contact the water in the test tank 2. The circulating temperature control device 5 transfers heat to the test tank 2 through a heat exchange structure. The heat exchange structure can use existing structures in the prior art, and this embodiment does not improve or optimize it. For example, a loop is set on the outer wall of the test tank 2, and water is circulated in the loop to achieve temperature control through heat conduction.

[0047] The circulating temperature control device 5 simulates the water-temperature coupling environment of the road surface. By combining the water-circulating temperature control device 5 with thermally conductive silicone, the temperature of the specimen 1 is uniformly controlled (fluctuation ≤ 0.3℃), which can simulate temperature changes in different climate zones and solve the problem that traditional equipment cannot simulate water-temperature coupling environments. In other embodiments, to simulate working conditions at even lower temperatures, air circulation can be used, achieving temperature control through heating and cooling the air.

[0048] Specifically, the accelerated loading device 6 includes an accelerated loader 61 and a device controller 62. The device controller 62 controls the accelerated loader 61, which is located above the test chamber 2 to enable loading of the specimen 1 on the test chamber 2. The accelerated loading device 6 is a small accelerated loading device in the prior art, such as the MMLS3 accelerated loading test device from MLS Test Systems Pty Ltd. The following embodiment only describes the important structures.

[0049] The accelerator 61 includes a tire 611 and a transmission mechanism 612. The transmission mechanism 612 drives the tire 611 to move and rotate along the length of the test groove 2, sequentially applying pressure to the specimen 1 located on the test groove 2, causing the specimen 1 to undergo strain. The pressure of the tire 611 and the loading frequency (number of times) can be adjusted as needed.

[0050] Small-scale accelerated loading equipment is used to simulate tire load (contact pattern and pressure distribution are close to those of actual vehicles), replacing the axial force of traditional single-axis compression, which is more in line with the actual stress state of asphalt mixture pavement.

[0051] In this embodiment, the test tank 2 and the accelerated loading device 6 are separate units. During the test, the accelerated loading device 6 has independent support (not shown in the figure) and can be set up independently.

[0052] The asphalt mixture dynamic modulus testing system provided in this embodiment, after loading, calculates the strain amplitude based on the strain gauge 31 at the loading number T. and the applied tire pressure (i.e., loading stress), the dynamic modulus of the asphalt mixture when the loading number is T is calculated as follows:

[0053] The asphalt mixture dynamic modulus testing system provided in this embodiment can adjust tire pressure and loading times according to the traffic load level of highway asphalt pavement, and adjust test temperature and loading frequency according to design and other needs, covering different operating scenarios. It is used to test the dynamic modulus of asphalt mixture under actual working conditions. The method has high reliability and accurate results. Example 2

[0054] As a specific embodiment of the present invention, this embodiment provides a method for testing the dynamic modulus of asphalt mixtures, using the testing system of Embodiment 1, and referring to... Figure 4 The specific steps are as follows: S101. Multiple asphalt mixture specimens are prepared, and strain acquisition components and temperature acquisition components are connected to the specimens respectively; S102. The test specimens are placed in the test chamber in sequence, and the test specimens are heated by the circulating temperature control device. The temperature of the test specimens is collected by the temperature acquisition component. After the target temperature is reached, the test specimens are kept warm for a set period of time. S103. Start the accelerated loading device and load the specimen under the set loading pressure and / or loading frequency, and use the strain acquisition component to acquire the strain amplitude of the specimen. S104. Calculate the dynamic modulus of the specimen based on the loading pressure and strain amplitude.

[0055] Specifically, in this embodiment S101, the specimen 1 can be prepared in the following two ways: 1. First, in the laboratory, prepare cylindrical asphalt mixture specimens with a diameter of 150mm according to the same design proportions as the road surface at the construction site. The height of the cylindrical specimens should not be less than 120mm, and the height should be 150mm. After cooling to room temperature, use a cutting machine to cut the top and bottom surfaces of the cylindrical specimens parallel to each other, making the height 100mm. Then, cut away 25mm thick portions of asphalt mixture parallel to both sides of the cylindrical specimen's axis, finally forming specimen 1, which is 150×100×100mm. Figure 3As shown, multiple specimens 1 are finally obtained. In this embodiment, there are 9 specimens 1. The asphalt mixture cylindrical sample is formed by rotary compaction using a rotary compactor with a compaction pressure of 600 kPa and a compaction number of 60-80 times. If the compaction number is 70 times, the porosity after compaction is between 3%-6%. If the porosity is 4.5%, the density fluctuation range of the specimen 1 does not exceed ±0.03 g / cm³. The cutting machine precision is controlled, and the parallelism of the two end faces of the specimen 1 is checked after cutting to ensure that the error is ≤0.2 mm, thus ensuring uniform stress in the later stage. 2. First, take a cylindrical asphalt mixture specimen with a diameter of 150mm from the road surface at the construction site. The height of the specimen should not be less than 120mm. Then, in the laboratory, use a cutting machine to cut the top and bottom surfaces of the cylindrical specimen to a height of 100mm. Cut away 25mm thick asphalt mixture along both sides of the cylindrical specimen's axis to finally form a 150×100×100mm specimen (sample 1). Figure 3 As shown, multiple specimens 1 are finally obtained. In this embodiment, there are 9 specimens 1. After cutting, the parallelism of the two end faces of the specimen 1 is tested to ensure that the error is ≤0.2mm, so as to ensure uniform stress in the later stage.

[0056] In other feasible embodiments, the specimen 1 may also be prepared in other sizes.

[0057] The raw material parameters of asphalt mixtures include asphalt content, viscosity at 135℃, aggregate lithology, aggregate sphericity, and gradation factor. For example, in one instance, specimen 1 is composed of 70# base asphalt and basalt aggregate (hard rock lithology). The asphalt content is 5%, the viscosity at 135℃ is 600 mPa·s, the basalt aggregate sphericity is 0.65, and the gradation factor is 1.0.

[0058] The strain gauges 31 of the strain acquisition assembly 3 and the temperature sensors 41 of the temperature acquisition assembly 4 are fixed to the side of the specimen 1. Multiple strain gauges 31 are each connected to a strain data acquisition unit 32, and multiple temperature sensors 41 are each connected to a temperature data acquisition unit 42. Preferably, the strain gauges 31 and temperature sensors 41 are bonded to the specimen 1. After 24 hours of curing, the conductivity of the temperature sensors 41 is checked to ensure stable and reliable temperature monitoring. Alternatively, other convenient connection methods can also be used for the connection of the strain gauges 31 and temperature sensors 41 to the specimen 1.

[0059] Preferably, the strain gauge 31 and the temperature sensor 41 are fixed at 1 / 2 height of the specimen 1.

[0060] Two measuring points are symmetrically marked radially at half the height of the side of the specimen 1. The surface of the measuring points is sanded with sandpaper, strain gauge adhesive is applied, and a 120Ω strain gauge 31 (sensitive grid size 3mm×5mm) is attached. The sensitive grid axis of the strain gauge 31 is aligned with the radial direction of the specimen 1. The strain gauge is left to cure for 24 hours. A PT100 temperature sensor is attached to the opposite side of the strain gauge 31 and fixed with high-temperature resistant tape. The gap between the sensor and the specimen surface is ≤0.1mm.

[0061] In other feasible embodiments, the connection of the strain acquisition component 3 and the temperature acquisition component 4 to the specimen 1 can also be performed after the specimen 1 is placed in the test tank 2.

[0062] Furthermore, in step 102 of this embodiment, the circulating temperature control device 5 is activated to heat the specimen 1, and after reaching the target temperature, it is kept at that temperature for a set period of time. The target temperature range is 0-60℃, such as 20℃, 50℃, 55℃, or 60℃, with a holding error ≤0.3℃, and the holding time range is 0.5h-6h, such as 3h, 4h, or 5h. This embodiment can simulate and calculate the dynamic modulus of asphalt mixture at different temperatures by adjusting the target temperature.

[0063] Preferably, the circulating temperature control device 5 is activated to heat the specimen 1, adjusting the temperature of the specimen 1 to the target temperature at a rate of 0.5-1℃ / min.

[0064] In S102 of this embodiment, the temperature sensor 41 is used to collect and feedback the temperature of the specimen 1 in real time to ensure the reliability of temperature monitoring.

[0065] Further, in S103 of this embodiment, the set loading pressure is based on the tire pressure of the tire 611 of the accelerated loading device 6. The tire pressure of the tire 611 is adjusted by an air compressor. The tire pressure range is 0.5~1.2MPa, such as 0.5MPa, 0.6MPa, 0.7MPa, 0.8MPa, or 0.9MPa. This embodiment can simulate normal and heavy load traffic conditions by adjusting the tire pressure. Optionally, the wheel width of the tire 611 is 60mm to ensure uniform contact between the tire 611 and the surface of the specimen 1, with a contact pressure distribution deviation ≤5%, and the tire 611 contacts the specimen 1 to simulate the tire ground contact pattern.

[0066] The loading frequency of the test is set by the device controller 62 of the accelerated loading device 6. The loading frequency range is 0~2.5Hz, such as 1Hz, 1.5Hz, 2Hz, and 2.5Hz. The loading frequency is matched with the driving speed; a speed of 20km / h corresponds to a loading frequency of 0.5Hz, and a speed of 100km / h corresponds to a loading frequency of 2.5Hz.

[0067] After adjusting the tire pressure and loading frequency, the specimen 1 is loaded, and the strain amplitude of the specimen 1 is collected using the strain acquisition component 3. The loading path is a reciprocating (circular rolling) motion along the central area of ​​the surface of the specimen 1. This embodiment can simulate the dynamic load effect of a vehicle on the road surface at different speeds by adjusting the loading frequency.

[0068] In S103 of this embodiment, the tire pressure (i.e., loading pressure) of the tire 611 is The strain amplitude of the specimen 1 when the loading number is T is recorded by the strain gauge 31 of the strain acquisition component 3. The stable strain amplitude (the average of 100 consecutive data points) is extracted every 5 minutes. Preferably, the strain data acquisition frequency of the strain acquisition component 3's strain data acquisition device 32 is 10kHz.

[0069] In S104 of this embodiment, according to the loading pressure and strain amplitude The dynamic modulus of specimen 1 was obtained as follows:

[0070] Since each of the specimens 1 is provided with a strain gauge 31, the strain data of each specimen 1 can be monitored and collected and its dynamic modulus can be calculated.

[0071] In one feasible embodiment, the parameters involved in the method are as follows: Specimen: Waist-shaped cross section, dimensions 150×100×100mm; Target temperature: 25℃; Insulation period: 0.5 hours; Tire pressure: 0.7 MPa; Loading frequency: 1.5Hz (driving speed 60km / h).

[0072] The accelerated loading device 6 is activated, and loading is performed for 5 minutes. The strain amplitude is... If the value stabilizes at 80 με, the calculated result of the dynamic modulus is: .

[0073] In one feasible embodiment, the parameters involved in the method are as follows: Specimen: Waist-shaped cross section, dimensions 150×100×100mm; Target temperature: 40℃; Insulation period: 0.5 hours; Tire pressure: 0.7 MPa; Loading frequency: 1.5Hz (driving speed 60km / h).

[0074] Using the same specimen 1, after the strain amplitude stabilized under loading, the dynamic modulus of the asphalt mixture measured under this condition was 6200 MPa. The dynamic modulus of the asphalt mixture decreased significantly with increasing temperature. Increased temperature reduces the viscosity of the asphalt mixture, weakens intermolecular forces, and consequently reduces the internal cohesion and resistance to deformation, leading to a decrease in the dynamic modulus.

[0075] In one feasible embodiment, the parameters involved in the method are as follows: Specimen: Waist-shaped cross section, dimensions 150×100×100mm; Target temperature: 25℃; Insulation period: 0.5 hours; Tire pressure: 1.0 MPa; Loading frequency: 1.5Hz (driving speed 60km / h).

[0076] Using the same specimen 1, after the strain amplitude stabilized under loading, the dynamic modulus of the asphalt mixture measured under this condition was 9500 MPa. The dynamic modulus of the asphalt mixture increased with increasing pressure. Increased pressure compacts the aggregate skeleton within the asphalt mixture, reducing the voids between aggregate particles, thereby enhancing the interlocking effect and contact area between particles, and ultimately strengthening the overall stiffness of the asphalt mixture, leading to a corresponding increase in the dynamic modulus.

[0077] In one feasible embodiment, the parameters involved in the method can also be set as follows: Specimen: Waist-shaped cross section, dimensions 150×100×100mm; Target temperature: 60℃; Insulation period: 4 hours; Tire pressure: 0.7 MPa; Loading frequency: 2.5Hz (driving speed 100km / h).

[0078] The dynamic modulus of asphalt mixtures was tested under higher temperatures, longer insulation times, and higher loading frequencies.

[0079] The dynamic modulus test method for asphalt mixtures in this embodiment allows for adjustments to tire pressure and loading cycles based on the traffic load level of the highway asphalt pavement, and adjustments to test temperature and loading frequency based on pavement design requirements, thereby obtaining the dynamic modulus of asphalt mixtures under different working conditions. Example 3

[0080] As a specific embodiment of the present invention, this embodiment provides a method for predicting the dynamic modulus of asphalt mixtures, referring to... Figure 5 The specific steps are as follows: S201. Obtain historical data of dynamic modulus testing of asphalt mixtures and construct a dynamic modulus dataset using the historical data. In S201, the historical data includes asphalt mixture raw material parameters, working condition parameters, and dynamic modulus data, which are the objects to be processed. The asphalt mixture raw material parameters include asphalt content (representing the proportion of asphalt in the mixture), asphalt viscosity at a specific temperature (representing the flow characteristics of asphalt under specific heating conditions), aggregate lithology (representing the mineral composition of the aggregate), aggregate sphericity (representing the degree to which the shape of aggregate particles approximates a sphere), and gradation factor (representing the uniformity of aggregate particle size distribution), reflecting the inherent properties of the material. The working condition parameters include temperature (representing the temperature of the external environment in which the mixture exists), speed (representing the speed at which vehicles pass over the road surface), load level (tire pressure, representing the pressure exerted by vehicle tires on the road surface), and number of loading cycles (representing the total number of times the road surface is subjected to vehicle loads), representing external conditions. The dynamic modulus data is affected by the asphalt mixture raw material parameters and working condition parameters. Using the asphalt mixture raw material parameters and working condition parameters from the historical data as features, and the dynamic modulus data as labels, a dynamic modulus dataset with no fewer than 1000 samples is formed.

[0081] In other feasible embodiments, the raw material parameters of the mixture also include the specific surface area of ​​mineral powder (reflecting the surface area of ​​mineral powder particles), and the operating condition parameters also include humidity (reflecting the air humidity of the external environment where the mixture is located) and the number of freeze-thaw cycles (reflecting the number of times the mixture undergoes alternating freezing and thawing).

[0082] Optionally, the specific temperature viscosity of the asphalt can be the viscosity at 135°C. In other feasible embodiments, viscosities at other temperatures may be used.

[0083] The aggregate lithology can be limestone, basalt, granite, or diabase, with an aggregate sphericity range of 0.5-0.9 and a gradation factor range of 0.6-3.0.

[0084] Temperature range: 0-60℃; speed range: 20-100km / h; load level (tire pressure) range: 0.5-1.2MPa; loading cycles range: 1000-10000.

[0085] Optionally, the dynamic modulus data of a certain asphalt mixture can be obtained using the asphalt mixture dynamic modulus testing system in Example 1 and the asphalt mixture dynamic modulus testing method in Example 2. In other embodiments, the dynamic modulus of a certain asphalt mixture can also be obtained using methods already available in the prior art.

[0086] For specimen 1, the asphalt content can be determined by extraction testing with an accuracy of ±0.1%; the viscosity of asphalt at 135℃ can be tested using a Brookfield rotational viscometer with a temperature fluctuation of ±0.5℃; the aggregate lithology, such as limestone, basalt, granite, and diabase, can be determined using X-ray diffraction; the sphericity of the aggregate can be calculated using image recognition technology, based on the comparison of the equivalent circle of the aggregate's projected area and perimeter; and the gradation factor can be calculated based on the gradation uniformity coefficient of the gradation curve.

[0087] Specifically, the testing methods for each parameter of the raw materials for asphalt mixtures are as follows.

[0088] For the viscosity of asphalt at 135℃, a Brookfield rotational viscometer was used as the measuring instrument. The instrument temperature was set to 135℃. After the instrument temperature stabilized, asphalt was added to the measuring container of the instrument. The instrument was started to make the rotor rotate in the asphalt sample. The instrument calculated and output the viscosity value of asphalt at 135℃ by detecting the resistance encountered by the rotor during rotation. This value is used as the viscosity of asphalt at a specific temperature.

[0089] For aggregate lithology, an X-ray diffractometer is used as the detection instrument. After the aggregate is ground into fine powder, it is placed on the sample stage of the instrument. The instrument is started to emit X-rays to irradiate the sample. The instrument analyzes and determines the mineral composition of the aggregate sample based on the diffraction pattern of the sample to X-rays, thereby classifying it into four types of aggregate lithology: limestone, basalt, granite, and diabase.

[0090] For aggregate sphericity, a two-dimensional projection image of the aggregate particles is acquired through an image acquisition device. Image recognition technology is used to process the image to extract the projected area and perimeter of each aggregate particle. Then, the area of ​​a circle with the same perimeter (i.e., the equivalent circle area) is calculated according to a geometric formula. Finally, the ratio of the projected area of ​​the aggregate particle to the equivalent circle area is used as the aggregate sphericity. The value ranges from 0 to 1, and the closer it is to 1, the more regular the shape.

[0091] For the gradation factor, the gradation data of the aggregate (passing rate corresponding to different sieve sizes) is obtained through sieving tests. The aggregate gradation curve is plotted based on the gradation data, and then the uniformity coefficient of the curve is calculated, which is the ratio of the 85% sieve size to the 15% sieve size. This uniformity coefficient is used as the gradation factor.

[0092] When the raw material parameters of the mixture include the specific surface area of ​​mineral powder, the specific surface area of ​​the mineral powder is determined using the Blaine method. Specifically, a dry mineral powder sample is selected, and the Blaine method is used as the determination method. The mineral powder sample is placed in a specific sample tube, and the specific surface area of ​​the mineral powder sample is calculated by measuring the time it takes for air to pass through the sample tube under certain pressure and temperature, combined with relevant formulas.

[0093] Furthermore, the parameters in the dynamic modulus dataset are classified and standardized preprocessed to generate a standardized dynamic modulus dataset, for example: Numerical parameters (such as asphalt content and temperature) are normalized by mapping them to a preset numerical range according to a pre-defined normalization formula. For example, the normalization formula x'=(xx) is used. min ) / (x max -x min This method maps numerical data to the [0,1] interval to eliminate the influence of different parameters due to differences in units, and generates normalized numerical parameters. One-Hot Encoding is performed on category parameters (such as aggregate lithology) to convert their category information into computer-recognizable binary data and generate encoded category parameters. For example, when the aggregate lithology is limestone, basalt, granite, or diabase, its one-hot encoding is 1000, 0100, 0010, or 0001. Outlier processing is performed based on the 3σ principle, removing data that deviates from the mean of the parameters by more than three standard deviations to eliminate outliers and improve data quality.

[0094] The detailed process of normalizing numerical parameters is as follows: The process involves obtaining all the original values ​​of a numerical parameter, calculating its minimum and maximum values, subtracting the minimum value from each original value to obtain the difference, and finally dividing the difference by the difference between the maximum and minimum values ​​to obtain the normalized value. This process maps all numerical parameters to a preset range of 0 to 1, generating normalized numerical parameters. This avoids the model becoming overly sensitive to parameters with large values ​​due to different parameter units (such as asphalt content in percentage units and asphalt viscosity at a specific temperature in millipascals per second units).

[0095] The detailed process of performing unique thermal coding on aggregate lithology is as follows: Establish a mapping table for aggregate lithology categories (limestone, basalt, granite, diabase). Create an aggregate lithology-code mapping table where each aggregate lithology corresponds to a unique binary code, and the binary codes of different aggregate lithologies do not overlap. Limestone corresponds to 1000, basalt to 0100, granite to 0010, and diabase to 0001. Generate the aggregate lithology-code mapping table. Based on the aggregate lithology-code mapping table, each aggregate lithology data in the raw material parameters is converted into a corresponding binary code to generate the coded aggregate lithology parameters; The coding results are verified for completeness. All aggregate lithology data that need to be coded have been coded to ensure that no aggregate lithology data is missing. If any data is missing, it is coded to obtain the complete coded aggregate lithology parameters.

[0096] After preprocessing, a standardized dynamic modulus dataset is obtained, which is used for subsequent model training.

[0097] S202. Divide the dynamic modulus dataset into a training set and a test set. Use the training set to train the improved neural network model, and use the test set to test the trained neural network model to obtain the dynamic modulus prediction model. Preferably, the neural network model is an improved BP neural network model. A BP neural network is a multi-layer feedforward network that achieves self-optimization through an error backpropagation algorithm. It consists of an input layer, hidden layers (which may contain multiple layers), and an output layer. Its core mechanism includes two stages: forward propagation and backward error correction. Forward propagation transmits the input signal to the output layer through weighted connections, while backward propagation adjusts the network parameters according to the steepest descent method to minimize the sum of squared errors. The improved BP neural network model of this application incorporates an attention calculation layer.

[0098] Training the improved BP neural network model requires sufficient data, with a sample size of ≥1000. For easier understanding, this embodiment uses a smaller dynamic modulus dataset for illustration, but it does not mean that only the example dynamic modulus dataset can be used.

[0099] In the example dynamic modulus dataset of this embodiment, there are 243 sets of data, of which 170 sets are training sets and 73 sets are test sets. The training set accounts for no less than two-thirds of the dynamic modulus dataset. In this embodiment, the ratio of training set to test set is 7:3.

[0100] Specifically, the example dynamic modulus dataset in this embodiment contains 243 sets of data for 9 different asphalt mixtures under 3 temperatures, 3 tire pressures (load levels), 3 speeds, and 1 loading cycle.

[0101] Specifically, the improved BP neural network model consists of an input layer, an attention calculation layer, a hidden layer, and an output layer. The number of nodes in the input layer is consistent with the number of feature parameters in the dynamic modulus dataset. The number of nodes in the hidden layer is determined based on the number of nodes in the input layer and engineering practice experience. The output layer has only one node and is used solely to output the predicted dynamic modulus of asphalt mixtures.

[0102] In this embodiment, since there are 9 feature parameters in the dynamic modulus dataset, the input layer has 9 nodes, which are used to input the 9 features in the dynamic modulus dataset, namely asphalt content, asphalt viscosity at 135℃, aggregate lithology, aggregate sphericity, gradation factor, temperature, velocity, load level, and number of loading times, to ensure that the input layer can completely receive all the information in the dynamic modulus dataset. The hidden layers are set to two layers, including a first hidden layer and a second hidden layer. In engineering practice, too few nodes in the hidden layers will cause the model to fail to learn the relationship between parameters, while too many nodes will cause the model to overfit. Therefore, considering all factors, the first hidden layer is designed with 12 nodes and uses the Leaky ReLU activation function (leaky linear rectified function), and the second hidden layer is designed with 8 nodes and uses the ELU activation function (exponential linear function). The number of nodes in the first and second hidden layers can be adjusted according to the training effect of the model in engineering practice (such as the size of the prediction error). For example, the first hidden layer can have 12-20 nodes and the second hidden layer can have 8-12 nodes to ensure that the model can learn fully without overfitting. The output layer consists of one node, which is the dynamic modulus, used to output the predicted dynamic modulus of asphalt mixtures; the loss function is the mean squared error (MSE).

[0103] After receiving parameters from the dynamic modulus dataset, the input layer transmits them to the attention calculation layer. The attention calculation layer assigns weights to each parameter, strengthening the weight of parameters with a greater impact on the dynamic modulus (such as temperature and load level) and weakening the weight of parameters with a smaller impact, generating a set of weighted parameters. The first hidden layer processes this set of weighted parameters, using the Leaky ReLU activation function to perform a nonlinear transformation on the set, converting linearly inseparable parameter relationships into linearly separable relationships, generating first-order nonlinear transformation parameters. The second hidden layer processes these first-order nonlinear transformation parameters, using the ELU activation function to perform a second-order nonlinear transformation on the parameters, further optimizing the feature representation of the parameters, generating second-order nonlinear transformation parameters. The output layer processes these second-order nonlinear transformation parameters, using a linear activation function to perform a linear mapping on the parameters, mapping the transformed parameters to predicted values ​​of the dynamic modulus, generating the predicted dynamic modulus of asphalt mixture.

[0104] Furthermore, when the attention calculation layer assigns weights to each feature parameter, it first calculates the importance coefficient of each feature parameter in the standardized dynamic modulus dataset. Then, it substitutes the importance coefficient of each parameter into the exponential function and divides it by the sum of the exponential function values ​​of all parameter importance coefficients to obtain the weight coefficient corresponding to each parameter. In the Leaky ReLU activation function of the first hidden layer, the negative slope is set to a fixed value; in the ELU activation function of the second hidden layer, the decay coefficient is set to a fixed value.

[0105] Specifically, the process by which the attention computation layer assigns weights to each parameter in the standardized dynamic modulus dataset is as follows: Determine the degree of correlation between each feature parameter in the dynamic modulus dataset and the dynamic modulus of asphalt mixture (e.g., through correlation analysis, engineering experience, etc.), assign a corresponding importance coefficient to each feature parameter according to the degree of correlation, and generate the importance coefficient of each parameter by reflecting the degree of influence of the feature parameter on the dynamic modulus prediction result. Substitute the importance coefficient of each feature parameter into an exponential function (such as the natural exponential function) to obtain the exponential function value of each parameter. Then calculate the sum of the exponential function values ​​of all parameters. Finally, divide the exponential function value of each parameter by the sum to obtain the weight coefficient corresponding to each parameter. This generates a set of parameters with weight coefficients, which enables the model to pay more attention to the parameters with greater influence in subsequent calculations.

[0106] The first hidden layer uses the Leaky ReLU activation function to perform a nonlinear transformation on the set. The difference between this activation function and the traditional linear rectified function is that when the input value is negative, the function output is not zero but a small fixed value, that is, the value corresponding to the negative slope. This negative slope is set to a fixed value, such as 0.01. This process avoids the problem of neuron "death" (some neurons always output zero and cannot participate in subsequent calculations) during model training, and converts the set of parameters with weight coefficients into parameters of a nonlinear transformation.

[0107] The second hidden layer uses the ELU activation function for a second nonlinear transformation. The characteristic of this function is that when the input value is negative, the output value exponentially approaches a fixed value as the input value decreases, which is the value corresponding to the attenuation coefficient. This attenuation coefficient is set to a fixed value, such as 1.0. This process can make the feature expression of the parameters more stable and reduce the sensitivity of the model to outliers. The parameters of the first nonlinear transformation are converted into parameters of the second nonlinear transformation. These parameters are transmitted to the output layer to generate the preliminary results of the dynamic modulus of asphalt mixture.

[0108] The improved BP neural network model was iteratively trained, and training ended when the test set loss stabilized, generating a dynamic modulus prediction model. The Adam optimizer was used to train the model with a learning rate of 0.001 and at least 400 iterations. Early stopping (training stopped when the test set loss did not decrease for 10 consecutive iterations) was employed to avoid overfitting, and the final model prediction error was controlled within 5%.

[0109] Specifically, in the example dynamic modulus dataset of this embodiment, the improved BP neural network model was iteratively trained. After 420 iterations, the test set loss stabilized and the prediction error was 4.2%, which is less than 5%.

[0110] S203. Input the raw material parameters and working condition parameters of the asphalt mixture to be predicted into the dynamic modulus prediction model, and calculate the dynamic modulus prediction result of the asphalt mixture. In one example, the raw material parameters of the asphalt mixture are: asphalt content 5.0%, asphalt viscosity at 135℃ 600 mPa·s, aggregate lithology hard rock, aggregate sphericity 0.65, and gradation factor 0.4; Operating parameters are: temperature 30℃, speed 80km / h, tire pressure 0.8MPa.

[0111] After inputting the above-mentioned asphalt mixture raw material parameters and working condition parameters into the constructed dynamic modulus prediction model, the dynamic modulus prediction result is 7800MPa.

[0112] The asphalt mixture in the above example was tested under the same working parameters, and the dynamic modulus was 7950 MPa with an error of only 1.9%, which meets the accuracy requirements.

[0113] By inputting asphalt mixture raw material parameters and working condition parameters, and utilizing neural networks to capture the nonlinear relationships between different information in the data for multi-parameter coupled prediction, the prediction accuracy is improved. In scenarios lacking experimental equipment or requiring rapid evaluation (such as the design stage), the dynamic modulus prediction model of this embodiment can directly predict the dynamic modulus of asphalt mixture under target conditions, reducing experimental costs and time.

[0114] Furthermore, the prediction method may also include correcting the prediction results of the dynamic modulus of the asphalt mixture by combining the engineering scenario correction coefficient, so as to obtain the target dynamic modulus value of the asphalt mixture.

[0115] Specifically, the engineering scenario correction coefficient is determined based on the measured dynamic modulus data of similar projects in the target project area and the prediction results output by the trained dynamic modulus prediction model. By calculating the ratio of the measured data to the prediction results, an engineering scenario correction coefficient that can adapt to the characteristics of the target project area is obtained.

[0116] Subsequently, the dynamic modulus of asphalt mixtures in similar projects is predicted using a dynamic modulus prediction model. The prediction results are then multiplied by the engineering scenario correction coefficient to obtain the target dynamic modulus value of the asphalt mixture, thus eliminating the influence of regional differences on the predicted value.

[0117] In one feasible embodiment, the engineering scenario correction coefficient can be determined using the following process: Collect measured data of dynamic modulus of asphalt mixture under the same parameter conditions from similar projects in the target project area, and compare the predicted dynamic modulus of asphalt mixture output by the trained dynamic modulus prediction model. Calculate the ratio of multiple sets of measured data to the corresponding predicted data, and take the average of all ratios as the engineering scenario correction coefficient. The value range of the engineering scenario correction coefficient is 0.95 to 1.05. This engineering scenario correction coefficient is used to correct the predicted dynamic modulus of asphalt mixture, ensuring that the corrected prediction result will not deviate too much from the actual value due to the coefficient being too large or too small.

[0118] Furthermore, in one feasible embodiment, before correcting the dynamic modulus prediction results of asphalt mixture by incorporating engineering scenario correction coefficients, a consistency verification is performed on the dynamic modulus prediction results of asphalt mixture.

[0119] Specifically, based on the characteristic parameters in the dynamic modulus dataset and the parameter-dynamic modulus correlation law obtained from the dynamic modulus test of asphalt mixture, the theoretical range of dynamic modulus values ​​for asphalt mixture under the parameter combination is determined. The predicted dynamic modulus of asphalt mixture is compared with the theoretical range. If the predicted dynamic modulus of asphalt mixture is within the theoretical range, the predicted dynamic modulus of asphalt mixture is retained and further correction is performed. If the predicted dynamic modulus of asphalt mixture exceeds the theoretical range, the dynamic modulus prediction model is called again for prediction to avoid abnormal prediction results.

[0120] Furthermore, based on the target dynamic modulus value of asphalt mixtures and the corresponding characteristic parameters, a parameter-dynamic modulus correlation database is established: The original parameters of the asphalt mixture, standardized parameters, dynamic modulus prediction results, engineering scenario correction coefficients, and target dynamic modulus values ​​corresponding to the current prediction task are associated and stored to form a parameter-dynamic modulus association database. This database can record complete data for each prediction process and provide data support for subsequent projects.

[0121] Traditional dynamic modulus measurements require indoor testing, a process that is time-consuming, material-intensive, and costly. The asphalt mixture dynamic modulus prediction method in this embodiment, through a dynamic modulus prediction model trained on a neural network, achieves accurate and efficient prediction of the asphalt mixture's dynamic modulus. The prediction model can quickly output results after inputting parameters, reducing the time and cost associated with traditional testing methods. It can flexibly adapt to different types of asphalt mixtures, varying environmental conditions, and loading conditions, providing reliable data support for asphalt pavement structure design and performance evaluation, and possesses significant engineering application value.

[0122] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any other way. Any modifications or equivalent changes made based on the technical essence of the present invention shall still fall within the scope of protection claimed by the present invention.

Claims

1. A dynamic modulus testing system for asphalt mixtures, characterized in that, The system includes a test chamber (2), a strain acquisition component (3), a temperature acquisition component (4), a circulating temperature control device (5), and an accelerated loading device (6), wherein: The test chamber (2) is used to place multiple asphalt mixture specimens (1). The strain acquisition component (3) is used to acquire the strain of the specimen (1) placed on the test tank (2) due to the load; The temperature acquisition component (4) is used to acquire the temperature of the specimen (1) placed on the test tank (2); The circulating temperature control device (5) is connected to the test tank (2) and is used to control the temperature of the specimen (1) placed on the test tank (2) so that the specimen (1) can be tested for dynamic modulus at the target temperature. The accelerated loading device (6) includes a tire (611), which accelerates the loading of the specimen (1) placed on the test groove (2) by adjusting the tire (611).

2. The asphalt mixture dynamic modulus testing system according to claim 1, characterized in that, The strain acquisition component (3) includes a strain gauge (31) and a strain data acquisition device (32). The strain gauge (31) is used to connect with the specimen (1) to monitor the strain of the specimen (1). The strain data acquisition device (32) is connected with the strain gauge (31) and is used to acquire the strain of the specimen (1) monitored by each strain gauge (31).

3. The asphalt mixture dynamic modulus testing system according to any one of claims 1-2, characterized in that, The temperature acquisition component (4) includes a temperature sensor (41) and a temperature data acquisition unit (42). The temperature sensor (41) is connected to the specimen (1) to monitor the temperature of the specimen (1). The temperature data acquisition unit (42) is connected to the temperature sensor (41) and is used to acquire the temperature of the specimen (1) monitored by each of the temperature sensors (41).

4. A method for testing the dynamic modulus of asphalt mixtures, characterized in that, The system described in any one of claims 1-3 is used for testing, and the method includes: Multiple asphalt mixture specimens were prepared, and strain acquisition components and temperature acquisition components were connected to the specimens respectively. The specimens are placed sequentially in the test chamber, and the specimens are heated using a circulating temperature control device. The temperature of the specimens is collected using the temperature acquisition component, and the specimens are kept warm for a set period of time after the target temperature is reached. The accelerated loading device is activated to load the specimen at a set loading pressure and / or loading frequency, and the strain amplitude of the specimen is acquired using the strain acquisition component. The dynamic modulus of the specimen was calculated based on the loading pressure and strain amplitude.

5. The method for testing the dynamic modulus of asphalt mixtures according to claim 4, characterized in that, The target temperature range is 0-60℃, the heat preservation time range is 0.5h-6h, the loading pressure range is 0.5~1.2MPa, and the loading frequency range is 0~2.5Hz.

6. The method for testing the dynamic modulus of asphalt mixtures according to claim 4, characterized in that, The strain acquisition component and temperature acquisition component are connected to the specimen at a position that is half the height of the specimen.

7. A method for predicting the dynamic modulus of asphalt mixtures, characterized in that, The method includes: Historical data of dynamic modulus testing of asphalt mixtures are obtained, and a dynamic modulus dataset is constructed using the historical data. The historical data includes raw material parameters, working condition parameters, and dynamic modulus data of asphalt mixtures. The dynamic modulus data in the historical data is obtained by measurement using the system described in any one of claims 1-3 or the method described in any one of claims 4-6. The dynamic modulus dataset is divided into a training set and a test set. The improved neural network model is trained using the training set, and the trained neural network model is tested using the test set to obtain a dynamic modulus prediction model. The raw material parameters and working condition parameters of the asphalt mixture to be predicted are input into the dynamic modulus prediction model to calculate the dynamic modulus prediction result of the asphalt mixture.

8. The method for predicting the dynamic modulus of asphalt mixtures according to claim 7, characterized in that, The raw material parameters include asphalt content, asphalt viscosity at a specific temperature, aggregate lithology, aggregate sphericity, and gradation factor. The working condition parameters include temperature, speed, load level, and number of loading cycles.

9. The method for predicting the dynamic modulus of asphalt mixtures according to claim 7, characterized in that, After constructing a dynamic modulus dataset using the historical data, the dynamic modulus dataset is preprocessed to generate a standardized dynamic modulus dataset.

10. The method for predicting the dynamic modulus of asphalt mixtures according to claim 7, characterized in that, The improved neural network model includes an input layer, an attention calculation layer, a hidden layer, and an output layer. The hidden layer includes a first hidden layer and a second hidden layer. The first hidden layer uses the Leaky ReLU activation function, and the second hidden layer uses the ELU activation function.

Images