An asphalt pavement paving quality automatic detection and closed-loop control method and system

Abstract

The present application belongs to the technical field of asphalt pavement paving quality detection, in particular to an asphalt pavement paving quality automatic detection and closed-loop control method and system, step one, collecting pavement data by using a multi-sensor assembly integrated in the rear part of the paver, the multi-sensor assembly comprising a laser scanner, an infrared thermal imager, an industrial line array camera and a pose sensor; dynamically adjusting the sampling frequency of the sensor according to the real-time driving speed of the paver to ensure constant longitudinal sampling resolution. The asphalt pavement paving quality automatic detection and closed-loop control method and system integrate multiple sensors such as laser, infrared and vision, realize synchronous real-time detection of flatness, temperature, segregation and thickness, fully reflect the paving quality, break the barrier between detection and control, map different quality defects to corresponding actuators through decoupling control matrix, realize targeted automatic adjustment, and avoid the lag and error of manual adjustment.

Description

Technical Field

[0001] This invention relates to the field of asphalt pavement paving quality testing technology, and in particular to an automatic detection and closed-loop control method and system for asphalt pavement paving quality. Background Technology

[0002] Asphalt pavement paving is a crucial step in highway construction, and its quality directly determines the pavement's lifespan and driving comfort. Traditional paving quality inspection relies primarily on manual measurements, such as using a three-meter straightedge to check smoothness and drilling core samples to measure thickness and compaction. This method is not only inefficient and lacks representativeness but also suffers from significant time lag. By the time quality problems are discovered, the section of road is often already paved, requiring rework or repairs, resulting in substantial material waste and economic losses.

[0003] In recent years, although some sensor-based road surface inspection equipment has emerged, such as vehicle-mounted evenness gauges and infrared thermometers, most of these devices can only detect a single indicator or record simple data, failing to achieve comprehensive analysis of multi-dimensional quality indicators. More importantly, existing inspection systems are often disconnected from the paver's control system, essentially "inspecting but not managing." The detected data cannot be fed back to the paver in real time for parameter adjustments, making it impossible to correct quality deviations promptly during construction.

[0004] Furthermore, the properties of asphalt mixtures (such as aggregate void ratio and porosity) and environmental factors have a significant impact on the final compaction effect. Existing paving control strategies are mostly based on fixed empirical parameters, lacking self-learning and adaptive capabilities, making it difficult to cope with changes in different mixture types and complex environmental conditions, resulting in difficulties in ensuring consistent construction quality.

[0005] Therefore, an automatic detection and closed-loop control method and system for asphalt pavement paving quality is needed. Summary of the Invention

[0006] Based on existing technical problems, this invention proposes an automatic detection and closed-loop control method and system for asphalt pavement paving quality.

[0007] The present invention proposes an automatic detection and closed-loop control method for asphalt pavement paving quality, comprising the following steps: Step 1: Collect pavement data using a multi-sensor assembly integrated at the rear of the paver. The multi-sensor assembly includes a laser scanner, an infrared thermal imager, an industrial line array camera, and a pose sensor; dynamically adjust the sampling frequency of the sensors according to the real-time travel speed of the paver to ensure constant longitudinal sampling resolution. Step 2: Based on the Euler angles and position information output by the pose sensor, construct a transformation matrix from the sensor coordinate system to the world coordinate system, and perform spatiotemporal registration and fusion on the laser point cloud, infrared image and visible light image acquired at the same time. Step 3: Based on the fused multi-source data, calculate the road surface smoothness index, temperature index, texture segregation index, and geometric thickness index respectively. Step 4: Compare each indicator with the preset standard value to identify the main quality defect types and defect levels of the current road surface; Step 5: Based on the identified quality defect type, call the corresponding decoupling control strategy model, calculate the adjustment amount of each actuator of the paver, and send the adjustment command to the paver controller for execution; Step 6: Obtain road surface retest data after the compaction process, and correct the parameters in the decoupled control strategy model in reverse based on the deviation between the retest data and the predicted data of the paving stage.

[0008] Preferably, in step one, the real-time travel speed of the paver is considered. Dynamically adjust sampling frequency The calculation formula is: ; in, The preset vertical sampling interval, The preset anti-aliasing offset frequency is set according to the inherent vibration frequency of the paver chassis and is used to suppress signal spectrum aliasing caused by mechanical vibration.

[0009] Preferably, the transformation matrix constructed in step two includes using the roll angle measured by the pose sensor. Pitch angle and heading angle The sensor data is subjected to rigid transformation correction, and the correction formula is as follows: ; in, Coordinates in the world coordinate system The coordinates of the sensor at the origin in the world coordinate system. The original measurement coordinates of the sensor, It is a rotation matrix consisting of three attitude angles.

[0010] Preferably, the calculation of the smoothness index in step three includes detrending the elevation sequence to eliminate the influence of longitudinal slope, and then calculating the local smoothness index. The calculation formula is as follows: ; in, This is the detrended road surface elevation function. To analyze window length, This represents the mean elevation within the window after detrending. The coordinates are longitudinal, representing the distance variable along the road surface. For integral infinitesimal elements, it represents the mathematical integral variable, which means dividing the continuous road surface into countless tiny segments and accumulating them.

[0011] Preferably, the calculation of the temperature index in step three includes performing environmental radiation compensation on the infrared data to calculate the true road surface temperature. The formula is: ; in, The radiation temperature measured by the infrared sensor. For the emissivity of asphalt pavement, The ambient reflected temperature, Atmospheric transmittance is calculated based on atmospheric distance and humidity. This refers to the atmospheric temperature.

[0012] Preferably, in step three, the texture separation index is calculated using texture contrast based on the gray-level co-occurrence matrix, and the formula is: ; in, Position in the gray-level co-occurrence matrix The normalized probability value at that point. and For pixel grayscale levels, when When the value is higher than the first threshold, it is determined to be coarse aggregate segregation; when it is lower than the second threshold, it is determined to be fine aggregate segregation.

[0013] Preferably, the calculation of the paving geometric thickness index in step three incorporates a dynamic compaction correction based on the mixture design parameters, and the calculation formula is as follows: ; in, The elevation of the top surface of the paving layer. The elevation of the underlying layer. The current road surface temperature, As the reference initial pressure temperature, The aggregate void ratio of the mixture in the current construction section The corresponding first regression coefficient, To match the porosity of the mixture The corresponding second regression coefficient is automatically retrieved from the database based on the input mixture design number.

[0014] Preferably, step five specifically includes: S1, establishing a quality deviation vector. ,in, For flatness deviation, For the degree of separation deviation, Temperature deviation; S2. Establish the execution mechanism adjustment vector. ,in, Adjustment amount for hydraulic cylinder leveling of ironing board. This refers to the adjustment amount of the spiral feeder speed. This refers to the adjustment amount of the hammer frequency; S3. Through the preset decoupling control matrix The formula for mapping the deviation vector to the adjustment vector is as follows: .

[0015] Preferably, in step six, the decoupling control matrix is ​​reverse-corrected. The method for obtaining the parameters is as follows: obtain the pavement quality test data after compaction by the road roller as the true value, and calculate the error vector between the true value and the predicted value in step three. ; Using the recursive least squares algorithm Decoupling control matrix Gain coefficient in The update is performed using the following formula: ; in, The control matrix updated at the current moment. The control matrix from the previous time step. Let covariance matrix be the variance matrix. Forgetting factor, Let this be the quality deviation vector at the current moment. This is the prediction error vector at the current moment.

[0016] The present invention proposes an automatic detection and closed-loop control system for asphalt pavement paving quality, specifically comprising: a multi-sensor acquisition module for performing data acquisition in step one; The core of the data processing is used to execute steps two and three, including a data registration unit, a feature extraction unit, and an indicator calculation unit. Defect Identification and Decision Module: Used to execute steps four and five, identify the main quality defect types, and calculate the adjustment vectors for each actuator; Self-learning correction module: used to execute step six, updating control parameters online using the recursive least squares algorithm based on the retest data after compaction.

[0017] The beneficial effects of this invention are as follows: This invention achieves real-time detection across all dimensions: integrating multiple sensors such as laser, infrared, and vision, it enables simultaneous real-time detection of flatness, temperature, segregation, and thickness, comprehensively reflecting the paving quality.

[0018] Intelligent closed-loop control: It breaks down the barriers between detection and control. By decoupling the control matrix, different quality defects are mapped to the corresponding actuators (such as leveling cylinders, material distributors, and tampers), achieving targeted automatic adjustment and avoiding the lag and errors of manual adjustment.

[0019] High-precision prediction and correction: The thickness calculation model introduces dynamic compaction correction based on temperature and mixture parameters (VMA, porosity), which significantly improves the prediction accuracy of the conversion from loose thickness to compacted thickness.

[0020] Self-evolution capability: An innovative feedback mechanism based on post-compaction retest data is introduced, and the control matrix parameters are corrected online using the recursive least squares (RLS) algorithm, enabling the system to continuously learn and adapt to changes in different working conditions and mixtures, thereby continuously improving the control effect. Attached Figure Description

[0021] Figure 1 A schematic diagram of the overall process of an automatic detection and closed-loop control method and system for asphalt pavement paving quality; Figure 2 This is a schematic diagram of the multi-sensor arrangement and connection relationship of an automatic detection and closed-loop control method and system for asphalt pavement paving quality. Figure 3 A schematic diagram illustrating the principle of data processing and spatiotemporal registration for an automatic detection and closed-loop control method and system for asphalt pavement paving quality. Figure 4 A schematic diagram of the physical model for dynamic correction of paving thickness in an automatic detection and closed-loop control method and system for asphalt pavement paving quality. Figure 5 This is a block diagram of the closed-loop control logic based on a decoupled control matrix for an automatic detection and closed-loop control method and system for asphalt pavement paving quality. Figure 6 This is a schematic diagram of the self-learning correction algorithm of an automatic detection and closed-loop control method and system for asphalt pavement paving quality. Detailed Implementation

[0022] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0023] Example 1

[0024] An automatic detection and closed-loop control system for asphalt pavement paving quality includes a hardware layer and a software layer.

[0025] The hardware layer includes: a multi-sensor acquisition module: a laser scanner: employing a 3D laser profile scanner, mounted directly above and behind the screed, used to scan the three-dimensional cross-section of the freshly paved road surface to acquire high-precision point cloud data. an infrared thermal imager: mounted to the side and rear, overlapping the field of view of the laser scanner, used to acquire infrared thermal images of the road surface to obtain temperature field data. an industrial linear array camera: working with a high-brightness LED ring light, mounted at the end of a rigid bracket, vertically photographing the road surface to acquire high-resolution texture images. a pose sensor: including a high-precision RTK-GNSS receiver and an inertial measurement unit (IMU), mounted at the center of the sensor integration bracket, used to provide accurate position, velocity, and attitude (roll, pitch, heading) information.

[0026] Industrial control computer: As the data processing core and decision-making center, it is located in the paver cab or in a dedicated waterproof control cabinet and is connected to various sensors through a high-speed data acquisition card; The software layer runs on an industrial control computer and includes a data registration unit, a feature extraction unit, an index calculation unit, a defect identification and decision-making module, and a self-learning correction module. The defect identification and decision-making module identifies the main quality defect types and calculates the adjustment vectors of each actuator. The self-learning correction module updates the control parameters online using a recursive least squares algorithm based on the retest data after compaction.

[0027] This invention, by integrating a laser scanner, an infrared thermal imager, an industrial linear array camera, and a pose sensor, breaks through the limitations of traditional single-method detection. The system can simultaneously acquire three-dimensional geometric information (smoothness, thickness), thermal field information (temperature distribution), and texture information (aggregate segregation) of the road surface, realizing comprehensive and multi-dimensional real-time monitoring of asphalt pavement paving quality. This solves the technical problems of low efficiency, poor representativeness, and lag in traditional manual inspection.

[0028] Example 2

[0029] Reference Figure 1-6 An automatic detection and closed-loop control method for asphalt pavement paving quality includes the following steps: Step 1, data acquisition and dynamic sampling; when the paver starts working, multiple sensor components are activated synchronously. To prevent uneven density of pavement data due to changes in paver speed (e.g., data stretching at high speeds and data overlapping at low speeds), the system reads the speed output from the RTK-GNSS in real time. And according to the formula Dynamically adjust the trigger frequency.

[0030] in, The preset longitudinal sampling interval is set to 2cm to ensure that a cross-section is sampled every 2cm. The preset anti-aliasing bias frequency is set to 10Hz to counteract the high-frequency noise interference caused by the vibration of the paver engine.

[0031] Step 2: Spatiotemporal Registration and Fusion; Due to the different installation positions and viewing angles of various sensors, and the paver being in a bumpy state, the directly collected data has coordinate deviations. The system utilizes attitude angles provided by the IMU, such as roll angle. Pitch angle and heading angle Construct rotation matrix By combining the translation vectors provided by RTK, the laser point cloud, infrared image pixels, and visible light image pixels are uniformly mapped to the world coordinate system (UTM coordinate system or route-relative coordinate system). In this way, the same coordinate system... The data contains both elevation information and temperature and texture information, achieving pixel-level / point cloud-level data fusion. Constructing the transformation matrix includes using the roll angle measured by the pose sensor. Pitch angle and heading angle The sensor data is subjected to rigid transformation correction, and the correction formula is as follows: ; in, Coordinates in the world coordinate system The coordinates of the sensor at the origin in the world coordinate system. The original measurement coordinates of the sensor, It is a rotation matrix consisting of three attitude angles.

[0032] This invention designs a dynamic spatiotemporal registration method based on pose information and employs dynamic sampling frequency control that varies with vehicle speed and an anti-aliasing bias frequency design. This not only eliminates data distortion and coordinate misalignment caused by mechanical vibration, vehicle body bumps, and changes in driving speed, but also ensures high-level fusion of laser, infrared, and visual data at the pixel / point cloud level, providing a solid data foundation for subsequent accurate calculations and control decisions.

[0033] Step 3: Quality index calculation: Based on the fused multi-source data, calculate the pavement smoothness index, temperature index, texture segregation index and geometric thickness index respectively. Smoothness index calculation: Extracting the fused elevation data. To remove the interference of road design longitudinal slope (such as a 3% uphill slope) on the smoothness calculation, the elevation sequence is first linearly detrended to extract the pure fluctuation component. Subsequently, within a 10-meter window length Calculate the root mean square value internally; This includes detrending the high-slope column, eliminating the influence of longitudinal slope, and then calculating the local flatness index. The calculation formula is as follows: ; in, This is the detrended road surface elevation function. To analyze window length, This represents the mean elevation within the window after detrending. The coordinates are longitudinal, representing the distance variable along the road surface. For integral infinitesimal elements, it represents the mathematical integral variable, which means dividing the continuous road surface into countless tiny segments and accumulating them.

[0034] Temperature index calculation: Extract the original grayscale value of the infrared image and combine it with the ambient temperature sensor data, then use a variant of the Stefan-Boltzmann law for compensation.

[0035] This includes performing environmental radiation compensation on infrared data and calculating the true road surface temperature. The formula is: ; in, The radiation temperature measured by the infrared sensor. For the emissivity of asphalt pavement, The ambient reflected temperature, Atmospheric transmittance is calculated based on atmospheric distance and humidity. This refers to the atmospheric temperature.

[0036] Taking into account the attenuation of thermal radiation in the air and the reflection from surrounding heat sources, this formula can accurately reproduce the true paving temperature of the road surface.

[0037] Texture segregation index calculation: The visible light image is converted to grayscale. This is achieved using the gray-level co-occurrence matrix. Extract texture features. The texture of a normal road section is uniform. If coarse aggregate segregation occurs, the image will appear coarse and grainy, affecting texture contrast. The value will spike; if fine aggregate segregation occurs, the image surface will be smooth. The value will decrease significantly.

[0038] Texture contrast is calculated based on the gray-level co-occurrence matrix, using the following formula: ; in, Position in the gray-level co-occurrence matrix The normalized probability value at that point. and For pixel grayscale levels, when When the value is higher than the first threshold, it is determined to be coarse aggregate segregation; when it is lower than the second threshold, it is determined to be fine aggregate segregation.

[0039] Geometric thickness calculation: The current elevation of the loose paved surface is obtained using laser point cloud data. Combined with the stored elevation data of the underlying layer Calculate the loose paving thickness. However, asphalt mixtures settle after compaction by a roller. To predict the final compacted thickness during the paving stage, this invention introduces a modified model based on the physical properties of the mixture; A dynamic compaction correction based on the mixture design parameters is introduced, and the calculation formula is as follows: ; in, The elevation of the top surface of the paving layer. The elevation of the underlying layer. The current road surface temperature, As the reference initial pressure temperature, The aggregate void ratio of the mixture in the current construction section The corresponding first regression coefficient, To match the porosity of the mixture The corresponding second regression coefficient is automatically retrieved from the database based on the input mixture design number.

[0040] Before construction, the system inputs the type of aggregate for this project (e.g., SMA-13), and the system automatically retrieves the corresponding aggregate void ratio from the database. and porosity .

[0041] Meanwhile, combined with the current measured temperature If the paving temperature is low and the mixture is hard, the settlement after compaction will be smaller (i.e., (If the exponential term decreases), the system will automatically reduce the predicted settlement correction value, thereby prompting the user that the current loose paving thickness may be insufficient.

[0042] Step 4: Defect Identification: The system compares the calculated four indicators with the set standard values ​​to identify the main quality defect types and defect levels of the current road surface. For example, the smoothness threshold is set at 1.2mm, and the lower temperature limit is 140℃. If the smoothness exceeds the standard, it is judged as a "smoothness defect"; if the temperature in a certain area is generally below 140℃ and... If the value is abnormal, it is determined to be "temperature segregation".

[0043] Step 5: Decoupling Control Decisions: Establishing a Quality Deviation Vector and actuator adjustment vector The system determines the type of defect by using a decoupling control matrix. Calculate the adjustment amount.

[0044] Specifically, this includes: S1, establishing the quality deviation vector. ,in, For flatness deviation, For the degree of separation deviation, Temperature deviation; S2. Establish the execution mechanism adjustment vector. ,in, Adjustment amount for hydraulic cylinder leveling of ironing board. This refers to the adjustment amount of the spiral feeder speed. This refers to the adjustment amount of the hammer frequency; S3. Through the preset decoupling control matrix The formula for mapping the deviation vector to the adjustment vector is as follows: .

[0045] For example: If a flatness deviation is detected Positive (road surface protrusion). The matrix corresponding terms calculate the required descent value of the hydraulic cylinder for leveling the ironing board. This reduces the height of the ironing board.

[0046] If a separation deviation is detected (Rough texture) Matrix calculation of spiral feeder speed The adjustment value makes the fabric more even.

[0047] If a temperature deviation is detected (Too low) Matrix calculation of hammer frequency The increase in compaction value is achieved by increasing the impact force to enhance the compaction potential of the low-temperature soft section.

[0048] This invention innovatively constructs a decoupled control model between the quality deviation vector and the actuator adjustment vector. Through the decoupled control matrix, the system can accurately identify different types of quality defects and map them to the corresponding actuators for precise adjustment. For example, it adjusts the leveling hydraulic cylinder for flatness defects, adjusts the speed of the auger distributor for segregation defects, and adjusts the frequency of the tamping hammer for temperature defects. This targeted control strategy effectively avoids indiscriminate adjustments, significantly improving the effectiveness and response speed of quality intervention.

[0049] Step Six: Self-Learning and Model Correction: After the paver passes, the following roller completes the compaction operation. Subsequently, the thickness and compaction degree of the compacted pavement are remeasured using vehicle-mounted lightweight testing equipment or drones, and these values ​​are taken as the "true values".

[0050] The system compares the predicted values ​​during the paving stage with the true values ​​after compaction to obtain the error vector. .

[0051] Obtain road surface retest data after compaction, and based on the deviation between the retest data and the predicted data of the paving stage, reversely correct the parameters in the decoupled control strategy model; Using the recursive least squares algorithm Decoupling control matrix Gain coefficient in The update is performed using the following formula: ; in, The control matrix updated at the current moment. The control matrix from the previous time step. Let covariance matrix be the variance matrix. Forgetting factor, Let this be the quality deviation vector at the current moment. This is the prediction error vector at the current moment.

[0052] For example, if it is found that after several consecutive low-temperature conditions, despite increasing the tamping frequency according to the original strategy, the final compaction degree is still insufficient, the system will automatically learn that for this type of mixture, the contribution coefficient of the tamping frequency to the compaction degree (the corresponding element in the matrix) needs to be increased. In this way, the system will issue a stronger adjustment command the next time a similar situation occurs.

[0053] Through the aforementioned closed loop, this system can not only control the current paving quality, but also become increasingly "intelligent" and easier to use as the project progresses.

[0054] To address the error problem caused by simply estimating compaction thickness based on loose paving thickness in existing technologies, this invention introduces a dynamic compaction correction mechanism into the thickness calculation model. This mechanism not only considers the influence of paving temperature on compaction settlement but also creatively incorporates the aggregate void ratio, a key design parameter for the mixture. and porosity By using regression coefficients related to physical properties, the system can accurately predict the compacted thickness of different mixtures under different temperature conditions, providing high-precision data support for construction control and effectively preventing excessive or insufficient thickness.

[0055] This invention overcomes the limitations of fixed parameters in traditional control systems by using retest data after the compaction process as truth feedback and introducing a recursive least squares algorithm. The decoupled control matrix is ​​corrected online. This gives the system self-learning capabilities, enabling it to continuously optimize control parameters as construction progresses. This adapts to changes in operating conditions caused by different batches of mixes, ambient temperatures, and mechanical wear, ensuring continuous stability and optimal control performance throughout the entire construction cycle.

[0056] This invention achieves real-time detection across all dimensions: integrating multiple sensors such as laser, infrared, and vision, it enables simultaneous real-time detection of flatness, temperature, segregation, and thickness, comprehensively reflecting the paving quality.

[0057] Intelligent closed-loop control: It breaks down the barriers between detection and control. By decoupling the control matrix, different quality defects are mapped to the corresponding actuators (such as leveling cylinders, material distributors, and tampers), achieving targeted automatic adjustment and avoiding the lag and errors of manual adjustment.

[0058] High-precision prediction and correction: The thickness calculation model introduces dynamic compaction correction based on temperature and mixture parameters (VMA, porosity), which significantly improves the prediction accuracy of the conversion from loose thickness to compacted thickness.

[0059] Self-evolution capability: An innovative feedback mechanism based on post-compaction retest data is introduced, and the control matrix parameters are corrected online using the recursive least squares (RLS) algorithm, enabling the system to continuously learn and adapt to changes in different working conditions and mixtures, thereby continuously improving the control effect.

[0060] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for automatic detection and closed-loop control of asphalt pavement paving quality, characterized in that: The process includes the following steps: Step 1: Collect road surface data using a multi-sensor assembly integrated at the rear of the paver. The multi-sensor assembly includes a laser scanner, an infrared thermal imager, an industrial line array camera, and a pose sensor. The sampling frequency of the sensors is dynamically adjusted according to the real-time travel speed of the paver to ensure a constant longitudinal sampling resolution. Step 2: Based on the Euler angles and position information output by the pose sensor, construct a transformation matrix from the sensor coordinate system to the world coordinate system, and perform spatiotemporal registration and fusion on the laser point cloud, infrared image and visible light image acquired at the same time. Step 3: Based on the fused multi-source data, calculate the road surface smoothness index, temperature index, texture segregation index, and geometric thickness index respectively. Step 4: Compare each indicator with the preset standard value to identify the main quality defect types and defect levels of the current road surface; Step 5: Based on the identified quality defect type, call the corresponding decoupling control strategy model, calculate the adjustment amount of each actuator of the paver, and send the adjustment command to the paver controller for execution; Step 6: Obtain road surface retest data after the compaction process, and correct the parameters in the decoupled control strategy model in reverse based on the deviation between the retest data and the predicted data of the paving stage.

2. The automatic detection and closed-loop control method for asphalt pavement paving quality according to claim 1, characterized in that: In step one, the real-time travel speed of the paver is used. Dynamically adjust sampling frequency The calculation formula is: ; in, The preset vertical sampling interval, The preset anti-aliasing offset frequency is set according to the inherent vibration frequency of the paver chassis and is used to suppress signal spectrum aliasing caused by mechanical vibration.

3. The automatic detection and closed-loop control method for asphalt pavement paving quality according to claim 1, characterized in that: The second step of constructing the transformation matrix includes using the roll angle measured by the pose sensor. Pitch angle and heading angle The sensor data is subjected to rigid transformation correction, and the correction formula is as follows: ； in, Coordinates in the world coordinate system The coordinates of the sensor at the origin in the world coordinate system. The original measurement coordinates of the sensor, It is a rotation matrix consisting of three attitude angles.

4. The automatic detection and closed-loop control method for asphalt pavement paving quality according to claim 1, characterized in that: The calculation of the smoothness index in step three includes detrending the elevation sequence to eliminate the influence of longitudinal slope, and then calculating the local smoothness index. The calculation formula is as follows: ; in, This is the detrended road surface elevation function. To analyze window length, This represents the mean elevation within the window after detrending. The coordinates are longitudinal, representing the distance variable along the road surface. For integral infinitesimal elements, it represents the mathematical integral variable, which means dividing the continuous road surface into countless tiny segments and accumulating them.

5. The automatic detection and closed-loop control method for asphalt pavement paving quality according to claim 1, characterized in that: The calculation of the temperature index in step three includes performing environmental radiation compensation on the infrared data to calculate the true road surface temperature. The formula is: ; in, The radiation temperature measured by the infrared sensor. For the emissivity of asphalt pavement, The ambient reflected temperature, Atmospheric transmittance is calculated based on atmospheric distance and humidity. This refers to the atmospheric temperature.

6. The automatic detection and closed-loop control method for asphalt pavement paving quality according to claim 1, characterized in that: In step three, the texture decontamination index is calculated using texture contrast based on the gray-level co-occurrence matrix, and the formula is: ; in, Position in the gray-level co-occurrence matrix The normalized probability value at that point. and For pixel grayscale levels, when When the value is higher than the first threshold, it is determined to be coarse aggregate segregation; when it is lower than the second threshold, it is determined to be fine aggregate segregation.

7. The automatic detection and closed-loop control method for asphalt pavement paving quality according to claim 1, characterized in that: In step three, the calculation of the paving geometric thickness index incorporates a dynamic compaction correction based on the mixture design parameters. The calculation formula is as follows: ； in, The elevation of the top surface of the paving layer. The elevation of the underlying layer. The current road surface temperature, As the reference initial pressure temperature, The aggregate void ratio of the mixture in the current construction section The corresponding first regression coefficient, To match the porosity of the mixture The corresponding second regression coefficient is automatically retrieved from the database based on the input mixture design number.

8. The automatic detection and closed-loop control method for asphalt pavement paving quality according to claim 1, characterized in that: Step five specifically includes: S1, establishing a quality deviation vector. ,in, For flatness deviation, For the degree of separation deviation, Temperature deviation; S2. Establish the execution mechanism adjustment vector. ,in, Adjustment amount for hydraulic cylinder leveling of ironing board. This refers to the adjustment amount of the spiral feeder speed. This refers to the adjustment amount of the hammer frequency; S3. Through the preset decoupling control matrix The formula for mapping the deviation vector to the adjustment vector is as follows: .

9. The automatic detection and closed-loop control method for asphalt pavement paving quality according to claim 1, characterized in that: In step six, the decoupling control matrix is ​​reverse-corrected. The method for obtaining the parameters is as follows: obtain the pavement quality test data after compaction by the road roller as the true value, and calculate the error vector between the true value and the predicted value in step three. ; Using the recursive least squares algorithm Decoupling control matrix Gain coefficient in The update is performed using the following formula: ; in, The control matrix updated at the current moment. The control matrix from the previous time step. Let covariance matrix be the variance matrix. Forgetting factor, Let this be the quality deviation vector at the current moment. This is the prediction error vector at the current moment.

10. The automatic detection and closed-loop control method for asphalt pavement paving quality according to claim 1, characterized in that: It also includes a control system, specifically: a multi-sensor acquisition module: used to perform data acquisition in step one; The core of the data processing is used to execute steps two and three, including a data registration unit, a feature extraction unit, and an indicator calculation unit. Defect Identification and Decision Module: Used to execute steps four and five, identify the main quality defect types, and calculate the adjustment vectors for each actuator; Self-learning correction module: used to execute step six, updating control parameters online using the recursive least squares algorithm based on the retest data after compaction.

Images