Road field flatness surveying and monitoring method and system

Abstract

The invention discloses a road field flatness investigation monitoring method and system, and relates to the technical field of road flatness monitoring, and the method comprises the steps: integrating a support vector machine model at an edge calculation node, carrying out the quantitative prediction analysis through employing an IRI value, an RMS value and a crack score as input features, and obtaining a preliminary road surface flatness monitoring result; the method comprises the steps of obtaining environment parameters corresponding to a current monitored road and historical road surface flatness data, constructing a multivariable compensation function, dynamically correcting a preliminary flatness monitoring result, obtaining a final road surface flatness monitoring result, generating a road surface monitoring report based on the final road surface flatness monitoring result, and uploading the road surface monitoring report to a cloud server for real-time monitoring. According to the road flatness monitoring method disclosed by the invention, the precision and efficiency of road flatness monitoring are improved, and meanwhile, the strong capabilities of road management level and maintenance efficiency are shown.

Description

Technical Field

[0001] This invention relates to the field of road smoothness monitoring technology, and in particular to a method and system for on-site road smoothness survey and monitoring. Background Technology

[0002] With the continuous development of transportation infrastructure construction, road quality and maintenance have become key factors in ensuring traffic safety, improving driving comfort, and extending road service life. As an important indicator for measuring road quality, road surface smoothness has received widespread attention and research in its monitoring methods. Early road smoothness detection relied on manual visual inspection and simple tool measurement. However, with the advancement of technology, especially the development of sensors, computer vision, and data processing algorithms, automatic road smoothness monitoring mechanisms based on laser scanners, accelerometers, and high-resolution cameras have gradually matured and are widely used.

[0003] Nevertheless, existing road smoothness monitoring methods still have room for improvement. Current monitoring mechanisms mostly focus on single-dimensional data analysis and fail to fully integrate multiple types of data for comprehensive evaluation, which limits the ability to fully understand the overall road surface condition. In addition, existing data analysis models usually use fixed parameter settings and cannot dynamically adapt to different environmental conditions and road characteristics, thereby reducing monitoring accuracy. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a method for on-site road smoothness survey and monitoring to solve the problem of low accuracy in road smoothness monitoring.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a method for road surface smoothness surveying and monitoring, comprising, Collect road surface height change data, vehicle vibration data, and road surface RGB images, and perform timestamp synchronization and spatial coordinate unification processing; IRI was used to calculate the IRI value of road surface height variation data, RMS value of vehicle vibration data was calculated using root mean square (RMS) data, and crack score of road surface RGB image was calculated using Canny. A support vector machine model is integrated into the edge computing node, and IRI value, RMS and crack score are used as input features for quantitative prediction and analysis to obtain preliminary road surface smoothness monitoring results. Obtain the environmental parameters and historical road surface smoothness data corresponding to the currently monitored road, construct a multivariate compensation function, dynamically correct the preliminary smoothness monitoring results, and obtain the final road surface smoothness monitoring results; Based on the final road surface smoothness monitoring results, a road surface monitoring report is generated and uploaded to the cloud server for real-time monitoring.

[0007] As a preferred embodiment of the road surface smoothness survey and monitoring method of the present invention, the road surface height change data includes road longitudinal elevation, local road surface unevenness information and road surface roughness; The vehicle vibration data includes acceleration and vibration frequency; Using spatiotemporal information provided by GPS positioning and inertial navigation equipment, time-stamp synchronization processing is performed on road surface height change data, vehicle vibration data, and road surface RGB images; Geographic information units are used to convert road surface height change data, vehicle vibration data, and road surface RGB images into a unified spatial coordinate framework.

[0008] As a preferred embodiment of the road surface smoothness survey and monitoring method of the present invention, the method includes: using IRI to calculate the IRI value of road surface height change data, using root mean square to calculate the RMS value of vehicle vibration data, and using Canny to calculate the crack score of the pavement RGB image. Based on road surface height variation data, the ratio of the relative displacement of the vehicle suspension to the travel distance is calculated using the IRI (Industrial Roughness Index) to obtain the roughness index IRI value. Perform point-by-point squaring on the time-domain signal of vehicle vibration data to obtain the squared time-domain signal sequence; The average value of vehicle vibration data is calculated by dynamic convolution filtering of time-domain signal sequence, and the RMS value of vehicle vibration data is obtained by convolution root mean square transformation. Canny was used to perform gradient calculation, nonmaximum suppression, and double threshold detection on the RGB image of the road surface to extract the binary image of crack edges in the RGB image of the road surface. Based on the binary image of the crack edge, the crack length is measured by skeletonization, and the crack width is calculated by distance transformation. The number of crack edges in the RGB image of the road surface is counted, and the crack length and width are combined for quantitative calculation to obtain the crack score of the RGB image of the road surface.

[0009] As a preferred embodiment of the road surface smoothness survey and monitoring method of the present invention, wherein: the integration of the support vector machine model at the edge computing nodes specifically involves... Historical IRI values, RMS, and crack scores were collected as a training set; The training set is input into the support vector machine model, and the grid search method is used to exhaustively search for the kernel function type, penalty coefficient and kernel parameters of the support vector machine model to form the trained support vector machine model. The trained support vector machine model is converted into a format recognizable by the lightweight inference engine, and then loaded onto the support vector machine model on the edge computing node.

[0010] As a preferred embodiment of the road surface smoothness survey and monitoring method of the present invention, the step of using IRI value, RMS and crack score as input features for quantitative prediction and analysis to obtain preliminary road surface smoothness monitoring results specifically involves: The IRI value, RMS, and crack score are transmitted as input features to the edge computing node. The support vector machine model integrated by the edge computing node normalizes the IRI value, RMS, and crack score, and transforms the input features into a high-dimensional space through kernel function mapping for pattern recognition and regression analysis, generating preliminary pavement smoothness monitoring results.

[0011] As a preferred embodiment of the road surface smoothness survey and monitoring method of the present invention, the step of acquiring the environmental parameters and historical road surface smoothness data corresponding to the currently monitored road, constructing a multivariate compensation function, dynamically correcting the preliminary smoothness monitoring results, and obtaining the final road surface smoothness monitoring results specifically involves: Historical environmental parameters and road surface smoothness data of monitored roads were collected, and a multivariate regression analysis method was used to establish the mapping relationship between environmental parameters and changes in road surface smoothness. Based on the mapping relationship, a multivariate compensation function is constructed using a nonlinear kernel regression method. Environmental parameters are input into the multivariate compensation function for calculation to obtain the environmental compensation coefficient. After Bayesian probability fusion of the environmental compensation coefficient and the preliminary road surface smoothness monitoring results, dynamic adjustment is performed using Kalman filtering to form the final road surface smoothness monitoring results.

[0012] As a preferred embodiment of the road surface smoothness survey and monitoring method of the present invention, the step of generating a road surface monitoring report based on the final road surface smoothness monitoring results and uploading it to a cloud server for real-time monitoring specifically involves: The final road surface smoothness monitoring results are fused with GPS coordinates, timestamps, and environmental parameters to generate structured monitoring data. The structured monitoring data is encapsulated in JSON format, and the edge computing node calls the WebSocket protocol to establish a persistent connection with the cloud server. The cloud server receives the structured monitoring data and performs real-time monitoring.

[0013] Secondly, the present invention provides a road surface smoothness survey and monitoring system, comprising, The data acquisition module collects data on road surface height changes, vehicle vibration data, and road surface RGB images, and performs timestamp synchronization and spatial coordinate unification processing. The crack scoring module uses IRI to calculate the IRI value of road surface height variation data, uses root mean square to calculate the RMS value of vehicle vibration data, and uses Canny to calculate the crack score of the road surface RGB image. The monitoring module integrates a support vector machine model at the edge computing node and uses IRI value, RMS and crack score as input features for quantitative prediction and analysis to obtain preliminary road surface smoothness monitoring results. The adjustment module obtains the environmental parameters and historical road surface smoothness data corresponding to the currently monitored road, constructs a multivariate compensation function, dynamically corrects the preliminary smoothness monitoring results, and obtains the final road surface smoothness monitoring results. The reporting module generates a road surface monitoring report based on the final road surface smoothness monitoring results and uploads it to the cloud server for real-time monitoring.

[0014] As a preferred embodiment of the road on-site smoothness survey and monitoring method of the present invention, wherein: Thirdly, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the road surface smoothness survey and monitoring method as described in the first aspect of the present invention.

[0015] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the road surface smoothness survey and monitoring method as described in the first aspect of the present invention.

[0016] The beneficial effects of this invention are as follows: by forming a complete chain from data collection to decision support through the road smoothness monitoring method, the accuracy and efficiency of road smoothness monitoring are improved, thereby effectively ensuring road traffic safety and driving comfort, and extending the service life of roads. It demonstrates the huge potential of the deep integration of modern sensing technology and intelligent computing, and has also proven its powerful ability to improve road management and maintenance efficiency in practical applications. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of the method for on-site survey and monitoring of road smoothness.

[0019] Figure 2This is a flowchart illustrating the methodology for road surface smoothness survey and monitoring.

[0020] Figure 3 This is a flowchart for data time synchronization.

[0021] Figure 4 This is a flowchart for calculating crack scores. Detailed Implementation

[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0023] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0025] Reference Figures 1-4 As one embodiment of the present invention, this embodiment provides a method for road surface smoothness survey and monitoring, comprising the following steps: S1. Collect road surface height change data, vehicle vibration data, and road surface RGB images, and perform timestamp synchronization and spatial coordinate unification processing, specifically: Road surface height variation data includes road longitudinal elevation, local road surface unevenness information, and road surface roughness; A high-precision laser rangefinder was used to continuously scan the road longitudinally to acquire longitudinal elevation data. Subsequently, a 3D laser scanner was used to capture local unevenness information of the road surface, recording the depth and distribution of minute undulations. At the same time, a non-contact laser profilometer was used to measure the road surface roughness. The longitudinal elevation, local unevenness information, and road surface roughness data formed a complete dataset of road surface height variation.

[0026] Vehicle vibration data includes acceleration and vibration frequency; When collecting vehicle vibration data, a triaxial accelerometer is installed on the vehicle body to measure the acceleration during vehicle movement in real time. At the same time, a dynamic signal analyzer is used to record the vibration frequency. The acceleration and vibration frequency data are stored synchronously through the data acquisition equipment.

[0027] When acquiring RGB images of the road, a high-resolution digital camera is used to continuously capture images along the direction of the monitored road to ensure stable lighting conditions and clear images. During the shooting process, the focal length and exposure parameters are adjusted so that the RGB images can accurately reflect the road surface texture, cracks, and color characteristics. Subsequently, the RGB images are numbered and stored according to mileage information, and distortion correction and color calibration are performed using image processing software to finally form the road surface RGB image, which is used for road surface condition detection or defect identification and analysis.

[0028] Using spatiotemporal information provided by GPS positioning and inertial navigation equipment, time-stamp synchronization processing is performed on road surface height change data, vehicle vibration data, and road surface RGB images; It should be noted that road surface height change data is synchronized to the timestamps of GPS positioning and inertial navigation equipment via a high-precision clock. Vehicle vibration data is matched with the sampling time of the inertial navigation equipment using a hardware trigger signal. The road surface RGB image is calibrated with the UTC time of GPS positioning via frame timestamps. Subsequently, interpolation methods are used to unify the road surface height change data, vehicle vibration data, and road surface RGB image to the same time series, eliminating time deviations. Ultimately, the road surface height change data, vehicle vibration data, and road surface RGB image are strictly synchronized in the time dimension.

[0029] Geographic information units are used to convert road surface height change data, vehicle vibration data, and road surface RGB images into a unified spatial coordinate framework.

[0030] Furthermore, latitude, longitude, altitude, and attitude data provided by GPS positioning and inertial navigation equipment serve as a spatial reference. Road surface height variation data is transformed into the UTM coordinate system through Geographic Information Units (GIS). Vehicle vibration data, combined with the heading angle and acceleration information from the inertial navigation equipment, undergoes coordinate rotation and is projected onto the same UTM grid. The road surface RGB image is geometrically corrected using camera intrinsic and extrinsic parameters and GPS positioning data through collinearity equations and mapped to the UTM coordinate system. After coordinate transformation, the elevation values ​​of the road surface height variation data, the spatial trajectory of the vehicle vibration data, and the pixel coordinates of the road surface RGB image achieve sub-meter level spatial alignment within the UTM spatial coordinate framework, ensuring that the road surface height variation data, vehicle vibration data, and road surface RGB image have a unified spatial reference benchmark.

[0031] S2. Calculate the IRI value of the road surface height variation data using IRI, calculate the RMS value of the vehicle vibration data using root mean square (RMS) data, and calculate the crack score of the pavement RGB image using Canny. Specifically... Based on road surface height variation data, the ratio of the relative displacement of the vehicle suspension to the travel distance is calculated using the IRI (Industrial Roughness Index) to obtain the roughness index IRI value. It should be noted that the road surface height change data is first low-pass filtered to eliminate high-frequency components with wavelengths less than, for example, 0.5m, retaining the effective wavelength range that affects vehicle dynamic response. A quarter-vehicle model is used to dynamically simulate the filtered road surface height change data at a standard speed of 80km / h, calculating the relative displacement between sprung and unsprung mass. The IRI algorithm uses 10cm as the basic calculation unit, taking the absolute value of the relative displacement of the vehicle suspension and accumulating it in segments. The accumulated value is then divided by the driving distance of 100m within each 100m segment and magnified 1000 times to obtain the roughness index (IRI) value.

[0032] Perform point-by-point squaring on the time-domain signal of vehicle vibration data to obtain the squared time-domain signal sequence; It should be noted that the time-domain signal of vehicle vibration data is extracted point-by-point as input according to the sampling time sequence. Each sampling point is traversed in chronological order, and the amplitude value of the current sampling point is arithmetically squared. The squared amplitude values ​​are then recombine according to the sampling time sequence to generate the squared time-domain signal sequence.

[0033] The average value of vehicle vibration data is calculated by dynamic convolution filtering of time-domain signal sequence, and the RMS value of vehicle vibration data is obtained by convolution root mean square transformation. Furthermore, the time-domain signal sequence is first processed using a dynamic convolutional filtering method. This method employs a convolutional kernel to perform a moving average calculation on the time-domain signal sequence, generating a sequence of average vehicle vibration data. The average vehicle vibration data sequence is then input into a root mean square (RMS) convolutional transformation process. The RMS value is obtained by performing a point-by-point squaring operation on the average vehicle vibration data sequence, followed by a moving average calculation using a convolutional kernel. Finally, the square root of the moving average is taken to obtain the RMS value of the vehicle vibration data.

[0034] Canny was used to perform gradient calculation, nonmaximum suppression, and double threshold detection on the RGB image of the road surface to extract the binary image of crack edges in the RGB image of the road surface. It should be noted that the RGB image of the road surface is first converted to a grayscale image before being input into the Canny algorithm processing flow. The Canny algorithm uses a 5×5 Gaussian kernel to smooth the grayscale image, eliminating high-frequency noise interference. The Sobel operator calculates the gradient components in the horizontal and vertical directions of the grayscale image and synthesizes the gradient magnitude image and the gradient direction image. The Canny algorithm performs non-maximum suppression on the gradient magnitude image, comparing the gradient magnitude of each pixel with its neighboring pixels along the gradient direction, and retaining the gradient maxima within each 3×3 neighborhood. The Canny algorithm uses a dual-threshold detection with a high threshold and a low threshold. The high threshold is set to 30% of the maximum gradient magnitude (e.g., 76), and the low threshold is set to 40% of the high threshold (e.g., 30). Pixels with gradient magnitudes higher than the high threshold are marked as strong edges, and pixels between the high and low thresholds are marked as weak edges. The Canny algorithm connects the strong edge pixels and the weak edge pixels through edge tracking, and finally outputs a complete binary image of the crack edges in the RGB image of the road surface. Based on the binary image of the crack edge, the crack length is measured by skeletonization, and the crack width is calculated by distance transformation. Furthermore, the binary image of the crack edge is first processed by a skeletonization algorithm. This algorithm uses iterative erosion to refine the connected regions in the binary image of the crack edge into skeleton images with a width of one pixel. Each connected branch in the skeleton image output by the skeletonization algorithm represents the centerline of an independent crack. The crack length is measured by counting the number of skeleton pixels and multiplying it by the skeleton image resolution, expressed by the formula: ; in, This indicates the measured crack length. This indicates the resolution of the binary image at the crack edge. Represents the skeleton image in coordinates pixel values, Represents a skeleton image; The distance transform algorithm calculates the distance to the binary image of the crack edge, generating a distance field image that records the Euclidean distance from each skeleton pixel to the nearest edge pixel. The distance values ​​of the skeleton pixels in the distance field image are then compared with the image resolution to obtain the crack width measurement, expressed by the formula: ; in, This indicates the measured value of the crack width. Binary image resolution at crack edge. Represents coordinates after distance transformation The pixel distance from the skeleton point to the nearest crack edge; The number of crack edges in the RGB image of the road surface is counted, and the crack length and width are combined for quantitative calculation to obtain the crack score of the RGB image of the road surface.

[0035] It should be noted that the RGB image of the road surface is processed by the Canny edge detection algorithm to generate a binary image of crack edges. The number of all independently connected crack edge regions in the binary image of crack edges is counted as the crack edge count. Simultaneously, a weighted calculation formula is used to calculate the crack score, combining crack length and width. The weighting coefficients are set according to the road engineering evaluation standards. For example, the crack edge count has a weight of 0.5, the crack length has a weight of 0.3, and the crack width has a weight of 0.2. Therefore, the crack score can be expressed as... ; in, Indicates crack score, Indicates the number of crack edges. This indicates the measured crack length. This indicates the measured value of the crack width; Crack score is a quantitative indicator based on RGB image analysis of the road surface, used to assess the severity of road surface cracks.

[0036] S3. Integrate a support vector machine model at the edge computing nodes, and use IRI values, RMS, and crack scores as input features for quantitative predictive analysis to obtain preliminary pavement smoothness monitoring results. Specifically... Historical IRI values, RMS, and crack scores were collected as a training set; It should be noted that historical data meeting the criteria are selected from the pavement inspection database to ensure that each record contains IRI value, RMS and crack score. Then, the IRI value, RMS and crack score are cleaned to remove outliers and missing samples to ensure data integrity. Next, the training set (70%) and validation set (30%) are divided according to the ratio to ensure balanced sample distribution. Finally, the processed IRI value, RMS and crack score are stored in a structured format for subsequent machine learning algorithm training.

[0037] The training set is input into the support vector machine model, and the grid search method is used to exhaustively search for the kernel function type, penalty coefficient and kernel parameters of the support vector machine model to form the trained support vector machine model. It should be noted that, firstly, the candidate range of kernel function types (e.g., linear kernel, polynomial kernel, Gaussian kernel) is defined, and the search range of the penalty coefficient is set to [0.001, 1000] and the search range of the kernel parameters is set to [0.00001, 100]. A grid search method is used to traverse all parameter combinations. For each set of kernel function type, penalty coefficient, and kernel parameters, cross-validation is used to evaluate the performance metrics (e.g., accuracy, recall) of the support vector machine model on the training set. The optimal combination of kernel function type, penalty coefficient, and kernel parameters is selected, and the support vector machine model is retrained, ultimately obtaining the trained support vector machine model.

[0038] The trained support vector machine model is converted into a format recognizable by the lightweight inference engine, and then loaded onto the support vector machine model on the edge computing node.

[0039] Furthermore, the trained Support Vector Machine (SVM) model is converted into a Neural Network Exchange Format (NNES) file supported by the lightweight inference engine using a model serialization tool. The NNES file fully preserves the kernel function type, support vectors, and decision function parameters of the SVM model. After conversion, the NNES file is deployed to the storage space of the edge computing node via a file transfer protocol. Once the edge computing node starts the lightweight inference engine, it calls the SVM model loading interface to read and parse the contents of the NNES file, reconstructs the computational graph structure of the SVM model, and initializes the kernel function computation units and classification decision units. Finally, the lightweight inference engine completes the memory mapping and parameter preloading of the SVM model, enabling it to process input data in real time. This entire process ensures the efficient deployment and stable operation of the SVM model in the edge computing environment.

[0040] The IRI value, RMS, and crack score are transmitted as input features to the edge computing node. The support vector machine model integrated by the edge computing node normalizes the IRI value, RMS, and crack score, and transforms the input features into a high-dimensional space through kernel function mapping for pattern recognition and regression analysis, generating preliminary pavement smoothness monitoring results.

[0041] Furthermore, the IRI value, RMS, and crack score are used as input feature vectors and linearly transformed to the [0,1] interval using the minimax normalization method. Then, the normalized feature vectors are mapped to a high-dimensional regenerative kernel Hilbert space using the radial basis function of the support vector machine (SVM) model. After mapping, the high-dimensional spatial distance between the input feature vectors and the support vectors is calculated using the radial basis function, generating a similarity matrix. Subsequently, the similarity matrix is ​​weighted and summed with the weight coefficients determined during the SVM model training phase using the SVM model's decision function. Finally, the preliminary pavement smoothness monitoring results are output using the SVM model's regression decision function. The preliminary pavement smoothness monitoring results represent the pavement smoothness status in continuous numerical form.

[0042] Furthermore, pattern recognition refers to the technical process of automatically discovering and classifying regular patterns from multidimensional feature data using mathematical modeling and machine learning methods. In road monitoring scenarios, feature vectors are mapped to a category space through feature space transformation and decision boundary construction, realizing the transformation from raw data to semantic labels. Specifically, this manifests as pattern mining and damage classification of features such as IRI values, RMS vibration data, and crack scores. Regression analysis refers to the statistical method of studying the quantitative relationship between independent variables (predictor variables) and continuous dependent variables (response variables) by establishing mathematical models.

[0043] S4. Obtain the environmental parameters and historical road surface smoothness data corresponding to the currently monitored road, construct a multivariate compensation function, dynamically correct the preliminary smoothness monitoring results, and obtain the final road surface smoothness monitoring results. Specifically, Historical environmental parameters and road surface smoothness data of monitored roads were collected, and a multivariate regression analysis method was used to establish the mapping relationship between environmental parameters and changes in road surface smoothness. Mapping relationship refers to the statistical correlation or functional dependence between environmental parameters such as pavement smoothness data (e.g., smoothness index IRI). For example, a 1°C increase in temperature leads to a 0.05 increase in the average IRI value, a 10% increase in humidity leads to a 0.03 increase in the IRI value, and a 50mm increase in monthly cumulative rainfall leads to a 0.1 increase in the IRI value.

[0044] It should be noted that historical environmental parameters include temperature, humidity, and monthly cumulative rainfall, while historical pavement smoothness data refers to the historical pavement smoothness index (IRI) value. Based on historical environmental parameters and historical pavement smoothness index (IRI) values, a linear relationship is established according to the Gauss-Markov theorem, specifically as follows: ; in, Indicates the baseline flatness. Indicates the temperature coefficient. Indicates the humidity coefficient. Indicates the rainfall erosion coefficient. Indicates the error term; The above relationships form the final mapping model between environmental parameters and road surface smoothness. The purpose of this model is to provide a baseline linear relationship for the subsequent multivariate compensation function, allowing it to further correct for the nonlinear effects of environmental factors based on this baseline linear relationship.

[0045] Based on the mapping relationship, a multivariate compensation function is constructed using a nonlinear kernel regression method. Environmental parameters are input into the multivariate compensation function for calculation to obtain the environmental compensation coefficient. It should be noted that the environmental parameters are mapped to a high-dimensional feature space using a radial basis function kernel, and an optimization problem with an ε-insensitive band is established within the support vector regression framework. The Lagrange multiplier is solved using a sequential minimum optimization algorithm to obtain the multivariate compensation function, expressed by the formula: ; in, Indicates the bias term. Represents the radial basis kernel function. Represents a multivariable compensation function. Denotes the upper limit of Lagrange multipliers. Denotes the lower bound of the Lagrange multipliers. Indicates the first Historical environmental parameters, Indicates the current environmental parameters; Substitute the current environmental parameters as the input vector into the multivariate compensation function to obtain the preliminary compensation value. Normalize the preliminary compensation value to obtain the environmental compensation coefficient η. When η>1, it means that environmental factors cause a decrease in smoothness, and when η<1, it means that environmental factors are conducive to maintaining the smoothness of the road surface.

[0046] After Bayesian probability fusion of the environmental compensation coefficient and the preliminary road surface smoothness monitoring results, dynamic adjustment is performed using Kalman filtering to form the final road surface smoothness monitoring results.

[0047] Furthermore, a Bayesian probabilistic fusion method is used to fuse the compensation coefficients with the preliminary pavement evenness monitoring results to obtain observed values, thereby improving the accuracy and stability of the pavement evenness monitoring results. Subsequently, a Kalman filter algorithm is used to dynamically adjust the fused monitoring data, that is, the preliminary pavement evenness monitoring results are used as predicted values ​​and compared with the observed values ​​after Bayesian probabilistic fusion to obtain observation residuals. Based on the observation residuals and the noise covariance and observation noise covariance of the Kalman filter algorithm, the Kalman gain is calculated, and the predicted values ​​are then corrected to obtain updated pavement evenness estimates. The state estimation principle of the Kalman filter algorithm is to continuously update the current estimated state of pavement evenness based on new observation data, thereby reducing the influence of random errors and biases.

[0048] S5. Based on the final road surface smoothness monitoring results, generate a road surface monitoring report and upload it to the cloud server for real-time monitoring.

[0049] The final road surface smoothness monitoring results are fused with GPS coordinates, timestamps, and environmental parameters to generate structured monitoring data. It should be noted that the final road surface evenness monitoring results are used as the core monitoring indicator through spatiotemporal data fusion, with GPS coordinates as the spatial reference benchmark, timestamps as the time-series index, and environmental parameters as auxiliary variables. Structured monitoring data is constructed according to the OGC geographic information standard. The structured monitoring data is organized using a four-dimensional spatiotemporal data model. The spatial dimension is divided into 100m × 100m grids according to the UTM projection coordinate system, the temporal dimension uses a 5-minute sampling interval, and the attribute dimension contains 21 standardized data fields. The final result is a structured dataset for road surface health monitoring with complete spatiotemporal attributes and environmental characteristics.

[0050] The structured monitoring data is encapsulated in JSON format, and the edge computing node calls the WebSocket protocol to establish a persistent connection with the cloud server. The cloud server receives the structured monitoring data and performs real-time monitoring.

[0051] Furthermore, the structured monitoring data is encapsulated according to the JSON Schema specification, including GPS coordinates in the WGS84 coordinate system, ISO8601 standard timestamps, road surface smoothness values, and environmental parameter fields. Edge computing nodes initialize WebSocket client connections, establishing a persistent full-duplex communication channel with the cloud server based on the RFC 6455 standard, using TLS 1.3 for transport layer encryption. The encapsulated JSON data is transmitted via the WebSocket protocol in binary frames, with a default setting of 5 seconds per transmission. The WebSocket server endpoint deployed on the cloud server receives the structured monitoring data in real time, updates the road surface status records in the spatiotemporal database after parsing the field values ​​using the JSON parser. The edge computing nodes and the cloud server maintain connection activity through the WebSocket Ping / Pong mechanism. In case of network interruption, the local SQLite cache is automatically activated, and incremental data synchronization is performed upon reconnection to ensure the integrity and timeliness of the structured monitoring data. The cloud server provides the field values ​​extracted from the structured monitoring data to the visualization platform through a RESTful API interface, enabling real-time monitoring of road surface conditions.

[0052] This embodiment also provides a road surface smoothness survey and monitoring system, including: The data acquisition module collects data on road surface height changes, vehicle vibration data, and road surface RGB images, and performs timestamp synchronization and spatial coordinate unification processing. The crack scoring module uses IRI to calculate the IRI value of road surface height variation data, uses root mean square to calculate the RMS value of vehicle vibration data, and uses Canny to calculate the crack score of the road surface RGB image. The monitoring module integrates a support vector machine model at the edge computing node and uses IRI value, RMS and crack score as input features for quantitative prediction and analysis to obtain preliminary road surface smoothness monitoring results. The adjustment module obtains the environmental parameters and historical road surface smoothness data corresponding to the currently monitored road, constructs a multivariate compensation function, dynamically corrects the preliminary smoothness monitoring results, and obtains the final road surface smoothness monitoring results. The reporting module generates a road surface monitoring report based on the final road surface smoothness monitoring results and uploads it to the cloud server for real-time monitoring.

[0053] This embodiment also provides a computer device applicable to the road on-site smoothness survey and monitoring method, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the road on-site smoothness survey and monitoring method proposed in the above embodiment.

[0054] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0055] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the method for road surface smoothness survey and monitoring as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0056] In summary, this invention, through its road smoothness monitoring method, forms a complete chain from data acquisition to decision support, improving the accuracy and efficiency of road smoothness monitoring. This effectively ensures road traffic safety and driving comfort, extends the service life of roads, demonstrates the enormous potential of the deep integration of modern sensing technology and intelligent computing, and has also proven its powerful ability to improve road management and maintenance efficiency in practical applications.

[0057] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for monitoring the road site flatness survey, characterized in that: The application relates to a road surface roughness monitoring method and device based on edge computing and cloud computing. Road surface height change data, vehicle vibration data and road surface RGB images are collected, time stamp synchronization and space coordinate unification processing are carried out; IRI values of the road surface height change data are calculated using IRI, RMS values of the vehicle vibration data are calculated using root mean square, and crack scores of the road surface RGB images are calculated using Canny; A support vector machine model is integrated on an edge computing node, IRI values, RMS and crack scores are taken as input features for quantitative prediction analysis, and preliminary road surface roughness monitoring results are obtained; Environment parameters corresponding to the current monitoring road and historical road surface roughness data are obtained, a multivariate compensation function is constructed, the preliminary roughness monitoring results are dynamically corrected, and final road surface roughness monitoring results are obtained; Based on the final road surface roughness monitoring results, a road surface monitoring report is generated and uploaded to a cloud server for real-time monitoring.

2. The method for monitoring the road site flatness survey according to claim 1, characterized in that: The road surface height change data comprises road longitudinal elevation, road local concave-convex information and road roughness; The vehicle vibration data comprises acceleration and vibration frequency; Time stamp synchronization processing is performed on the road surface height change data, vehicle vibration data and road surface RGB images by using space-time information provided by GPS positioning and inertial navigation equipment; The road surface height change data, vehicle vibration data and road surface RGB images are converted to a unified space coordinate frame by a geographic information unit.

3. The method for monitoring the road site flatness survey according to claim 2, characterized in that: The IRI values of the road surface height change data are calculated using IRI, the RMS values of the vehicle vibration data are calculated using root mean square, and the crack scores of the road surface RGB images are calculated using Canny, and the specific process is as follows, Based on the road surface height change data, the ratio of the relative displacement of the vehicle suspension to the driving distance is calculated by IRI to obtain the roughness index IRI value; The time domain signal of the vehicle vibration data is subjected to point-by-point square operation to obtain a squared time domain signal sequence; The time domain signal sequence is subjected to dynamic convolution filtering to calculate the average value of the vehicle vibration data, and the RMS value of the vehicle vibration data is obtained through convolution root mean square conversion; The road surface RGB image is subjected to gradient calculation, non-maximum suppression and double threshold detection by Canny to extract a crack edge binary image in the road surface RGB image; Based on the crack edge binary image, the skeletonization is used to measure the crack length, and the crack width is calculated by distance transformation; The number of crack edges in the road surface RGB image is counted, and the crack length and width are combined for quantitative calculation to obtain the crack score of the road surface RGB image.

4. The method for monitoring the road site flatness survey according to claim 3, characterized in that: The support vector machine model is integrated on the edge computing node, and the specific process is as follows, The historical IRI values, RMS and crack scores are collected as a training set; The training set is input into the support vector machine model, and a grid search method is used to exhaustively search the kernel function type, penalty coefficient and kernel parameter of the support vector machine model to form a trained support vector machine model; The trained support vector machine model is converted into a format recognizable by a lightweight inference engine, and the support vector machine model is loaded on the edge computing node.

5. The method for monitoring the road site flatness survey according to claim 4, characterized in that: The IRI values, RMS and crack scores are taken as input features for quantitative prediction analysis, and the specific process is as follows, The IRI value, RMS and crack score are transmitted to the edge computing node as input features, the support vector machine model integrated by the edge computing node performs normalization processing on the IRI value, RMS and crack score, and converts the input features to a high-dimensional space through kernel function mapping for pattern recognition and regression analysis, thereby generating a preliminary road flatness monitoring result.

6. The method for monitoring the road site flatness survey according to claim 5, characterized in that: The environment parameters corresponding to the current monitoring road and the historical road flatness data are acquired, a multivariate compensation function is constructed, the preliminary flatness monitoring result is dynamically corrected, and a final road flatness monitoring result is obtained, specifically, The environment parameters and road flatness data of the historical monitoring road are collected, and a mapping relationship between the environment parameters and the change of road flatness is established by using a multivariate regression analysis method; Based on the mapping relationship, a multivariate compensation function is constructed by using a nonlinear kernel regression method, the environment parameters are input into the multivariate compensation function for calculation, and an environment compensation coefficient is obtained; After the environment compensation coefficient and the preliminary road flatness monitoring result are fused by using Bayesian probability, the final road flatness monitoring result is dynamically adjusted by using Kalman filtering.

7. The method for monitoring the road site flatness survey according to claim 6, characterized in that: Based on the final road flatness monitoring result, a road monitoring report is generated, and the cloud server is uploaded for real-time monitoring, specifically, The final road flatness monitoring result, GPS coordinates, time stamp and environment parameters are fused to generate structured monitoring data; After the structured monitoring data is packaged in JSON format, the edge computing node calls the WebSocket protocol to establish a persistent connection with the cloud server, and the cloud server receives the structured monitoring data for real-time monitoring.

8. A road site flatness survey monitoring system based on the road site flatness survey monitoring method according to any one of claims 1 to 7, characterized in that: Comprise, A data acquisition module acquires road surface height change data, vehicle vibration data and road RGB images, and performs time stamp synchronization and spatial coordinate unification processing; A crack score module calculates the IRI value of the road surface height change data using IRI, calculates the RMS value of the vehicle vibration data using root mean square, and calculates the crack score of the road RGB image using Canny; A monitoring module integrates a support vector machine model in the edge computing node, and performs quantitative prediction analysis on the IRI value, RMS and crack score as input features to obtain a preliminary road flatness monitoring result; An adjustment module acquires environment parameters corresponding to the current monitoring road and historical road flatness data, constructs a multivariate compensation function, dynamically corrects the preliminary flatness monitoring result, and obtains a final road flatness monitoring result; A report module generates a road monitoring report based on the final road flatness monitoring result, and uploads the cloud server for real-time monitoring. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is characterized in that: The processor executes the computer program to realize the steps of the road site flatness surveying and monitoring method of any one of claims 1-7.

10. A computer readable storage medium having stored thereon a computer program, characterized in that: The computer program is executed by the processor to realize the steps of the road site flatness surveying and monitoring method of any one of claims 1-7.

Images