Road hidden disease geometric parameter inversion method, device and equipment and storage medium

Abstract

The application discloses a road hidden disease geometric parameter inversion method, device and equipment and a storage medium. The method comprises the following steps: fusing strain and stress double-parameter mechanical response data collected by a multi-sensor array, constructing a parameterized driving road damage three-dimensional finite element model and defining a disease geometric parameter vector, iteratively correcting the disease geometric parameter through a regularized geometric-mechanical residual error objective function until the simulation response approximates the measured response, and finally realizing accurate inversion and digital twin reconstruction of the road hidden disease geometric parameter. The strain and stress sensors are fused, the specific three-dimensional position, geometric shape and accurate volume of the disease are accurately inverted from the discrete point mechanical response data collected by the sensors, and the problem of over-maintenance or insufficient maintenance of the road caused by unknown disease geometric parameters is solved.

Description

Technical Field

[0001] This invention relates to the field of road engineering, and in particular to a method, apparatus, equipment, and storage medium for inverting geometric parameters of hidden road defects. Background Technology

[0002] With the development of smart highways, embedding sensors within road structures has become a common method for monitoring road health. Current monitoring systems primarily rely on "threshold alarm" logic, meaning that an alarm is triggered when sensor readings, such as tensile strain or vibration amplitude, exceed preset safety values. However, sensors can only acquire discrete points of mechanical response data and cannot directly reflect the continuous spatial distribution characteristics of road defects. Existing technologies can only provide a simple assessment of the presence of defects, but cannot accurately determine their specific location, geometry, and precise volume. Because the exact geometric parameters of the defects are unknown, maintenance personnel cannot accurately calculate grouting volume or determine the excavation area, often leading to over- or under-maintenance. Summary of the Invention

[0003] The main objective of this invention is to provide a method, apparatus, equipment, and storage medium for inverting the geometric parameters of hidden road defects, aiming to solve the technical problem that existing technologies cannot accurately determine the location, shape, and volume of defects, and that the unknown geometric parameters of defects lead to over- or under-maintenance of roads.

[0004] To achieve the above objectives, the present invention provides a method for inverting geometric parameters of hidden road defects, the method comprising the following steps:

[0005] A multi-source heterogeneous sensing network for road structure is constructed and road mechanical response data is collected. The data is then processed to obtain the target residual vector for inversion. The multi-source heterogeneous sensing network includes strain sensors and stress sensors.

[0006] A three-dimensional continuous medium finite element model is constructed based on the pavement design information of the road structure. The geometric parameter vector of the disease in the three-dimensional continuous medium finite element model is defined and the initial value is assigned to the geometric parameter vector of the disease, so as to obtain a parameterized driven three-dimensional finite element model of road damage.

[0007] Based on the target residual vector, a regularized geometric-mechanical residual objective function is constructed, and the values ​​of the disease geometric parameter vector are iteratively corrected based on the geometric-mechanical residual objective function.

[0008] The iterative process is converged. If the convergence condition is met, the numerical value of the disease geometric parameter vector obtained in the current iteration is output as the disease geometric optimal parameter vector. The disease geometric optimal parameter vector is then mapped back to three-dimensional space to realize the geometric parameter inversion of hidden road diseases.

[0009] Optionally, the construction of a multi-source heterogeneous sensing network for the road structure and the acquisition of road mechanical response data, followed by processing to obtain a target residual vector for inversion, includes:

[0010] A grid-like sensor array consisting of strain sensors and stress sensors is deployed in the mechanically sensitive layers of the road structure to form a multi-source heterogeneous sensing network.

[0011] Under the vehicle test load, the original mechanical response time history signal of the sensor array is collected, and the peak response data of each sensor is extracted when the center of the vehicle test load passes directly above it.

[0012] Obtain the baseline mechanical response vector under road health conditions and the measured mechanical response vector within the current monitoring period;

[0013] Construct a weighted diagonal matrix, perform difference processing on the measured mechanical response vector and the reference mechanical response vector, and then perform weighted calculation using the weighted diagonal matrix to obtain the target residual vector for inversion, as shown in the following formula:

[0014]

[0015]

[0016]

[0017] in, Represents the target residual vector. This represents the measured mechanical response vector. Represents the reference mechanical response vector. This represents the weighting coefficient of the k-th sensor. This represents the weight diagonal matrix. Represents a diagonal matrix. This indicates the number of sensors in a multi-source heterogeneous sensing network. This represents the maximum value of all elements in the reference mechanical response vector. This represents the signal-to-noise ratio coefficient of the k-th sensor.

[0018] Optionally, the construction of a three-dimensional continuous medium finite element model based on pavement design information of road structure, the definition of the defect geometric parameter vector in the three-dimensional continuous medium finite element model and the assignment of initial values ​​to the defect geometric parameter vector to obtain a parameterized driven three-dimensional finite element model of road damage, including:

[0019] A three-dimensional continuous medium finite element model with a preset size is established based on the pavement design information of the road structure. Symmetrical boundaries are set for the three-dimensional continuous medium finite element model, and road structure layers are divided according to the depth direction. Constitutive parameters of standard pavement material are assigned to each structure layer.

[0020] The three-dimensional spatial coordinates of the sensor array in the multi-source heterogeneous sensing network are mapped to the mesh of the three-dimensional continuous medium finite element model, and the corresponding set of virtual monitoring points is marked.

[0021] Using the material weakening method, hidden road defects are equivalent to spatial ellipsoids, and the defect geometric parameter vector of the spatial ellipsoid in the three-dimensional continuous medium finite element model is defined.

[0022] Establish a discrimination function for the diseased unit, and set a material property assignment strategy for the three-dimensional continuous medium finite element model based on the discrimination function;

[0023] Based on the material property assignment strategy and the virtual monitoring point set traversing each element point in the three-dimensional continuous medium finite element model, initial values ​​are assigned to the disease geometric parameter vector to obtain a parameterized driven three-dimensional finite element model of road damage.

[0024] Optionally, the iterative correction of the disease geometric parameter vector based on the geometric-mechanical residual objective function includes:

[0025] A forward simulation calculation is performed based on the aforementioned three-dimensional finite element model of road damage to obtain the simulation response vector;

[0026] A sensitivity matrix is ​​constructed based on the dimension of the disease geometric parameter vector and the dimension of the sensor data collected by the multi-source heterogeneous sensing network.

[0027] The finite difference perturbation method is used to individually perturb each parameter in the simulation response vector to obtain the perturbed simulation response vector;

[0028] Based on the forward difference formula and the simulated response vector after the disturbance, the matrix elements in the sensitivity matrix are calculated to obtain the solved sensitivity matrix.

[0029] The values ​​of the disease geometric parameter vector are iteratively corrected based on the sensitivity matrix and the geometric-mechanical residual objective function.

[0030] Optionally, the geometric-mechanical residual objective function is used to optimize the weighted Euclidean distance between the simulated response vector and the measured mechanical response vector, and the mathematical expression of the geometric-mechanical residual objective function is:

[0031]

[0032] in, This represents the value of the objective function after regularization. Let represent the square of the Euclidean norm of a vector. Represents the regularization coefficient. Represents the regularization matrix. This represents the simulated response vector under the current values ​​of the disease's geometric parameter vectors. This represents the measured mechanical response vector. Represents the vector of geometric parameters of the disease. A vector representing prior information about the geometric parameters of the disease. This represents the weight diagonal matrix, used to configure sensor weights in a multi-source heterogeneous sensing network.

[0033] Optionally, the iterative correction of the disease geometric parameter vector based on the sensitivity matrix and the geometric-mechanical residual objective function includes:

[0034] Using the Levenberg-Marquardt optimization algorithm, a system of linear equations is constructed based on the sensitivity matrix. The mathematical expression of the system of linear equations is as follows:

[0035]

[0036] in, This represents the sensitivity matrix at the k-th iteration. Indicates transpose. This represents the adaptive damping factor of the optimization algorithm in the k-th iteration. Represents the identity matrix. This represents the parameter adjustment increment at the k-th iteration. This represents the vector of geometric parameters of the disease at the k-th iteration.

[0037] Solving the system of linear equations yields the parameter correction increments of the disease geometric parameter vector;

[0038] The current value of the disease geometric parameter vector is added to the parameter correction increment to obtain the trial parameter vector;

[0039] The proposed parameter vector is subjected to projection constraint checks and corrections to obtain an updated disease geometric parameter vector, thus completing one iteration of the numerical correction of the disease geometric parameter vector.

[0040] Optionally, after determining the convergence of the iterative process, the following may be included:

[0041] Calculate the root mean square error between the current simulated response vector and the measured mechanical response vector;

[0042] Calculate the Euclidean norm of the parameter correction increment for the current iteration;

[0043] Record the current iteration number;

[0044] If the convergence condition is met, the numerical values ​​of the disease geometric parameter vector obtained in the current iteration are output as the disease geometric optimal parameter vector. The convergence condition includes at least one of the following:

[0045] The relative root mean square error is lower than a preset accuracy threshold;

[0046] The Euclidean norm is lower than a preset step size threshold;

[0047] If the current iteration count reaches the preset iteration upper limit threshold and the convergence condition is not met, the circuit breaker mechanism is triggered, and the value of the disease geometric parameter vector with the smallest residual in the historical iteration process is output as the disease geometric optimal parameter vector.

[0048] The optimal geometric parameter vector of the road defects is mapped back to three-dimensional space to realize the geometric parameter inversion of hidden road defects.

[0049] Furthermore, to achieve the above objectives, the present invention also proposes a road hidden defect geometric parameter inversion device that applies the road hidden defect geometric parameter inversion method described above, the road hidden defect geometric parameter inversion device comprising:

[0050] The data acquisition module is used to construct a multi-source heterogeneous sensing network for the road structure and collect road mechanical response data, and process it to obtain the target residual vector for inversion. The multi-source heterogeneous sensing network includes strain sensors and stress sensors.

[0051] The damage modeling module is used to construct a three-dimensional continuous medium finite element model based on the pavement design information of the road structure, define the disease geometric parameter vector in the three-dimensional continuous medium finite element model and assign initial values ​​to the disease geometric parameter vector to obtain a parameterized driven three-dimensional finite element model of road damage.

[0052] The iterative correction module is used to construct a regularized geometric-mechanical residual objective function based on the target residual vector, and to iteratively correct the values ​​of the disease geometric parameter vector based on the geometric-mechanical residual objective function.

[0053] The parameter inversion output module is used to determine the convergence of the iteration process. If the convergence condition is met, the numerical value of the disease geometric parameter vector obtained by the current iteration is output as the disease geometric optimal parameter vector. The disease geometric optimal parameter vector is then mapped back to three-dimensional space to realize the geometric parameter inversion of hidden road diseases.

[0054] Furthermore, to achieve the above objectives, this application also proposes a road hidden defect geometric parameter inversion device, the device comprising: a memory, a processor, and a road hidden defect geometric parameter inversion program stored in the memory, the processor being used to run the road hidden defect geometric parameter inversion program, the computer program being configured to implement the steps of the road hidden defect geometric parameter inversion method as described above.

[0055] In addition, to achieve the above objectives, this application also proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the road hidden defects geometric parameter inversion method described above.

[0056] This invention improves the comprehensiveness and reliability of data acquisition through a multi-source heterogeneous sensing network, establishes a bridge between defects and mechanical responses through a parameterized finite element model, ensures the rationality and accuracy of the inversion results through regularized iterative optimization, and realizes the intuitive presentation of defect information through three-dimensional visualization. It effectively solves the core pain point of not being able to accurately obtain the geometric parameters of defects, realizes the accurate inversion and digital reconstruction of the geometric parameters of hidden defects, effectively improves the detection accuracy and efficiency of hidden road defects, provides accurate decision-making basis for road maintenance, avoids the problems of over-maintenance or under-maintenance, reduces road maintenance costs, extends the service life of roads, and promotes the intelligent and precise development of smart highway maintenance. Attached Figure Description

[0057] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0058] Figure 1 This is a schematic diagram of the structure of the road hidden defects geometric parameter inversion device in the hardware operating environment involved in the embodiments of the present invention;

[0059] Figure 2 This is a flowchart illustrating the first embodiment of the method for inverting geometric parameters of hidden road defects according to the present invention.

[0060] Figure 3 This is a flowchart illustrating the second embodiment of the method for inverting geometric parameters of hidden road defects according to the present invention.

[0061] Figure 4 This is a structural block diagram of the first embodiment of the road hidden defects geometric parameter inversion device of the present invention.

[0062] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0063] It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention.

[0064] Reference Figure 1 , Figure 1 This is a schematic diagram of the road hidden defects geometric parameter inversion device in the hardware operating environment involved in the embodiment of the present invention.

[0065] like Figure 1 As shown, the road hidden defects geometric parameter inversion device may include: a processor 1001, such as a central processing unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. The communication bus 1002 is used to enable communication between these components. The user interface 1003 may include a display screen and an input unit such as a keyboard; the user interface 1003 may also include standard wired and wireless interfaces. The network interface 1004 may optionally include standard wired and wireless interfaces (such as Wireless-Fidelity (Wi-Fi) interfaces). The memory 1005 may be high-speed random access memory (RAM) or stable non-volatile memory (NVM), such as a disk storage device. The memory 1005 may also optionally be a storage device independent of the aforementioned processor 1001.

[0066] Those skilled in the art will understand that Figure 1 The structure shown does not constitute a limitation on the geometric parameter inversion device for hidden road defects. It may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0067] like Figure 1 As shown, the memory 1005, which is a computer-readable storage medium, may include an operating system, a network communication module, a user interface module, and a road hidden defects geometric parameter inversion program.

[0068] exist Figure 1In the road hidden defect geometric parameter inversion device shown, the network interface 1004 is mainly used for data communication with the network server; the user interface 1003 is mainly used for data interaction with the user; the processor 1001 and memory 1005 in the road hidden defect geometric parameter inversion device of the present invention can be set in the road hidden defect geometric parameter inversion device. The road hidden defect geometric parameter inversion device calls the road hidden defect geometric parameter inversion program stored in the memory 1005 through the processor 1001 and executes the road hidden defect geometric parameter inversion method provided in the embodiment of the present invention.

[0069] This invention provides a method for inverting geometric parameters of hidden road defects, referring to... Figure 2 , Figure 2 This is a flowchart illustrating the first embodiment of the method for inverting geometric parameters of hidden road defects according to the present invention.

[0070] In this embodiment, the method for inverting the geometric parameters of hidden road defects includes the following steps:

[0071] Step S10: Construct a multi-source heterogeneous sensing network for the road structure and collect road mechanical response data, process it to obtain the target residual vector for inversion, wherein the multi-source heterogeneous sensing network includes strain sensors and stress sensors.

[0072] It should be understood that the executing entity of this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or a terminal electronic device capable of performing the above functions. The following description uses a road hidden defect geometric parameter inversion device (hereinafter referred to as the inversion device) as an example to illustrate this embodiment and the following embodiments.

[0073] It should be noted that the multi-source heterogeneous sensing network refers to a monitoring network composed of strain sensors and stress sensors. Through multi-parameter collaborative monitoring, it can achieve comprehensive capture of anomalies in the internal mechanical field of the road, which is different from the monitoring method of a single type of sensor and improves the comprehensiveness and reliability of data acquisition.

[0074] Road mechanical response data refers to the mechanical signal data such as strain and stress generated inside the road structure under external loads (such as vehicle loads). It is used to reflect the health status of the road structure and capture the impact of hidden defects. It includes baseline data under healthy conditions and measured data under current monitoring.

[0075] The target residual vector can be the difference vector between the current measured mechanical response vector and the healthy baseline vector after preprocessing, normalization and weighted calculation. It is used to quantify the degree of deviation between the current road condition and the healthy condition. It is the core target data for subsequent inversion calculation and can effectively eliminate systematic errors from non-disease factors such as ambient temperature and material aging.

[0076] In practical implementation, sensitive layers of road structure mechanics are selected (preferably the bottom of the asphalt surface layer and the top surface of the base layer, which are prone to bending and tensile failure). A sensor array consisting of strain sensors and stress sensors is deployed, with the sensors arranged in a grid to cover the road section to be monitored. The three-dimensional spatial coordinates of each sensor are recorded, and the sensor deployment density is reasonably set according to the road's service condition to ensure comprehensive capture of mechanical field anomalies caused by hidden defects within the road. Among them, strain sensors are used to capture the triaxial tensile deformation of the road structural layers, and stress sensors (such as earth pressure cells) are used to capture the triaxial compressive stress distribution between the road structural layers.

[0077] In some embodiments, the inversion device can acquire the raw time history signal of the sensor network under a standard test load (such as the BZZ-100 standard axle load), extract the peak response when the center of the load passes directly above the sensor as the basic data for inversion, and avoid interference from dynamic vehicle loads. At the same time, calibration tests are performed in the early stage of road completion or when the health status is confirmed, and baseline response data under the health status is acquired. The acquired raw data is low-pass filtered to eliminate noise interference. The strain data and stress data are normalized to eliminate the difference in their dimensions. A healthy baseline vector and the current measured response vector are constructed, and the difference between the two is calculated by weighting to obtain the target residual vector used for inversion. The weight matrix is ​​assigned according to the sensor signal quality and the distance from the defect area. Sensors closer to the anomaly area and with good signal quality have higher weights, while those with lower weights have lower weights, in order to enhance the accuracy of defect location.

[0078] Understandably, this embodiment solves the problem that existing monitoring systems can only acquire discrete point mechanical data and cannot eliminate interference from non-disease factors. By deploying a multi-source heterogeneous sensor array, it achieves comprehensive and accurate acquisition of road mechanical response data. Through data preprocessing and the construction of target residual vectors, it eliminates dimensional differences and systematic errors, filters out effective data directly related to hidden defects, provides a reliable and accurate data source for subsequent inversion calculations, avoids interference from invalid data on inversion results, and improves the basic reliability of inversion.

[0079] Furthermore, to improve the comprehensiveness, accuracy, and relevance of data collection, step S10 above may include:

[0080] Step S101: Deploy a grid-like sensor array consisting of strain sensors and stress sensors in the mechanically sensitive layers of the road structure to form a multi-source heterogeneous sensing network.

[0081] Understandably, in order to comprehensively capture the mechanical field anomalies caused by hidden defects within the road, this invention employs a dual-parameter collaborative monitoring network of strain and stress. Strain sensors are used to capture the triaxial tensile deformation of the road structural layers, while stress sensors (such as earth pressure cells) are used to capture the triaxial compressive stress distribution between the road structural layers.

[0082] N sensors are embedded in the mechanically sensitive layers of the road structure, and the sensor density is selected based on service conditions. Typically, the bottom of the asphalt surface layer and the top surface of the base course are selected as the areas most prone to bending and tensile damage.

[0083] The sensors are arranged in a grid pattern, covering the road section to be monitored, and the three-dimensional spatial coordinates of each sensor are recorded. , where i = 1, 2, ..., N.

[0084] Step S102: Under the vehicle test load, collect the original mechanical response time history signal of the sensor array, and extract the peak response data of each sensor when the center of the vehicle test load passes directly above it.

[0085] In the specific implementation, we define time t as the raw dataset collected by the sensor network under the standard test load.

[0086] Define a measurement set M. Let the total number of sensors be N, which includes... One strain sensor and Earth pressure box.

[0087]

[0088] For the i-th strain sensor, its reading is For the j-th earth pressure cell, its reading is .

[0089] Since the vehicle load is dynamic, we need to extract the peak response when the center of the load passes directly above the sensor as the basis for inversion.

[0090] Extract the peak data from the i-th sensor :

[0091]

[0092] in, The original time-history signal is denoted as Filter(·), which is a low-pass filter function, and T is the time window in which the vehicle passes.

[0093] Step S103: Obtain the baseline mechanical response vector under the road health condition and the measured mechanical response vector within the current monitoring period.

[0094] It should be noted that the baseline mechanical response vector refers to the peak mechanical response vector obtained by collecting and processing data through standard test loads when the road is in a healthy state (no hidden defects and intact structure). It is used to quantify the deviation of the current state of the road from the healthy state.

[0095] The measured mechanical response vector refers to the peak mechanical response vector collected and processed under the same standard test load within the current monitoring period. It reflects the actual mechanical state of the current road structure, and the difference between it and the baseline vector directly reflects changes in the road structure's health. The monitoring period refers to a fixed time interval for road health monitoring, such as one month for routine road sections and 15 days for sections with high rates of road defects, ensuring timely detection of abnormal mechanical changes in the road structure.

[0096] Step S104: Construct a weighted diagonal matrix, perform differential processing on the measured mechanical response vector and the reference mechanical response vector, and perform weighted calculation in combination with the weighted diagonal matrix to obtain the target residual vector for inversion.

[0097] In practical implementation, to eliminate systematic errors caused by non-pathogenic factors such as ambient temperature and material aging, this invention uses the differential response vector as the sole objective for subsequent finite element inversion. Strain and pressure data need to be normalized to eliminate dimensional differences.

[0098] Calibration tests are conducted at the initial stage of road completion or when the road is confirmed to be in a healthy state to construct a health baseline vector. :

[0099]

[0100] Within the current monitoring period, collect measured response vectors under the same load conditions. :

[0101]

[0102] Calculate the target residual vector for inversion. :

[0103]

[0104] in, This is a weighted diagonal matrix.

[0105]

[0106] in, Represents the target residual vector. This represents the measured mechanical response vector. Represents the reference mechanical response vector. This represents the weighting coefficient of the k-th sensor. This represents the weight diagonal matrix. Represents a diagonal matrix. This indicates the number of sensors in a multi-source heterogeneous sensing network. This represents the maximum value of all elements in the reference mechanical response vector. The first term is used to eliminate the order-of-magnitude difference between strain (10^-6) and stress (10^3). This represents the signal-to-noise ratio coefficient of the k-th sensor. For sensors with poor signal quality or that are too far from the affected area, their weight is reduced.

[0107] Step S20: Construct a three-dimensional continuous medium finite element model based on the pavement design information of the road structure, define the defect geometric parameter vector in the three-dimensional continuous medium finite element model and assign initial values ​​to the defect geometric parameter vector to obtain a parameterized driven three-dimensional finite element model of road damage.

[0108] It should be noted that pavement design information refers to the various technical parameters determined during the road design phase, including the division of road structural layers, the thickness of each layer, material constitutive parameters (elastic modulus, Poisson's ratio, etc.), and road dimensions.

[0109] The three-dimensional continuous medium finite element model refers to a three-dimensional model based on the theory of continuous medium mechanics, which discretizes the road structure into a finite number of grid elements and simulates the mechanical response of the road structure under external loads through numerical calculation. It can accurately reflect the mechanical distribution characteristics inside the road.

[0110] It should be noted that the defect geometric parameter vector refers to a set of parameters used to describe the equivalent geometric shape of hidden road defects. For example, it may include the center coordinates of the defect, the length of the ellipsoid semi-axis, and the rotation angle. This vector can be used to quantitatively describe the geometric shape of the defect and provide operable variables for subsequent parameter optimization.

[0111] Parametrically driven three-dimensional finite element model of road damage refers to a finite element model that uses the geometric parameter vector of the disease as the driving variable and can dynamically update the distribution and morphology of the disease units in the model by adjusting the parameter vector, thus realizing a one-to-one correspondence between the disease morphology and the parameter vector.

[0112] In some embodiments, the inversion device can establish a three-dimensional solid model consistent with the actual road based on road design data (including the thickness of each structural layer of the road, material parameters, etc.). The model size is set according to the actual situation of the road section to be monitored, and symmetrical boundaries are set to reduce boundary effects in dynamic analysis. The road model is layered along the depth direction according to the actual structure (such as upper layer, middle layer, lower layer, upper base layer, etc.), and each layer is assigned corresponding constitutive parameters such as elastic modulus, Poisson's ratio, and density to ensure that the model is consistent with the mechanical properties of the actual road. At the same time, the coordinates of the real sensors deployed in the multi-source heterogeneous sensing network are mapped to the finite element mesh to mark the corresponding virtual monitoring point set for subsequent simulation data extraction.

[0113] Furthermore, in order to accurately quantify and model hidden road defects and improve the model's realism and reliability, step S20 above may include:

[0114] Step S201: Based on the pavement design information of the road structure, establish a three-dimensional continuous medium finite element model with a preset size, set symmetrical boundaries for the three-dimensional continuous medium finite element model, divide the road structure into layers according to the depth direction, and assign constitutive parameters of standard pavement material to each structural layer;

[0115] Step S202: Map the three-dimensional spatial coordinates of the sensor array in the multi-source heterogeneous sensing network to the mesh of the three-dimensional continuous medium finite element model, and mark the corresponding set of virtual monitoring points;

[0116] Step S203: Using the material weakening method, the hidden road defects are equivalent to a spatial ellipsoid, and the defect geometric parameter vector of the spatial ellipsoid in the three-dimensional continuous medium finite element model is defined.

[0117] Step S204: Establish the discrimination function of the defective unit, and set the material property assignment strategy for the three-dimensional continuous medium finite element model based on the discrimination function;

[0118] Step S205: Based on the material property assignment strategy and the virtual monitoring point set, traverse each element point in the three-dimensional continuous medium finite element model, assign initial values ​​to the disease geometric parameter vector, and obtain the parameterized driven three-dimensional finite element model of road damage.

[0119] In practice, the inversion equipment establishes a three-dimensional continuous medium finite element model based on road design data.

[0120] Create a three-dimensional solid model with dimensions L × W × H. Set symmetrical boundaries for the model to reduce boundary effects in dynamic analysis.

[0121] The model was designed as a K-layer structure along the depth direction based on actual road conditions. This study adopted a 7-layer structure, including the top layer, middle layer, bottom layer, upper base layer, lower base layer, subbase layer, and subgrade. Constitutive parameters such as elastic modulus, Poisson's ratio, and density of each layer material were set.

[0122] The real sensor coordinates recorded in the multi-source heterogeneous sensing network Mapped onto the finite element mesh, the mesh node closest to that coordinate is identified and marked as the virtual monitoring point set. .

[0123] To enable the disease morphology to be optimized by algorithms, a region-based material weakening method is adopted, and parametric geometry is introduced to describe the disease.

[0124] We assume that hidden defects inside the road (such as voids or looseness) are approximated as a spatial ellipsoid. Although the actual shape of the defect is irregular, the ellipsoid has enough degrees of freedom parameters such as center position, size, and flatness to equivalently represent its mechanical influence domain.

[0125] Define the geometric parameter vector P of the disease, and construct the unknown parameter vector to be inverted:

[0126]

[0127] It represents the global coordinates of the geometric center of the lesion. The lengths of the semi-axis of the ellipsoid along the X, Y, Z axes in the local coordinate system. The angle of rotation of the principal axis of the disease in the XY plane;

[0128] Define a discriminant function for any point (x, y, z) in space relative to the disease and its geometric parameters P:

[0129]

[0130] in( , , The coordinates are the local coordinates obtained through coordinate transformation.

[0131]

[0132] In order to dynamically generate models containing different defects during the iteration process, the following material property assignment logic is established.

[0133] Traverse all integration points in the finite element model, based on their coordinates Determine material properties;

[0134] Calculate the discriminant function value ;

[0135] like If a point is located inside an ellipsoid, the element is identified as a diseased element. It is then assigned either air or loose medium properties.

[0136] like If a point is located outside the ellipsoid, the element is considered a healthy road element. Standard pavement material properties are then assigned to the element in the corresponding layer.

[0137] Step S30: Construct a regularized geometric-mechanical residual objective function based on the target residual vector, and iteratively correct the values ​​of the disease geometric parameter vector based on the geometric-mechanical residual objective function.

[0138] It should be noted that the regularized geometric-mechanical residual objective function is a target loss function used to measure the difference between the simulated guess of the defect geometry and the actual measured defect geometry. It combines data fitting terms and regularization terms, which not only ensures the consistency between the simulation response and the measured data, but also avoids inversion results that are not physically meaningful.

[0139] It is understandable that this embodiment, through the construction of a regularized objective function, not only ensures the consistency between the inversion results and the measured data, but also avoids the occurrence of non-physical interpretations, improves the rationality of the inversion results, gradually narrows the gap between simulation and real diseases, and provides core support for finally obtaining accurate disease geometric parameters.

[0140] In some embodiments, the inversion device constructs a nonlinear least squares objective function with the target residual vector as the core. This function consists of two parts: first, a data fitting term, which measures the degree of agreement between the simulation response of the finite element model and the measured target residual vector, calculated by weighted Euclidean distance; and second, a Tikhonov regularization term, which is used to prevent ill-conditioned solutions (such as negative volume, distorted shape, and other non-physical results) from appearing in the inversion results. By introducing a regularization coefficient, the degree of data fitting and model smoothness are balanced. Prior geometric information (set as zero vector when there is no prior information) can also be introduced to optimize the objective function.

[0141] Step S40: Perform convergence determination on the iteration process. If the convergence condition is met, output the numerical value of the disease geometric parameter vector obtained in the current iteration as the disease geometric optimal parameter vector, and map the disease geometric optimal parameter vector back to three-dimensional space to realize the geometric parameter inversion of hidden road diseases.

[0142] It should be noted that the optimal geometric parameter vector of the disease refers to the geometric parameter vector of the disease obtained after iterative convergence. This vector enables the simulation response of the finite element model to achieve the best match with the measured data and can most closely approximate the geometric shape of the real disease.

[0143] Understandably, geometric parameter inversion refers to the process of reversely deriving the geometric parameters (location, shape, volume, etc.) of hidden road defects through measured data, finite element simulation, and iterative optimization, thereby realizing the digital reconstruction of hidden defects and transforming invisible defects into quantifiable and visualized parameters and models.

[0144] It should be understood that this embodiment ensures the reliability and efficiency of the inversion results through multi-dimensional convergence judgment criteria, which avoids the waste of resources caused by invalid iterations and prevents the result deviation caused by insufficient iterations. The analysis and three-dimensional mapping of the optimal parameter vector transforms the abstract parameters into intuitive three-dimensional models and core indicators, realizing the digital and visual reconstruction of hidden diseases and providing maintenance personnel with accurate data on the location, shape and volume of diseases.

[0145] Furthermore, in order to accurately determine whether convergence has occurred, thereby avoiding resource waste caused by invalid iterations and preventing result deviations due to insufficient iterations, in one embodiment, after determining the convergence of the iteration process, the following may be included:

[0146] Calculate the root mean square error between the current simulated response vector and the measured mechanical response vector;

[0147] Calculate the Euclidean norm of the parameter correction increment for the current iteration;

[0148] Record the current iteration number;

[0149] If the convergence condition is met, the numerical values ​​of the disease geometric parameter vector obtained in the current iteration are output as the disease geometric optimal parameter vector. The convergence condition includes at least one of the following:

[0150] The relative root mean square error is lower than a preset accuracy threshold;

[0151] The Euclidean norm is lower than a preset step size threshold;

[0152] If the current iteration count reaches the preset iteration upper limit threshold and the convergence condition is not met, the circuit breaker mechanism is triggered, and the value of the disease geometric parameter vector with the smallest residual in the historical iteration process is output as the disease geometric optimal parameter vector.

[0153] The optimal geometric parameter vector of the road defects is mapped back to three-dimensional space to realize the geometric parameter inversion of hidden road defects.

[0154] It should be noted that the circuit breaker mechanism is an emergency handling mechanism triggered when the number of iterations reaches a preset upper limit and still fails to converge. It selects the parameter vector with the smallest residual in the historical iterations as the approximate optimal solution, thereby avoiding infinite iteration and ensuring the integrity of the inversion process.

[0155] The minimum historical iteration residual means that the magnitude of the target residual vector is the smallest in each iteration, corresponding to the smallest deviation between the simulation response and the measured data. The parameter vector of this iteration is closest to the actual geometric parameters of the disease, which is the optimal choice under the circuit breaker mechanism.

[0156] Understandably, this embodiment balances inversion accuracy (RRMSE threshold) and computational efficiency (norm threshold) through multi-condition convergence determination, solving the problem of insufficient or excessive iteration caused by a single convergence condition; the circuit breaker mechanism effectively avoids inversion failure due to model complexity and slow convergence, ensuring the integrity and stability of the inversion process; the validity verification of the optimal parameter vector further ensures the physical rationality of the output parameters, avoiding the influence of non-physical parameters on subsequent inversion results.

[0157] In the specific implementation, the determination of whether the iteration has ended involves setting termination criteria, including:

[0158] 1. Determine whether the relative root mean square error (RRMSE) between the current simulation response and the measured data is lower than the preset accuracy threshold. This criterion guarantees the reliability of the inversion results in the observation space.

[0159]

[0160] in, The Euclidean norm of a vector;

[0161] 2. Determine if the parameter adjustment step size in step k has become small, i.e., the model state has entered the stable region. This prevents unnecessary computational resource consumption in extremely flat gradient regions.

[0162]

[0163] 3. If the number of iterations k reaches the preset upper limit If convergence is still not achieved, the system triggers the circuit breaker mechanism and automatically outputs the set of parameters with the smallest residuals from all iterations. As an approximate solution.

[0164] Once the iteration satisfies the convergence condition, the parameter vector at the final time step is locked. The vector needs to be analyzed using engineering semantics to extract scalar indicators describing the characteristics of the disease.

[0165] Disease location coordinate calculation and extraction To determine the precise three-dimensional location of the centroid of the defect in the road coordinate system, including the mileage marker, lateral offset, and absolute burial depth;

[0166] Based on the finally converged ellipsoidal semi-axis parameters (a,b,c), the theoretical void volume of the defect is calculated. This serves as the basis for calculating the amount of grouting reinforcement material used:

[0167]

[0168] in This is a correction factor for irregularity coefficients;

[0169] The optimal parameter vector Mapping back to three-dimensional space, the damage scalar field inside the road is reconstructed.

[0170] Using computer graphics algorithms, generate in 3D visualization software The zero-level set isosurface is used to generate a 3D rendering map that includes road structure texture and perspective effects of road defects. The stress concentration coefficient around the defect area is displayed in the map using color gradients, visually showing potential areas at risk of collapse.

[0171] This embodiment improves the comprehensiveness and reliability of data acquisition through a multi-source heterogeneous sensing network, establishes a bridge between defects and mechanical responses through a parameterized finite element model, ensures the rationality and accuracy of the inversion results through regularized iterative optimization, and realizes the intuitive presentation of defect information through 3D visualization. It effectively solves the core pain point of not being able to accurately obtain the geometric parameters of defects, realizes the accurate inversion and digital reconstruction of the geometric parameters of hidden defects, effectively improves the detection accuracy and efficiency of hidden road defects, provides accurate decision-making basis for road maintenance, avoids the problems of over-maintenance or under-maintenance, reduces road maintenance costs, extends the service life of roads, and promotes the intelligent and precise development of smart highway maintenance.

[0172] refer to Figure 3 , Figure 3 This is a flowchart illustrating the second embodiment of the method for inverting geometric parameters of hidden road defects according to the present invention.

[0173] Based on the first embodiment described above, in this embodiment, step S30 further includes:

[0174] Step S301: Perform forward simulation calculations based on the three-dimensional finite element model of road damage to obtain the simulation response vector.

[0175] It should be noted that the simulation response vector refers to the vector formed by extracting the peak mechanical response (strain, stress) of virtual monitoring points through forward simulation calculation and arranging them in a fixed order. It is used to compare with the measured response vector and the target residual vector to measure the degree of agreement between the model and the actual road condition.

[0176] In practical implementation, based on the three-dimensional finite element model of road damage, the current geometric parameter vector is calculated. The pavement mechanical response under (kth iteration) control.

[0177] At the center of the lane on the model surface, apply a vehicle test load consistent with the actual test conditions in step S101, such as BZZ-100, single wheel load 50kN, ground pressure 0.7MPa.

[0178] Call the finite element kernel to perform static or dynamic implicit analysis and calculate the full-field displacement and stress data.

[0179] Virtual monitoring point set in the model Data is extracted and assembled into a simulation response vector:

[0180]

[0181]

[0182] in The peak strain or stress calculated for the i-th virtual sensor.

[0183] Step S302: Construct a sensitivity matrix based on the dimension of the disease geometric parameter vector and the dimension of the sensor data collected by the multi-source heterogeneous sensing network.

[0184] Understandably, to guide the automatic correction of geometric parameters, it is necessary to calculate the rate of change of sensor readings with respect to geometric parameters. Since the finite element model is a complex nonlinear black-box function and cannot be solved analytically, this invention uses the finite difference perturbation method to numerically approximate the sensitivity matrix.

[0185] Let the dimension of the disease geometric parameter vector P be M (e.g., M=7), and the dimension of the sensor data be N. The sensitivity matrix J is an N×M matrix:

[0186]

[0187] Step S303: The finite difference perturbation method is used to individually perturb each parameter in the simulation response vector to obtain the perturbed simulation response vector.

[0188] It should be noted that the perturbed simulation response vector refers to the peak mechanical response vector obtained by forward simulation calculation after a certain defect geometric parameter is perturbed. The difference between the peak vector and the original simulation response vector reflects the degree of influence of the perturbed parameter on the mechanical response.

[0189] Understandably, this embodiment solves the problem of directly calculating sensitivity in complex nonlinear finite element models by employing the finite difference perturbation method, simplifying the calculation process, reducing computational complexity, and ensuring the accuracy of sensitivity calculation. The operation of individual perturbation effectively isolates the influence of various disease parameters, avoids the distortion of sensitivity calculation caused by multiple parameter superposition perturbations, and ensures that the sensitivity of each parameter can be accurately captured. By fixing the perturbation amplitude and direction, the consistency and comparability of sensitivity calculations for different parameters are ensured.

[0190] Step S304: Calculate each matrix element in the sensitivity matrix based on the forward difference formula and the simulated response vector after the disturbance, and obtain the solved sensitivity matrix.

[0191] In practical implementation, the inversion device for parameter vectors Each element in (j=1…M), perform individual perturbations in sequence:

[0192] (1) Add a small increment to the j-th parameter. (For example, take 1% of the current value or a fixed value):

[0193]

[0194] (2) According to Update the material property distribution in the finite element model and re-divide the diseased elements and healthy road elements.

[0195] (3) Run the finite element method to obtain the response vector after disturbance. ;

[0196] (4) Calculate the j-th column element of the matrix using the forward difference formula:

[0197]

[0198] in, This represents the change in the reading of the i-th sensor when the j-th geometric parameter (such as the cavity radius a) changes slightly.

[0199] Step S305: Iteratively correct the values ​​of the disease geometric parameter vector based on the sensitivity matrix and the geometric-mechanical residual objective function.

[0200] In some embodiments, the inversion device may employ the Levenberg-Marquardt (LM) algorithm for iterative correction, specifically including: First, substituting the solved sensitivity matrix and geometric-mechanical residual objective function into the LM algorithm to solve for the correction increment ΔP of the defect geometric parameter vector. The solution for the correction increment needs to be combined with a regularization coefficient to balance the iteration convergence speed and the stability of the results; Second, calculating a new trial parameter vector P = P + ΔP (P is the parameter vector of the current iteration, and P is the trial parameter vector); Then, performing a physical constraint check on the trial parameter vector to ensure the ellipsoidal semi-axis Lengths a, b, and c are all greater than 0 (to avoid negative volumes). The z-coordinate of the lesion center is less than the road surface elevation but greater than the roadbed elevation (to ensure the lesion is inside the road structure). The rotation angle θ is within the range of 0 to π. If it exceeds the constraints, the parameters are truncated to reasonable boundary values. Finally, the corrected trial parameter vector is substituted into the three-dimensional finite element model of road damage, and the simulation response vector, sensitivity matrix, and objective function value are recalculated. The current objective function value is compared with the objective function value of the previous iteration. If the difference does not meet the convergence condition, the iteration continues until the convergence condition is met, thus completing the iterative correction of the lesion geometric parameter vector.

[0201] Furthermore, to improve the accuracy of disease inversion, the geometric-mechanical residual objective function is used to optimize the weighted Euclidean distance between the simulated response vector and the measured mechanical response vector. The mathematical expression of the geometric-mechanical residual objective function is as follows:

[0202]

[0203] in, The first term is the data fitting term, and the second term is the Tikhonov regularization term, which prevents ill-conditioned solutions from appearing in the inversion results, such as calculating non-physical results with negative volume or extremely distorted shape. Let represent the square of the Euclidean norm of a vector. This represents the regularization coefficient, used to balance the degree of data fit with the smoothness of the model; Represents the regularization matrix; The simulated response vector represents the current value of the geometric parameters of the defect; Represents the measured mechanical response vector; Represents the vector of geometric parameters of the disease; This represents a vector of prior information about the geometric parameters of the disease. If there is no prior information, it can be set to a zero vector. This represents a weighted diagonal matrix used to configure sensor weights in a multi-source heterogeneous sensing network. Sensors located in regions with anomalous signal peaks are assigned larger weights. This is to enhance the ability to locate the center of the disease.

[0204] Furthermore, in order to balance the fast convergence and global stability of iterative correction, and to avoid the problems of slow convergence and easy getting trapped in local optima by a single algorithm, step S305 above may include:

[0205] Step S3051: Using the Levenberg-Marquardt optimization algorithm, construct a system of linear equations based on the sensitivity matrix;

[0206] Step S3052: Solve the system of linear equations to obtain the parameter correction increment of the disease geometric parameter vector;

[0207] Step S3053: Add the current value of the disease geometric parameter vector to the parameter correction increment to obtain the trial parameter vector;

[0208] Step S3054: Perform projection constraint checks and corrections on the trial parameter vector to obtain the updated disease geometric parameter vector, thus completing one iteration correction of the value of the disease geometric parameter vector.

[0209] In the specific implementation, the Levenberg-Marquardt (LM) algorithm is used to solve the above nonlinear least squares problem. The LM algorithm combines the fast convergence of the Gauss-Newton method with the global stability of the gradient descent method, making it suitable for such highly nonlinear inversion problems.

[0210] In the k-th iteration, the system of linear equations is solved to obtain the parameter correction increment. :

[0211]

[0212] in, This represents the sensitivity matrix at the k-th iteration. Indicates transpose. This represents the adaptive damping factor of the optimization algorithm at k iterations. Represents the identity matrix. This represents the parameter adjustment increment at iteration k. This represents the vector of geometric parameters of the disease at k iterations;

[0213] Obtain the correction increment Then, calculate the new trial parameter vector:

[0214]

[0215] To ensure the physical meaning of the solution, it is necessary to... Perform projection constraint checks:

[0216] The lengths of the semi-axis must be positive (a, b, c > 0);

[0217] The top of the defect must be below the road surface. ;

[0218] The extent of the damage must not exceed the roadbed boundary. If it does, the parameters will be forcibly truncated to the boundary value to obtain an effective result. .

[0219] This embodiment obtains a benchmark through forward simulation, constructs a sensitivity matrix with matching dimensions, accurately captures the influence of parameters through individual perturbations, solves the sensitivity through forward difference, iteratively corrects and constrains the parameters, improves the accuracy, efficiency and rationality of parameter iterative correction, gradually reduces the objective function value, and makes the parameter vector continuously approach the actual geometric parameters of the disease, providing a core guarantee for finally obtaining the optimal parameter vector and significantly improving the accuracy of the inversion results.

[0220] Furthermore, this embodiment of the invention also proposes a computer-readable storage medium storing a road hidden defect geometric parameter inversion program. When the road hidden defect geometric parameter inversion program is executed by a processor, it implements the steps of the road hidden defect geometric parameter inversion method as described above.

[0221] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0222] The aforementioned computer-readable storage medium may be included in the road hidden defect geometric parameter inversion device; or it may exist independently and not be assembled into the road hidden defect geometric parameter inversion device.

[0223] Furthermore, this invention also proposes a computer program product, including a road hidden defect geometric parameter inversion program, which, when executed by a processor, implements the steps of the road hidden defect geometric parameter inversion method as described above.

[0224] The specific implementation of the computer program product of the present invention is basically the same as the embodiments of the above-mentioned method for inverting geometric parameters of hidden road defects, and will not be repeated here.

[0225] Reference Figure 4 , Figure 4 This is a structural block diagram of the first embodiment of the road hidden defects geometric parameter inversion device of the present invention.

[0226] like Figure 4 As shown, the road hidden defects geometric parameter inversion device proposed in this embodiment of the invention includes:

[0227] Data acquisition module 10 is used to construct a multi-source heterogeneous sensing network for road structure and acquire road mechanical response data, and process it to obtain the target residual vector for inversion. The multi-source heterogeneous sensing network includes strain sensors and stress sensors.

[0228] Damage modeling module 20 is used to construct a three-dimensional continuous medium finite element model based on the pavement design information of the road structure, define the disease geometric parameter vector in the three-dimensional continuous medium finite element model and assign initial values ​​to the disease geometric parameter vector to obtain a parameterized driven three-dimensional finite element model of road damage.

[0229] The iterative correction module 30 is used to construct a regularized geometric-mechanical residual objective function based on the target residual vector, and iteratively correct the values ​​of the disease geometric parameter vector based on the geometric-mechanical residual objective function;

[0230] The parameter inversion output module 40 is used to determine the convergence of the iteration process. If the convergence condition is met, the numerical value of the disease geometric parameter vector obtained by the current iteration is output as the disease geometric optimal parameter vector. The disease geometric optimal parameter vector is mapped back to three-dimensional space to realize the geometric parameter inversion of hidden road diseases.

[0231] This embodiment improves the comprehensiveness and reliability of data acquisition through a multi-source heterogeneous sensing network, establishes a bridge between defects and mechanical responses through a parameterized finite element model, ensures the rationality and accuracy of the inversion results through regularized iterative optimization, and realizes the intuitive presentation of defect information through 3D visualization. It effectively solves the core pain point of not being able to accurately obtain the geometric parameters of defects, realizes the accurate inversion and digital reconstruction of the geometric parameters of hidden defects, effectively improves the detection accuracy and efficiency of hidden road defects, provides accurate decision-making basis for road maintenance, avoids the problems of over-maintenance or under-maintenance, reduces road maintenance costs, extends the service life of roads, and promotes the intelligent and precise development of smart highway maintenance.

[0232] The road hidden defect geometric parameter inversion device provided in this application, employing the road hidden defect geometric parameter inversion method described in the above embodiments, can solve the technical problem of road hidden defect geometric parameter inversion. Compared with the prior art, the beneficial effects of the road hidden defect geometric parameter inversion device provided in this application are the same as those of the road hidden defect geometric parameter inversion method provided in the above embodiments, and other technical features in the road hidden defect geometric parameter inversion device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0233] It should be understood that the above are merely illustrative examples and do not constitute any limitation on the technical solutions of the present invention. In specific applications, those skilled in the art can make settings as needed, and the present invention does not impose any restrictions on this.

[0234] It should be noted that the workflow described above is merely illustrative and does not limit the scope of protection of this invention. In practical applications, those skilled in the art can select some or all of the workflow to achieve the purpose of this embodiment according to actual needs, and no restrictions are imposed here.

[0235] In addition, for technical details not described in detail in this embodiment, please refer to the method for inverting geometric parameters of hidden road defects provided in any embodiment of the present invention, which will not be repeated here.

[0236] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0237] It should be noted that the user information (including but not limited to user device information, user personal information, user location information, user behavior information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0238] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0239] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as read-only memory / random access memory, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0240] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A method for inverting geometric parameters of hidden road defects, characterized in that, The method includes: A multi-source heterogeneous sensing network for road structure is constructed and road mechanical response data is collected. The data is then processed to obtain the target residual vector for inversion. The multi-source heterogeneous sensing network includes strain sensors and stress sensors. A three-dimensional continuous medium finite element model is constructed based on the pavement design information of the road structure. The geometric parameter vector of the disease in the three-dimensional continuous medium finite element model is defined and the initial value is assigned to the geometric parameter vector of the disease, so as to obtain a parameterized driven three-dimensional finite element model of road damage. Based on the target residual vector, a regularized geometric-mechanical residual objective function is constructed, and the values ​​of the disease geometric parameter vector are iteratively corrected based on the geometric-mechanical residual objective function. The iterative process is converged. If the convergence condition is met, the numerical value of the disease geometric parameter vector obtained in the current iteration is output as the disease geometric optimal parameter vector. The disease geometric optimal parameter vector is mapped back to three-dimensional space to realize the geometric parameter inversion of hidden road diseases. The numerical values ​​of the disease geometric parameter vector iteratively corrected based on the geometric-mechanical residual objective function include: A forward simulation calculation is performed based on the aforementioned three-dimensional finite element model of road damage to obtain the simulation response vector; A sensitivity matrix is ​​constructed based on the dimension of the disease geometric parameter vector and the dimension of the sensor data collected by the multi-source heterogeneous sensing network. The finite difference perturbation method is used to individually perturb each parameter in the simulation response vector to obtain the perturbed simulation response vector; Based on the forward difference formula and the simulated response vector after the disturbance, the matrix elements in the sensitivity matrix are calculated to obtain the solved sensitivity matrix. The numerical values ​​of the disease geometric parameter vector are iteratively corrected based on the sensitivity matrix and the geometric-mechanical residual objective function; The geometric-mechanical residual objective function is used to optimize the weighted Euclidean distance between the simulated response vector and the measured mechanical response vector. The mathematical expression of the geometric-mechanical residual objective function is as follows: in, This represents the value of the objective function after regularization. Let represent the square of the Euclidean norm of a vector. Represents the regularization coefficient. Represents the regularization matrix. This represents the simulated response vector under the current values ​​of the disease's geometric parameter vectors. This represents the measured mechanical response vector. Represents the vector of geometric parameters of the disease. A vector representing prior information about the geometric parameters of the disease. This represents the weight diagonal matrix, used to configure sensor weights in a multi-source heterogeneous sensing network.

2. The method for inverting geometric parameters of hidden road defects as described in claim 1, characterized in that, The construction of a multi-source heterogeneous sensing network for the road structure and the acquisition of road mechanical response data, followed by processing to obtain the target residual vector for inversion, includes: A grid-like sensor array consisting of strain sensors and stress sensors is deployed in the mechanically sensitive layers of the road structure to form a multi-source heterogeneous sensing network. Under the vehicle test load, the original mechanical response time history signal of the sensor array is collected, and the peak response data of each sensor is extracted when the center of the vehicle test load passes directly above it. Obtain the baseline mechanical response vector under road health conditions and the measured mechanical response vector within the current monitoring period; Construct a weighted diagonal matrix, perform difference processing on the measured mechanical response vector and the reference mechanical response vector, and then perform weighted calculation using the weighted diagonal matrix to obtain the target residual vector for inversion, as shown in the following formula: in, Represents the target residual vector. This represents the measured mechanical response vector. Represents the reference mechanical response vector. This represents the weighting coefficient of the k-th sensor. This represents the weight diagonal matrix. Represents a diagonal matrix. This indicates the number of sensors in a multi-source heterogeneous sensing network. This represents the maximum value of all elements in the reference mechanical response vector. This represents the signal-to-noise ratio coefficient of the k-th sensor.

3. The method for inverting geometric parameters of hidden road defects as described in claim 2, characterized in that, The method involves constructing a three-dimensional continuous medium finite element model based on pavement design information of the road structure, defining the defect geometric parameter vector in the three-dimensional continuous medium finite element model and assigning initial values ​​to the defect geometric parameter vector, thereby obtaining a parameterized driven three-dimensional finite element model of road damage, including: A three-dimensional continuous medium finite element model with a preset size is established based on the pavement design information of the road structure. Symmetrical boundaries are set for the three-dimensional continuous medium finite element model, and road structure layers are divided according to the depth direction. Constitutive parameters of standard pavement material are assigned to each structure layer. The three-dimensional spatial coordinates of the sensor array in the multi-source heterogeneous sensing network are mapped to the mesh of the three-dimensional continuous medium finite element model, and the corresponding set of virtual monitoring points is marked. Using the material weakening method, hidden road defects are equivalent to spatial ellipsoids, and the defect geometric parameter vector of the spatial ellipsoid in the three-dimensional continuous medium finite element model is defined. Establish a discrimination function for the diseased unit, and set a material property assignment strategy for the three-dimensional continuous medium finite element model based on the discrimination function; Based on the material property assignment strategy and the virtual monitoring point set traversing each element point in the three-dimensional continuous medium finite element model, initial values ​​are assigned to the disease geometric parameter vector to obtain a parameterized driven three-dimensional finite element model of road damage.

4. The method for inverting geometric parameters of hidden road defects as described in claim 3, characterized in that, The numerical values ​​of the disease geometric parameter vector iteratively corrected based on the sensitivity matrix and the geometric-mechanical residual objective function include: Using the Levenberg-Marquardt optimization algorithm, a system of linear equations is constructed based on the sensitivity matrix. The mathematical expression of the system of linear equations is as follows: in, This represents the sensitivity matrix at the k-th iteration. Indicates transpose. This represents the adaptive damping factor of the optimization algorithm in the k-th iteration. Represents the identity matrix. This represents the parameter adjustment increment at the k-th iteration. This represents the vector of geometric parameters of the disease at the k-th iteration. Solving the system of linear equations yields the parameter correction increments of the disease geometric parameter vector; The current value of the disease geometric parameter vector is added to the parameter correction increment to obtain the trial parameter vector; The proposed parameter vector is subjected to projection constraint checks and corrections to obtain an updated disease geometric parameter vector, thus completing one iteration of the numerical correction of the disease geometric parameter vector.

5. The method for inverting geometric parameters of hidden road defects as described in any one of claims 1 to 4, characterized in that, After determining the convergence of the iterative process, the following may be included: Calculate the root mean square error between the current simulated response vector and the measured mechanical response vector; Calculate the Euclidean norm of the parameter correction increment for the current iteration; Record the current iteration number; If the convergence condition is met, the numerical values ​​of the disease geometric parameter vector obtained in the current iteration are output as the disease geometric optimal parameter vector. The convergence condition includes at least one of the following: The relative root mean square error is lower than a preset accuracy threshold; The Euclidean norm is lower than a preset step size threshold; If the current iteration count reaches the preset iteration upper limit threshold and the convergence condition is not met, the circuit breaker mechanism is triggered, and the value of the disease geometric parameter vector with the smallest residual in the historical iteration process is output as the disease geometric optimal parameter vector. The optimal geometric parameter vector of the road defects is mapped back to three-dimensional space to realize the geometric parameter inversion of hidden road defects.

6. A road hidden defect geometric parameter inversion device applying the road hidden defect geometric parameter inversion method according to any one of claims 1 to 5, characterized in that, The device includes: The data acquisition module is used to construct a multi-source heterogeneous sensing network for the road structure and collect road mechanical response data, and process it to obtain the target residual vector for inversion. The multi-source heterogeneous sensing network includes strain sensors and stress sensors. The damage modeling module is used to construct a three-dimensional continuous medium finite element model based on the pavement design information of the road structure, define the disease geometric parameter vector in the three-dimensional continuous medium finite element model and assign initial values ​​to the disease geometric parameter vector to obtain a parameterized driven three-dimensional finite element model of road damage. The iterative correction module is used to construct a regularized geometric-mechanical residual objective function based on the target residual vector, and to iteratively correct the values ​​of the disease geometric parameter vector based on the geometric-mechanical residual objective function. The parameter inversion output module is used to determine the convergence of the iteration process. If the convergence condition is met, the numerical value of the disease geometric parameter vector obtained by the current iteration is output as the disease geometric optimal parameter vector. The disease geometric optimal parameter vector is then mapped back to three-dimensional space to realize the geometric parameter inversion of hidden road diseases.

7. A device for inverting geometric parameters of hidden road defects, characterized in that, The road hidden defects geometric parameter inversion device includes: a memory, a processor, and a road hidden defects geometric parameter inversion program stored in the memory. The processor is used to run the road hidden defects geometric parameter inversion program, which is configured to implement the road hidden defects geometric parameter inversion method as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a road hidden defect geometric parameter inversion program, which, when executed by a processor, implements the road hidden defect geometric parameter inversion method as described in any one of claims 1 to 5.

Images