Complex pavement mechanical response monitoring device and method

Abstract

The invention relates to the technical field of mechanical response monitoring, in particular to a complex pavement mechanical response monitoring device and method, and the method comprises the steps: collecting sensor data of each section of pavement of a road; the method comprises the following steps: constructing a graph structure by utilizing the correlation among mechanics, heat and humidity among sensors, and mapping the graph structure into a graph frequency domain feature vector by utilizing graph Fourier transform to form a hypergraph space; constructing an objective function based on the spatial proximity and the time continuity of the graph frequency domain feature vector in the hypergraph space so as to perform global optimization on the graph frequency domain feature vector by using a particle swarm optimization algorithm; and inversely transforming the optimized graph frequency domain feature vector into a data space, generating a sensor signal coupling curve, and identifying road surface damage according to abnormal change of the curve, thereby realizing real-time monitoring of the road surface health state. The invention aims to improve the accuracy and real-time performance of road surface health state monitoring.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of mechanical response monitoring, in particular to a complex pavement mechanical response monitoring device and method. BACKGROUND

[0002] In modern highway construction, pavement mechanical response devices monitor the pavement state under the action of complex traffic load and environmental factors in real time by burying various sensors. These sensors can capture the mechanical response of different layers, such as the tensile deformation of the asphalt layer, the fatigue stress of the base layer, and the vertical pressure of the soil base, and analyze the changes in material properties in combination with temperature and humidity data. When the monitoring data is abnormal, the system can evaluate the health status of the pavement through multi-source data fusion technology, thereby providing a basis for maintenance decisions.

[0003] However, with the increase in the number of sensors and the expansion of the data size, precise monitoring faces significant challenges. On the one hand, the data of different sensors (such as strain gauges and pressure gauges) have complex correlations in space and time. On the other hand, the massive time series data accumulated through long-term monitoring makes it difficult for traditional analysis methods to effectively mine the implicit rules between the data. SUMMARY

[0004] To solve the above technical problems, the present application provides a complex pavement mechanical response monitoring device and method, and the technical solutions adopted are as follows: In a first aspect, the embodiments of the present application provide a complex pavement mechanical response monitoring method, which comprises the following steps: Multiple source sensors are buried at different layers of the highway, and mechanical response values, temperature values, and humidity values are collected in real time at each pavement section; A graph structure is constructed with sensors as nodes and the data coupling correlation between mechanical-thermal-humidity of the nodes as edge weights, and a graph Fourier transform is used to map the graph structure into a graph frequency domain feature vector. The graph frequency domain feature vectors of all road sections at all collection times form a hypergraph space; Based on the spatial proximity and temporal continuity of the graph frequency domain feature vectors in the hypergraph space, a target function is constructed containing data reconstruction error terms, spatial smoothing terms, and temporal smoothing terms, and a particle swarm optimization algorithm is used to globally optimize the graph frequency domain feature vectors; The optimized graph frequency domain feature vectors are inversely transformed to the data space to generate sensor signal coupling curves, and pavement damage is identified according to the abnormal changes of the curves, thereby realizing real-time monitoring of the pavement health status.

[0005] Preferably, the sensors include but are not limited to asphalt strain gauges, base strain gauges, pressure gauges, temperature sensors, and humidity sensors.

[0006] Preferably, the edge weight value between nodes in the graph structure is determined by the following formula: wherein represents the edge weight value between the first node and the second node of the first road section at the first time point, e is a natural constant, , represents the edge weight value between the first node and the second node of the first road section at the first time point, e is a natural constant, , represents the edge weight value between the first node and the second node of the first road section at the first time point, e is a natural constant, , represents the data collected by the sensor corresponding to the first node and the second node of the first road section at the first time point, , represents the data collected by the sensor corresponding to the first node and the second node of the first road section at the first time point, , represents the data collected by the sensor corresponding to the first node and the second node of the first road section at the first time point, represents that there is a correlation between the data of two nodes, if the data of two nodes is positively correlated, 0 is taken; if the data of two nodes is negatively correlated, 1 is taken; represents that there is no correlation between the data of two nodes.

[0007] Preferably, the data collected by the asphalt strain gauge, the base strain gauge and the pressure gauge is positively correlated, the data collected by the asphalt strain gauge and the temperature sensor is positively correlated, the data collected by the base strain gauge and the humidity sensor is positively correlated, the data collected by the temperature sensor and the humidity sensor is negatively correlated, and there is no correlation between the rest of the data.

[0008] Preferably, the construction method of the objective function is: using the difference between the graph frequency domain feature vectors and the distance between any two road sections to determine the spatial edge weight value of the graph frequency domain feature vectors of any two road sections at the same time point; using the difference between the graph frequency domain feature vectors to determine the time edge weight value of the graph frequency domain feature vectors at any two time points on the same road section; combining the spatial edge weight value and the time edge weight value to construct the objective function.

[0009] Preferably, the calculation method of the spatial edge weight value is: ; wherein represents the spatial edge weight value of the graph frequency domain feature vectors of the first road section and the second road section at the same time point , , represents the spatial edge weight value of the graph frequency domain feature vectors of the first road section and the second road section at the same time point , represents the graph frequency domain feature vectors of the first road section and the second road section at the same time point , represents the graph frequency domain feature vectors of the first road section and the second road section at the same time point represents the L2 norm between two vectors, represents the L2 norm between two vectors, Indicates in At this moment The road section and the first The Euclidean spatial distance between the center points of each road segment, where e is a natural constant. It is a preset spatial attenuation factor.

[0010] Preferably, the time edge weights are calculated as follows: ;in, Indicates in At this moment The frequency domain feature vector of each road segment It is a preset smoothing factor for the time edge weights.

[0011] Preferably, the objective function is calculated as follows: Where T represents all times collected, and N represents all road segments collected. It is the first At this moment The diagram structure of each road segment, The corresponding value is obtained from the eigenvalue decomposition of the Laplace matrix during the graphical Fourier transform. The graph Fourier basis matrix.

[0012] Preferably, the particle swarm optimization algorithm uses each graph frequency domain feature vector as an individual particle, the fitness function as the objective function, the particle velocity update factor as 2, and the number of iterations as 100.

[0013] Secondly, embodiments of this application also provide a complex road surface mechanical response monitoring device, the device comprising: The integrated housing houses an asphalt strain gauge, a base strain gauge, a pressure gauge, a temperature sensor, and a humidity sensor. The wireless communication module is used to collect sensor data at 30-second sampling intervals and upload it to a remote system. The data processing unit is used to execute the steps of the complex pavement mechanical response monitoring method described above.

[0014] This application has at least the following beneficial effects: The application can simultaneously obtain three key physical quantities of mechanics-heat-moisture in the internal structure layer of the pavement through one-time burying, eliminate the construction secondary excavation and sensor position deviation caused by layered arrangement, reduce the installation cost and improve the data space consistency. The application uses a graph structure to quantitatively couple the physical coupling relationship between sensors, which can improve the data correlation dimension and reduce the subsequent calculation complexity without adding additional sensors compared with the traditional matrix splicing method. The application uses a space-time joint constraint objective function, so that the particle swarm optimization can converge to a globally better solution within a limited number of iterations, significantly reducing the interference of abnormal values on the damage identification result and improving the robustness of the algorithm. The coupled curve after inverse transformation of the application restores the high-dimensional frequency domain features to the physical quantity curve which is easy to explain, which can improve the accuracy and real-time performance of the on-site staff to intuitively judge the damage location and degree. BRIEF DESCRIPTION OF DRAWINGS

[0015] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and those skilled in the art can obtain other drawings according to these drawings without creative labor.

[0016] Figure 1 The steps flow chart of a complex pavement mechanical response monitoring method of the present application. DETAILED DESCRIPTION

[0017] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the specific embodiments, structures, features and effects of a complex pavement mechanical response monitoring device and method according to the present application are described in detail as follows. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs.

[0019] The specific scheme of a complex pavement mechanical response monitoring device and method provided by the present application is specifically described below with reference to the drawings.

[0020] An embodiment of the present application provides a complex pavement mechanical response monitoring device and method.

[0021] Specifically, a complex pavement mechanical response monitoring method is provided as follows, please refer to Figure 1 The method comprises the following steps: Step one: bury multiple source sensors at different levels of the highway, and collect real-time mechanical response values, temperature values and humidity values at each section of the road.

[0022] The pavement mechanical response device is an integrated device with multiple sensors. According to the details of the highway project, corresponding sensors are installed at different depths of the device, including but not limited to asphalt strain gauges, base strain gauges, pressure gauges, temperature sensors and humidity sensors. Then the device is buried in the highway to ensure that the sensors of key layers such as asphalt layer, base layer and subbase layer can accurately measure the mechanical response sensor data of the corresponding layer.

[0023] Arranged longitudinally along the direction of the highway, a set of devices is buried at every predetermined interval, and each set of devices is used to capture various mechanical response sensor data of a section of the road. The interval of the device can be set by the implementer according to the implementation scene, and no special limitation is made here. In this embodiment, the interval between devices is 10 meters.

[0024] The sensor data in the device is wirelessly transmitted through the Internet of Things technology, and the data of the sensors in the device is sampled every 30 seconds and uploaded to the system for storage, which is used for subsequent long-term real-time monitoring of the pavement state. The sampling time can be set by the implementer according to the implementation scene, and no special limitation is made here. In this embodiment, the sampling interval is 30 seconds.

[0025] The sensor data is cleaned. For transient noise outliers, such as transient noise caused by road maintenance and construction interference, the outliers can be removed by interpolation method. The inserted value can be obtained by taking the median value of the continuous The size can be set by the implementer according to the implementation scene, and in this embodiment, The size is set to 3. The interpolation method is a prior art, and its specific process will not be described here.

[0026] At this point, the pavement mechanical response device is installed, and the sensor data is acquired and preprocessed.

[0027] Step two: construct a graph structure with sensors as nodes and data coupling correlation between mechanical-thermal-humidity as edge weight, map the graph structure to graph frequency domain feature vector using graph Fourier transform, and construct a hypergraph space with all graph frequency domain feature vectors of all road sections at all collection times.

[0028] With the expansion of the construction scale of highways and other roads, the number of pavement state monitoring sensors and the scale of data are also expanding, and the traditional pavement mechanical response device is facing challenges in accurate monitoring.

[0029] ​First, the data of different sensors (such as strain gauges and pressure gauges) on the same section of the road have complex correlations in space and time, for example, temperature changes will affect strain readings, and the dynamic distribution of traffic loads will affect the response mode of pressure sensors.

[0030] From the long-term monitoring of all sensor data, a large amount of time series data is accumulated, which makes it difficult for traditional analysis methods to effectively mine the implicit rules between the data, and it is difficult to analyze the current road state by comprehensively analyzing the time series changes of a large amount of sensor data, thereby making it difficult to achieve accurate long-term monitoring.

[0031] In view of the above problems, the present application models and maps the pavement mechanical response value data to the graph signal space by constructing a graph structure, analyzes the relationship of the graph frequency domain feature vector in the time-space dimension to construct a target function, and uses a particle swarm optimization algorithm to optimize the graph frequency domain feature vector representation of each data, thereby improving the accuracy of complex pavement data monitoring.

[0032] Specifically, taking a single mechanical response device as a unit, the sensor data is a node, the weight value between the connection lines of two nodes is determined based on the correlation between the data collected by the two nodes, the graph structure of the sensor data is constructed, the mechanical response value data is mapped to the graph signal space using graph Fourier transform, the graph frequency domain feature vector hypergraph space is constructed by analyzing the characteristic performance of various pavement data in time-space, then a target constraint function is constructed, and finally a particle swarm optimization algorithm is used to optimize the graph frequency domain feature vector representation of each data, thereby improving the accuracy of complex pavement data monitoring.

[0033] a) The graph structure is constructed with sensors as nodes and the data coupling correlation between mechanics-heat-moisture as edge weight, the graph structure is mapped to graph frequency domain feature vector using graph Fourier transform, and a hypergraph space containing time-space relationship is formed.

[0034] For a monitored highway, according to the interval of device embedding, it is assumed that devices are installed to detect the mechanical response data of each section of the road, and a total of time points of monitoring data of the entire highway are monitored. A graph structure is constructed with the mechanical response device as a unit, and finally data graph structures are obtained.

[0035] Regarding the construction of a single data graph structure, it is assumed that the mechanical response device contains kinds of sensors, so that types of sensor data can be obtained. The sensor data is taken as a node, and the weight value between the connection lines of two nodes is determined based on the correlation between the data collected by the two nodes, so that at the moment, the type of sensor data is obtained.The graph structure of the mechanical response data of the segment pavement can be expressed as wherein represents the first node of the first segment pavement at the first time point, represents the data collected by the sensor corresponding to the first node of the first segment pavement at the first time point, represents the edge weight between the first node and the second node of the first segment pavement at the first time point.

[0036] wherein the node represents a sensor data, and the edge weight is calculated based on the correlation between the data collected by two nodes, and the analysis and construction method is as follows: For example, the asphalt strain gauge and the base strain gauge measure the micro strain of the asphalt layer and the base layer caused by the change of the strain gauge under pressure and , and the pressure data measured by the pressure gauge are positively correlated. The asphalt strain is positively correlated with the temperature . High temperature leads to a decrease in the modulus of the asphalt, and the micro strain increases under the same pressure, and the two are positively correlated. The base strain is positively correlated with the humidity . High humidity, especially in the case of freezing and thawing, leads to a decrease in the modulus of the base layer, and the micro strain increases, and the two are positively correlated. The temperature is negatively correlated with the humidity . The pressure data and the temperature and humidity have no relationship, the asphalt strain has no relationship with the humidity, and the base strain has no relationship with the temperature.

[0037] Based on the above correlation analysis between different types of data, the edge weight between the nodes can be calculated by the following formula: wherein represents the edge weight between the first node and the second node of the first segment pavement at the first time point, e is a natural constant, , respectively represent the data collected by the sensor corresponding to the first node and the second node of the first segment pavement at the first time point, ​​​​​​​​​​​​​​​​This indicates a correlation between the data of two nodes. If the data of two nodes are positively correlated, Set to 0; if the data of two nodes are negatively correlated, Take 1; If there is no correlation between the data of two nodes, then the edge weight between the two nodes in the graph structure is 0.

[0038] By constructing a graph structure through analyzing the correlations between different types of sensor data, the relationships between different sensor data under a mechanical response device can be clearly modeled. Modeling monitoring data for the same road segment in the form of a graph structure not only records the numerical performance of sensor data but also quantifies the correlations between sensor data, thus reflecting the overall correlation of monitoring data for the same road segment. This provides the smallest data unit for subsequent analysis of the temporal-spatial relationships of monitoring data for the entire highway.

[0039] This completes the construction of the graphical structure for the pavement mechanical response data.

[0040] To facilitate subsequent analysis of the temporal-spatial relationships of the monitoring data, the graph structure is then transformed using a graph Fourier transform. Mapping to the graph signal space, we obtain information about Graph frequency domain feature vector ,in Indicates the first Each node is mapped to a feature vector in the graph signal space, and there are a total of [number] feature vectors. Each sensor node reflects the energy distribution of sensor data signals on the graph structure. Graph Fourier transform is a current technology, and its specific process will not be elaborated here.

[0041] The graph Fourier transform encodes data from the graph node structure to the frequency domain space, which can map the graph structure, which is difficult to compute, to a structured vector space, providing a foundation for the quantization of time-space relationships in the subsequent hypergraph space and the construction of objective constraint functions.

[0042] Mapping the graph structure of all time-space pavement mechanical response data to the graph signal space yields a... Hypergraph space of dimensionality Each coordinate in the hypergraph space represents a graph frequency domain feature vector representation of the pavement mechanical response data of different road segments at different times: in, This represents the frequency domain feature vector of the data map for the first road segment at time 1. Indicates the first time step The frequency domain feature vector of the data map for each road segment Indicates the first the data graph frequency domain feature vector of the first road section at the time point, the data graph frequency domain feature vector of the first road section at the time point, the data graph frequency domain feature vector of the first road section at the time point, the data graph frequency domain feature vector of the first road section at the time point.

[0043] Step three: based on the spatial proximity and time continuity of the graph frequency domain feature vectors in the hypergraph space, a target function containing data reconstruction error terms, spatial smoothing terms and time smoothing terms is constructed to globally optimize the graph frequency domain feature vectors using a particle swarm optimization algorithm.

[0044] In particular, each graph frequency domain feature vector in the hypergraph space is associated in a separate row or a separate column, i.e. in the spatial or temporal level, two graph frequency domain feature vectors at the same time or the same space have an edge weight relationship, which can be represented by and to represent the edge weight value of the graph frequency domain feature vector in the hypergraph space: First, in the spatial level, assuming that vehicles drive on the road surface, the adjacent two road sections can capture approximately the same mechanical response data in a short time, so their graph frequency domain feature vectors have similarity; while the mechanical response data on the road sections far apart have no correlation, and their graph frequency domain feature vectors have low similarity.

[0045] In addition, assuming that the section of the road surface has road damage (such as depression, pothole), which will cause the monitoring value of strain and pressure in the mechanical response data to be high, due to the influence of physical mechanics conduction and associated vibration, the mechanical response data at the section of the road surface adjacent to the road section will also fluctuate, the mechanical response data at the section of the road surface will also fluctuate less than , so this abnormal relationship will produce a wave-like relationship with spatial decay, with the wave peak at the center abnormal value.

[0046] According to the above analysis, the spatial edge weight value of the graph frequency domain feature vectors of the and the sections of the road at the same time can be represented as: wherein , represent the graph frequency domain feature vectors of the and the sections of the road at the time point , respectively, represents the calculation of the L2 norm between the two vectors, represents the calculation of the L2 norm between the two vectors, At this moment The road section and the first The Euclidean spatial distance between the center points of two road segments indicates that the correlation between their mechanical response data decreases as the distance between the two road segments increases. The smaller the value, the higher the correlation between the mechanical data of two adjacent road sections. Also bigger. This is the spatial decay term, where e is the natural constant. This is a preset spatial attenuation factor, set to 0.2 in this embodiment. When the distance between two road sections is large, the spatial attenuation term is smaller, further reducing... When two road sections are adjacent, the spatial attenuation term is slightly larger, and the spatial attenuation term is always less than 1. This setting of the spatial edge weights of the graph frequency domain feature vectors of the two road sections at the same time conforms to the spatial transmission properties of mechanics.

[0047] Next, from a temporal perspective, the mechanical response values ​​on the same road segment also exhibit similarities at two adjacent time points, and the... The moment of the first If the response data of the graph structure is abnormal, then in the first uninterrupted graph structure... The graph structure subsequently exhibited a high abnormal signal state for a continuous period of time, affecting the normal operation of the sensor. On the same road segment... Next The and the first The temporal side weights of the graph frequency domain feature vector at time t It can be represented as: Where e is the natural constant. Indicates in At this moment The frequency domain feature vector of each road segment Indicates in At this moment The frequency domain feature vector of each road segment It is a preset smoothing factor for the time edge weights, which is set to 0.5 in this embodiment.

[0048] Based on the above analysis of the time-space relationship of the graph frequency domain eigenvectors of mechanical response data in the hypergraph space, a concept regarding the hypergraph space can be obtained. The objective function, used as the objective function of the particle swarm optimization algorithm, optimizes the feature representation of pavement mechanical response data by minimizing the objective function, thereby improving the accuracy of pavement condition monitoring. The objective function includes the reconstruction error term, spatial smoothing term, and temporal smoothing term of the original mechanical response sensor data and the image frequency domain feature vector: Wherein, T is all the time of collection, N is all the road section of collection, is the first time of the first road section, is the graph Fourier basis matrix corresponding to the graph Fourier transform by the eigen decomposition of the Laplacian matrix, herein represents the inverse transform of the frequency domain signal to realize the reconstruction of the graph signal. represents the calculation of the L2 norm between two vectors, is the spatial edge weight of the graph frequency domain feature vector of the first and the first road section at the same time in the hypergraph space, is the spatial edge weight of the graph frequency domain feature vector of the first and the first time at the same road section in the hypergraph space, is the time edge weight of the graph frequency domain feature vector of the first and the first time at the same road section in the hypergraph space.

[0049] The design of the objective function guarantees the fidelity of the reconstruction of the mechanical response data through the reconstruction error term, prevents distortion caused by excessive smoothing of the spatial and temporal smoothing terms, and ensures the composite physical and mechanical conduction characteristics of the mechanical response data on different road sections through the spatial edge weight constraint in the hypergraph space, and ensures the consistency of the change of the mechanical response data signal in time through the time edge weight constraint in the hypergraph space.

[0050] The particle individual of the particle swarm optimization algorithm is set to each graph frequency domain feature vector , the fitness function is the above objective function, the particle velocity update factor is set to 2, and the iteration number is 100. The particle swarm optimization algorithm is a prior art, and its specific process will not be described again.

[0051] Up to now, the optimization of the graph frequency domain feature vector in the hypergraph space is completed.

[0052] Step four: inverse transform the optimized graph frequency domain feature vector to the data space to generate the sensor signal coupling curve, and identify the pavement damage according to the abnormal change of the curve to realize the real-time monitoring of the pavement health state.

[0053] The graph frequency domain feature vector of the optimized pavement response data obtained in the above step is then inversely mapped to the data space by the graph Fourier basis matrix obtained by the graph Fourier transform and visualized, thereby improving the accuracy of long-term monitoring of the pavement state.

[0054] The method uses a multi-axis broken line chart to draw a time sequence coupling curve of road surface mechanical response data, and identifies road surface damage according to an abnormal value of the curve, analyzes the influence of pressure strain, temperature cycle, humidity penetration on highway interlayer damage, and realizes real-time monitoring of the road surface health state.

[0055] In addition, the road surface mechanical response device is regularly checked, the sensor accuracy is verified, and the reliability of long-term monitoring is improved.

[0056] Based on the same inventive concept as the above-mentioned complex road surface mechanical response monitoring method, an embodiment of the present application provides a complex road surface mechanical response monitoring device, which comprises: An integrated shell internally integrates an asphalt strain gauge, a base layer strain gauge, a pressure gauge, a temperature sensor and a humidity sensor; A wireless communication module is configured to collect sensor data at a sampling interval of 30 seconds and upload the sensor data to a remote system; A data processing unit is configured to execute the steps of the above-mentioned complex road surface mechanical response monitoring method.

[0057] Each embodiment in the present application is described in a progressive manner, and the same or similar parts between each embodiment can be referred to each other, and each embodiment mainly explains the difference from other embodiments.

[0058] It should be noted that, unless otherwise specified and limited, terms such as "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that the circuit structure, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such article or device. Without more limitation, the element limited by the statement "including a" does not exclude the presence of another identical element in the article or device including the element. In addition, the term "and / or" used herein includes any and all combinations of one or more related listed items.

[0059] Other embodiments of the present application will be readily apparent to those skilled in the art upon considering the description herein, with the present application intended to cover any variations, uses, or adaptations of the application following, in general, the principles of the application and including such steps and features as come within the purview of the present application.

[0060] It should be understood that the present application is not limited to the precise construction that has been described and shown in the accompanying drawings, and that various modifications and changes can be made without departing from the scope thereof.

Claims

1. A method for monitoring complex pavement mechanical response, characterized in that, The method comprises the following steps: Multiple sensors are buried at different layers of the highway, and the mechanical response value, temperature value and humidity value are collected in real time at each road section; A graph structure is constructed by taking the sensors as nodes and the data coupling correlation between the mechanical-thermal-humidity of the nodes as edge weights, the graph structure is mapped into a graph frequency domain feature vector by using graph Fourier transform, and the graph frequency domain feature vectors of all road sections at all collection times form a hypergraph space; Based on the spatial proximity and time continuity of the graph frequency domain feature vectors in the hypergraph space, a target function is constructed including a data reconstruction error term, a spatial smoothing term and a time smoothing term, and a particle swarm optimization algorithm is used to globally optimize the graph frequency domain feature vectors; The optimized graph frequency domain feature vectors are inversely transformed into the data space to generate a sensor signal coupling curve, and the curve is used to identify road damage and realize real-time monitoring of the road health status.

2. The method of claim 1, wherein the method comprises: The sensors include but are not limited to asphalt strain gauges, base strain gauges, pressure gauges, temperature sensors and humidity sensors.

3. The method of claim 2, wherein the method further comprises: The edge weight values between the nodes in the graph structure are determined by the following formula: in Indicates the first The first moment The first section of road surface The and the first The edge weights between nodes, where e is the natural constant. , They represent the first The first moment The first section of road surface The, the Each node corresponds to data collected by the sensor. This indicates a correlation between the data of two nodes. If the data of two nodes are positively correlated, Set to 0; if the data of two nodes are negatively correlated, Take 1; This indicates that there is no correlation between the data of the two nodes.

4. The method for monitoring the mechanical response of a complex pavement according to claim 3, characterized in that the asphalt The data collected between the strain gauges, base strain gauges and pressure gauges are positively correlated, the data collected between the asphalt strain gauges and the temperature sensors are positively correlated, the data collected between the base strain gauges and the humidity sensors are positively correlated, the data collected between the temperature sensors and the humidity sensors are negatively correlated, and there is no correlation between the remaining data.

5. The method of claim 1, wherein the method comprises: The construction method of the target function is: The spatial edge weight values of the graph frequency domain feature vectors of any two road sections at the same time are determined by the difference between the graph frequency domain feature vectors and the distance between any two road sections; The time edge weight values of the graph frequency domain feature vectors of any two times at the same road section are determined by the difference between the graph frequency domain feature vectors; The target function is constructed by combining the spatial edge weight values and the time edge weight values.

6. The method of claim 5, wherein the method comprises: The spatial edge weight is calculated in the following manner: ; wherein, represents the graph frequency domain feature vector of the i-th road segment at the same time point, , represents the L2 norm between two vectors, is a preset spatial decay factor.​​​​​​​​​​​ 7. The method of claim 6, wherein the method further comprises: The calculation method for the time edge weights is as follows: ;in, Indicates in At this moment The frequency domain feature vector of each road segment It is a preset smoothing factor for the time edge weights.

8. The method of claim 7, wherein the method comprises: The objective function is calculated as follows: Where T represents all times collected, and N represents all road segments collected. It is the first At this moment The diagram structure of each road segment, The corresponding value is obtained from the eigenvalue decomposition of the Laplace matrix during the graphical Fourier transform. The graph Fourier basis matrix.

9. The method of claim 8, wherein the method further comprises: The particle swarm optimization algorithm takes each graph frequency domain feature vector as a particle individual, the fitness function is the target function, the particle velocity update factor is 2, and the iteration number is 100.

10. A complex pavement mechanical response monitoring device, characterized in that, The device comprises: An integrated shell integrating asphalt strain gauges, base strain gauges, pressure gauges, temperature sensors and humidity sensors inside; A wireless communication module for collecting sensor data at a sampling interval of 30 seconds and uploading the data to a remote system; A data processing unit for executing the steps of the complex road mechanical response monitoring method according to any one of claims 1-9.

Images