Road underground cavity detection method and device based on falling weight deflectometer

Abstract

The invention provides a road underground cavity detection method and device based on a drop weight deflectometer, and relates to the technical field of underground engineering. The method comprises the following steps: acquiring a first deflection data set acquired by each displacement sensor when each stage of hammering load is applied; and under the same-level hammering load, a second deflection data set corresponding to each second displacement sensor is obtained according to the first deflection data set collected by the first displacement sensor and the first deflection data set collected by each second displacement sensor. Inputting the thickness of the road structure layer, the road type, a plurality of first deflection data sets collected by the first displacement sensor under the multi-stage hammering load and a plurality of second deflection data sets corresponding to each second displacement sensor under the multi-stage hammering load into a cavity prediction model, and obtaining a first judgment result output by the cavity prediction model. And determining whether there is a hole in the test point according to the first determination result, thereby improving the accuracy of road underground hole detection.

Description

Technical Field

[0001] The present application relates to the field of underground engineering technology, and in particular to a method and device for detecting underground cavities in roads based on a drop-weight deflectometer. Background Art

[0002] Underground voids in roads are one of the main causes of urban road collapse accidents. They are highly concealed and difficult to detect. The commonly used method for void detection is the ground penetrating radar method, which requires a high level of professionalism in the interpretation of the map and is greatly affected by the subjectivity of the inspectors. The Falling Weight Deflectometer (FWD) is a widely used road inspection tool. Its working principle is to evaluate the overall performance of the road structure by applying a hammer load and measuring the deflection response of the road surface. The FWD test results are simple and intuitive, and have been successfully applied to the detection of voids developed between the road structure layer and the foundation soil. However, the deflection value mainly reflects the mechanical properties of the surface structure of the road surface and the roadbed, and is less sensitive to changes in the deep structure. The traditional FWD test has obvious limitations in the application of underground void detection in roads.

[0003] Therefore, how to improve the accuracy and reliability of FWD method for detecting underground cavities in roads is a technical problem that needs to be solved urgently by those skilled in the art. Summary of the Invention

[0004] The purpose of this application is to provide a method and device for detecting underground cavities in roads based on a drop weight deflectometer, so as to improve the accuracy and reliability of underground cavity detection in roads.

[0005] In order to achieve the above objectives, the technical solutions adopted in this application are as follows:

[0006] On the one hand, the present application provides a method for detecting underground cavities in roads based on a drop-weight deflectometer. The drop-weight deflectometer includes: a hammer load generating device, a first displacement sensor, and a plurality of second displacement sensors; the first displacement sensor is located at the hammer center of the hammer load generating device, and the plurality of second displacement sensors are distributed around the hammer center, and each second displacement sensor has a different horizontal distance from the hammer center; after the hammer center of the hammer load generating device is aligned with a test point on a test road, the hammer load generating device applies a multi-level hammer load to the test point; the method includes:

[0007] Each time a hammer load is applied, a first deflection data group collected by each displacement sensor is obtained;

[0008] Under the same level of hammer load, obtaining a second deflection data group corresponding to each second displacement sensor based on the first deflection data group collected by the first displacement sensor and the first deflection data group collected by each second displacement sensor;

[0009] Obtaining the road structure layer thickness and road type of the test road;

[0010] Inputting the thickness of the road structure layer, the road type, a plurality of first deflection data sets collected by the first displacement sensor under multi-level hammer loads, and a plurality of second deflection data sets corresponding to each second displacement sensor under the multi-level hammer loads into a cavity prediction model, and obtaining a first discrimination result output by the cavity prediction model;

[0011] Determine whether the test point has a hole according to the first discrimination result.

[0012] Furthermore, the first deflection data set includes a deflection peak value D loc，load and time Dt loc，load ;

[0013] D loc，load It represents the peak vertical displacement of the road surface collected by the displacement sensor at a horizontal distance of loc from the impact center when the impact load is load;

[0014] Dt loc，load It indicates the time when the peak vertical displacement of the road surface detected by the displacement sensor at the horizontal distance loc from the impact center occurs when the impact load is load;

[0015] Wherein, 0≤loc≤max1, max1 is the maximum horizontal distance between the hammer center and the plurality of second displacement sensors.

[0016] Furthermore, under the same level of hammer load, the step of obtaining a second deflection data group corresponding to each second displacement sensor based on the first deflection data group collected by the first displacement sensor and the first deflection data group collected by each second displacement sensor includes:

[0017] The second deflection data set corresponding to the current second displacement sensor includes the relative deflection peak value ND li，load and the time Dt collected by the current second displacement sensor li，load ; Where li is the horizontal distance between the hammer center and the current second displacement sensor, and 0 <li≤max1;

[0018] Relative deflection peak ND li，load Satisfies the following formula:

[0019] ND li，load =D li，liad / D 0，load

[0020] Among them, D li，load is the deflection peak value collected by the second displacement sensor, D 0，load is the peak deflection value collected by the first displacement sensor.

[0021] Furthermore, the cavity prediction model includes a two-layer neural network and multiple bottom-level decision trees, and the number of the bottom-level decision trees is equal to the number of the second displacement sensors;

[0022] The steps of inputting the thickness of the road structure layer, the road type, a plurality of first deflection data sets collected by the first displacement sensor under multi-level hammer loads, and a plurality of second deflection data sets corresponding to each second displacement sensor under the multi-level hammer loads into a cavity prediction model, and obtaining a first discrimination result output by the cavity prediction model include:

[0023] Inputting the road structure layer thickness, road type, multiple first deflection data sets collected by the first displacement sensor under multi-level hammer loads, and multiple second deflection data sets corresponding to each second displacement sensor under the multi-level hammer loads into a corresponding bottom-level decision tree, and obtaining a second discrimination result output by each bottom-level decision tree;

[0024] The second discrimination results output by all the bottom-level decision trees are input into the double-layer neural network, and the first discrimination results output by the double-layer neural network are obtained.

[0025] Furthermore, the cavity prediction model includes a two-layer neural network and multiple bottom-level decision trees, and the number of the bottom-level decision trees is equal to the number of the second displacement sensors;

[0026] Before the step of obtaining a first deflection data set collected by each displacement sensor when applying each level of hammer load, the method further includes:

[0027] Acquire multiple reference road sample sets; each reference road sample set includes a road structure layer thickness of the reference road, a road type, multiple first deflection data sets collected by a first displacement sensor under multiple hammer loads, and multiple second deflection data sets corresponding to a second displacement sensor under the multiple hammer loads;

[0028] Performing one-to-one training on multiple underlying decision trees based on multiple reference road sample sets to obtain multiple trained underlying decision trees;

[0029] The double-layer neural network is trained according to the multiple second discrimination results output by the multiple trained bottom-level decision trees to obtain a trained double-layer neural network.

[0030] Furthermore, the step of determining whether a hole exists at the test point according to the first discrimination result includes:

[0031] If the first discrimination result is 0, it is determined that there is no hole at the test point;

[0032] If the first discrimination result is 1, it is determined that there is a hole at the test point.

[0033] Furthermore, the hammer load generating device applies four levels of hammer loads to the test point, and the four levels of hammer loads are 25KN, 50KN, 75KN and 100KN respectively.

[0034] Furthermore, after the step of obtaining a first deflection data set collected by each displacement sensor when applying each level of hammer load, the method further includes:

[0035] Establishing a first regression equation based on multiple deflection peak values ​​collected by the first displacement sensor under multiple levels of hammer loads; wherein the first regression equation represents the deflection peak values ​​collected by the first displacement sensor under different levels of hammer loads;

[0036] A second regression equation is established based on multiple deflection peaks collected by the second displacement sensor at a horizontal distance of max1 from the hammer center under multiple levels of hammer loads. The second regression equation represents the deflection peaks collected by the second displacement sensor at a horizontal distance of max1 from the hammer center under different levels of hammer loads.

[0037] Obtain the maximum unilateral load max2 of a single axle of a vehicle passing through the test road;

[0038] According to the first regression formula, when the hammer load is max2, the first deflection peak value D collected by the first displacement sensor is obtained. 0，max2 ;

[0039] According to the second regression formula, when the hammer load is max2, the second deflection peak value D collected by the second displacement sensor at the horizontal distance max1 from the hammer center is obtained. max1，max2 ;

[0040] According to D 0，max2 and D max1，max2 The difference between the two determines whether the test road has a risk of collapse.

[0041] Furthermore, according to D 0，max2 and D max1，max2 The steps to determine if a test road is at risk of collapse include:

[0042] According to D 0，max2 and D max1，max2 Get the deflection difference △Dmax2 ; Among them, △D max2 =D 0，max2 -D max1，max2 ;

[0043] If △D max2 If the value is greater than the preset threshold, it is determined that the test road is at risk of collapse;

[0044] If △D max2 If the value is less than or equal to the preset threshold, it is determined that there is no risk of collapse of the test road.

[0045] On the other hand, the present application also provides a road underground cavity detection device based on a drop weight deflectometer, the drop weight deflectometer comprising: a hammer load generating device, a first displacement sensor, and a plurality of second displacement sensors; the first displacement sensor is located at the hammer center of the hammer load generating device, and the plurality of second displacement sensors are distributed around the hammer center, and the horizontal distance between each second displacement sensor and the hammer center is different; after the hammer center of the hammer load generating device is aligned with a test point on a test road, the hammer load generating device applies a multi-level hammer load to the test point; the road underground cavity detection device based on the drop weight deflectometer comprises:

[0046] A first acquisition module is configured to acquire a first deflection data group collected by each displacement sensor when each level of hammer load is applied;

[0047] a calculation module configured to obtain, under the same level of hammer load, a second deflection data group corresponding to each second displacement sensor based on the first deflection data group collected by the first displacement sensor and the first deflection data group collected by each second displacement sensor;

[0048] A second acquisition module is used to obtain the road structure layer thickness and road type of the test road;

[0049] an output module, configured to input the thickness of the road structure layer, the road type, a plurality of first deflection data sets collected by the first displacement sensor under multi-level hammer loads, and a plurality of second deflection data sets corresponding to each second displacement sensor under multi-level hammer loads into a cavity prediction model, and obtain a first discrimination result output by the cavity prediction model;

[0050] A determination module is configured to determine whether a hole exists at the test point according to the first discrimination result.

[0051] Compared with the prior art, this application has the following beneficial effects:

[0052] The present application provides a method and device for detecting underground cavities in roads based on a drop hammer deflectometer. The drop hammer deflectometer includes a hammer load generating device, a first displacement sensor, and a plurality of second displacement sensors. The first displacement sensor is located at the hammer center of the hammer load generating device, and the plurality of second displacement sensors are distributed around the hammer center, and the horizontal distance between each second displacement sensor and the hammer center is different. After the hammer center of the hammer load generating device is aligned with the test point on the test road, the hammer load generating device applies a multi-level hammer load to the test point. The method comprises: obtaining a first deflection data group collected by each displacement sensor each time a level of hammer load is applied. Under the same level of hammer load, the second deflection data group corresponding to each second displacement sensor is obtained based on the first deflection data group collected by the first displacement sensor and the first deflection data group collected by each second displacement sensor. The road structure layer thickness and road type of the test road are obtained. The cavity prediction model inputs the thickness of the road structure layer, the road type, multiple first deflection data sets collected by the first displacement sensor under multiple hammer loads, and multiple second deflection data sets corresponding to each second displacement sensor under multiple hammer loads. The model then outputs a first determination result. Compared to existing technologies, this application can more accurately determine whether a cavity exists at a test point based on the first determination result output by the cavity prediction model, thereby improving the accuracy and reliability of underground cavity detection in roads. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application for which protection is claimed, but merely represents selected embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0054] Figure 1 A schematic structural diagram of a drop weight deflectometer provided in an embodiment of the present application;

[0055] Figure 2 This is a flow chart of a method for detecting underground cavities in roads based on a drop weight deflectometer provided in an embodiment of the present application;

[0056] Figure 3 A second flow chart of a method for detecting underground cavities in roads based on a drop weight deflectometer is provided in an embodiment of the present application;

[0057] Figure 4 A schematic diagram of a method for detecting underground cavities in roads based on a drop weight deflectometer provided in an embodiment of the present application;

[0058] Figure 5 A third flow chart of a method for detecting underground cavities in roads based on a drop weight deflectometer is provided in an embodiment of the present application;

[0059] Figure 6 A fourth flow chart of a method for detecting underground cavities in roads based on a drop weight deflectometer is provided in an embodiment of the present application;

[0060] Figure 7 A schematic diagram of the functional modules of a road underground cavity detection device based on a drop weight deflectometer provided in an embodiment of the present application.

[0061] Icons: 10-hammer load generating device; 20-first displacement sensor; 30-second displacement sensor; 40-road underground cavity detection device based on drop hammer deflectometer; 41-first acquisition module; 42-calculation module; 43-second acquisition module; 44-output module; 45-determination module. DETAILED DESCRIPTION

[0062] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work fall within the scope of protection of the present application.

[0063] In the description of this application, it should be noted that relational terms such as first and second are used solely to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. The term "connected" should be understood broadly, for example, it can mean fixed connection, detachable connection, or integral connection; it can be directly connected or indirectly connected through an intermediate medium.

[0064] The following describes some embodiments of the present application in detail with reference to the accompanying drawings. The following embodiments and features thereof may be combined with each other unless there is any conflict.

[0065] As mentioned in the background, the deflection values ​​obtained by traditional FWD testing methods primarily reflect the mechanical properties of the surface structure of the road surface and subgrade, but are less sensitive to changes in deeper structures. This significantly reduces the reliability of these analysis and evaluation results when detecting underground voids, making it difficult to meet the requirements for high-precision detection. Therefore, improving the accuracy and reliability of FWD methods for detecting underground voids in roads is a pressing technical challenge for those skilled in the art.

[0066] In order to solve the above technical problems, the embodiment of the present application provides a method for detecting underground cavities in roads based on a drop weight deflectometer. For better understanding, the specific structure of the drop weight deflectometer is first described below.

[0067] In the examples of this application, please refer to Figure 1 The drop-weight deflectometer includes an impact load generator 10, a first displacement sensor 20, and multiple second displacement sensors 30. The impact load generator 10 comprises a drop weight, a lifting device, and a supporting plate. The lifting device, based on computer instructions, lifts the drop weight to a predetermined height and then releases it. The drop weight, through free fall, generates an impact load at the impact center A of the supporting plate, which then acts on the road surface.

[0068] The first displacement sensor 20 is located at the impact center A of the impact load generating device 10, i.e., the horizontal distance between the first displacement sensor 20 and the impact center A is 0. Multiple second displacement sensors 30 are distributed at different locations around the impact center A, i.e., each second displacement sensor 30 has a different horizontal distance from the impact center A. Therefore, the data collected by each second displacement sensor 30 is different.

[0069] Optionally, the horizontal distance between each second displacement sensor 30 and the hammer center A is greater than 0 and less than or equal to 2000 mm.

[0070] When testing underground cavities on roads, a drop-weight deflectometer vehicle strikes points along the test road, each 3-5m apart, and multiple strikes are performed. During the strikes, the drop-weight deflectometer activates all displacement sensors to collect surface deformation curves during the impact process.

[0071] To better understand the technical solution of this application, the following description uses a striking point as an example. A drop-hammer deflectometer is driven on the driving lane of a test road. After the striking center A of the hammer load generating device 10 is aligned with a test point (i.e., striking point) on the test road, the hammer load generating device 10 strikes the test point multiple times to apply multiple levels of hammer loads. The magnitude of the hammer load can be adjusted by adjusting the mass and drop height of the drop hammer.

[0072] Optionally, the hammer load generating device 10 applies four levels of hammer loads to the test point, and the four levels of hammer loads are 25KN, 50KN, 75KN and 100KN respectively.

[0073] Based on the above design, please refer to Figure 2 The present application provides a method for detecting underground cavities in roads based on a drop weight deflectometer, which includes the following steps.

[0074] Step S200: When each level of hammer load is applied, a first deflection data set collected by each displacement sensor is obtained.

[0075] It can be understood that when the hammer load generating device strikes the test point once, that is, applies a first-level hammer load, each displacement sensor can collect a first deflection data set at the corresponding road surface position.

[0076] In the embodiment of the present application, the first deflection data group includes: deflection peak value D loc，load and time Dt loc，load .

[0077] D loc，load It represents the peak vertical displacement of the road surface collected by the displacement sensor at a horizontal distance of loc mm from the hammer center when the hammer load is load KN. And D loc，load The unit is μm.

[0078] Dt loc，load It represents the time value of the peak vertical displacement of the road surface collected by the displacement sensor at a horizontal distance of loc mm from the hammer center when the hammer load is load KN. And Dt loc，load The unit is ms.

[0079] Wherein, 0≤loc≤max1, max1 is the maximum horizontal distance between the hammer center and the plurality of second displacement sensors.

[0080] That is, D 0，load and Dt 0，load It represents the first deflection data set collected by the first displacement sensor under the hammer load of load KN. max1，load and Dt max1，load It represents the first deflection data set collected by the second displacement sensor with the largest horizontal distance from the hammer center under the hammer load of load KN.

[0081] Step S300: Under the same level of hammer load, according to the first deflection data group collected by the first displacement sensor and the first deflection data group collected by each second displacement sensor, obtain a second deflection data group corresponding to each second displacement sensor.

[0082] In the embodiment of the present application, the second deflection data group corresponding to the current second displacement sensor includes: relative deflection peak ND li，load and the time Dt collected by the current second displacement sensor li，load Where li is the horizontal distance between the hammer center and the current second displacement sensor, and 0 <li≤max1。

[0083] Relative deflection peak ND li，load Satisfies the following formula:

[0084] ND li，load =D li，liad / D 0，load

[0085] Among them, D li，load is the deflection peak value collected by the second displacement sensor, D 0，load is the peak deflection value collected by the first displacement sensor.

[0086] It can be seen that under the same hammer load, the ND li，load The deflection peak value D in the first deflection data set collected by the first displacement sensor is 0，load and the deflection peak value D in the first deflection data set collected by the second displacement sensor li，load The Dt in the second deflection data set corresponding to the current second displacement sensor is li，load This is the time Dt in the first deflection data set collected by the second displacement sensor. li，load .

[0087] Step S400: Obtain the road structure layer thickness and road type of the test road.

[0088] Step S500: Input the thickness of the road structure layer, the road type, multiple first deflection data sets collected by the first displacement sensor under multi-level hammer loads, and multiple second deflection data sets corresponding to each second displacement sensor under multi-level hammer loads into the void prediction model, and obtain a first discrimination result output by the void prediction model.

[0089] It should be noted that, according to the contents of steps S200 and S300, each time the hammer load generating device strikes the test point once, that is, when a first-level hammer load is applied to the test point, a first deflection data set collected by the first displacement sensor under the hammer load of that level can be obtained, as well as a second deflection data set corresponding to each second displacement sensor under the hammer load of that level.

[0090] Therefore, when the hammer load generating device strikes the test point multiple times, that is, applies multi-level hammer loads to the test point, multiple first deflection data groups collected by the first displacement sensor under the multi-level hammer loads and multiple second deflection data groups corresponding to each second displacement sensor under the multi-level hammer loads can be obtained.

[0091] In an optional embodiment, the hole prediction model includes a double-layer neural network and multiple bottom-level decision trees, and the number of bottom-level decision trees is equal to the number of second displacement sensors. That is, the multiple bottom-level decision trees serve as the bottom-level model of the hole prediction model, and the double-layer neural network serves as the upper-level model of the hole prediction model. Figure 3 As shown, step S500 includes sub-steps S510 and S520.

[0092] Step S510: Input the road structure layer thickness, road type, multiple first deflection data sets collected by the first displacement sensor under multi-level hammer loads, and multiple second deflection data sets corresponding to each second displacement sensor under multi-level hammer loads into a corresponding bottom-level decision tree, and obtain a second discrimination result output by each bottom-level decision tree.

[0093] For example, see Figure 4 Assume that the hammer load generating device applies four levels of hammer load to the test point (the four levels of hammer load are 25kN, 50kN, 75kN, and 100kN respectively). The number of second displacement sensors is three (i.e., second displacement sensor A, second displacement sensor B, and second displacement sensor C, with second displacement sensor C being the largest horizontal distance from the hammer center).

[0094] Then, three underlying decision trees can be established, namely, the underlying decision tree A (established by the data of the first displacement sensor and the second displacement sensor A), the underlying decision tree B (established by the data of the first displacement sensor and the second displacement sensor B), and the underlying decision tree C (established by the data of the first displacement sensor and the second displacement sensor C).

[0095] In addition, the four first deflection data sets (D 0，25 and Dt 0，25 、D 0，50 and Dt 0，50 、D 0，75 and Dt 0，75 、D 0，100 and Dt 0，100 );

[0096] The four second deflection data sets (ND liA，25 and Dt liA，25 ,ND liA，50 and Dt liA，50 ,NDliA，75 and Dt liA，75 ,ND liA，100 and Dt liA，100 );

[0097] The four second deflection data sets (ND liB，25 and Dt liB，25 ,ND liB，50 and Dt liB，50 ,ND liB，75 and Dt liB，75 ,ND liB，100 and Dt liB，100 );

[0098] The four second deflection data sets (ND liA，25 and Dt max1，25 ,ND max1，50 and Dt max1，50 ,ND max1，75 and Dt max1，75 ,ND max1，100 and Dt max1，100 ).

[0099] The thickness of the road structure layer, the road type, the four first deflection data groups collected by the first displacement sensor, and the four second deflection data groups corresponding to the second displacement sensor A are input into the trained bottom-level decision tree A, and the second discrimination result output by the bottom-level decision tree A is obtained.

[0100] The thickness of the road structure layer, the road type, the four first deflection data groups collected by the first displacement sensor, and the four second deflection data groups corresponding to the second displacement sensor B are input into the trained bottom-level decision tree B, and the second discrimination result output by the bottom-level decision tree B is obtained.

[0101] The thickness of the road structure layer, the road type, the four first deflection data groups collected by the first displacement sensor, and the four second deflection data groups corresponding to the second displacement sensor C are input into the trained bottom-level decision tree C, and the second discrimination result output by the bottom-level decision tree C is obtained.

[0102] The second discrimination result is used to indicate whether there is an underground cavity at the test point. For example, if the second discrimination result is 0, it indicates that there is no underground cavity at the test point. If the second discrimination result is 1, it indicates that there is an underground cavity at the test point.

[0103] Step S520: input the second discrimination results output by all bottom-level decision trees into the double-layer neural network, and obtain the first discrimination results output by the double-layer neural network.

[0104] As can be seen, the second discrimination result output by each bottom-level decision tree is obtained based on the data from the first displacement sensor and a corresponding second displacement sensor. Because each second displacement sensor is located at a different position around the hammer center (i.e., the test point), the input data for each bottom-level decision tree is not exactly the same.

[0105] By establishing a bottom-level decision tree based on the first displacement sensor and one of the multiple second displacement sensors, the number of bottom-level decision trees is expanded, increasing the input data for the two-layer neural network. The two-layer neural network then performs weighted intelligent recognition on the second discrimination results output by the bottom-level decision trees at various locations, significantly improving the error correction capability and recognition accuracy of the cavity prediction model.

[0106] Step S600: Determine whether there is a hole at the test point according to the first determination result.

[0107] In an embodiment of the present application, the value of the first discrimination result output by the two-layer neural network (i.e., the first discrimination result finally output by the cavity prediction model) is between 0 and 1, representing the probability of the existence of an underground cavity at the test point.

[0108] Optionally, if the value of the first discrimination result is close to 0, it is considered that the first discrimination result is 0 and there is no underground cavity at the test point. If the value of the first discrimination result is close to 1, it is considered that the first discrimination result is 1 and there is an underground cavity at the test point.

[0109] Based on the above design, since the cavity prediction model in this application is composed of multiple trained underlying decision trees and a trained two-layer neural network, the second discrimination results output by the multiple underlying decision trees can reflect the road surface conditions at different locations. Therefore, the first discrimination result output by the two-layer neural network after weighted intelligent identification of the multiple second discrimination results more accurately reflects the presence of a cavity at the test point. The cavity prediction model provided in this application has a simple structure. By combining the two-layer neural network with multiple underlying decision trees, it has the advantages of high cavity recognition accuracy, strong reliability, and fast computing speed.

[0110] The following combination Figure 5 The training process of the cavity prediction model provided in this application is described. Before step S200, the road underground cavity detection method based on the falling weight deflectometer further includes the following steps.

[0111] Step S110: Acquire multiple reference road sample sets.

[0112] Among them, each reference road sample set includes the road structure layer thickness of the reference road, the road type, multiple first deflection data groups collected by a first displacement sensor under multi-level hammer loads, and multiple second deflection data groups corresponding to a second displacement sensor under multi-level hammer loads.

[0113] Step S120: performing one-to-one training on multiple bottom-level decision trees according to multiple reference road sample sets to obtain multiple trained bottom-level decision trees.

[0114] Step S130: training the double-layer neural network according to the multiple second discrimination results output by the multiple trained bottom-level decision trees to obtain a trained double-layer neural network.

[0115] For example, assuming that the number of second displacement sensors is 3 (second displacement sensor A, second displacement sensor B, and second displacement sensor C), the number of reference road sample sets is also 3 (reference road sample set A, reference road sample set B, and reference road sample set C), and the number of underlying decision trees is also 3 (bottom-level decision tree A, bottom-level decision tree B, and bottom-level decision tree C).

[0116] Among them, the reference road sample set A includes: the road structure layer thickness of the reference road, the road type, multiple first deflection data groups collected by the first displacement sensor under multi-level hammer loads, and multiple second deflection data groups corresponding to the second displacement sensor A under multi-level hammer loads.

[0117] The reference road sample set B includes: the road structure layer thickness of the reference road, the road type, multiple first deflection data groups collected by the first displacement sensor under multi-level hammer loads, and multiple second deflection data groups corresponding to the second displacement sensor B under multi-level hammer loads.

[0118] The reference road sample set C includes: the road structure layer thickness of the reference road, the road type, multiple first deflection data groups collected by the first displacement sensor under multi-level hammer loads, and multiple second deflection data groups corresponding to the second displacement sensor C under multi-level hammer loads.

[0119] The underlying decision tree A is trained based on the reference road sample set A to obtain the trained underlying decision tree A. Similarly, the underlying decision tree B is trained based on the reference road sample set B to obtain the trained underlying decision tree B. The underlying decision tree C is trained based on the reference road sample set C to obtain the trained underlying decision tree C.

[0120] The double-layer neural network is trained according to the second discrimination result output by the trained bottom-level decision tree A, the second discrimination result output by the trained bottom-level decision tree B, and the second discrimination result output by the trained bottom-level decision tree C to obtain the trained double-layer neural network, and finally complete the training of the entire void prediction model.

[0121] Furthermore, in actual road construction, some underground cavities do not pose a significant risk, and even if a road has underground cavities, it does not necessarily lead to collapse immediately. However, if a road is at risk of collapse, it is definitely because of the presence of underground cavities.

[0122] In view of this, in order to further improve the diagnostic accuracy of underground cavities in roads, the present application also provides a method for assessing the risk of road collapse based on the data obtained in step S200.

[0123] Specifically, see Figure 6 After step S200, the road underground cavity detection method based on the drop weight deflectometer further includes the following steps.

[0124] Step S710: establishing a first regression equation based on multiple deflection peak values ​​collected by the first displacement sensor under multi-level hammer loads.

[0125] The first regression equation represents the peak deflection value D collected by the first displacement sensor under different levels of hammer load. 0，load .

[0126] Step S720: establishing a second regression equation based on multiple deflection peak values ​​collected by the second displacement sensor at a horizontal distance max1 from the hammer center under multiple hammer loads.

[0127] The second regression equation represents the peak deflection value D collected by the second displacement sensor at a horizontal distance of max1 from the hammer center under different levels of hammer load. max1，load .

[0128] Step S730: Obtain the maximum unilateral load max2 of a single axle of a vehicle traveling on the test road.

[0129] Step S740: Obtain the first deflection peak value D collected by the first displacement sensor when the hammer load is max2 (ie, load=max2) according to the first regression equation. 0，max2 .

[0130] Step S750: Obtain the second deflection peak value D collected by the second displacement sensor at a horizontal distance of max1 from the hammer center when the hammer load is max2 (i.e. load = max2) according to the second regression equation. max1，max2 .

[0131] Step S760: According to D 0，max2 and D max1，max2 The difference between the two determines whether the test road has a risk of collapse.

[0132] Specifically, according to D 0，max2 and D max1，max2 Get the deflection difference △D max2 .

[0133] Among them, △D max2 =D 0，max2 -D max1，max2 .

[0134] If △D max2 If the value is greater than the preset threshold, it is determined that the test road is at risk of collapse.

[0135] If △D max2 If the value is less than or equal to the preset threshold, it is determined that there is no risk of collapse of the test road.

[0136] The size of the preset threshold is related to the road type of the test road.

[0137] Based on the above design, this application uses the extrapolated results of the empirical data of the drop weight deflectometer to judge the risk of road collapse, further improving the diagnostic accuracy of underground cavities in the road.

[0138] Further, see Figure 7 , Figure 7 A functional module diagram of a road underground cavity detection device 40 based on a drop hammer deflectometer provided in an embodiment of the present application. It should be noted that the basic principle of the device and the technical effects produced are the same as those in the above-mentioned embodiment. For the sake of brief description, for parts not mentioned in this embodiment, reference can be made to the corresponding contents in the above-mentioned embodiment. The drop hammer deflectometer includes: a hammer load generating device, a first displacement sensor and a plurality of second displacement sensors. The first displacement sensor is located at the hammer center of the hammer load generating device, and the plurality of second displacement sensors are distributed around the hammer center, and the horizontal distance between each second displacement sensor and the hammer center is different. After the hammer center of the hammer load generating device is aligned with the test point on the test road, the hammer load generating device applies a multi-level hammer load to the test point. The road underground cavity detection device 40 based on the drop hammer deflectometer includes:

[0139] The first acquisition module 41 is configured to acquire a first deflection data set collected by each displacement sensor each time a level of hammer load is applied.

[0140] The calculation module 42 is configured to obtain a second deflection data set corresponding to each second displacement sensor based on the first deflection data set collected by the first displacement sensor and the first deflection data set collected by each second displacement sensor under the same level of hammer load.

[0141] The second acquisition module 43 is used to obtain the road structure layer thickness and road type of the test road.

[0142] The output module 44 is used to input the thickness of the road structure layer, the road type, multiple first deflection data sets collected by the first displacement sensor under the multi-level hammer load, and multiple second deflection data sets corresponding to each second displacement sensor under the multi-level hammer load into the void prediction model, and obtain a first judgment result output by the void prediction model.

[0143] The determination module 45 is configured to determine whether a hole exists at the test point according to the first determination result.

[0144] The above description is merely a preferred embodiment of the present application and is not intended to limit the present application. Various modifications and variations are possible for those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present application shall be included within the scope of protection of the present application.

[0145] It will be apparent to those skilled in the art that the present application is not limited to the details of the exemplary embodiments described above and that the present application can be implemented in other specific forms without departing from the spirit or essential characteristics of the present application. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the present application is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be included therein. Any reference sign in a claim should not be construed as limiting the claim to which it relates.

Claims

1. A method for detecting underground cavities in roads based on a drop weight deflectometer, characterized in that: The drop-weight deflectometer includes: a hammer load generating device, a first displacement sensor, and a plurality of second displacement sensors; the first displacement sensor is located at the hammer center of the hammer load generating device, and the plurality of second displacement sensors are distributed around the hammer center, and each of the second displacement sensors has a different horizontal distance from the hammer center; after the hammer center of the hammer load generating device is aligned with a test point on a test road, the hammer load generating device applies a multi-level hammer load to the test point; the method includes: Each time a hammer load is applied, a first deflection data group collected by each displacement sensor is obtained; Under the same level of hammer load, obtaining a second deflection data group corresponding to each second displacement sensor based on the first deflection data group collected by the first displacement sensor and the first deflection data group collected by each second displacement sensor; Obtaining the road structure layer thickness and road type of the test road; Inputting the thickness of the road structure layer, the road type, a plurality of first deflection data sets collected by the first displacement sensor under multi-level hammer loads, and a plurality of second deflection data sets corresponding to each second displacement sensor under the multi-level hammer loads into a cavity prediction model, and obtaining a first discrimination result output by the cavity prediction model; Determine whether the test point has a hole according to the first discrimination result.

2. The method for detecting underground cavities in roads based on a drop weight deflectometer according to claim 1, characterized in that: The first deflection data set includes a deflection peak value D loc，load and time Dt loc，load ; D loc，load It represents the peak vertical displacement of the road surface collected by the displacement sensor at a horizontal distance of loc from the impact center when the impact load is load; Dt loc，load It indicates the time when the peak vertical displacement of the road surface detected by the displacement sensor at the horizontal distance loc from the impact center occurs when the impact load is load; Wherein, 0≤loc≤max1, max1 is the maximum horizontal distance between the hammer center and the plurality of second displacement sensors.

3. The method for detecting underground cavities in roads based on a drop weight deflectometer according to claim 2, characterized in that: Under the same level of hammer load, the step of obtaining a second deflection data group corresponding to each second displacement sensor based on the first deflection data group collected by the first displacement sensor and the first deflection data group collected by each second displacement sensor includes: The second deflection data set corresponding to the current second displacement sensor includes the relative deflection peak value ND li，load and the time Dt collected by the current second displacement sensor li，load ; Where li is the horizontal distance between the hammer center and the current second displacement sensor, and 0 <li≤max1; Relative deflection peak ND li，load Satisfies the following formula: ND li，load =D li，load / D 0，load Among them, D li，load is the deflection peak value collected by the second displacement sensor, D 0，load is the peak deflection value collected by the first displacement sensor.

4. The method for detecting underground cavities in roads based on a drop weight deflectometer according to claim 1, characterized in that: The cavity prediction model includes a two-layer neural network and multiple bottom-level decision trees, and the number of the bottom-level decision trees is equal to the number of the second displacement sensors; The steps of inputting the thickness of the road structure layer, the road type, a plurality of first deflection data sets collected by the first displacement sensor under multi-level hammer loads, and a plurality of second deflection data sets corresponding to each second displacement sensor under the multi-level hammer loads into a cavity prediction model, and obtaining a first discrimination result output by the cavity prediction model include: Inputting the road structure layer thickness, road type, multiple first deflection data sets collected by the first displacement sensor under multi-level hammer loads, and multiple second deflection data sets corresponding to each second displacement sensor under the multi-level hammer loads into a corresponding bottom-level decision tree, and obtaining a second discrimination result output by each bottom-level decision tree; The second discrimination results output by all the bottom-level decision trees are input into the double-layer neural network, and the first discrimination results output by the double-layer neural network are obtained.

5. The method for detecting underground cavities in roads based on a drop weight deflectometer according to claim 1, characterized in that: The cavity prediction model includes a two-layer neural network and multiple bottom-level decision trees, and the number of the bottom-level decision trees is equal to the number of the second displacement sensors; Before the step of obtaining a first deflection data set collected by each displacement sensor when applying each level of hammer load, the method further includes: Acquire multiple reference road sample sets; each reference road sample set includes a road structure layer thickness of the reference road, a road type, multiple first deflection data sets collected by a first displacement sensor under multiple hammer loads, and multiple second deflection data sets corresponding to a second displacement sensor under the multiple hammer loads; Performing one-to-one training on multiple underlying decision trees based on multiple reference road sample sets to obtain multiple trained underlying decision trees; The double-layer neural network is trained according to the multiple second discrimination results output by the multiple trained bottom-level decision trees to obtain a trained double-layer neural network.

6. The method for detecting underground cavities in roads based on a drop weight deflectometer according to claim 1, characterized in that: The step of determining whether there is a hole at the test point according to the first discrimination result includes: If the first discrimination result is 0, it is determined that there is no hole at the test point; If the first discrimination result is 1, it is determined that there is a hole at the test point.

7. The method for detecting underground cavities in roads based on a drop weight deflectometer according to claim 1, characterized in that: The hammer load generating device applies four levels of hammer loads to the test point, and the four levels of hammer loads are 25KN, 50KN, 75KN and 100KN respectively.

8. The method for detecting underground cavities in roads based on a drop weight deflectometer according to claim 2, characterized in that: After the step of obtaining a first deflection data set collected by each displacement sensor when applying a level of hammer load, the method further includes: Establishing a first regression equation based on multiple deflection peak values ​​collected by the first displacement sensor under multiple levels of hammer loads; wherein the first regression equation represents the deflection peak values ​​collected by the first displacement sensor under different levels of hammer loads; A second regression equation is established based on multiple deflection peaks collected by the second displacement sensor at a horizontal distance of max1 from the hammer center under multiple levels of hammer loads. The second regression equation represents the deflection peaks collected by the second displacement sensor at a horizontal distance of max1 from the hammer center under different levels of hammer loads. Obtain the maximum unilateral load max2 of a single axle of a vehicle passing through the test road; According to the first regression formula, when the hammer load is max2, the first deflection peak value D collected by the first displacement sensor is obtained. 0，max2 ; According to the second regression formula, when the hammer load is max2, the second deflection peak value D collected by the second displacement sensor at the horizontal distance max1 from the hammer center is obtained. max1，max2 ; According to D 0，max2 and D max1，max2 The difference between the two determines whether the test road has a risk of collapse.

9. The method for detecting underground cavities in roads based on a drop weight deflectometer according to claim 8, characterized in that: According to D 0，max2 and D max1，max2 The steps to determine if a test road is at risk of collapse include: According to D 0，max2 and D max1，max2 Get the deflection difference △D max2 ; Among them, △D max2 =D 0，max2 -D max1，max2 ; If △D max2 If the value is greater than the preset threshold, it is determined that the test road is at risk of collapse; If △D max2 If the value is less than or equal to the preset threshold, it is determined that there is no risk of collapse of the test road.

10. A road underground cavity detection device based on a drop weight deflectometer, characterized in that: The drop-weight deflectometer includes: a hammer load generating device, a first displacement sensor, and a plurality of second displacement sensors; the first displacement sensor is located at the hammer center of the hammer load generating device, and the plurality of second displacement sensors are distributed around the hammer center, and the horizontal distance between each second displacement sensor and the hammer center is different; after the hammer center of the hammer load generating device is aligned with a test point on a test road, the hammer load generating device applies a multi-level hammer load to the test point; the road underground cavity detection device based on the drop-weight deflectometer includes: A first acquisition module is configured to acquire a first deflection data group collected by each displacement sensor when each level of hammer load is applied; a calculation module configured to obtain, under the same level of hammer load, a second deflection data group corresponding to each second displacement sensor based on the first deflection data group collected by the first displacement sensor and the first deflection data group collected by each second displacement sensor; A second acquisition module is used to obtain the road structure layer thickness and road type of the test road; an output module, configured to input the thickness of the road structure layer, the road type, a plurality of first deflection data sets collected by the first displacement sensor under multi-level hammer loads, and a plurality of second deflection data sets corresponding to each second displacement sensor under multi-level hammer loads into a cavity prediction model, and obtain a first discrimination result output by the cavity prediction model; A determination module is configured to determine whether a hole exists at the test point according to the first discrimination result.

Images