Running posture recognition method of high-precision tunnel boring machine
By collecting and processing point cloud data in the shield machine, using structural confidence and distance error adjustment technology, the distance measurement between data points is adaptively adjusted, and the matching error problem of shield machine attitude detection in the PL-ICP algorithm is solved. , improves detection efficiency and credibility.
Patent Information
- Application Number
- CN202411404438.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-10
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2044-10-10
AI Technical Summary
In the prior art, during the real-time detection of the shield machine attitude, the PL-ICP algorithm causes data point matching errors due to point cloud data interference in some actual defect areas on the shield machine, which reduces the credibility and detection efficiency of the matching results.
Point cloud data in the cutter plate area and palm area of the shield machine, and adaptively adjust the distance measurement between data points to perform high-precision matching detection through structural confidence, position change consistency and distance error adjustment necessity.
The probability of data point matching errors is reduced, the credibility of matching results and the pose detection efficiency are improved, and the correlation characteristics of the operating posture of the shield machine and the impact of the actual defect area are clarified.
Smart Images

Figure CN118918578A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of image processing, and in particular to a method for identifying the running posture of a high-precision tunnel boring machine. Background Art
[0002] In the process of tunneling in soft soil, tunnel boring machines are used to ensure the smooth progress of engineering projects such as road reconstruction. Among them, shield machines are often used as a tunnel boring machine suitable for tunneling in soft soil. Then, micro-motion radar detection and other related technologies are used to dynamically extract point cloud data of the environmental conditions of the shield machine's operating area and the operating posture of the shield machine itself, so as to achieve accurate identification and control of the shield machine's operating status.
[0003] The existing technology usually uses the PL-ICP algorithm to match the point cloud data. In the process of real-time detection of the shield machine posture, the point cloud data of some actual defect areas on the shield machine will interfere with the distance relationship between different data points in the point cloud data, causing some data points to match incorrectly, reducing the credibility of the matching results and reducing the efficiency of posture detection. Summary of the invention
[0004] The present invention provides a high-precision tunnel boring machine operation posture recognition method to solve the existing problem: in the process of PL-ICP algorithm matching point cloud data to perform real-time detection of the shield machine posture, the point cloud data of some actual defective areas on the shield machine will interfere with the distance relationship between different data points in the point cloud data, causing some data points to match incorrectly.
[0005] A high-precision tunnel boring machine operation posture recognition method of the present invention adopts the following technical solution: The following steps are involved: Collecting point cloud data of the cutter head area and the tunnel face area of the shield machine in several monitoring frames, wherein the point cloud data includes a plurality of shield image points with three-dimensional spatial position information, and each shield image point corresponds to a data point in each monitoring frame; The cutterhead area and the tunnel face area of the shield machine are both recorded as a shield monitoring area. According to the overall movement of the position of the same shield image point in different monitoring frames, the structural confidence of each shield image point in each monitoring frame is obtained. According to the change of position information of different shield image points in the same monitoring frame and the structural confidence, the consistency of position change of each shield image point in the cutter head area of the shield machine in each monitoring frame is obtained; according to the structural difference of the consistency of position change between different data points of each shield image point in each monitoring frame, the necessity of adjusting the distance error between any two data points of each shield image point in each monitoring frame is obtained; according to the necessity of adjusting the distance error between different data points, the distance metric between different data points is adjusted to obtain the adaptive distance metric between any two data points; Match detection is performed based on an adaptive distance metric between any two data points.
[0006] Preferably, the structural confidence of each shield image point in each monitoring frame is obtained according to the overall movement of the position of the same shield image point in different monitoring frames, and the specific method includes: Get the tunnel face area in The weight coefficient of the tunnel face of the frame monitoring frame and the weight coefficient of each shield image point in the first Confidence coefficient of structure in frame monitoring frame; For any shield image point in the cutterhead area of the shield machine, The structural confidence factor in the frame monitoring frame is related to the tunnel face area in the first The product of the weight coefficients of the tunnel face of the frame monitoring frame is recorded as the shield image point at the Frame Monitoring Confidence of structures in frames.
[0007] Preferably, the obtaining of the tunnel face area is performed in the The weight coefficient of the tunnel face of the frame monitoring frame and the weight coefficient of each shield image point in the first The structure confidence coefficient in the frame monitoring frame includes the following specific methods: For any shield image point in the cutterhead area of the shield machine, The data point in the frame monitoring frame is consistent with the shield image point in the The Euclidean distance between the data points in the frame monitoring frame is recorded as the first distance; the shield image point is recorded as the The data point in the frame monitoring frame is consistent with the shield image point in the The Euclidean distance between the data points in the frame monitoring frame is recorded as the second distance; the absolute value of the difference between the first distance and the second distance is recorded as the shield image point in the first The structural confidence factor in the frame monitoring frame; the structural confidence factor of the shield image point in all monitoring frames is linearly normalized, and the normalized structural confidence factor is recorded as the structural confidence coefficient; Any shield image point in the tunnel face area is recorded as the target tunnel face image point; the reference shield image point is recorded in the The acquisition method of the structural confidence factor in the frame monitoring frame is to obtain the target face image point in the first The structural confidence factor in the frame monitoring frame is calculated by The cumulative sum of the structural confidence factors in the frame monitoring frame is recorded as the tunnel face area in the first The tunnel face weight of the frame monitoring frame; the tunnel face weight of the tunnel face area in all monitoring frames is linearly normalized, and the normalized weight of each tunnel face is recorded as the tunnel face weight coefficient.
[0008] Preferably, the position change consistency of each shield image point in the cutter head area of the shield machine in each monitoring frame is obtained according to the change size of the position information of different shield image points in the same monitoring frame and the structural confidence, and the specific method includes: Obtain the offset vector of each monitoring frame of the shield machine cutterhead area according to the structural confidence; According to the shield machine cutter head area The shield image point is The difference between the structural confidence of the frame monitoring frame and other monitoring frames, as well as the difference between the offset vectors of adjacent monitoring frames, is used to obtain the first The shield image point is The frame monitors the consistency of the position change of the frame, including the specific methods: In the formula, Indicates the shield machine cutterhead area The shield image point is Frame monitoring: consistency of position changes between frames; Represents the number of all clusters in the cutterhead area of the shield machine; Indicates the shield machine cutterhead area The offset vector of the clusters; Indicates the shield machine cutterhead area The offset vector of the clusters; represents the preset denominator hyperparameter; Indicates the shield machine cutterhead area The shield image point is Frame monitoring frame structure confidence; Indicates the shield machine cutterhead area The mean value of the structural confidence of each shield image point in all monitoring frames; represents an exponential function with a natural constant as base; Indicates taking the absolute value.
[0009] Preferably, the specific method of obtaining the offset vector of each monitoring frame of the cutter head area of the shield machine according to the structural confidence level includes: For any monitoring frame of the shield machine cutter head area, the mean value of the structural confidence of all shield image points in the shield machine cutter head area in the monitoring frame is recorded as the structural confidence coefficient of the monitoring frame, and the structural confidence coefficients of all monitoring frames are obtained; the absolute value of the difference between the structural confidence coefficients of different monitoring frames is used as the distance metric, and all monitoring frames are hierarchically clustered according to the distance metric to obtain several clusters; for any cluster, the least squares method is used to fit the data points of all shield image points in the shield machine cutter head area in all monitoring frames in the cluster to obtain the fitting straight line of the shield machine cutter head area in the cluster; the vector formed by the fitting straight line is recorded as the offset vector.
[0010] Preferably, the necessity of adjusting the distance error between any two data points of each shield image point in each monitoring frame is obtained according to the structured difference of the consistency of the position change between different data points of each shield image point in each monitoring frame, and the specific method includes: For all shield image points in the shield machine cutterhead area, Frame monitoring frame and the The data points in the frame monitoring frame are clustered by density to obtain several clusters; each cluster is recorded as a monitoring cluster; any shield image point in the shield machine cutter head area in any frame monitoring frame is recorded as a target data point, and a reference range is preset. ; Take the data point as the center and the radius as The area is recorded as the neighborhood monitoring area of the target data point; Any data point in the neighborhood monitoring area of the target data point is recorded as a monitoring reference data point; according to the difference in the consistency of position changes between the target data point and the monitoring cluster to which the monitoring reference data point belongs, the necessary factor for adjusting the distance error between the target data point and the monitoring reference data point is obtained; Obtain necessary factors for distance error adjustment between all data points in the neighborhood monitoring area of the target data point and the target data point, perform linear normalization on all necessary factors for distance error adjustment, and record each normalized necessary factor for distance error adjustment as the necessity for distance error adjustment.
[0011] Preferably, the necessary factor for adjusting the distance error between the target data point and the monitoring reference data point is obtained according to the difference in consistency of position change between the target data point and the monitoring cluster to which the monitoring reference data point belongs, and the specific method includes: In the formula, Indicates the necessary factor for adjusting the distance error between the target data point and the monitoring reference data point; Indicates the consistency of position change of target data points; Indicates the consistency of position change of monitoring reference data points; Represents the mean of the consistency of position changes of all data points in the monitoring cluster to which the monitoring reference data point belongs; Represents the Euclidean distance between the target data point and the monitoring reference data point; Represents the preset denominator reference hyperparameter.
[0012] Preferably, the distance metric between different data points is adjusted according to the necessity of adjusting the distance error between different data points to obtain an adaptive distance metric between any two data points, including the specific method of: Obtain all pairs of data points to be adjusted and fixed pairs of data points according to the necessity of distance error adjustment; For any two data points in a pair of data points to be adjusted, the product of the Euclidean distance between the two data points and the necessity of distance error adjustment is recorded as the adaptive distance metric of the two data points; the adaptive distance metric of the two data points in each pair of data points to be adjusted is obtained; and the Euclidean distance of the two data points in each fixed data point pair is used as the adaptive distance metric of the two data points in each fixed data point pair.
[0013] Preferably, the method of obtaining all the data point pairs to be adjusted and the fixed data point pairs according to the necessity of adjusting the distance error includes: Preset a distance error adjustment necessity threshold For any two shield image points in the shield machine cutter head area in any monitoring frame, if the distance error between the two data points is greater than , the data point pair formed by these two data points is recorded as the data point pair to be adjusted; if the distance error between the two data points is less than or equal to the necessity of adjustment , the data point pair formed by these two data points is recorded as a fixed data point pair.
[0014] Preferably, the matching detection is performed based on the adaptive distance metric between any two data points, and the specific method includes: The adaptive distance metric between any two different data points is used as the distance metric, and the point cloud data matching result is obtained by matching through the PL-ICP algorithm according to the distance metric.
[0015] The technical solution of the present invention has the following beneficial effects: according to the overall position movement of the same shield image point in different monitoring frames, the structural confidence of each shield image point in each monitoring frame is obtained; according to the change size of the position information of different shield image points in the same monitoring frame, the consistency of the position change of each shield image point in each monitoring frame is obtained; according to the structural difference of the consistency of the position change between different data points in each monitoring frame, the necessity of adjusting the distance error between any two data points in each monitoring frame is obtained; according to the necessity of adjusting the distance error between different data points, the distance metric between different data points is adjusted to obtain the distance error between any two data points in each monitoring frame. Adaptive distance measurement between two data points; matching detection is performed based on the adaptive distance measurement between any two data points; the consistency of position change is used to describe the obvious degree of change trend of slow settlement or displacement of the running posture of the shield machine, so that the correlation characteristics of the running posture of the shield machine and the point cloud data are more clearly represented; the necessity of distance error adjustment is used to describe the structural changes between different data points affected by the wear area of the shield machine or other environmental factors, so that the influence of the actual defect area on the distance measurement when matching point cloud data is clearer; the probability of data point matching errors is reduced, the credibility of the matching results is improved, and the efficiency of posture detection is improved. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0017] Figure 1 The present invention is a flowchart of the steps of a method for identifying the operating posture of a high-precision tunnel boring machine. DETAILED DESCRIPTION
[0018] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the operation posture recognition method of a high-precision tunnel boring machine proposed by the present invention, its specific implementation method, structure, features and effects, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.
[0019] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
[0020] The specific scheme of the high-precision tunnel boring machine operation posture recognition method provided by the present invention is described in detail below with reference to the accompanying drawings.
[0021] See also Figure 1 , which shows a flowchart of a method for identifying the running posture of a high-precision tunnel boring machine provided by an embodiment of the present invention, the method comprising the following steps: Step S001: Collect point cloud data of the cutter head area and the tunnel face area of the shield machine in several monitoring frames.
[0022] It should be noted that in the process of real-time detection of the shield machine posture, the existing technology usually uses the PL-ICP algorithm to match the point cloud data. The point cloud data of some actual defect areas on the shield machine will interfere with the distance relationship between different data points in the point cloud data, causing some data points to match incorrectly, reducing the credibility of the matching results, and reducing the efficiency of posture detection.
[0023] Specifically, it is necessary to first collect the point cloud data and several monitoring frames of each shield monitoring area. The specific process is: the shield cutter area and the face area of the shield machine are both used as the shield machine monitoring area; the point cloud data of each shield machine monitoring area is obtained through three-dimensional laser, and each data point in the point cloud data of each shield monitoring area is recorded as a shield image point; taking any shield machine monitoring area as an example, in the shield machine historical monitoring frame database, every 5 seconds is taken as a monitoring frame, and several monitoring frames of the shield machine monitoring area in the past day are obtained; and several monitoring frames of each shield machine monitoring area are obtained. The time interval for recording the monitoring frames and the total time length for recording the monitoring frames can be determined according to the specific implementation situation.
[0024] It should be noted that the cutterhead area of the shield machine is mainly composed of the main beam, auxiliary beam, roller cutter, scraper, side scraper and super-excavation cutter; the face area is mainly composed of the working surface area when the shield machine is advancing; the point cloud data of each shield monitoring area contains multiple data points, each data point corresponds to a data point with three-dimensional spatial position information in each monitoring frame, and the three-dimensional spatial position corresponding to each data point in different monitoring frames is not always the same.
[0025] So far, the point cloud data of the cutter head area and the tunnel face area of the shield machine in several monitoring frames are obtained through the above method.
[0026] Step S002: according to the overall movement of the position of the same shield image point in different monitoring frames, the structural confidence of each shield image point in each monitoring frame is obtained.
[0027] It should be noted that the image data related to the shield machine's operating environment mainly includes the point cloud data of the tool assembly structure in the shield machine's cutter head area and the point cloud data of the corresponding soil structure in the face area. The positions of different data points corresponding to the point cloud data will be affected by environmental factors such as vibration and the soil structure of the face in different operating segments, resulting in difficulty in judging the degree of wear on different data point positions in each shield monitoring area. When the position of the corresponding data point of the point cloud data changes within multiple monitoring frames, the position of the data point caused by wear will slowly change over time. Therefore, the structural confidence of each data point can be obtained in combination with the change characteristics of the structural points in the shield monitoring area, so as to achieve accurate cleaning of the data points and complete accurate visualization of the shield machine's tool and operating environment.
[0028] Preferably, in one embodiment of the present invention, according to the overall movement of the position of the same shield image point in different monitoring frames, the structural confidence of each shield image point in each monitoring frame is obtained, and the specific method includes: Take any shield image point in the cutterhead area of the shield machine as an example, and The data point in the frame monitoring frame is consistent with the shield image point in the The Euclidean distance between the data points in the frame monitoring frame is recorded as the first distance; the shield image point is The data point in the frame monitoring frame is consistent with the shield image point in the The Euclidean distance between the data points in the frame monitoring frame is recorded as the second distance; the absolute value of the difference between the first distance and the second distance is recorded as the distance between the shield image point in the first The structural confidence factor in the frame monitoring frame is linearly normalized for the structural confidence factor of the shield image point in the cutter head area of the shield machine in all monitoring frames, and the normalized structural confidence factor is recorded as the structural confidence coefficient.
[0029] Further, any shield image point in the tunnel face area is recorded as the target tunnel face image point; The acquisition method of the structural confidence factor in the frame monitoring frame is to obtain the target face image point in the first The structural confidence factor in the frame monitoring frame is calculated by The cumulative sum of the structural confidence factors in the frame monitoring frame is recorded as the tunnel face area in the first The tunnel face weight of the frame monitoring frame; the tunnel face weight of the tunnel face area in all monitoring frames is linearly normalized, and the normalized weight of each tunnel face is recorded as the tunnel face weight coefficient.
[0030] Further, taking any shield image point in the cutterhead area of the shield machine as an example, the shield image point is The structural confidence factor in the frame monitoring frame is related to the tunnel face area in the first The product of the weight coefficients of the tunnel face of the frame monitoring frame is recorded as the shield image point at the Frame Monitoring Confidence of structures in frames.
[0031] It is particularly noted that, in this embodiment, by default, the Euclidean distance between the data point of the shield image point in the first monitoring frame and the data point of the shield image point in the second monitoring frame is used as the structural confidence factor of the shield image point in the first monitoring frame; the Euclidean distance between the data point of the shield image point in the last monitoring frame and the data point of the shield image point in the second to last monitoring frame is used as the structural confidence factor of the shield image point in the last monitoring frame.
[0032] So far, the structural confidence of each shield image point in each monitoring frame is obtained through the above method.
[0033] Step S003: According to the change size of the position information of different shield image points in the same monitoring frame and the structural confidence, the consistency of the position change of each shield image point in the cutter head area of the shield machine in each monitoring frame is obtained; according to the structured difference of the consistency of the position change between different data points of each shield image point in each monitoring frame, the necessity of adjusting the distance error between any two data points of each shield image point in each monitoring frame is obtained; according to the necessity of adjusting the distance error between different data points, the distance metric between different data points is adjusted to obtain an adaptive distance metric between any two data points.
[0034] It should be noted that the structural confidence of each shield image point in each monitoring frame can be used to judge the running posture of the shield machine, but the actual running posture of the shield machine is greatly affected by the working environment. When environmental factors such as rainwater deposition occur, the soil structure changes, and the running posture of the shield machine has a trend of slow settlement or offset. Therefore, according to the structural change characteristics of each shield image point in each monitoring frame, the running posture of the shield machine can be accurately judged. And because the data cleaning process uses the matching of the corresponding position neighborhood points, only the more obvious abnormal point cloud data of the shield machine position due to environmental factors can be removed for cleaning. The data points in the actual defective area will also affect the judgment of the running posture of the shield machine. Therefore, it is necessary to analyze the running posture from the structural offset trend of all point cloud data in the shield machine area, and reduce the influence of the defective position point cloud data through the refined registration of the PL-ICP algorithm to achieve accurate judgment of the running posture of the shield machine.
[0035] Preferably, in one embodiment of the present invention, according to the change of the position information of different shield image points in the same monitoring frame, the position change consistency of each shield image point in the cutter head area of the shield machine in each monitoring frame is obtained, and the specific method includes: Taking any monitoring frame of the shield machine cutter disc area as an example, the mean value of the structural confidence of all shield image points in the shield machine cutter disc area in the monitoring frame is recorded as the structural confidence coefficient of the monitoring frame, and the structural confidence coefficients of all monitoring frames are obtained; the absolute value of the difference between the structural confidence coefficients of different monitoring frames is used as the distance metric, and all monitoring frames are hierarchically clustered according to the distance metric to obtain several clusters; taking any cluster as an example, the least squares method is used to fit the data points of all shield image points in the shield machine cutter disc area in all monitoring frames in the cluster to obtain the fitting straight line of the shield machine cutter disc area in the cluster; the vector formed by the fitting straight line is recorded as the offset vector. The process of obtaining the fitting straight line according to the data points is a well-known content of the least squares method, which will not be repeated in this embodiment.
[0036] Furthermore, the first The shield image point is The difference between the structural confidence of the frame monitoring frame and other monitoring frames, as well as the difference between the offset vectors of adjacent monitoring frames, is used to obtain the first The shield image point is The position change consistency of the frame monitoring frame. As an example, the first The shield image point is Frame monitoring frame position change consistency: In the formula, Indicates the shield machine cutterhead area The shield image point is Frame monitoring: consistency of position changes between frames; Represents the number of all clusters in the cutterhead area of the shield machine; Indicates the shield machine cutterhead area The offset vector of the clusters; Indicates the shield machine cutterhead area The offset vector of the clusters; Represents the preset denominator hyperparameter. In this embodiment, , used to prevent the denominator from being 0; Indicates the shield machine cutterhead area The shield image point is Frame monitoring frame structure confidence; Indicates the shield machine cutterhead area The mean value of the structural confidence of each shield image point in all monitoring frames; represents an exponential function with a natural constant as the base, and the embodiment adopts Model to present inverse proportional relationship and normalization processing, As the input of the model, the implementer can choose the inverse proportional function and normalization function according to the actual situation; Indicates taking the absolute value.
[0037] It should be noted that Indicates the difference in the position changes of all shield image points in all monitoring frames; if the first The shield image point is The greater the consistency of the position change of the frame monitoring frame, the more obvious the trend of slow settlement or deviation of the running posture of the shield machine is.
[0038] Preferably, in one embodiment of the present invention, the necessity of adjusting the distance error between any two data points of each shield image point in each monitoring frame is obtained according to the difference in consistency of position changes between different data points of each shield image point in each monitoring frame, and the specific method includes: For all shield image points in the shield machine cutterhead area, Frame monitoring frame and the The data points in the frame monitoring frame are clustered by density to obtain several clusters; each cluster is recorded as a monitoring cluster; any shield image point in the shield machine cutter head area in any frame monitoring frame is recorded as a target data point, and a reference range is preset. , wherein this embodiment is based on This example is described as an example, and this embodiment is not specifically limited. It can be determined according to the specific implementation situation; the target data point is the center and the radius is The area is recorded as the neighborhood monitoring area of the target data point; the monitoring frames included in different monitoring clusters are different; in addition, the density clustering algorithm is a well-known technology and will not be repeated in this embodiment.
[0039] Furthermore, any data point in the neighborhood monitoring area of the target data point is recorded as a monitoring reference data point; according to the difference in the consistency of the position change between the target data point and the monitoring cluster to which the monitoring reference data point belongs, the necessary factor for adjusting the distance error between the target data point and the monitoring reference data point is obtained. As an example, the necessary factor for adjusting the distance error between the target data point and the monitoring reference data point can be calculated by the following formula: In the formula, Indicates the necessary factor for adjusting the distance error between the target data point and the monitoring reference data point; Indicates the consistency of position change of target data points; Indicates the consistency of position change of monitoring reference data points; Represents the mean of the consistency of position changes of all data points in the monitoring cluster to which the monitoring reference data point belongs; Represents the Euclidean distance between the target data point and the monitoring reference data point; Represents the preset denominator reference hyperparameter. In this embodiment, , used to prevent the denominator from being 0.
[0040] It should be noted that the larger the necessary factor for adjusting the distance error between the target data point and the monitoring reference data point is, the more significant the structural change between the target data point and the monitoring reference data point is, which reflects that the structural change between the target data point and the monitoring reference data point is more likely to be caused by the wear area of the shield machine or other environmental factors.
[0041] Furthermore, necessary factors for distance error adjustment between all data points in the neighborhood monitoring area of the target data point and the target data point are obtained, all necessary factors for distance error adjustment are linearly normalized, and each normalized necessary factor for distance error adjustment is recorded as the necessity for distance error adjustment.
[0042] Preferably, in one embodiment of the present invention, according to the necessity of adjusting the distance error between different data points, the distance metric between different data points is adjusted to obtain an adaptive distance metric between any two data points, including the specific method of: Preset a distance error adjustment necessity threshold , wherein this embodiment is based on This example is described as an example, and this embodiment is not specifically limited. It can be determined according to the specific implementation situation; taking the data points of any two shield image points in the shield machine cutter head area in any frame of monitoring as an example, if the distance error between the two data points is greater than the necessity of adjusting , the data point pair formed by these two data points is recorded as the data point pair to be adjusted; if the distance error between the two data points is less than or equal to the necessity of adjustment , record the data point pair formed by these two data points as a fixed data point pair; obtain all data point pairs to be adjusted and fixed data point pairs.
[0043] Furthermore, taking two data points in any pair of data points to be adjusted as an example, the product of the Euclidean distance between the two data points and the necessity of distance error adjustment is recorded as the adaptive distance metric of the two data points; the adaptive distance metric of the two data points in each pair of data points to be adjusted is obtained. The Euclidean distance of the two data points in each fixed data point pair is used as the adaptive distance metric of the two data points in each fixed data point pair.
[0044] So far, the adaptive distance metric between any two data points is obtained through the above method.
[0045] Step S004: Perform matching detection based on the adaptive distance metric between any two data points.
[0046] Preferably, in one embodiment of the present invention, matching detection is performed based on an adaptive distance metric between any two data points, and the specific method includes: The adaptive distance metric between any two different data points is used as the distance metric, and the point cloud data matching result is obtained according to the distance metric; posture detection is performed according to the point cloud data matching result. The process of obtaining the point cloud data matching result according to the distance metric is the well-known content of the PL-ICP (Point-to-line Iterative Closest Point) point-to-line iterative closest point algorithm, which will not be repeated in this embodiment; posture detection according to the point cloud data matching result is the content disclosed in the "Study on Autonomous Deviation Correction and Parameters of Shield Machine Learning Based on Machine Learning" published by authors Zhang Jun and Li Maopeng in the journal Chinese and Foreign Highways in 2024.
[0047] At this point, this embodiment is completed.
[0048] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.
Claims
1. A high-precision tunnel boring machine operation posture recognition method, characterized in that: The method comprises the following steps: Collecting point cloud data of the cutter head area and the tunnel face area of the shield machine in several monitoring frames, wherein the point cloud data includes a plurality of shield image points with three-dimensional spatial position information, and each shield image point corresponds to a data point in each monitoring frame; The cutterhead area and the tunnel face area of the shield machine are both recorded as a shield monitoring area. According to the overall movement of the position of the same shield image point in different monitoring frames, the structural confidence of each shield image point in each monitoring frame is obtained. According to the change of position information of different shield image points in the same monitoring frame and the structural confidence, the consistency of position change of each shield image point in the cutter head area of the shield machine in each monitoring frame is obtained; according to the structural difference of the consistency of position change between different data points of each shield image point in each monitoring frame, the necessity of adjusting the distance error between any two data points of each shield image point in each monitoring frame is obtained; according to the necessity of adjusting the distance error between different data points, the distance metric between different data points is adjusted to obtain the adaptive distance metric between any two data points; Match detection based on an adaptive distance metric between any two data points; The method of obtaining the structural confidence of each shield image point in each monitoring frame according to the overall movement of the position of the same shield image point in different monitoring frames includes the following specific methods: Get the tunnel face area in The weight coefficient of the tunnel face of the frame monitoring frame and the weight coefficient of each shield image point in the first Confidence coefficient of structure in frame monitoring frame; For any shield image point in the cutterhead area of the shield machine, The structural confidence factor in the frame monitoring frame is related to the tunnel face area in the first The product of the weight coefficients of the tunnel face of the frame monitoring frame is recorded as the shield image point at the Frame monitoring: Structural confidence in frames; The tunnel face area is obtained in the The weight coefficient of the tunnel face of the frame monitoring frame and the weight coefficient of each shield image point in the first The structure confidence coefficient in the frame monitoring frame includes the following specific methods: For any shield image point in the cutterhead area of the shield machine, The data point in the frame monitoring frame is consistent with the shield image point in the The Euclidean distance between the data points in the frame monitoring frame is recorded as the first distance; the shield image point is recorded as the The data point in the frame monitoring frame is consistent with the shield image point in the The Euclidean distance between the data points in the frame monitoring frame is recorded as the second distance; the absolute value of the difference between the first distance and the second distance is recorded as the shield image point in the first The structural confidence factor in the frame monitoring frame; the structural confidence factor of the shield image point in all monitoring frames is linearly normalized, and the normalized structural confidence factor is recorded as the structural confidence coefficient; Any shield image point in the tunnel face area is recorded as the target tunnel face image point; the reference shield image point is recorded in the The acquisition method of the structural confidence factor in the frame monitoring frame is to obtain the target face image point in the first The structural confidence factor in the frame monitoring frame is calculated by The cumulative sum of the structural confidence factors in the frame monitoring frame is recorded as the tunnel face area in the first The face weight of each monitoring frame is calculated; the face weight of the face area in all monitoring frames is linearly normalized, and each face weight after normalization is recorded as the face weight coefficient; The method of obtaining the consistency of the position change of each shield image point in the cutter head area of the shield machine in each monitoring frame according to the change of the position information of different shield image points in the same monitoring frame and the structural confidence includes the following specific methods: Obtain the offset vector of each monitoring frame of the shield machine cutterhead area according to the structural confidence; According to the shield machine cutter head area The shield image point is The difference between the structural confidence of the frame monitoring frame and other monitoring frames, as well as the difference between the offset vectors of adjacent monitoring frames, is used to obtain the first The shield image point is The frame monitors the consistency of the position change of the frame, including the specific methods: In the formula, Indicates the shield machine cutterhead area The shield image point is Frame monitoring frame position change consistency; Represents the number of all clusters in the cutterhead area of the shield machine; Indicates the shield machine cutterhead area The offset vector of the clusters; Indicates the shield machine cutterhead area The offset vector of the clusters; represents the preset denominator hyperparameter; Indicates the shield machine cutterhead area The shield image point is Frame monitoring frame structure confidence; Indicates the shield machine cutterhead area The mean value of the structural confidence of each shield image point in all monitoring frames; represents an exponential function with a natural constant as base; Indicates taking the absolute value; The specific method of obtaining the offset vector of each monitoring frame of the cutter head area of the shield machine according to the structural confidence is as follows: For any monitoring frame in the cutter head area of the shield machine, the mean value of the structural confidence of all shield image points in the cutter head area of the shield machine in the monitoring frame is recorded as the structural confidence coefficient of the monitoring frame, and the structural confidence coefficients of all monitoring frames are obtained; the absolute value of the difference between the structural confidence coefficients of different monitoring frames is used as the distance metric, and all monitoring frames are hierarchically clustered according to the distance metric to obtain several cluster clusters; for any cluster cluster, the least squares method is used to fit the data points of all shield image points in the cutter head area of the shield machine in all monitoring frames in the cluster cluster to obtain the fitting straight line of the cutter head area of the shield machine in the cluster cluster; the vector formed by the fitting straight line is recorded as the offset vector; The necessity of adjusting the distance error between any two data points of each shield image point in each monitoring frame is obtained based on the structural difference in the consistency of the position change between different data points of each shield image point in each monitoring frame, and the specific method includes: For all shield image points in the shield machine cutterhead area, Frame monitoring frame and the The data points in the frame monitoring frame are clustered by density to obtain several clusters; each cluster is recorded as a monitoring cluster; any shield image point in the shield machine cutter head area in any frame monitoring frame is recorded as a target data point, and a reference range is preset. ; Take the data point as the center and the radius as The area is recorded as the neighborhood monitoring area of the target data point; Any data point in the neighborhood monitoring area of the target data point is recorded as a monitoring reference data point; according to the difference in the consistency of position changes between the target data point and the monitoring cluster to which the monitoring reference data point belongs, the necessary factor for adjusting the distance error between the target data point and the monitoring reference data point is obtained; Obtain necessary factors for distance error adjustment between all data points in the neighborhood monitoring area of the target data point and the target data point, perform linear normalization on all necessary factors for distance error adjustment, and record each necessary factor for distance error adjustment after normalization as necessity for distance error adjustment; The necessary factor for adjusting the distance error between the target data point and the monitoring reference data point is obtained according to the difference in the consistency of the position change between the target data point and the monitoring cluster to which the monitoring reference data point belongs, and the specific method includes: In the formula, Indicates the necessary factor for adjusting the distance error between the target data point and the monitoring reference data point; Indicates the consistency of position change of target data points; Indicates the consistency of position change of monitoring reference data points; Represents the mean of the consistency of position changes of all data points in the monitoring cluster to which the monitoring reference data point belongs; Represents the Euclidean distance between the target data point and the monitoring reference data point; represents the preset denominator reference hyperparameter; The distance metric between different data points is adjusted according to the necessity of adjusting the distance error between different data points to obtain an adaptive distance metric between any two data points, including the specific method of: Obtain all pairs of data points to be adjusted and fixed pairs of data points according to the necessity of distance error adjustment; For any two data points in a pair of data points to be adjusted, the product of the Euclidean distance between the two data points and the necessity of distance error adjustment is recorded as the adaptive distance metric of the two data points; the adaptive distance metric of the two data points in each pair of data points to be adjusted is obtained; and the Euclidean distance of the two data points in each fixed data point pair is used as the adaptive distance metric of the two data points in each fixed data point pair.
2. The method for identifying the running posture of a high-precision tunnel boring machine according to claim 1, characterized in that: The specific method of obtaining all the data point pairs to be adjusted and the fixed data point pairs according to the necessity of adjusting the distance error includes: Preset a distance error adjustment necessity threshold For any two shield image points in the shield machine cutter head area in any monitoring frame, if the distance error between the two data points is greater than , the data point pair formed by these two data points is recorded as the data point pair to be adjusted; if the distance error between the two data points is less than or equal to the necessity of adjustment , the data point pair formed by these two data points is recorded as a fixed data point pair.
3. The method for identifying the running posture of a high-precision tunnel boring machine according to claim 1, characterized in that: The specific method of performing matching detection based on the adaptive distance metric between any two data points includes: The adaptive distance metric between any two different data points is used as the distance metric, and the point cloud data matching result is obtained by matching through the PL-ICP algorithm according to the distance metric.
Citation Information
Patent Citations
Method of shielding rectification based on statistical analysis and XGboost
CN108868805A
Comprehensive geological forecasting method for shield tunneling machine
CN112558178A
Intelligent human shape recognition alarm system and method based on monitoring camera
CN117809379A
Real-time road condition monitoring method based on target recognition
CN118262521A
Registration of 3D point cloud data by creation of filtered density images
US20090232388A1
