Shield machine segment assembly quality deviation detection method based on laser scanning

Abstract

The present invention discloses a method for detecting the quality deviation of the assembly of shield machine segments based on laser scanning, which includes the following steps: 1) deploying multiple station equipment for scanning to obtain three-dimensional point cloud data of the shield tunnel structure; 2) denoising the three-dimensional point cloud data; 3) registering and splicing the denoised three-dimensional point cloud data; 4) adjusting the coordinate system of the three-dimensional point cloud data so that the central axis of the adjusted three-dimensional point cloud data is parallel to or coincides with a coordinate axis in the world coordinate system; 5) after the spatial position of the three-dimensional point cloud data is adjusted, a non-average segmentation algorithm is then used to segment and extract the segment data. The technical solution of the present invention solves the problem of the insufficiency of traditional segmentation methods for the straight-line point cloud data of the tunnel by using a non-average segmentation method along the axial direction. The segmentation method can realize the efficient and rapid segmentation of the segment data and the annular seam data in the tunnel point cloud, and quickly distinguish and classify the segmented data according to the characteristics of the segmented data.

Description

Technical Field

[0001] The invention relates to a method for detecting the assembly quality deviation of shield machine segments based on laser scanning, and belongs to the field of tunnel structure safety technology detection. Background Art

[0002] The scale of subway construction is large, and the proportion of shield method application is high. Shield method construction has the advantages of small disturbance to the surrounding environment, high safety, high degree of automation, fast construction speed, and no influence of climate and traffic. The construction principle of shield method is that the shield machine uses the lining blocks installed at the tail as a fulcrum to move forward, cuts the soil with a cutter disc, and assembles the segments and grouts them. In this process, the shield segment is one of the force-bearing control components, and its construction quality will directly affect the overall quality of the tunnel. The relevant standards stipulate that misalignment, ovality and center axis deviation are the three main indicators of segment assembly quality.

[0003] There are different detection methods for misalignment, ellipticity and central axis deviation. In terms of misalignment detection, existing technical personnel such as Ding Lieyun and others have proposed a shield segment assembly quality detection system and method based on image recognition. By attaching markers to the detection object, collecting image information of the detection object, and using 3Dmax and other software for image processing, the segment assembly misalignment value can be identified and the processing results can be displayed. Existing technical personnel such as Zhao Yabo and Wang Zhi use mobile three-dimensional scanning technology to scan tunnel data and process and detect tunnel misalignment values. Analysis of the processing results shows that this method has a large amount of detection data and high detection efficiency. In the existing technology, augmented reality technology is also used to realize the rapid detection of misalignment during tunnel construction. This technology automatically evaluates structural safety by establishing a virtual quality control baseline model and covering it on the real pipe segment in AR to detect misalignment. Feedback can be used to explore improving the stability of the tunnel.

[0004] In terms of ovality detection, existing technicians have proposed a tunnel segment ovality detection technology based on a new mobile 3D laser scanning system based on the analysis of the insufficient detection of tunnel ovality by plumb bob method, total station section method and laser section instrument method. This technology has comprehensive detection information, high operating efficiency and improved the reliability of ovality detection. Existing technicians in this field, such as Meng Qingnian, compared and analyzed the ovality detection results of the mobile 3D laser scanning system and the total station, and believed that the detection results of the two were similar, but the mobile 3D laser scanning system had higher operating efficiency.

[0005] In terms of central axis detection, the existing technology transforms the tunnel point cloud projection into planar binary image data, and uses the skeleton extraction algorithm in image processing technology to extract the tunnel central axis. Cheng Yunjian and other technicians in this field horizontally project the tunnel point cloud, extract boundary points, use the random sampling consistency algorithm to extract the inner points in the boundary points for multi-model fitting, and calculate the tunnel centerline based on the global optimization algorithm of random sampling again. Existing technicians in this field, such as Fekete and Diederichs, have proposed a tunnel central axis extraction method based on cylindrical surface fitting.

[0006] In the prior art, Chinese patent publication number CN112504062A discloses a device for detecting shield segments, but the device can only detect the misalignment value of the shield segments, but cannot detect the ovality and axis deviation. Moreover, the device requires technicians to move on the segments when realizing the continuity detection of the shield segments, and the misalignment value detected is limited to the place where the device is placed, and it is not possible to better detect the misalignment value of the entire segment range.

[0007] Another prior art, Chinese patent publication number CN113150692A discloses a piezoelectric coating for shield tunnel segment detection. The piezoelectric material is mainly used to detect the health of the shield tunnel segments and determine the damage status of the shield tunnel segments. It is mainly for status detection after the segments are assembled or during tunnel operation. Moreover, the material needs to be applied to the segments in advance. If the material is applied to each segment in advance, the construction cost will increase and the construction time will be extended. However, if the material is only applied to some segments, the detection data will be incomplete.

[0008] Similarly, traditional manual inspection methods have problems such as low measurement accuracy, insufficient informationization, difficulty in efficiently and comprehensively inspecting tunnel segments, and poor objectivity of inspection data.

[0009] In view of this, it is obviously necessary to develop a new method for detecting quality deviation of shield machine segment assembly based on laser scanning to solve the above-mentioned defects. Summary of the invention

[0010] The purpose of the invention is to provide a method for detecting quality deviation of shield machine segment assembly based on laser scanning, so as to solve the defects of the prior art such as long detection time, high cost of manual detection, incomplete detection data and large error.

[0011] In order to achieve the above-mentioned invention object, the technical solution adopted by the present invention is: a method for detecting the assembly quality deviation of shield machine segments based on laser scanning, comprising the following steps:

[0012] 1) Deploy multiple site equipment for scanning to obtain three-dimensional point cloud data of the shield tunnel structure;

[0013] 2) De-noising the 3D point cloud data;

[0014] 3) Register and stitch the denoised 3D point cloud data;

[0015] 4) Adjust the coordinate system of the 3D point cloud data: First, the central axis of the 3D point cloud data is obtained by the cylindrical fitting method, and then the rotation matrix of the central axis to a certain coordinate system is calculated. The rotation matrix makes the central axis of the adjusted 3D point cloud data parallel or coincident with any axis in the world coordinate system;

[0016] 5) After the spatial position of the 3D point cloud data is adjusted, a non-average segmentation algorithm is used to segment and extract the annular seam and annular segment data;

[0017] 6) Use the least squares plane circle fitting algorithm to eliminate invalid data in the three-dimensional point cloud data, planarize the three-dimensional point cloud data, obtain the tunnel fitting circle equation, improve the calculation speed, and then calculate the data planarized by the least squares method by finding the major and minor semi-axis method to obtain the detected ellipticity;

[0018] 7) Perform annular average segmentation on each annular seam and segment data according to the slope annular segmentation algorithm, perform inter-annular calculation on the segmented data, and obtain the detected misalignment value;

[0019] 8) Calculate the tunnel center point by extracting the annular seam data and the annular segment data, calculate the deviation value of the center points of adjacent segments, and then fit the central axis of the point cloud data of the entire tunnel to obtain the complete center deviation of the entire tunnel.

[0020] In a further technical solution, the denoising process in step 2) is based on Cloud Compare software, specifically including selecting and deleting noise points that are not related to the scanned object and directly removing noise through the software's own filtering and denoising function to process the collected data to obtain denoised three-dimensional point cloud data.

[0021] In a further technical solution, the registration and stitching in step 3) is based on Cloud Compare software, specifically using the ICP rigid body registration principle, and the steps are to load the two sets of point clouds to be registered in the Cloud Compare point cloud software, and display them in different colors to distinguish them, then select the ICP registration method, set the parameter conditions, the parameter conditions include the number of iterations, threshold, overlap degree, and then complete the post-registration.

[0022] In a further technical solution, in step 5), a non-average segmentation algorithm is used to separate different structural objects in the three-dimensional point cloud data, and the steps are as follows: Assume that the pipe segment width is l1, and the data starting position z value is 0. Since the data starting position is the first annular seam between the first ring pipe segment and the second ring pipe segment, the i-th joint position is:

[0023] z i =(i-1)×l1

[0024] Then, two types of segmentation points are set on the left and right sides of the segment annular seam:

[0025] Type ① segmentation point:

[0026] z i1 =(i-1)×l1+s

[0027] The second type of segmentation point:

[0028] z i2 =il1-s

[0029] In the above formula: s is a constant, which is half of the width of the data at the extraction seam.

[0030] After obtaining the first and second types of segmentation points, the two types of segmentation points are sequentially spaced and arranged from small to large on the Z axis of the data. The two types of segmentation points are arranged from small to large to form a set K:

[0031] K={(i-1)l1+s,il1-s},i=1,2,…,n

[0032] The data in K are combined into pairs of data for segmentation in sequence, so that the tunnel point cloud data can be segmented non-evenly.

[0033] In a further technical solution, in step 6), the ellipticity of the detected data is calculated by finding the major and minor semi-axis method through the least squares planarization method, and the steps are as follows: assuming q i (x i ,y i ) is any point in the point cloud data, point q i The distance from the fitting circle center O(X,Y) is R i ,but:

[0034]

[0035] R i Maximum R max With the minimum value R min Solve it, that is:

[0036]

[0037] Maximum R max With the minimum value R min are considered as the semi-major and semi-minor axes of the tunnel ellipse, then the major and minor axes of the ellipse are:

[0038] D max =2×R max

[0039] D min =2×R min

[0040] According to the theoretical formula of ellipticity, the tunnel ellipticity can be obtained as:

[0041]

[0042] In a further technical solution, the specific steps of step 7) are as follows:

[0043] Suppose a point cloud after the average segmentation is p, and any point p in p i The two-dimensional distance to the origin of the XY plane of the fitted circle, that is:

[0044]

[0045] Where: d i For p i Distance to the origin of the XY plane, x i ,y i They are point p i The horizontal and vertical coordinates of

[0046] According to the above formula, the data corresponding to the peak point with the maximum two-dimensional distance is obtained, and the z value data of the point corresponding to the distance is extracted, which is the z value corresponding to the peak value:

[0047]

[0048] z=z i

[0049] Where: d max is the maximum distance from the midpoint data of p to the origin of the XY plane, z i The z value of the point corresponding to the maximum distance is extracted, and the maximum and minimum z value data of p in the point cloud are recorded as z max , z min , then there must be z max >z>z min Relationship, create set O:

[0050]

[0051] Extract the data in the set O in pairs to perform Z-axis segmentation on the point cloud p, and finally divide the data into four segments. The front and back segments are the point cloud data of parts A and B. After extracting the point cloud data of parts A and B, the misalignment value can be detected. The average two-dimensional distance of the point cloud data of parts A and B to the origin in the XY plane is calculated respectively.

[0052]

[0053]

[0054] Where: d A , d B are the average distances between parts A and B and the origin in the XY plane, and d is calculated. A , d B The difference between the values ​​of can be used to get the misalignment value at that location, namely:

[0055] △=|d A -d B |

[0056] Where: Δ is the misalignment value calculated at this location.

[0057] In a further technical solution, the specific calculation steps of step 8) are as follows:

[0058] Set point cloud p i is the maximum two-dimensional distance to the origin of the XY plane of the fitting circle. This point is located in the annular seam data. In order to retain as much annular seam data as possible, the distance threshold δ is set, that is:

[0059] |d i -d j |≤δ

[0060] Where: d i is the maximum distance to the origin, d j is the distance from any point to the origin,

[0061] When the absolute value of the difference between the distance from any point to the origin and the maximum distance is less than the threshold, the data point is considered to be located in the annular seam data. After the annular seam is extracted, the least squares cylinder fitting is performed on the data to obtain the center point of the annular seam data.

[0062] The steps for extracting the center point data between rings are as follows:

[0063] Extract the middle part of the ring segment data to calculate the center point, and extract the maximum and minimum z value data of the ring segment, which are z max , z min , calculate the z value of the midpoint data and set the distance threshold δ to calculate and extract the data in the ring, that is:

[0064]

[0065] |z i -z mind |≤δ

[0066] Where: z mid is the midpoint data z value, z i is the z value of the data at any point. After extracting the data in the ring, the least squares cylindrical fitting is performed on the data to obtain the center point of the ring segment data.

[0067] Then the center point deviation is calculated, and the steps are as follows:

[0068] The center point of a ring circle fitting point cloud is p k (x k ,y k ), the center point of the adjacent circle fitting point cloud is p k+1 (x k+1 ,y k+1 ), the deviation of the center points of any adjacent ring segments is solved as follows:

[0069]

[0070] Where: d is the deviation of the center points of adjacent ring segments.

[0071] Due to the application of the above technical solution, the present invention has the following advantages compared with the prior art:

[0072] 1. Compared with traditional manual detection, the technical solution of the present invention reduces the detection time by more than 70%, and the segment data are collected during the period when the shield machine is shut down after the shield machine tunnel is excavated and the segments are assembled into a ring. The efficient collection of segment data can prevent the situation where the collection of data affects the normal operation of the shield machine and thus affects the construction progress to the greatest extent.

[0073] 2. The detection cost of the technical solution of the present invention is lower than that of manual detection. Compared with manual detection, machine detection only requires the initial investment in machinery and equipment and the development of the detection system, and then regular maintenance of the relevant equipment is required, without the need to pay additional other detection fees.

[0074] 3. The technical solution of the present invention can also be used for detection in some relatively harsh environments where human vision cannot meet the requirements, and maintain relatively accurate detection results, with a higher cost-effectiveness.

[0075] 4. Under the same detection time, the technical solution of the present invention can obtain and process a large amount of data, greatly improving the detection efficiency. During manual detection, the detection results will be affected by factors such as subjectivity and physical energy, and the detection efficiency will decrease with the increase of the number of detections and the detection time. The present invention solves the problem of reduced efficiency very well. It only needs to regularly maintain the detection equipment to achieve long-term and efficient detection.

[0076] 5. The technical solution of the present invention solves the shortcomings of traditional segmentation methods for tunnel linear point cloud data through a non-uniform segmentation method along the axial direction. This segmentation method can achieve efficient and rapid segmentation of ring segment data and annular seam data in the tunnel point cloud, and quickly distinguish and classify the data based on the characteristics of the segmented data.

[0077] 6. The technical solution of the present invention calculates the ellipticity by finding the major and minor semi-axes of the ellipse. The calculation result is more rigorous than the ideal state algorithm of the plumb bob method. By improving the algorithm for segmentation by slope, the annular point cloud data can be quickly and accurately segmented. The central axis of the point cloud data of the entire tunnel can be fitted based on the calculated center point and central axis of the annular segment and annular seam data, and the complete center deviation of the entire tunnel can be obtained. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] Figure 1 This is a comparison diagram of point cloud data denoising.

[0079] Figure 2 It is the point cloud coordinate adjustment diagram.

[0080] Figure 3 This is the result of non-even segmentation of point cloud data.

[0081] Figure 4 It is a three-dimensional visualization digital twin model of the fitting curve.

[0082] Figure 5 It is a three-dimensional model of the fitting axis tunnel. DETAILED DESCRIPTION

[0083] The present invention will be further described below in conjunction with the accompanying drawings and embodiments:

[0084] Embodiment 1: In this example, the site data collection device is a Leica P40 3D laser scanner. The instrument has an accuracy of up to 50m and a detection error of less than 3mm.

[0085] The steps are as follows: The initial sampling data contains a lot of irrelevant data other than the tunnel point cloud, which is interference data. For this, this embodiment uses Cloud Compare software to remove the interference data. Figure 1 As shown, it is a schematic comparison diagram before and after noise removal, and the boxes before removal are irrelevant point data.

[0086] After the irrelevant data is removed, the spatial coordinates of the point cloud data obtained are still the internal coordinates of the 3D laser scanner. To facilitate subsequent processing, this embodiment adjusts the coordinates of the point cloud data so that the axis direction of the tunnel data center is parallel to or coincides with a coordinate axis of the world coordinate system after adjustment. Figure 2 The coordinate system on the left side of the figure is not conducive to the subsequent algorithm processing. Therefore, the point cloud data needs to be adjusted so that the axis direction of the tunnel data center is parallel to or coincides with a coordinate axis of the world coordinate system after adjustment. No matter which coordinate axis is parallel to or coincides with after adjustment, the coordinate adjustment is essentially the same. In this embodiment, the coordinate axis is coincident with the Z axis. Figure 2 As shown in the figure on the right.

[0087] After the spatial position of the point cloud data is adjusted, the segmented data of the whole data is segmented and extracted according to the improved non-average segmentation algorithm, and the different structural objects in the three-dimensional point cloud are separated to obtain the annular seam data to calculate the misalignment rate. Assuming the segment width is l1, the data starting position z value is 0, since the data starting position is the first annular seam between the first ring segment and the second ring segment, the i-th joint position is:

[0088] z i =(i-1)×l1

[0089] Then, two types of segmentation points are set on the left and right sides of the segment annular seam:

[0090] Type ① segmentation point:

[0091] z i1 =(i-1)×l1+s

[0092] The second type of segmentation point:

[0093] z i2 =il1-s

[0094] In the above formula: s is a constant, which is half of the width of the data at the extracted seam.

[0095] After obtaining the first and second types of segmentation points, the two types of segmentation points are sequentially spaced and arranged from small to large on the Z axis of the data. The two types of segmentation points are arranged from small to large to form a set K:

[0096] K={(i-1)l1+s,il1-s},i=1,2,…,n

[0097] The data in K are combined into pairs of data for segmentation, and the tunnel point cloud data can be segmented non-evenly. The schematic diagram of non-even segmentation is as follows: Figure 3 (a) shown.

[0098] In this example, the parameter s is set to 10 cm, that is, after the segmentation is completed, the width of each ring seam segment in the Z-axis direction is 20 cm, and the ring segment is 180 cm. The schematic diagram of the ring seam is as follows Figure 3 (c) as shown.

[0099] After obtaining the ring segment data, the ring segment schematic diagram is as follows: Figure 3 As shown in (b), it is projected and circle-fitted, and the point cloud data of the inner seam of the ring and some connecting components are eliminated. The ellipticity is calculated using the ideal state, the major and minor semi-axis method, and the ellipse fitting method. The results are shown in Table 1.

[0100] Table 1 Ellipticity calculation results

[0101]

[0102] Note: φ1 is the ideal state calculation result; φ2 is the long and short semi-axis calculation result; φ3 is the ellipse fitting result. It can be seen from the above table that the ellipticity calculated by the three methods (φ1φ2φ3) is within the range of the standard allowable value of 5‰. Among them, the calculation results of the ring length and short semi-axis methods are relatively close. The maximum calculation results of the three methods are within the allowable error value and the detection value is relatively stable. In addition, the ellipticity calculated by the long and short semi-axis methods will be larger. This method can increase the strictness of the detection and improve the construction quality.

[0103] In order to calculate the inter-ring misalignment, this method uses the improved slope segmentation algorithm to perform annular average segmentation on the fitting circle data at the annular gap between the segments. The steps are as follows: let the point cloud at a certain point after annular segmentation be p, and any point p in p i The two-dimensional distance to the origin of the XY plane of the fitted circle, that is:

[0104]

[0105] Where: d i For p i Distance to the origin of the XY plane, x i ,y i They are point p i The horizontal and vertical coordinates of .

[0106] According to the above formula, the maximum two-dimensional distance is obtained as the corresponding data of the peak point, and the z value data of the corresponding point under this distance is extracted, which is the z value corresponding to the peak:

[0107]

[0108] z=z i

[0109] Where: d max is the maximum distance from the midpoint data of p to the origin of the XY plane, zi is the z value of the point corresponding to the maximum distance.

[0110] Extract the maximum and minimum z value data in the point cloud and record it as z max , z min . Then there must be z max >z>z min Relationship, create set O:

[0111]

[0112] Extract the data in the set O in pairs to perform Z-axis segmentation on the point cloud p, and finally divide the data into four segments. The front and back segments are the point cloud data of parts A and B. After extracting the point cloud data of parts A and B, the misalignment value can be detected. Calculate the average two-dimensional distance of the point cloud data of parts A and B to the origin in the XY plane.

[0113]

[0114]

[0115] Where: d A , d B are the average distances between parts A and B and the origin in the XY plane. Calculate d A , d B The difference between the values ​​of can be used to get the misalignment value at that location, namely:

[0116] △=|d A -d B |

[0117] Where: Δ is the misalignment value calculated at this location.

[0118] The detected misalignment value was obtained and compared with the manual measurement value, as shown in Table 2.

[0119] Table 2 Calculation results of the maximum misalignment between rings

[0120]

[0121] Table 2 shows that most of the maximum misalignment values ​​calculated by the algorithm have certain errors with the manual measured values, with the maximum error being 1.26 mm. Only four points are selected for measurement in each ring for manual inspection, and the inspection locations are mostly at the bottom of the tunnel. However, this technical solution divides each section of the annular seam data into 360 parts, each with an angle of 1 degree. By dividing the data, it is easier to find the maximum misalignment value. In addition, there is a detection ring in the calculation result where the algorithm detection value is smaller than the manual measured value. The reason for this may be that the measurement error in the manual inspection makes the manual inspection result larger, while the actual misalignment of the ring may be close to the algorithm detection value, resulting in the manual measured value being larger than the algorithm inspection value. This analysis result also exposes the deficiency of large manual inspection errors.

[0122] Analysis of the results in Table 2 shows that under normal circumstances, regardless of algorithm detection or manual detection, the misalignment between segments exceeds 7 mm, which exceeds the standard allowable value for misalignment of highway shield tunnels. This phenomenon also often occurs in other engineering examples. Comprehensive calculation results can infer that the area between shield lining rings should be a frequent area where the misalignment value of shield tunnel segments exceeds the standard, which should attract attention in subsequent construction and be the focus of shield tunnel quality inspection and control.

[0123] The tunnel center point can be calculated based on the segmented annular seam data and segment data. i is the maximum two-dimensional distance to the origin of the XY plane of the fitting circle, and this point is located in the annular seam data. In order to retain as much annular seam data as possible, set the distance threshold δ, that is:

[0124] |d i -d j |≤δ

[0125] Where: d i is the maximum distance to the origin, d j is the distance from any point to the origin.

[0126] When the absolute value of the difference between the distance from any point to the origin and the maximum distance is less than the threshold, the data point is considered to be located in the annular seam data. After extracting the annular seam data, the least squares cylinder fitting is performed on the data to obtain the center point of the annular seam data.

[0127] The steps for extracting the center point data between rings are as follows:

[0128] Extract the middle part of the ring segment data to calculate the center point.

[0129] Extract the maximum and minimum z value data of the ring segment, which are z max , z min , calculate the z value of the midpoint data and set the distance threshold δ to calculate and extract the data in the ring, that is:

[0130]

[0131] |z i -z mind |≤δ

[0132] Where: z mid is the midpoint data z value, z i is the z value of the data at any point. After extracting the data in the ring, the least squares cylindrical fitting is performed on the data to obtain the center point of the ring segment data.

[0133] Then the center point deviation is calculated, and the steps are as follows:

[0134] The center point of a ring circle fitting point cloud is p k (x k ,y k ), the center point of the adjacent circle fitting point cloud is p k+1 (x k+1 ,y k+1 ), the deviation of the center points of any adjacent ring segments is solved as follows:

[0135]

[0136] Where: d is the deviation of the center points of adjacent ring segments.

[0137] In this example, the annular seam data and ring data of the supporting project are extracted respectively, and on this basis, the deviation calculation formulas of the tunnel center point and the adjacent ring center point are calculated respectively to calculate the deviation, as shown in Table 3.

[0138] Table 3 Calculation results based on the center points and deviations of the ring data

[0139]

[0140] From the calculation data in the above table, we can see that the deviation of the center point calculated based on the ring segment data is within the standard range of ±50mm, and the deviation of the numerical results is not large. Visual reference of the fitting axis and the design axis Figure 4 There are two curves in the tunnel segment, the upper one is the design axis and the lower one is the fitting axis. In this example, the center point is calculated using ring data, and the axis after fitting can also more realistically reflect the direction of the actual center axis.

[0141] The central axis reflects the direction of the actual centerline of the tunnel after the segments are assembled. It can be used to compare the error with the design axis and determine the actual axis, providing a reference for the axis control of subsequent shield construction. At the same time, the fitting axis can be used to create a BIM model for the construction tunnel and obtain a three-dimensional model of the actual state of the tunnel. This example project is east-west oriented, and the tunnel modeling after fitting the axis is as follows Figure 5 shown.

[0142] In summary, the present invention proposes a method for detecting the quality deviation of shield machine segment assembly based on laser scanning, which can realize the detection of three main indicators of misalignment, ovality and centerline deviation in the shield segment assembly quality inspection, so that the shield excavation quality is effectively guaranteed and the occurrence of tunnel safety hazards is greatly reduced.

Claims

1. A method for detecting the quality deviation of shield machine segment assembly based on laser scanning, characterized in that: The steps include: 1) Deploy multiple site equipment for scanning to obtain three-dimensional point cloud data of the shield tunnel structure; 2) De-noising the 3D point cloud data; 3) Register and stitch the denoised 3D point cloud data; 4) Adjust the coordinate system of the 3D point cloud data: First, the central axis of the 3D point cloud data is obtained by the cylindrical fitting method, and then the rotation matrix of the central axis to a certain coordinate system is calculated. The rotation matrix makes the central axis of the adjusted 3D point cloud data parallel or coincident with any axis in the world coordinate system; 5) After the spatial position of the 3D point cloud data is adjusted, a non-average segmentation algorithm is used to segment and extract the annular seam and annular segment data; 6) Use the least squares plane circle fitting algorithm to eliminate invalid data in the three-dimensional point cloud data, planarize the three-dimensional point cloud data, obtain the tunnel fitting circle equation, improve the calculation speed, and then calculate the data planarized by the least squares method by finding the major and minor semi-axis method to obtain the detected ellipticity; 7) Perform annular average segmentation on each annular seam and segment data according to the slope annular segmentation algorithm, perform inter-annular calculation on the segmented data, and obtain the detected misalignment value; 8) Calculate the tunnel center point by extracting the annular seam data and the annular segment data, calculate the deviation value of the center points of adjacent segments, and then fit the central axis of the point cloud data of the entire tunnel to obtain the complete center deviation of the entire tunnel.

2. The method for detecting the quality deviation of shield machine segment assembly based on laser scanning according to claim 1 is characterized in that: The denoising process in step 2) is based on Cloud Compare software, which specifically includes processing the collected data in a combination of two methods: selecting and deleting noise points that are not related to the scanned object and directly removing noise through the software's built-in filtering and denoising function to obtain denoised three-dimensional point cloud data.

3. The method for detecting the quality deviation of shield machine segment assembly based on laser scanning according to claim 1 is characterized in that: The registration and stitching in step 3) is based on Cloud Compare software, specifically using the ICP rigid registration principle, and the steps are to load the two sets of point clouds to be registered in Cloud Compare point cloud software, and distinguish them by displaying them in different colors, then select the ICP registration method, set the parameter conditions, including the number of iterations, threshold, overlap degree, and then complete the post-registration.

4. The method for detecting the quality deviation of shield machine segment assembly based on laser scanning according to claim 1 is characterized in that: In step 5), a non-average segmentation algorithm is used to separate different structural objects in the three-dimensional point cloud data. The steps are as follows: Assume that the segment width is l1, the data starting position z value is 0, and since the data starting position is the first annular seam between the first ring segment and the second ring segment, the i-th joint position is: z i =(i-1)×l1 Then, two types of segmentation points are set on the left and right sides of the segment annular seam: Type ① segmentation point: z i1 =(i-1)×l1+s The second type of segmentation point: z i2 =il1-s In the above formula: s is a constant, which is half of the width of the data at the extraction seam. After obtaining the first and second types of segmentation points, the two types of segmentation points are sequentially spaced and arranged from small to large on the Z axis of the data. The two types of segmentation points are arranged from small to large to form a set K: K={(i-1)l1+s,il1-s},i=1,2,…,n The data in K are combined into pairs of data for segmentation in sequence, so that the tunnel point cloud data can be segmented non-evenly.

5. The method for detecting quality deviation of shield machine segment assembly based on laser scanning according to claim 1 is characterized in that: In step 6), the ellipticity of the test is obtained by calculating the data planarized by the least squares method by finding the major and minor semi-axis method. The steps are as follows: assuming q i (x i ,y i ) is any point in the point cloud data, point q i The distance from the fitting circle center O(X,Y) is R i ,but: R i Maximum R max With the minimum value R min Solve it, that is: Maximum R max With the minimum value R min are considered as the semi-major and semi-minor axes of the tunnel ellipse, then the major and minor axes of the ellipse are: D max =2×R max D min =2×R min According to the theoretical formula of ellipticity, the tunnel ellipticity can be obtained as:

6. The method for detecting the quality deviation of shield machine segment assembly based on laser scanning according to claim 1 is characterized in that: The specific steps of step 7) are as follows: Suppose a point cloud after the average segmentation is p, and any point p in p i The two-dimensional distance to the origin of the XY plane of the fitted circle, that is: Where: d i For p i Distance to the origin of the XY plane, x i ,y i They are point p i The horizontal and vertical coordinates of According to the above formula, the maximum two-dimensional distance is obtained as the corresponding data of the peak point, and the z value data of the corresponding point under this distance is extracted, which is the z value corresponding to the peak: z=z i Where: d max is the maximum distance from the midpoint data of p to the origin of the XY plane, z i is the z value of the point corresponding to the maximum distance, Extract the maximum and minimum z value data in the point cloud and record it as z max , z min , then there must be z max >z>z min Relationship, create set O: Extract the data in the set O in pairs to perform Z-axis segmentation on the point cloud p, and finally divide the data into four segments. The front and back segments are the point cloud data of parts A and B. After extracting the point cloud data of parts A and B, the misalignment value can be detected. The average two-dimensional distance of the point cloud data of parts A and B to the origin in the XY plane is calculated respectively. Where: d A , d B are the average distances between parts A and B and the origin in the XY plane, and d is calculated. A , d B The difference between the values ​​of can be used to get the misalignment value at that location, namely: △=|d A -d B | Where: Δ is the misalignment value calculated at this location.

7. The method for detecting quality deviation of shield machine segment assembly based on laser scanning according to claim 1 is characterized in that: The specific calculation steps of step 8) are as follows: Set point cloud p i is the maximum two-dimensional distance to the origin of the XY plane of the fitting circle. This point is located in the annular seam data. In order to retain as much annular seam data as possible, the distance threshold δ is set, that is: |d i -d j |≤δ Where: d i is the maximum distance to the origin, d j is the distance from any point to the origin, When the absolute value of the difference between the distance from any point to the origin and the maximum distance is less than the threshold, the data point is considered to be located in the annular seam data. After the annular seam is extracted, the least squares cylinder fitting is performed on the data to obtain the center point of the annular seam data. The steps for extracting the center point data between rings are as follows: Extract the middle part of the ring segment data to calculate the center point. Extract the maximum and minimum z value data of the ring segment, which are z max , z min , calculate the z value of the midpoint data and set the distance threshold δ to calculate and extract the data in the ring, that is: |from i -With mind |≤δ Where: z mid is the midpoint data z value, z i is the z value of the data at any point. After extracting the data in the ring, the least squares cylindrical fitting is performed on the data to obtain the center point of the ring segment data. Then the center point deviation is calculated, and the steps are as follows: The center point of a ring circle fitting point cloud is p k (x k ,y k ), the center point of the adjacent circle fitting point cloud is p k+1 (x k+1 ,y k+1 ), the deviation of the center points of any adjacent ring segments is solved as follows: Where: d is the deviation of the center points of adjacent ring segments.

Images