Intelligent identification system and method for shield machine discharge volume

Abstract

The present invention relates to an intelligent identification system and method for the discharge volume of a shield machine. It mainly solves the technical problems of large measurement errors, unstable calculations, and large environmental impacts of existing methods. The intelligent identification system for the discharge volume of a shield machine of the present invention is mainly composed of a laser system, a camera system, a micro industrial computer, and an integrated control box; the identification method of the present invention includes: 1) construction of a laser cross-section scanning point coordinate system; 2) removal of scanning noise data; 3) calculation of discrete scanning point spatial Kriging interpolation volume objects and volume objects; 4) automatic identification of soil types based on machine learning; 5) calculation of shield excavation mass; the method integrates laser scanning, image AI (Artificial Intelligence) recognition, and stratum three-dimensional reconstruction technology. Through the spatial Kriging interpolation algorithm, the shield machine excavation body can be subjected to three-dimensional refined modeling and measurement with higher accuracy.

Description

Technical Field

[0001] The invention relates to a shield machine excavation measurement system, and in particular discloses a shield machine soil discharge volume intelligent identification system and method, which are applied to the field of shield tunnel engineering. Background Art

[0002] In the process of subway construction, the earth pressure balance shield construction method is widely used. During the excavation process, improper control of the excavation volume can easily cause surface uplift or subsidence and collapse. Therefore, controlling the excavation volume is crucial to the safety and efficiency of shield construction. However, the common methods currently have large measurement errors, the calculation is not stable enough, and are easily affected by the environment. How to achieve accurate measurement of the excavation volume is the key.

[0003] Based on the rapid development of sensor technologies such as laser scanning and machine vision, there are many methods on the market. Patent CN112211672A discloses a method for measuring the excavation volume of a shield machine. By receiving the belt speed of the belt conveyor that transmits the slag, the density of the slag, the cross-sectional area, the current temperature of the slag and the excavation stroke, the excavation volume is calculated, and the over-excavation and under-excavation are monitored and alarmed. CN202011604888.1 discloses a shield construction excavation volume monitoring device using laser scanning. A laser scanner and a leveling mechanism are installed at the upper end of the belt conveyor to calculate the amount of slag discharged per unit time. Patent CN102720510B installs a pinwheel device at the excavation port of the screw conveyor of the shield machine, which is installed in a plane perpendicular to the direction of the slag speed to calculate the excavation volume. It is suitable for shield machines to pass through moist soil layers with a certain viscosity. However, its device is complex, the installation equipment is large, and the types of sensors required are many. It is difficult to accurately calculate the excavation volume based on two-dimensional parameter calculation.

[0004] In summary, there is an urgent need for a system and method for high-precision modeling and calculation of excavated volumes, so as to achieve real-time and accurate calculation of excavated volumes. Summary of the invention

[0005] The purpose of the present invention is to design an intelligent identification system and method for the earth discharge of a shield machine, mainly to solve the technical problems of large measurement errors, unstable calculations, and great environmental influence.

[0006] The purpose of the present invention can be achieved by the following technical solutions: a shield machine soil discharge intelligent recognition system, including a laser cross-section scanner, a camera system, a micro industrial computer and an integrated control box; the laser cross-section scanner, the camera system, the micro industrial computer and the integrated control box are installed on an integrated bottom plate at intervals, wherein the laser cross-section scanner has an IP protection level not less than IP65, and the system distance measurement accuracy is not less than 1mm; the camera system includes an industrial camera and a fill light; the fill light is a ring light source; the micro industrial computer is mainly used for data processing and transmission, and measurement and acquisition software and data processing and analysis software are installed in the micro industrial computer, the micro industrial computer has an industrial display, and the micro The analysis results of the industrial computer are sent to the background control center through the router wirelessly or wiredly; the integrated control box includes a voltage conversion module, a relay controller, a switch and a control panel. The voltage conversion module converts the AC mains power into the DC power required by the equipment to power all equipment. The relay controller is connected to the switch via a network cable to establish communication with the industrial control. The industrial computer polls the working status of each acquisition device and communication system, and connects to control the switch output of the relay controller to realize the indicator light on the control panel to indicate the operating status; the integrated control box provides power supply, operation control, status display, etc. for the laser cross-section scanner, camera system, and micro industrial computer.

[0007] Furthermore, the laser cross-section scanner and camera system are connected to the micro-industrial computer via a data cable, and the micro-industrial computer operates and controls the laser scanner and camera via built-in software.

[0008] Furthermore, the trigger and stop operation signals of the real-time measurement device for the shield excavation volume are connected to the shield machine excavation system through the integrated control box, and the actions remain consistent.

[0009] A recognition method for a shield machine discharge volume intelligent recognition system comprises the following steps:

[0010] S1. The acquisition frequency F of the laser cross-section scanner is determined by two factors: the running speed V of the shield machine crawler and the distance interval L of the measured section. The calculation method is as follows:

[0011] F=V / L (1)

[0012] Among them, the unit of V is m / s, the unit of L is m, and the unit of F is Hz. To ensure the accuracy of soil measurement, L can generally be set to 0.01m, that is, 10mm;

[0013] Furthermore, the laser cross-section scanner is installed just above the center of the unearthed track of the shield machine, that is, the line connecting the scanning center of the laser cross-section scanner and the center point of the unearthed track is perpendicular to the track plane. The installation height H of the laser cross-section scanner is determined by two factors: the width W of the unearthed track of the shield machine and the scanning angle θ of the laser cross-section scanner. The calculation method is as follows:

[0014] H=0.5W / tan(θ / 2) (2)

[0015] Where, the unit of H and W is m, and the unit of θ is radian;

[0016] S2. In order to reduce the amount of data collection and match the data collected by the laser cross-section scanner, the camera collection is triggered by external program control. The collection trigger interval is mainly determined by the length L of the object photographed by the camera (the direction of track movement). There are two triggering methods:

[0017] (1) Timing trigger: The acquisition trigger interval is determined by the length L of the object photographed by the camera (in the direction of track movement) and the track movement speed V. The trigger time interval T is calculated as follows:

[0018] T=V / L (3)

[0019] Among them, the unit of L is m, and the unit of T is second;

[0020] (2) Fixed distance trigger: The track is triggered every time the track moves a distance L. When the track moves at a constant speed V, the trigger is equivalent to a timed trigger.

[0021] S3. The subsequent identification method comprises the following steps:

[0022] S3.1 Construction of laser cross-section scanning point coordinate system: Take the center point of the laser cross-section scanner as the origin of the coordinate system, the midline of the scanning surface as the y-axis, and rotate 90° counterclockwise as the x-axis to construct the overall rectangular coordinate system of the laser scanning point. The laser cross-section scanner's own device coordinates are polar coordinates. The polar coordinates of the laser cross-section scanner need to be converted into a rectangular coordinate system. The conversion method is:

[0023]

[0024] Among them, D i is the measured distance value of the i-th laser scanning point, θ i is the angle between the i-th laser scanning point and the Y axis, (x i ,y i ) is the coordinate point of the i-th laser scanning point in the rectangular coordinate system; the laser scanning point set of a single cross section is denoted by P.

[0025] S3.2 Laser cross-section scanner cross-section scanning data processing: Extract the laser scanning point set P of a single cross section, and perform coordinate transformation on the point set P through equation (4). The transformed point set is recorded as P o , due to the complex environmental interference of shield excavation, the point set P o There are some noise points, which need to be removed by 3 times the standard deviation σ kThe statistical filtering method is used to remove noise points, and the laser scanning point set after removal is recorded as P';

[0026] S3.3 extracts N groups of laser scanning point sets P' in sequence, and performs spatial interpolation fitting on the area composed of N groups of point sets through the spatial Kriging interpolation method, and the spatial body between the soil surface of the area and the XOZ plane of the overall rectangular coordinate system can be obtained, which is recorded as B c , when the crawler is in the initial state (no soil is transported), the initial space between the crawler surface and the XOZ plane of the overall rectangular coordinate system is denoted as B o ;

[0027] Steps of spatial kriging interpolation method based on discrete point set:

[0028] (1) Determine the spatial interpolation bottom surface area, which is the projection area of ​​the track surface composed of N sets of points on the XOZ plane of the overall rectangular coordinate system. This plane area is denoted as F;

[0029] (2) Taking the point set P' as the constraint point, the plane area F is triangularly constrained to mesh, and the basic mesh of the plane area F is obtained, which is denoted as M0. The point set of mesh M0 is denoted as N. The point set N is all located in the XOZ plane, that is, the Y coordinate value of the point set N is zero;

[0030] (3) Determine the variation function γ(h), take the point set P' as the known sample points, perform spatial Kriging interpolation on the other unknown points in the grid point set N, and calculate the Y coordinate value of the unknown point;

[0031] The estimation formula of Kriging is:

[0032]

[0033] Among them, Z * (x0) is the estimated value at position x0, Z(x i ) is the measured value of position, λ i is assigned to Z(x i ), and n is the number of measurements used to estimate the process.

[0034] The variogram γ(h) is used to characterize the spatial variation structure or spatial continuity of random variables. It is the basis of geostatistics and the Gaussian model is used.

[0035]

[0036] Among them, C0 is the nugget value, which represents the variation value of the variogram within a very short distance, which is usually determined by the sample error and the variability of the short distance; C0+C1 is the base value, and a is the variation distance (range) or related scale, indicating that when the sample is greater than this distance, the spatial correlation of the sample can be ignored, and the samples are considered to be completely independent;

[0037] (4) After the spatial interpolation of the unknown points is completed, a new spatial surface triangle mesh can be generated, denoted as M K ,

[0038] (5) Basic grid M0 and interpolation grid M K The nodes in the Y-axis direction are one-to-one corresponding. K By stitching the meshes around the object, a hollow closed object represented by triangular facets can be formed. The space object is denoted as B. The triangular mesh of object B is represented by a half-edge data structure, and it is ensured that the normal of each triangular facet faces outward according to the right-hand coordinate system.

[0039] S3.4 Since the surface of a spatial object is represented by a triangular mesh, the volume of the object can be obtained by the simplex method.

[0040] Through the above method, the space body B c and B o The volume of c and V o , for a fixed measuring device, V o is a certain value, then the volume of the soil composed of N sets of points P' is V o -V c , denoted as V oc .

[0041] S3.5 Automatic identification of soil types based on machine learning:

[0042] (1) The collected image data are divided into N categories according to different soil layer types as training sets;

[0043] (2) Use the training set to train the Inceptionv3 model to classify soil layers;

[0044] (3) Input the detection image and determine the soil layer type.

[0045] S3.6 Calculation and correction of soil mass: Based on the soil type identified by the machine learning method, the corresponding soil density ρ can be obtained. The mass G of the soil in this area can be calculated as follows:

[0046] G=αρV (7)

[0047] Where α is the soil mass correction coefficient, the default value is 1.0, and the specific value can be determined by comparative tests on the engineering site.

[0048] S3.7 Through the cycle of steps S3.2, S3.3, S3.4, S3.5 and S3.6, the volume and mass of the excavated materials by the shield machine during the period can be measured in real time.

[0049] The beneficial effects of the present invention are:

[0050] (1) The present invention can realize the automatic and digital real-time measurement of the shield machine's excavation volume;

[0051] (2) The present invention integrates laser scanning, image AI recognition and stratum three-dimensional reconstruction technology. Through the spatial Kriging interpolation algorithm, the shield machine excavation body can be finely reconstructed in three dimensions, with higher measurement accuracy;

[0052] (3) The acquisition system of the technical solution of the present invention is relatively simple, has good protection capabilities, the algorithm is easy to program and implement, and has strong applicability in application scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is a schematic diagram of a system architecture diagram of an embodiment of the present invention.

[0054] Figure 2 It is a diagram of the equipment layout of an embodiment of the present invention.

[0055] Figure 3 Schematic diagram of rectangular coordinates of a laser scanning system according to an embodiment of the present invention.

[0056] Figure 4 Schematic diagram of polar coordinate and rectangular coordinate conversion according to an embodiment of the present invention.

[0057] Figure 5 It is a schematic diagram of the unearthed volume measurement layout of an embodiment of the present invention.

[0058] Figure 6 It is a schematic diagram of the normal direction of a triangular facet using the simplex method to calculate the volume of a volume object according to an embodiment of the present invention.

[0059] Figure 7 Schematic diagram of a variation function model according to an embodiment of the present invention.

[0060] In the figure: 1- shield machine excavation crawler, 2- measurement system mounting bracket, 3- excavation measurement device, 4- camera shooting range, 5- laser measurement line, 6- relay controller, 7- switch, 8- control panel, 9- industrial camera, 10- ring light source, 11- industrial display, 12- micro industrial computer, 13- laser section scanner. DETAILED DESCRIPTION

[0061] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0062] like Figure 1 , Figure 2 As shown, a shield machine soil discharge intelligent identification system includes a laser cross-section scanner, a camera system, a micro industrial computer and an integrated control box; the laser cross-section scanner, the camera system, the micro industrial computer and the integrated control box are installed on the integrated bottom plate at intervals, wherein the laser cross-section scanner has an IP protection level not less than IP65, and the system distance measurement accuracy is not less than 1mm; the camera system includes an industrial camera and a fill light; the fill light is a ring light source; the micro industrial computer is mainly used for data processing and transmission, the micro industrial computer is installed with measurement acquisition software and data processing and analysis software, the micro industrial computer has an industrial display, and the analysis results of the micro industrial computer are sent to the background control center through a router by wireless or wired means; the integrated control box includes a voltage conversion module, a relay controller, a switch and a control panel, and the voltage conversion module converts the AC mains into the DC power required by the equipment to power all the equipment. The relay controller is connected to the switch through a network cable to establish communication with the industrial control. The industrial computer polls the working status of each acquisition device and communication system, and connects the switch output of the control relay controller to realize the indicator light on the control panel to indicate the operating status; the integrated control box provides power supply, operation control, status display, etc. for the laser cross-section scanner, camera system, and micro industrial computer.

[0063] Furthermore, the laser cross-section scanner and the camera are connected to the micro industrial computer via a data cable, and the laser scanner and the camera are operated and controlled via measurement and acquisition software.

[0064] Furthermore, the trigger and stop operation signals of the real-time measurement device for the shield excavation volume are connected to the shield machine excavation system through the integrated control box, and the actions remain consistent.

[0065] The laser cross-sectional scanner, camera, and micro industrial computer can all be commercially available.

[0066] Laser cross-sectional scanner: Sick's LMS4121R line laser scanner can be used, with a scanning angle of 70°, a single-point laser distance measurement accuracy of 1mm, 841 scanning points, and a scanning frequency of 600Hz;

[0067] Camera: BASLER brand acA1600-20um / uc can be used, camera resolution 1626×1236 (H×V pixels), frame rate 20, shell size (length×width×height): 29.3mm×29mm×29mm, synchronous control through hardware trigger, software trigger or automatic operation;

[0068] Micro industrial computer: You can choose Advantech ARK-1551 fanless embedded industrial computer, which uses the eighth-generation Core i5 series CPU, up to 32GB DDR4 2400MHz memory, and has dual-channel Gigabit network ports. The shell size (length × width × height) is 195mm × 55mm × 140mm, and it can work in an environment of -20 to 55℃, and has strong vibration and shock resistance.

[0069] Integrated control box: Select the Tongyan Technology TY-ICB01 integrated control box, which integrates power supply, operation control, status display and other functions, and supports touch screen operation. The shell size (length × width × height) is 360mm × 280mm × 220mm, and the cable connector uses an aviation plug with IP65 protection level and strong environmental adaptability.

[0070] like Figure 3 , Figure 4 , Figure 5 , Figure 6 , Figure 7 As shown: A recognition method of a shield machine discharge volume intelligent recognition system, comprising the following steps:

[0071] S1. The acquisition frequency F of the laser cross-section scanner is determined by two factors: the running speed V of the shield machine crawler and the distance interval L of the measured section. The calculation method is as follows:

[0072] F=V / L (1)

[0073] Among them, the unit of V is m / s, the unit of L is m, and the unit of F is Hz. To ensure the accuracy of soil measurement, L can generally be set to 0.01m, that is, 10mm;

[0074] For example, the crawler speed V of the shield machine is 3.0m / s, and the measurement section distance interval L is 0.01m, then the acquisition frequency is 300Hz, and Sick's LMS4121R line laser scanner can meet the requirements;

[0075] Furthermore, the laser cross-section scanner is installed just above the center of the unearthed track of the shield machine, that is, the line connecting the scanning center of the laser cross-section scanner and the center point of the unearthed track is perpendicular to the track plane. The installation height H of the laser cross-section scanner is determined by two factors: the width W of the unearthed track of the shield machine and the scanning angle θ of the laser cross-section scanner. The calculation method is as follows:

[0076] H=0.5W / tan(θ / 2) (2)

[0077] Where, the unit of H and W is m, and the unit of θ is radian;

[0078] For example, if the width W of the tunnel boring machine crawler is 0.8 m, and the scanning angle θ of the laser cross-section scanner is 70°, then the installation height H of the laser cross-section scanner is H = 0.5*0.8 / tan(35°) = 0.57 m;

[0079] S2. In order to reduce the amount of data collection and match the data collected by the laser cross-section scanner, the camera collection is triggered by external program control. The collection trigger interval is mainly determined by the length L of the object photographed by the camera (the direction of track movement). There are two triggering methods:

[0080] (1) Timing trigger: The acquisition trigger interval is determined by the length L of the object photographed by the camera (in the direction of track movement) and the track movement speed V. The trigger time interval T is calculated as follows:

[0081] T=L / V (3)

[0082] Among them, the unit of L is m, and the unit of T is second;

[0083] (2) Fixed distance trigger: The track is triggered every time the track moves a distance L. When the track moves at a constant speed V, the trigger is equivalent to a timed trigger.

[0084] For example, if the speed V of the shield machine crawler is 3.0m / s, and the interval L of the measuring section is 0.01m, the length of the 1-second scan is 3m, and the camera's shooting range is 0.6*0.8, of which the crawler movement direction is 0.6m. When the timing trigger is used, the camera's acquisition frequency can be set to 5Hz, and if the fixed-distance trigger is used, it can be set to 0.6m. The distance measurement can be triggered by a commercially available distance measuring wheel.

[0085] S3. The subsequent identification method comprises the following steps:

[0086] S3.1 Construction of laser cross-section scanning point coordinate system: Take the center point of the laser cross-section scanner as the origin of the coordinate system, the midline of the scanning surface as the y-axis, and rotate 90° counterclockwise as the x-axis to construct the overall rectangular coordinate system of the laser scanning point. The laser cross-section scanner's own device coordinates are polar coordinates. The polar coordinates of the laser cross-section scanner need to be converted into a rectangular coordinate system. The conversion method is:

[0087]

[0088] Among them, D i is the measured distance value of the i-th laser scanning point, θ i is the angle between the i-th laser scanning point and the Y axis, (x i ,y i ) is the coordinate point of the i-th laser scanning point in the rectangular coordinate system; the laser scanning point set of a single cross section is denoted by P.

[0089] S3.2 Laser cross-section scanner cross-section scanning data processing: Extract the laser scanning point set P of a single cross section, and perform coordinate transformation on the point set P through equation (4). The transformed point set is recorded as P o , due to the complex environmental interference of shield excavation, the point set P o There are some noise points, which need to be removed by 3 times the standard deviation σ k The statistical filtering method is used to remove noise points, and the laser scanning point set after removal is recorded as P';

[0090] Laser scanning noise removal method based on statistical filtering: First, for each point P i ∈P, first calculate the average distance d of its k nearest areas, and then i The variance U of the overall point cloud P in the vicinity of point k is k and standard deviation σ k The purpose of the estimation is to keep the point P i The average distance d of the point with the nearest k neighbors and the rest of the point is similar, and the valid point cloud P * Restrictions:

[0091]

[0092] Among them, α is the limiting factor, which can be 3.0. The distance value from each feature point to the center point of the laser scanner.

[0093] S3.3 extracts N groups of laser scanning point sets P' in sequence, and performs spatial interpolation fitting on the area composed of N groups of point sets through the spatial Kriging interpolation method, and the spatial body between the soil surface of the area and the XOZ plane of the overall rectangular coordinate system can be obtained, which is recorded as B c , when the crawler is in the initial state (no soil is transported), the initial space between the crawler surface and the XOZ plane of the overall rectangular coordinate system is denoted as B o ;

[0094] Steps of spatial kriging interpolation method based on discrete point set:

[0095] (1) Determine the spatial interpolation bottom surface area, which is the projection area of ​​the track surface composed of N sets of points on the XOZ plane of the overall rectangular coordinate system. This plane area is denoted as F;

[0096] (2) Taking the point set P' as the constraint point, the plane area F is triangularly constrained to mesh, and the basic mesh of the plane area F is obtained, which is denoted as M0. The point set of mesh M0 is denoted as N. The point set N is all located in the XOZ plane, that is, the Y coordinate value of the point set N is zero;

[0097] (3) Determine the variation function γ(h), take the point set P' as the known sample points, perform spatial Kriging interpolation on the other unknown points in the grid point set N, and calculate the Y coordinate value of the unknown point;

[0098] The estimation formula of Kriging is:

[0099]

[0100] Among them, Z * (x0) is the estimated value at position x0, Z(x i ) is the measured value of position, λ i is assigned to Z(x i ), and n is the number of measurements used to estimate the process.

[0101] The variogram γ(h) is used to characterize the spatial variation structure or spatial continuity of random variables. It is the basis of geostatistics and the Gaussian model is used.

[0102]

[0103] Among them, C0 is the nugget value, which represents the variation value of the variogram within a very short distance, which is usually determined by the sample error and the variability of the short distance; C0+C1 is the base value, and a is the variation distance (range) or related scale, indicating that when the sample is greater than this distance, the spatial correlation of the sample can be ignored, and the samples are considered to be completely independent;

[0104] (4) After the spatial interpolation of the unknown points is completed, a new spatial surface triangle mesh can be generated, denoted as M K ,

[0105] (5) Basic grid M0 and interpolation grid M K The nodes in the Y-axis direction are one-to-one corresponding. K By stitching the meshes around the object, a hollow closed object represented by triangular facets can be formed. The space object is denoted as B. The triangular mesh of object B is represented by a half-edge data structure, and it is ensured that the normal of each triangular facet faces outward according to the right-hand coordinate system.

[0106] S3.4 Since the surface of a spatial object is represented by a triangular mesh, the volume of the object can be obtained by the simplex method.

[0107] Through the above method, the space body B c and B o The volume of c and V o , for a fixed measuring device, V o is a certain value, then the volume of the soil composed of N sets of points P' is Vo -V c , denoted as V oc .

[0108] The simplex method for finding the volume of a spatial object:

[0109] (1) First, ensure that the normals of all triangular facets of the block are consistent, such as facing outward or inward;

[0110] (2) Select a reference point C. For the convenience of calculation, the origin (0,0,0) is usually selected.

[0111] (3) Traverse the triangular facets of the body one by one i , calculate each triangle △ i The volume V of the tetrahedron connected to the reference point C i ;

[0112]

[0113] (4) The volume of the block is: Where n is the number of triangular facets on the block surface.

[0114] S3.5 Automatic identification of soil types based on machine learning:

[0115] (1) The collected image data are divided into 15 categories according to different soil layer types, namely fill, silt, clay, sand, silty clay, silty silty clay, silty clay, silty sand, silty sand, silty sand, silty fine sand, clay silt, sandy silt, strongly weathered rock, weakly weathered rock, and slightly weathered rock as the training set;

[0116] (2) Use the training set to train the Inceptionv3 model to classify soil layers;

[0117] (3) Input the detection image and determine the soil layer type.

[0118] The main settings of Inceptionv3 hyperparameters are as follows:

[0119] training_steps: 12000

[0120] learning_rate: 0.06

[0121] testing_percentage: 10%

[0122] validation_percentage: 10%

[0123] train_batch_size: 300

[0124] test_batch_size: -1

[0125] validation_batch_size: 300

[0126] S3.6 Calculation and correction of soil mass: Based on the soil type identified by the machine learning method, the corresponding soil density ρ can be obtained. The mass G of the soil in this area can be calculated as follows:

[0127] G=αρV (9)

[0128] Where α is the soil mass correction coefficient, the default value is 1.0, and the specific value can be determined by comparative tests on the engineering site.

[0129] S3.7 Through the cycle of steps S3.2, S3.3, S3.4, S3.5 and S3.6, the volume and mass of the excavated materials by the shield machine during the period can be measured in real time.

[0130] The preferred specific embodiments of the present invention are described in detail above. It should be understood that a person skilled in the art can make many modifications and changes based on the concept of the present invention without creative work. Therefore, any technical solution that can be obtained by a person skilled in the art through logical analysis, reasoning or limited experiments based on the concept of the present invention on the basis of the prior art should be within the scope of protection required by the claims of the present invention.

Claims

1. A recognition method for a shield machine discharge volume intelligent recognition system, characterized in that: The shield machine soil discharge intelligent identification system includes a laser cross-section scanner, a camera system, a micro industrial computer and an integrated control box; the laser cross-section scanner, the camera system, the micro industrial computer and the integrated control box are installed at intervals on the integrated bottom plate, the laser cross-section scanner and the camera system are connected to the micro industrial computer through a data cable, and the camera system includes an industrial camera and a fill light; the micro industrial computer is installed with measurement and acquisition software and data processing and analysis software for data processing and transmission, and operates and controls the laser scanner and the industrial camera. The micro industrial computer has an industrial display, and the analysis results of the micro industrial computer are sent to the router by wireless or wired means. The integrated control box includes a voltage conversion module, a relay controller, a switch and a control panel. The voltage conversion module converts AC mains power into DC power required by the equipment to power all equipment. The relay controller is connected to the switch through a network cable to establish communication with the industrial control. The industrial control computer polls the working status of each acquisition device and the communication system, and connects to control the switch output of the relay controller to realize the indicator light on the control panel to indicate the operating status. The integrated control box provides power, operation control and status display for the laser cross-section scanner, the camera system and the micro industrial control computer. The identification method includes the following steps: S1. The acquisition frequency F of the laser cross-section scanner is determined by two factors: the running speed V of the shield machine crawler and the distance interval L of the measured section. The calculation method is as follows: F=V / L (1) Among them, the unit of V is m / s, the unit of L is m, and the unit of F is Hz; Furthermore, the laser cross-section scanner is installed just above the center of the unearthed track of the shield machine, that is, the line connecting the scanning center of the laser cross-section scanner and the center point of the unearthed track is perpendicular to the track plane. The installation height H of the laser cross-section scanner is determined by two factors: the width W of the unearthed track of the shield machine and the scanning angle θ of the laser cross-section scanner. The calculation method is as follows: H=0.5W / tan(θ / 2) (2) Where, the unit of H and W is m, and the unit of θ is radian; S2. In order to reduce the amount of data collection and match the data collected by the laser cross-section scanner, the camera collection is triggered by external program control. The collection trigger interval is mainly determined by the length L of the object photographed by the camera. There are two triggering methods: (1) Timing trigger: The acquisition trigger interval is determined by the length L of the object photographed by the camera and the track movement speed V. The trigger time interval T is calculated as follows: T=V / L (3) Among them, the unit of L is m, and the unit of T is second; (2) Fixed distance trigger: The trigger is triggered every time the track moves a distance L. When the track moves at a constant speed V, the trigger is equivalent to a timed trigger. S3. The subsequent identification method comprises the following steps: S3.1 Construction of laser cross-section scanning point coordinate system: Take the center point of the laser cross-section scanner as the origin of the coordinate system, the midline of the scanning surface as the y-axis, and rotate 90° counterclockwise as the x-axis to construct the overall rectangular coordinate system of the laser scanning point. The laser cross-section scanner's own device coordinates are polar coordinates. The polar coordinates of the laser cross-section scanner need to be converted into a rectangular coordinate system. The conversion method is: Among them, D i is the measured distance value of the i-th laser scanning point, θ i is the angle between the i-th laser scanning point and the Y axis, (x i ,y i ) is the coordinate point of the i-th laser scanning point in the rectangular coordinate system; the laser scanning point set of a single cross section is denoted as P; S3.2 Laser cross-section scanner cross-section scanning data processing: Extract the laser scanning point set P of a single cross section, and perform coordinate transformation on the point set P through equation (4). The transformed point set is recorded as P o , due to the complex environmental interference of shield excavation, the point set P o There are some noise points, which need to be removed by 3 times the standard deviation σ k The statistical filtering method is used to remove noise points, and the laser scanning point set after removal is recorded as P'; S3.3 extracts N groups of laser scanning point sets P' in sequence, and performs spatial interpolation fitting on the area composed of N groups of point sets through the spatial Kriging interpolation method, and the spatial body between the soil surface of the area and the XOZ plane of the overall rectangular coordinate system can be obtained, which is recorded as B c , when the crawler is in its initial state, the initial space between the crawler surface and the XOZ plane of the overall rectangular coordinate system is denoted as B o ; S3.4 Since the surface of a spatial object is represented by a triangular mesh, the volume of the object can be obtained by the simplex method. Through the above method, the space body B c and B o The volume of c and V o , for a fixed measuring device, V o is a certain value, then the volume of the soil composed of N sets of points P' is V o -V c , denoted as V oc ; S3.5 Automatic identification of soil types based on machine learning: (1) The collected image data are divided into N categories according to different soil layer types as training sets; (2) Use the training set to train the Inceptionv3 model to classify soil layers; (3) Input the detection image and determine the soil layer type; S3.6 Calculation and correction of soil mass: Based on the soil type identified by the machine learning method, the corresponding soil density ρ can be obtained. The mass G of the soil in this area can be calculated as follows: G=αρV (7) Where α is the soil mass correction coefficient; S3.7 Through the cycle of steps S3.2, S3.3, S3.4, S3.5 and S3.6, the volume and mass of the excavated materials by the shield machine during the period can be measured in real time.

2. The identification method of the shield machine discharge volume intelligent identification system according to claim 1 is characterized by: The fill light is a ring light source.

3. The identification method of the shield machine discharge volume intelligent identification system according to claim 1 is characterized by: The spatial Kriging interpolation method in step S3.3 is specifically: (1) Determine the spatial interpolation bottom surface area, which is the projection area of ​​the track surface composed of N sets of points on the XOZ plane of the overall rectangular coordinate system. This plane area is denoted as F; (2) Taking the point set P' as the constraint point, the plane area F is triangularly constrained to mesh, and the basic mesh of the plane area F is obtained, which is denoted as M0. The point set of mesh M0 is denoted as N. The point set N is all located in the XOZ plane, that is, the Y coordinate value of the point set N is zero; (3) Determine the variation function γ(h), take the point set P' as the known sample points, perform spatial Kriging interpolation on the other unknown points in the grid point set N, and calculate the Y coordinate value of the unknown point; The estimation formula of Kriging method is: Among them, Z * (x0) is the estimated value at position x0, Z(x i ) is the measured value of position, λ i is assigned to Z(x i ), n is the number of measurements used to estimate the process; The variogram γ(h) is used to characterize the spatial variation structure or spatial continuity of random variables. It is the basis of geostatistics and the Gaussian model is used. Among them, C0 is the nugget value, which represents the variation value of the variogram within a very short distance, which is usually determined by the sample error and the variability of the short distance; C0+C1 is the base value, and a is the variation distance or correlation scale, indicating that when the sample is greater than this distance, the spatial correlation of the sample can be ignored, and the samples are considered to be completely independent; (4) After the spatial interpolation of the unknown points is completed, a new spatial surface triangle mesh can be generated, denoted as M K , (5) Basic grid M0 and interpolation grid M K The nodes in the Y-axis direction are one-to-one corresponding. K By stitching the meshes around the object, a hollow closed body object represented by triangular facets can be formed. The triangular mesh of object B is represented by a half-edge data structure, and the normal of each triangular facet is ensured to face outwards according to the right-hand coordinate system.

Images