Portable wood shaving pressed cake sampling device and sampling method

Abstract

The invention discloses a portable wood shaving pressed cake sampling device and a sampling method, and belongs to the technical field of waste steel recovery processing equipment. The portable wood shaving pressed cake sampling device comprises a frame base, a telescopic box body is arranged on one side of the top of the frame base, a frame main body is arranged above the telescopic box body, an electric lifting mechanism is arranged on the frame main body, a telescopic mechanical arm is installed on the electric lifting mechanism, an electromagnetic chuck is arranged at the end of the telescopic mechanical arm, and the electromagnetic chuck is connected with the telescopic mechanical arm. A transverse driving device is arranged above the frame main body, a multi-stage telescopic arm is arranged at the bottom of the transverse driving device, a hydraulic crushing pickaxe is mounted at the bottom end of the multi-stage telescopic arm, and an image acquisition device is further arranged at the bottom of the transverse driving device. The problem that moving, crushing and sampling detection work of existing scrap steel pressing cakes and pressing blocks is not convenient and fast enough is solved. Automatic sampling is achieved through the three-stage technology of visual analysis, defect positioning and intelligent crushing.

Description

Technical Field

[0001] The present invention relates to the technical field of scrap steel recycling and processing equipment, and specifically to a portable shaving cake sampling device and a sampling method. Background Technique

[0002] A portable shaving cake sampling machine is a crusher used to process scrap steel cakes and particle steel blocks. In the iron and steel industry, the recycling and processing of scrap steel is an important link. Scrap steel is usually stored and transported in the form of blocks or cakes for subsequent reuse. Common forms of scrap steel include shaving cakes and particle steel blocks, which are made in the following ways:

[0003] Shaving cakes: Shavings are thin sheet-like waste materials generated during the steel processing process. Usually, a hydraulic press is used to press them into cake shapes. During the pressing process, the shavings are placed in a mold, and high pressure is applied to make them tightly combined, forming a flat cake-like structure. The size of the shaving cake is usually about 90 - 300 mm in diameter and 30 - 80 mm in thickness (the side length of the cube ≤ 300 mm), and the weight is about 0.5 - 100 kg, depending on the density and material composition of the cake.

[0004] Particle steel blocks: Particle steel is fine scrap steel particles generated during the steel production process. Usually, a hydraulic press or a mechanical press is used to press them into block shapes. During the pressing process, the particle steel is placed in a mold, and high pressure is applied to make them tightly combined, forming a regular block structure. The general size of the particle steel block is 300 - 600 mm in side length (cold pressing) or 260 mm in diameter × 240 mm in height (hot pressing), and the weight is between 25 - 80 kg, depending on the density and material composition of the block.

[0005] During the manufacturing, circulation, and application processes of shaving cakes and particle steel blocks, it is necessary to sample, crush, and re-inspect their quality multiple times. Currently, the sampling, crushing, and sampling processes are all manually carried out by operators, with small hammers for crushing and manual sampling. However, due to the large weight and tight combination of shaving cakes, manual operation is very inconvenient, laborious, time-consuming, inefficient, and there are safety hazards, making it difficult to meet the requirements of the modern scrap steel industry. Summary of the Invention

[0006] Automated sampling is achieved through three-level technologies of visual analysis - defect positioning - intelligent crushing, solving the problems raised in the above background technique.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] A convenient wood shavings cake sampling device comprises a frame base, a retractable box is provided on one side of the top of the frame base, a frame body is provided above the retractable box, a linear guide is provided on the side wall of the frame body facing the frame base, an electric lifting mechanism is provided on the linear guide, a retractable mechanical arm is installed on the electric lifting mechanism, an electromagnetic suction cup is provided at the end of the retractable mechanical arm, a transverse slide rail fixing frame is provided above the frame body, a transverse driving device is installed on the bottom surface of the transverse slide rail fixing frame through the slide rail, a multi-stage retractable arm is provided at the bottom of the transverse driving device, a hydraulic crushing pick is installed at the bottom end of the multi-stage retractable arm, an image acquisition device for real-time image acquisition of the scrap steel cake block on the electromagnetic suction cup is also provided at the bottom of the transverse driving device, and the image acquisition device is located on one side of the multi-stage retractable arm;

[0009] The image acquisition device is used to collect the surface cracks and iron oxide distribution characteristics of the scrap steel in real time, and transmit the image to the edge computing terminal;

[0010] The edge computing terminal is equipped with a deep learning model based on a convolutional neural network target detection framework and a metal defect detection algorithm based on an attention mechanism, which is used to analyze the morphological characteristics of scrap steel transmitted by an image acquisition device in real time and generate three-dimensional crushing path planning instructions.

[0011] Preferably, the retractable box includes a fixed box, a movable box, a folding connecting rod group and a hydraulic cylinder. The fixed box is fixed to the upper surface of the frame base by bolts. The folding connecting rod group consists of two groups of V-shaped hydraulic rods, one end of which is hinged to the inner wall of the fixed box and the other end is hinged to the top of the movable box. The hydraulic cylinder is horizontally arranged between the fixed box and the frame body, and the end of the piston rod is hinged to the intersection of the folding connecting rod group.

[0012] Preferably, a battery is connected to the side of the frame body facing away from the linear guide rail, an armrest assembly is installed on the side of the battery facing away from the frame body, a control handle and an edge computing terminal are provided on the top of the armrest assembly, and universal wheels with brakes are additionally installed at the bottom of the armrest assembly and the frame base.

[0013] Preferably, the edge computing terminal is signal-connected to the image acquisition device, the hydraulic breaker pick head, the lateral drive device and the multi-stage retractable arm.

[0014] A convenient wood chip cake sampling method is implemented based on a convenient wood chip cake sampling device, comprising the following steps:

[0015] Step 1: Use the electromagnetic suction cup to suck the scrap steel cake to the top of the retractable box;

[0016] Step 2: The image acquisition device obtains the scrap steel surface image and transmits it to the edge computing terminal;

[0017] Step 3: The edge computing terminal identifies the crack and iron oxide defect areas through a composite attention mechanism and generates a three-dimensional crushing path instruction;

[0018] Step 4: The edge computing terminal outputs an instruction and controls the lateral driving device to move along the lateral slide rail, and impacts the scrap steel defect area according to the path instruction through the cooperation of the multi-stage telescopic arm and the crushing pick;

[0019] Step 5: The crushed scrap steel falls into the telescopic box to complete sampling.

[0020] Preferably, for step 2, the surface image of the scrap steel cake is collected in real time by an RGB-D depth camera and input into the edge computing terminal, and the image is normalized by the following formula:

[0021]

[0022] where μ k , σ k are the training set statistics, i: the index in the height direction (1 ≤ i ≤ H), j: the index in the width direction (1 ≤ j ≤ W), k: the index in the channel direction (1 ≤ k ≤ C).

[0023] Preferably, for step 3, the defect features are extracted through a composite attention mechanism, and the following formula is used:

[0024] Spatial attention module calculation formula:

[0025] W s = Softmax(Γ(μ(F)||ε(F)))

[0026] Channel attention module calculation formula:

[0027] W c = σ(MLP(P_a(F)) + MLP'(P_a(F)))

[0028] where F is the input feature map, Γ(·) is a 3*3 depthwise separable convolution, || represents channel dimension concatenation, μ(·), ξ(·) represent average pooling and max pooling along the channel dimension respectively, and σ() is the sigmoid function;

[0029] P_a() represents global hybrid pooling: α·AvgPool(·) + (1 - α)·MaxPool(·), α ∈ [0.4, 0.6]; AvgPool calculates the average color of each area of the photo, MaxPool is the brightest point in each area of the photo, and MLP and MLP' are two-path fully connected layers with non-shared parameters;

[0030] When max(Fout⊙Ws)>θdefect, the area where the maximum value exceeds the threshold after the feature map is multiplied by the spatial weight is marked as the structurally weak area, where θdefect = 0.7 is determined by the ROC curve;

[0031] Preferably, for what is described in step three, the generation process of the three-dimensional fragmentation path is specifically as follows:

[0032] The scrap steel briquette is divided into 10mm×10mm×10mm voxel units;

[0033] Calculate the impact priority:

[0034] P(v) = α·D(v)+β·S(v)

[0035] Where: v is the voxel unit, generated based on the depth camera point cloud data; D(v) is the distance from the voxel v to the surface, obtained through the following formula:

[0036]

[0037] d max is the maximum thickness of the briquette, and d(v) is the actually measured distance;

[0038] S(v) is the defect confidence (0 - 1), which is the output value from the attention module. The larger the value, the higher the defect probability. The output value of the spatial attention weight W s α = 0.6, β = 0.4 are the weight coefficients, indicating that internal defects are preferred to surface defects for fragmentation, and are determined through experimental optimization;

[0039] Generate a B-spline curve path:

[0040]

[0041] Where: B(t) is the parametric curve coordinate control point, representing the movement trajectory of the fragmentation pick. P i is the coordinate of the i-th control point, corresponding to the center of the voxel with the highest priority. N i,p (t) is the p-th B-spline basis function, which determines the curve smoothness. The fragmentation pick is controlled to impact the briquette cake along the planned path B(t) through the above function.

[0042] Compared with the prior art, the beneficial effects of the present invention are:

[0043] 1. The present invention adsorbs scrap steel compacts through an electromagnetic chuck and automatically performs image acquisition, defect identification, and intelligent crushing. This process is fully automated, greatly reducing the need for manpower. Compared with manual operation, the present invention can significantly improve the speed of scrap steel processing. From the real-time acquisition of the surface cracks and the distribution characteristics of iron oxide on the scrap steel, to the analysis based on a deep learning model and the generation of three-dimensional crushing path planning instructions, and then to the final automatic crushing and sampling, the entire process is efficient and fast. The installation of universal wheels makes the device easier to push and carry, making it more convenient.

[0044] 2. By adopting a method combining a depth camera and a compound attention mechanism, the present invention can accurately identify the cracks and the distribution of iron oxide on the surface of scrap steel, ensuring the accuracy of the crushing position. At the same time, through B-spline curve path planning, an effective impact on the pressed cake of the briquette is achieved, further improving the accuracy and reliability of sampling. Whether it is a shaving briquette or a particle steel briquette, this device can effectively complete the extraction and crushing of samples, and is applicable to various types of scrap steel materials, with strong versatility and practicality. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 is an isometric view of the overall structure of the present invention;

[0046] Figure 2 is a schematic diagram of the telescopic box of the present invention;

[0047] Figure 3 is a flowchart of the working process of the present invention.

[0048] In the figure: 1. Frame base; 2. Universal wheel with brake; 3. Telescopic box; 3a. Fixed box; 3b. Movable box; 3c. Folding link group; 3d. Hydraulic cylinder; 4. Frame main body; 5. Edge computing terminal; 6. Battery; 7. Armrest assembly; 8. Control handle; 9. Electric lifting mechanism; 10. Telescopic robotic arm; 11. Electromagnetic chuck; 12. Linear guide rail; 13. Hydraulic crushing pick; 14. Multi-stage telescopic arm; 15. Lateral driving device; 16. Image acquisition device; 17. Lateral slide rail fixing bracket. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0049] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0050] To solve the problem that the existing work of moving, crushing, and sampling and detecting scrap steel compacts and briquettes is not convenient and fast enough, please refer to Figures 1-3, this embodiment provides the following technical solutions:

[0051] A portable particle cake sampling device, including a frame base 1. On one side of the top of the frame base 1, there is a telescopic box body 3. Above the telescopic box body 3, there is a frame main body 4. On the side wall of the frame main body 4 facing the frame base 1, there is a linear guide rail 12. The surface of the linear guide rail 12 is hardened, and the friction coefficient ≤ 0.15. An electric lifting mechanism 9 is arranged on the linear guide rail 12. The electric lifting mechanism 9 can adopt a scissor-type electric elevator, a wall-mounted electric elevator or a boom-type electric elevator. A telescopic robotic arm 10 is installed on the electric lifting mechanism 9. The telescopic robotic arm 10 has three degrees of freedom, namely telescopic movement, pitching movement and rotational movement. An electromagnetic chuck 11 is arranged at the end of the telescopic robotic arm 10 with movement function.

[0052] The electromagnetic chuck 11 adopts rare earth permanent magnet materials and is equipped with a power-off protection device. The power-off protection device includes a supercapacitor energy storage module, which can maintain the magnetic force ≥ 5 seconds after power-off. The electromagnetic chuck 11 is connected to the telescopic robotic arm 10 through a universal joint mechanism and has a multi-directional rotation freedom of ± 30°. When the scrap steel is at a lower position on the ground or on a non-horizontal surface, the electromagnetic chuck 11 can adaptively adjust the adsorption angle to ensure full contact with the surface of the scrap steel; during the adsorption operation, the telescopic robotic arm 10 can drive the electromagnetic chuck 11 to lower to a minimum height of 1.2 m along a hook-type movement trajectory. After the adsorption is completed, it automatically resets to a suitable working position through an electric push rod arranged at the connection between the electromagnetic chuck 11 and the telescopic robotic arm 10, and the reset accuracy is ± 2°.

[0053] Above the frame main body 4, there is a transverse slide rail fixing frame 17. On the bottom surface of the transverse slide rail fixing frame 17, a transverse driving device 15 is installed through a slide rail. The transverse driving device 15 can achieve displacement at any point on the x-axis and y-axis in cooperation with the slide rail, and can adopt the XY workbench system used in laser cutting machines and 3D printers.

[0054] At the bottom of the transverse driving device 15, there is a multi-stage telescopic arm 14. At the bottom end of the multi-stage telescopic arm 14, a hydraulic crushing pick 13 is installed. The hydraulic crushing pick 13 is made of high-carbon alloy steel to adapt to high-frequency impacts; at the bottom of the transverse driving device 15, there is also an image acquisition device 16. The image acquisition device 16 is located on one side of the multi-stage telescopic arm 14 and is used to collect the crack and iron oxide distribution characteristics on the surface of the scrap steel in real time and transmit the image to the edge computing terminal 5.

[0055] The edge computing terminal 5 is equipped with a deep learning model based on a target detection framework of a convolutional neural network and deploys a metal defect detection algorithm based on an attention mechanism, which is used to analyze the morphological characteristics of the scrap steel transmitted by the image acquisition device 16 in real time and generate a three-dimensional crushing path planning instruction.

[0056] The telescopic box 3 includes a fixed box 3a, a movable box 3b, a folding connecting rod group 3c and a hydraulic cylinder 3d. The fixed box 3a is fixed to the upper surface of the frame base 1 by bolts. The folding connecting rod group 3c is composed of two groups of V-shaped hydraulic rods, one end of which is hinged to the inner wall of the fixed box 3a and the other end is hinged to the top of the movable box 3b. The hydraulic cylinder 3d is horizontally arranged between the fixed box 3a and the frame body 4, and the end of the piston rod is hinged to the intersection of the folding connecting rod group 3c.

[0057] The frame body 4 is integrated with the frame base 1. The middle of the connection between the bottom of the frame body 4 and the frame base is hollowed out. The hollowed out part is provided with a fixed box 3a, a folding connecting rod group 3c, a hydraulic cylinder 3d, etc., and the movable box 3b is detachable on the fixed box 3a.

[0058] A battery 6 is connected to the side of the frame body 4 facing away from the linear guide rail 12. The battery module 6 can work continuously for 4-6 hours when fully charged. An armrest assembly 7 is installed on the side of the battery 6 facing away from the frame body 4. A control handle 8 and an edge computing terminal 5 for operating the equipment are provided on the top of the armrest assembly 7. The control panel 8 is provided with an emergency stop button, a mode selection switch and a parameter adjustment knob. A universal wheel 2 with a brake is additionally installed at the bottom of the armrest assembly 7 and the frame base 1.

[0059] The edge computing terminal 5 is signal-connected to the image acquisition device 16 , the hydraulic breaker pick head 13 , the lateral drive device 15 and the multi-stage retractable arm 14 .

[0060] A convenient wood chip cake sampling method is implemented based on a convenient wood chip cake sampling device, comprising the following steps:

[0061] Step 1: The scrap steel cake is sucked to the top of the retractable box 3 by the electromagnetic suction cup 11;

[0062] Step 2: The surface image of the scrap steel cake is collected in real time by using an RGB-D depth camera, input to the edge computing terminal 5, and the image is normalized by the following formula:

[0063]

[0064] where μ k , σ k is the statistics of the training set, i: index in the height direction (1≤i≤H), j: index in the width direction (1≤j≤W), k: index in the channel direction (1≤k≤C);

[0065] Step 3: The edge computing terminal 5 identifies cracks and iron oxide defect areas through a composite attention mechanism and generates a three-dimensional crushing path instruction, specifically:

[0066] Defect features are extracted through the composite attention mechanism using the following formula:

[0067] Calculation formula of spatial attention module:

[0068] W s = Softmax(Γ(μ(F)||ε(F)))

[0069] Calculation formula of channel attention module:

[0070] W c = σ(MLP(P_a(F)) + MLP′(P_a(F)))

[0071] Where F is the input feature map, Γ(·) is a 3×3 depthwise separable convolution, || represents concatenation along the channel dimension, μ(·) and ξ(·) represent average pooling and max pooling along the channel dimension respectively, and σ() is the sigmoid function;

[0072] P_a() represents global mixed pooling: α·AvgPool(·) + (1 - α)·MaxPool(·), α ∈ [0.4, 0.6]; AvgPool calculates the average color of each region of the photo, MaxPool is the brightest point in each region of the photo, and MLP and MLP' are two-path fully connected layers with non-shared parameters;

[0073] When the maximum value exceeds the threshold after multiplying the feature map by the spatial weight, i.e., max(Fout⊙Ws) > θdefect, it is marked as a structurally weak area, where θdefect = 0.7 is determined by the ROC curve;

[0074] The generation process of the three-dimensional fragmentation path is specifically as follows:

[0075] The scrap steel briquette is divided into 10mm×10mm×10mm voxel units;

[0076] Calculate the impact priority:

[0077] P(v) = α·D(v) + β·S(v)

[0078] Where: v is the voxel unit, generated based on the depth camera point cloud data; D(v) is the distance from voxel v to the surface, obtained through the following formula:

[0079]

[0080] d max is the maximum thickness of the briquette, and d(v) is the actual measured distance;

[0081] S(v) is the defect confidence (0 - 1), which is the output value from the attention module. The larger the value, the higher the defect probability, and the spatial attention weight W sThe output value, where α = 0.6 and β = 0.4 are weight coefficients, indicating that internal defects are given priority over surface defects in fragmentation, determined through experimental optimization;

[0082] Generate a B-spline curve path:

[0083]

[0084] Where: B(t) is the parametric curve coordinate control point, representing the movement trajectory of the crushing pick 13, P i is the coordinate of the i-th control point, corresponding to the center of the voxel with the highest priority, N i,p (t) is the p-th B-spline basis function, which determines the curve smoothness. The crushing pick 13 is controlled to impact the briquette and cake along the planned path B(t) through the above function;

[0085] Step 4: The edge computing terminal 5 outputs instructions and controls the lateral driving device 15 to move along the lateral slide rail, and impacts the scrap steel defect area according to the path instructions through the cooperation of the multi-stage telescopic arm 14 and the crushing pick 13;

[0086] Step 5: The crushed scrap steel falls into the telescopic box 3 to complete sampling.

[0087] In Steps 2 and 3, it is also necessary to establish a crushing model to improve the crushing generalization ability. The specific method is as follows:

[0088] S1. Establish a scrap steel briquette and cake dataset by collecting image data of different scrap steel briquettes and cakes on-site and image data of the on-site stacking of scrap steel briquettes and cakes; establish a scrap steel crushing dataset by collecting images of different scrap steel briquettes and cakes after crushing;

[0089] S2. Use labelimg to identify the scrap steel briquettes and cakes in each picture of the scrap steel briquette and cake dataset in Step S1;

[0090] S3. Use labelimg to identify the scrap steel crushing dataset in Step S1, and the annotation information is the scrap steel crushing situation in the picture;

[0091] S4. Use the Jetson-OPTIMIZED CutMix technology to perform image enhancement on the scrap steel briquette and cake dataset after identification in Step S2 and the scrap steel crushing dataset after identification in Step S3 to improve the diversity of samples and the generalization ability of the recognition model;

[0092] S5. Build a deep learning framework for object detection based on a convolutional neural network on a PC, and perform multiple rounds of training on the scrap steel briquette and cake dataset and the scrap steel crushing dataset after image enhancement in Step S4 respectively to obtain a scrap steel briquette and cake recognition model and a scrap steel crushing recognition model;

[0093] S6. Save the scrap cake briquetting recognition model and the scrap crushing recognition model in step S5 in the SD card storage module of the edge computing terminal (17).

[0094] In step S4, the CutMix image enhancement generation formula is as follows:

[0095]

[0096] In the formula: is the input vector, is the data label, (xi, yi) and (xj, yj) are two randomly selected samples and corresponding labels in the same batch, M is a binary mask matrix (0 / 1 matrix), generated by the CUDA kernel function, where ° represents the per-pixel multiplication accelerated by Tensor Core, N(0, 0.1 2 ) is Gaussian noise, the standard deviation σ = 0.1σ = 0.1, and λ is a number randomly sampled from the Beta distribution, λ ∈ Beta[0, 1].

[0097] When training in step S5, the Focal Loss loss function is used, and its expression is:

[0098]

[0099] In the formula: P t is the predicted probability of the model for the true class:

[0100] When the sample is a positive class (the area to be crushed): P t = P

[0101] When the sample is a negative class (background / non-target): P t = 1 - P

[0102] α t is the class weight coefficient, γ is the difficult sample focusing factor, ∈ = 0.1κ = 1.5 is the high-order balance term.

[0103] In step S2, the method for recognizing scrap cake briquetting and in step S3, the method for recognizing scrap crushing is as follows: Use the object detection model based on the convolutional neural network object detection framework as the backbone network, use BCELoss as the classification Loss function, and for each class, the predicted probabilities are p and 1 - p. The BCELoss function calculates the probabilities of both being this class and not being this class, and the losses calculated in these two cases will not be 0:

[0104]

[0105] N is the number of samples, yi is the true label of the i-th sample, and pi is the probability that the model predicts the i-th sample as the positive class (the base of the log is e).

[0106] And DFLLoss and EIoULoss are combined as the bounding box regression loss function. DFLLoss dynamically adjusts the weights of targets of different sizes, making the model more accurate in detecting targets of different sizes; while EIoULoss decouples the width and height errors based on CIoULoss, and further improves the localization accuracy of the broken area by independently optimizing the regression terms in the width and height directions, thus improving the accuracy of bounding box regression.

[0107]

[0108] yi and yi+1 are the left and right integer values of the floating-point value y, S is the output distribution, and DFL enables the network to more quickly focus on the values near the target y and increase their probabilities.

[0109]

[0110] ρ 2 (b, bgt) represents the square of the Euclidean distance between the centers of the predicted box and the ground truth box, ρ 2 (w, w gt ) represents the square difference between the predicted width and the ground truth width, ρ 2 (h, h gt ) represents the square difference between the predicted height and the ground truth height, c represents the diagonal distance of the smallest closed region that can contain both the predicted box and the ground truth box. Among them, the BCELoss function is used to accurately distinguish the briquette area from the background, avoiding misstriking non-target areas by the broken pickaxe. The DFL function is used to dynamically focus on the edge features of the briquette and reduce the error value of the bounding box determination; the EIoU Loss function is used to independently optimize the width / height regression, accurately select the defect area, and reduce the localization error of the briquette cavity.

[0111] Working principle: The portable shaving cake sampling device mainly focuses on automatic sampling and intelligent crushing, and realizes efficient processing through three-level technologies of visual analysis, defect positioning and intelligent crushing. First, the electromagnetic chuck 11 adsorbs the scrap steel cake above the telescopic box body 3; then, the image acquisition device 16 collects the surface crack and iron oxide distribution characteristics of the scrap steel in real time and transmits this information to the edge computing terminal 5. The edge computing terminal is equipped with a deep learning model based on the object detection framework of convolutional neural network, and deploys a metal defect detection algorithm based on the attention mechanism to analyze the image data and identify defect areas such as cracks and iron oxide, and generate a three-dimensional crushing path planning instruction. Subsequently, according to the instruction, the lateral driving device 15 moves along the lateral slide rail fixing frame 17, and the multi-stage telescopic arm 14 cooperates with the hydraulic crushing pick 13 to accurately impact and crush the scrap steel according to the planned path, and preferentially processes the internal defect areas. During the whole process, the crushed scrap steel automatically falls into the telescopic box body 3 to complete sampling. In addition, the braking universal wheels 2 and the handrail assembly 7 on the vehicle frame base 1 facilitate the overall movement and operation control of the equipment.

[0112] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device.

[0113] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principle and spirit of the present invention.

Claims

1. A convenient wood chip cake sampling device, characterized in that: The invention comprises a frame base (1), wherein a telescopic box (3) is provided on one side of the top of the frame base (1), a frame body (4) is provided above the telescopic box (3), a linear guide rail (12) is provided on the side wall of the frame body (4) facing the frame base (1), an electric lifting mechanism (9) is provided on the linear guide rail (12), a telescopic mechanical arm (10) is installed on the electric lifting mechanism (9), an electromagnetic suction cup (11) is provided at the end of the telescopic mechanical arm (10), and the frame body (4) A transverse slide rail fixing frame (17) is arranged on the top, a transverse driving device (15) is installed on the bottom surface of the transverse slide rail fixing frame (17) through a slide rail, a multi-stage telescopic arm (14) is arranged at the bottom of the transverse driving device (15), a hydraulic breaker head (13) is installed at the bottom end of the multi-stage telescopic arm (14), and an image acquisition device (16) is also arranged at the bottom of the transverse driving device (15), and the image acquisition device (16) is located on one side of the multi-stage telescopic arm (14); The image acquisition device (16) is used to collect the surface cracks and iron oxide distribution characteristics of the scrap steel in real time, and transmit the image to the edge computing terminal (5); The edge computing terminal (5) is equipped with a deep learning model based on a convolutional neural network target detection framework and is equipped with a metal defect detection algorithm based on an attention mechanism, which is used to analyze the morphological features of scrap steel transmitted by an image acquisition device (16) in real time and generate three-dimensional crushing path planning instructions.

2. A portable wood chip cake sampling device according to claim 1, characterized in that: The telescopic box (3) comprises a fixed box (3a), a movable box (3b), a folding connecting rod group (3c) and a hydraulic cylinder (3d); the fixed box (3a) is fixed to the upper surface of the frame base (1) by bolts; the folding connecting rod group (3c) is composed of two groups of V-shaped hydraulic rods, one end of which is hinged to the inner wall of the fixed box (3a) and the other end is hinged to the top of the movable box (3b); the hydraulic cylinder (3d) is horizontally arranged between the fixed box (3a) and the frame body (4); the end of the piston rod is hinged to the intersection of the folding connecting rod group (3c).

3. A portable wood chip cake sampling device according to claim 2, characterized in that: A battery (6) is connected to the side of the frame body (4) facing away from the linear guide rail (12); an armrest assembly (7) is installed on the side of the battery (6) facing away from the frame body (4); a control handle (8) and an edge computing terminal (5) are arranged on the top of the armrest assembly (7); and a universal wheel (2) with a brake is additionally installed at the bottom of the armrest assembly (7) and the frame base (1).

4. A portable wood chip cake sampling device according to claim 3, characterized in that: The edge computing terminal (5) is signal-connected to the image acquisition device (16), the hydraulic breaker pick head (13), the lateral drive device (15) and the multi-stage retractable arm (14).

5. A convenient wood chip cake sampling method, implemented based on a convenient wood chip cake sampling device according to any one of claims 1 to 4, characterized in that: The following steps are involved: Step 1: Using the electromagnetic suction cup (11) to suck the scrap steel cake to the top of the telescopic box (3); Step 2: The image acquisition device (16) acquires the surface image of the scrap steel and transmits it to the edge computing terminal (5); Step 3: The edge computing terminal (5) identifies cracks and iron oxide defect areas through a composite attention mechanism and generates a three-dimensional crushing path instruction; Step 4: The edge computing terminal (5) outputs instructions and controls the transverse driving device (15) to move along the transverse slide rail, and impacts the defective area of ​​the scrap steel according to the path instruction through the cooperation of the multi-stage retractable arm (14) and the crushing pick head (13); Step 5: The crushed scrap steel falls into the retractable box (3), completing the sampling.

6. A convenient wood chip cake sampling method according to claim 5, characterized in that: For step 2, the surface image of the scrap steel cake is collected in real time by using an RGB-D depth camera and input into the edge computing terminal (5). The image is normalized by the following formula: where μ k , σ k is the statistics of the training set, i: index in the height direction (1≤i≤H), j: index in the width direction (1≤j≤W), k: index in the channel direction (1≤k≤C).

7. A convenient wood chip cake sampling method according to claim 6, characterized in that: For step 3, the defect features are extracted through the composite attention mechanism using the following formula: Spatial attention module calculation formula: Channel attention module calculation formula: W c nσ(MLP(P_a(F))+MLP′(P_a(F))) Where F is the input feature map, is a 3*3 depthwise separable convolution, || represents channel dimension concatenation, μ(·) and ξ(·) represent average pooling and maximum pooling along the channel dimension, respectively, and σ() is a sigmoid function; P_a() represents global mixed pooling: α·AvgPool(·)+(1-α)·MaxPool(·), α∈[0.4,0.6]; AvgPool is used to calculate the average color of each area in the photo, MaxPool is the brightest point in each area of ​​the photo, MLP and MLP' are two-way fully connected layers with no shared parameters; When max(Fout⊙Ws)>θdefect, the maximum value after multiplication of the feature map and the spatial weight exceeds the threshold and is marked as a structural weak area, where θdefect=0.7 is determined by the ROC curve.

8. A convenient wood chip cake sampling method according to claim 7, characterized in that: For step 3, the generation process of the three-dimensional fragmentation path is as follows: Divide the scrap steel briquettes into 10mm×10mm×10mm voxel units; Calculate impact priority: P(v)=α·D(v)+β·S(v) Where: v is the voxel unit, generated based on the depth camera point cloud data; D(v) is the distance from voxel v to the surface, obtained by the following formula: d max is the maximum thickness of the pressing block, d(v) is the actual measurement distance; S(v) is the defect confidence (0-1) from the output value of the attention module. The larger the value, the higher the defect probability. The spatial attention weight W s The output value of , α=0.6, β=0.4 are weight coefficients, indicating that internal defects are crushed before surface defects, which are determined by experimental optimization; Generate a B-spline curve path: Where: B(t) is the parameterized curve coordinate control point, representing the motion trajectory of the crushing pick (13), P i is the coordinate of the ith control point, corresponding to the voxel center with the highest priority, N i,p (t) is a p-order B-spline basis function, which determines the smoothness of the curve. The above function is used to control the crushing pick (13) to impact the pressed cake along the planned path B(t).

Images