Hog house ground excrement cleaning equipment and target detection system for intelligently identifying residual excrement

Abstract

The invention discloses hog house ground excrement cleaning equipment and a target detection system for intelligently recognizing residual excrement. The hog house ground excrement cleaning equipment comprises a robot body, and an extrusion type collecting mechanism used for collecting ground lump-shaped excrement into a container and a flexible pier laying mechanism used for adhering and cleaning non-lump-shaped excrement on the ground are arranged at the bottom of the robot body. By the adoption of the scheme, fluid feces can be smoothly cleaned in the pig lump-shaped feces cleaning process, collection and cleaning of the fluid feces can be smoothly achieved, the flexible pier laying mechanism can be conveniently, rapidly and smoothly cleaned, meanwhile, the spreading difficulty of the flexible pier laying mechanism is remarkably reduced, and the working efficiency of the flexible pier laying mechanism is improved. The technical problems that lump excrement in the pig house is difficult to clean smoothly and the residual amount is large are solved; the ground image of the hog house is collected in real time through the camera, the position and the form of residual excrement are rapidly positioned, and data support is provided for secondary cleaning of equipment.

Description

Technical Field

[0001] This invention belongs to the technical field of pig farm manure cleaning equipment, specifically relating to a pigsty floor manure cleaning equipment and an intelligent target detection system for identifying residual manure. Background Technology

[0002] There are two main types of existing pigsty floor manure removal equipment: scraper-type and manure collection-type. Manure collection-type equipment primarily uses two methods: horizontal auger and negative pressure adsorption. The horizontal auger type uses a motor to drive a horizontal auger mechanism to push manure from the scraper into the storage bin. However, this method has a generally poor manure collection effect, with a large amount of residue (when cleaning clumps of manure, the residual manure accounts for approximately 30% of the initial manure volume). The negative pressure adsorption type is mainly suitable for fluid manure. However, due to the viscosity of pig manure, especially clumps, it often fails to adsorb the manure. Furthermore, this method is noisy and can easily cause stress to pigs. More importantly, under normal circumstances, pig manure is clump-shaped and sticky, and it doesn't disperse for a day or several days, making it difficult to clean effectively. In addition, the area around the clumps of pig feces contains a small amount of fluid fecal matter formed by pig urine and feces. How to effectively remove the fluid fecal matter when cleaning up the clumps of pig feces is also a problem that needs to be solved. Summary of the Invention

[0003] In order to solve the technical problems mentioned in the background art, the present invention aims to provide a pigsty floor manure cleaning device and an intelligent target detection system for identifying residual manure.

[0004] The present invention adopts the following technical solution.

[0005] A pigsty floor manure cleaning device includes a robot body, a squeezing collection mechanism at the bottom of the robot body for collecting clumps of manure on the ground into a container, and a flexible mopping mechanism for cleaning non-clumps of manure adhering to the ground.

[0006] Furthermore, the flexible pier paving mechanism includes vertically arranged connecting rods, with the upper end of the connecting rods connected to a connecting rod lifting mechanism, and the lower end of the connecting rods connected to several independent flexible strip materials. A sleeve is fitted on the connecting rods and is fixedly installed. When the connecting rods move to the upper target position, all the flexible strip materials are located in the sleeves in a compacted state. When the connecting rods move to the lower target position, all the flexible strip materials are radially dispersed.

[0007] Preferably, the flexible strip material is made of bundled cotton or bundled fibers with a diameter of no more than 5 mm.

[0008] One of the more preferred embodiments is that each flexible strip has a beryllium copper wire with a diameter of 1 to 2 mm arranged axially inside it.

[0009] A more preferred embodiment is that the flexible strip material is made of sponge with a diameter of no more than 15mm, and each flexible strip material is axially provided with a beryllium copper wire spring with a diameter of 3 to 8mm, and sponge material is provided inside and outside the spring.

[0010] Furthermore, when the flexible strip material is in its initial state without external force, the beryllium copper wire or beryllium copper wire spring has an arched structure, with the apex of the arch connected to the bottom of the connecting rod. Preferably, the angle between the inclined surface of the arched structure and the horizontal plane is 30–50°. This design not only enables the smooth collection and cleaning of fluid-state sewage, but also allows for the quick and easy removal of sewage adhering to the flexible paving mechanism. This facilitates rapid and smooth cleaning of the flexible paving mechanism and significantly reduces the difficulty of its deployment.

[0011] Furthermore, the compression collection mechanism includes a vertical telescopic mechanism. The lower end of the telescopic rod of the vertical telescopic mechanism is connected to the top of the cylinder. The cylinder is arranged vertically, and the bottom wall of the cylinder is provided with several arrayed through holes. The cylinder serves as a container for collecting and storing pig manure, and the through holes also serve as manure outlets. A piston is installed in the inner cavity of the cylinder, and the piston is connected to a telescopic device. The telescopic device drives the piston to move axially within the inner cavity of the cylinder. The through holes have a structure that is thick at both ends and thin in the middle. The cylinder is made of polytetrafluoroethylene material.

[0012] In order to promptly handle residual manure during the cleaning of clumps of manure, the flexible paving mechanism is located behind the cylinder during the movement of the manure cleaning equipment on the pigsty floor.

[0013] Furthermore, the device also includes a target detection system for intelligently identifying residual feces, comprising: an image acquisition module for acquiring image data of the pigsty floor in real time via a camera to generate a raw image dataset; an image preprocessing module for performing image enhancement processing on the raw image dataset; an image feature extraction module for inputting the normalized image into a lightweight detection model, the model being built based on the YOLOv11 architecture; a target detection module for performing multi-scale target recognition operations on the fused feature map, outputting the location coordinates and category probability distribution of the fecal targets, and deleting redundant boxes using a non-maximum suppression algorithm to obtain the final residual feces detection result; and a cleaning guidance module for sending the target location data and fecal category data from the detection result to the cleaning control module to drive the manure cleaning equipment to perform targeted secondary cleaning operations.

[0014] Beneficial effects: The solution of this invention can smoothly remove fluid manure during the cleaning of clumps of pig manure, and can smoothly collect and clean fluid manure. It facilitates quick and easy cleaning of the flexible manure paving mechanism, while significantly reducing the difficulty of spreading the flexible manure paving mechanism. This invention solves the technical problem of difficult cleaning of clumps of manure and large residual amount in pigsties with a minimal structure. It eliminates the need for pig manure identification sensors and corresponding algorithms, greatly reducing the cost and implementation difficulty of manure cleaning equipment. Moreover, it has the advantages of low noise and low stress on pigs during the manure cleaning process. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the external structure of the pigsty floor manure cleaning equipment in the embodiment; Figure 2 This is a schematic diagram of the internal structure of the pigsty floor manure cleaning equipment in the embodiment (the flexible paving mechanism is in its initial state); Figure 3 This is a schematic cross-sectional view of the pigsty floor manure cleaning equipment in the embodiment (the flexible pier mechanism is in its initial state); Figure 4 This is a schematic diagram of the flexible paving mechanism of the pigsty floor manure cleaning equipment in the embodiment (manure cleaning state); Figure 5 This is a three-dimensional schematic diagram (in its stowed state) of the flexible slab laying mechanism of the pigsty floor manure cleaning equipment in the embodiment; Figure 6 This is a side view (folded-up state) of the flexible paving mechanism of the pigsty floor manure cleaning equipment in the embodiment; Figure 7 This is a schematic diagram of the flexible strip material in the flexible pier laying mechanism of Example 2; Figure 8 This is a schematic diagram of the through hole in the bottom wall of the cylinder in the embodiment; Figure 9 This is a schematic diagram of the network structure in the embodiment. Detailed Implementation

[0016] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. In the present invention, the power source in the robot body can be a lithium battery (not shown in the figure), and the telescopic mechanism can be an electrically driven telescopic mechanism. Example 1

[0017] Combination Figures 1-6 and Figure 8As shown, a pigsty floor manure cleaning device includes a robot body 10. The robot body 10 is a tracked robot (i.e., the walking wheels of the robot body 10 adopt a tracked structure). A compression collection mechanism is set at the bottom of the robot body 10 to collect pig manure in a container by compression. In this embodiment, the compression collection mechanism includes a vertical telescopic mechanism 1. The lower end of the telescopic rod of the vertical telescopic mechanism 1 is connected to the top of the cylinder 2 (the lower end of the telescopic rod is fixedly connected to a connector 7, which is fixedly connected to the top of the cylinder 2 and has a gate-shaped structure). The cylinder 2 is vertically arranged, and the bottom wall of the cylinder 2 is provided with several arrayed through holes 3. The cylinder 2 is made of polytetrafluoroethylene material, the wall thickness of the side wall of the cylinder 2 is 8mm, and the wall thickness of the bottom wall of the cylinder 2 is 15mm. The cylinder 2 serves as a container for collecting and storing pig manure, and the through holes 3 also serve as manure outlets. A piston 4 is provided in the inner cavity of the cylinder 2, and the piston 4 is connected to a telescopic device 5. The telescopic device 5 is installed on the connector 7, and the telescopic rod of the telescopic device 5 penetrates the top wall of the cylinder 2. The piston 4 is driven to move axially in the inner cavity of the cylinder 2 through the telescopic device 5. The through holes 3 have a structure that is thick at both ends and thin in the middle. The maximum diameter of the through holes 3 is 22mm, the spacing between adjacent through holes 3 is 10mm, and the diameter of the middle part of the through holes 3 is 12mm.

[0018] In this embodiment, the pigsty floor manure cleaning equipment also includes a flexible paving mechanism for adhering and cleaning non-clump-shaped manure on the ground. The flexible paving mechanism is located behind the cylinder 2 (relative to the movement of the pigsty floor manure cleaning equipment). The flexible paving mechanism includes a vertically arranged connecting rod 11. The upper end of the connecting rod 11 is connected to a connecting rod lifting mechanism 15 (using an electric push rod), and the lower end of the connecting rod 11 is connected to several independent flexible strips 12. A sleeve 13 is fitted onto the connecting rod 11 and is fixedly mounted / connected to the housing. When the connecting rod 11 moves to the target position, all the flexible strips 12 are compacted within the sleeve 13 (i.e., the flexible strips 12 are housed within the sleeve 13, such as...). Figure 5 and Figure 6 As shown), when the connecting rod 11 moves to the lower target position, all the flexible strips 12 spread out radially, at which point... Figure 4 As shown. In this embodiment, a beryllium copper wire with a diameter of 2mm is axially arranged inside the flexible strip material 12. The flexible strip material 12 is made by winding a bundle of cotton material with a diameter of 5mm around the beryllium copper wire. When the flexible strip material 12 is in its initial state without being subjected to external force, the beryllium copper wire has an arched structure (as shown in the figure). Figure 2 and Figure 3 As shown in the initial state, the arch of the arch structure is connected to the bottom of the connecting rod 11, and the angle between the inclined surface of the arch structure and the horizontal plane is 30°.

[0019] The method of using the pigsty floor manure cleaning equipment in this embodiment includes the following steps: Step 1: Control the pigsty floor manure cleaning equipment to move along a preset path in the pigsty's defecation area. The pigsty's defecation area is a solid floor / panel structure without slatted panels. Step 2: Control the vertical telescopic mechanism 1 to operate, causing the cylinder 2 to intermittently press down and lift at a set frequency. The cylinder 2 presses down at a frequency of 30 times / minute, and immediately lifts up after each press. Whenever the cylinder 2 presses down on pig manure, it picks up the pig manure into the inner cavity of the cylinder 2. When the cylinder 2 presses down for the first time, control the linkage lifting mechanism 15 to move down to the lower target position and then maintain this state. During this process, the flexible strip material 12 spreads radially and contacts the ground (e.g., Figure 4 (as shown); Step 3: When the pigsty floor manure cleaning equipment reaches its endpoint, control the telescopic device 5 to push out the manure from the inner cavity of the cylinder 2. The manure is pushed downwards from the manure outlet 6. Control the linkage lifting mechanism 15 to repeatedly move "up" and "down" (when the linkage 11 moves to the upper target position, all the flexible strip materials 12 are in a compacted state inside the sleeve 13, such as...). Figure 5 (As shown), the feces and dirt attached to the flexible pier paving mechanism are cleaned off. Example 2

[0020] A pigsty floor manure cleaning device, referring to Embodiment 1, differs from Embodiment 1 mainly in that: the robot body 10's walking wheels are made of wheat straw, combined with... Figure 7 As shown, the flexible strip material 12 is made of sponge with a diameter of 15mm. A beryllium copper wire spring 14 with a diameter of 8mm (the diameter of the beryllium copper wire is 1mm) is axially arranged inside the flexible strip material 12. Sponge material 16 is arranged inside and outside the beryllium copper wire spring 14 (that is, the sponge material 16 fills the inner cavity of the beryllium copper wire spring 14 and covers the outer periphery of the beryllium copper wire spring 14). When the flexible strip material 12 is in the initial state without being subjected to external force, the beryllium copper wire spring 14 has an arched structure, and the angle between the inclined surface of the arched structure and the horizontal plane is 40°. Example 3

[0021] A pigsty floor manure cleaning device, referring to Embodiment 1, differs from Embodiment 1 mainly in that: a beryllium copper wire with a diameter of 1mm is axially arranged inside the flexible strip material 12; when the flexible strip material 12 is in the initial state without being subjected to external force, the beryllium copper wire has an arched structure, and the angle between the inclined surface of the arched structure and the horizontal plane is 50°.

[0022] Using the pigsty floor manure cleaning equipment described in the embodiment, the pig manure (normal clump-shaped manure from pigs within one day of defecation) in the experimental area was cleaned: first, normal clump-shaped manure from pigs within one day of defecation was collected, and then it was placed at intervals in the experimental area. After cleaning, the residual amount / percentage of manure was counted (residual amount / percentage of manure = weight of residual manure in the experimental area after cleaning / weight of initial clump-shaped manure). The results showed that after cleaning using the scheme in Example 1, the residual amount of manure was less than 3%.

[0023] Some key technical points of this invention: On the one hand, whenever the cylinder 2 presses down on the lumpy feces, the lumpy feces are squeezed and enter / drill into the inner cavity of the cylinder 2 through the through hole 3. Even if part of the lumpy feces is flattened (the flattened feces) will not adhere to the outer wall of the cylinder 2. At this time, it is only necessary to slightly change the position of the cylinder 2 and then press down on the flattened feces again to further press most of the feces into the inner cavity of the cylinder 2. The feces collected in the inner cavity of the cylinder 2 are subsequently pushed out / squeezed out by the movement of the piston 4; On the other hand, in During the process of cleaning up the lumps of pig manure, the fluid manure was also cleaned up smoothly and incidentally (adhered to the flexible strip material 12 in the spreading state). Because a beryllium copper wire skeleton with a specific arch structure was implanted inside the flexible strip material 12, the flexible strip material 12 still has excellent spreading and storage performance after being compressed and adhering to fluid manure. It can smoothly collect and clean up fluid manure, making it convenient to clean the flexible block laying mechanism quickly and smoothly, while significantly reducing the difficulty of spreading the flexible block laying mechanism.

[0024] During use, the preset path and frequency of the pigsty floor manure cleaning equipment are set by those skilled in the art based on the actual conditions of the pigsty, but it is necessary to ensure that the trajectory of the lower end / bottom of the cylinder 2 can cover the entire area where pigs defecate.

[0025] This invention solves the technical problem of difficult and large amounts of lumpy manure in pigsties with a minimalist structure and method. It eliminates the need for pig manure identification sensors and corresponding identification algorithms, significantly reducing the structural cost, software cost, and implementation difficulty of the manure cleaning robot. It also has advantages such as low noise (noise level below 60 decibels) and minimal stress on pigs during the manure cleaning process. In particular, it has the advantage of good stability during the manure cleaning process, as the stability of the equipment is further enhanced each time the cylinder is pressed down. Using the solution of this invention, liquid manure can be successfully cleaned up during the cleaning of lumpy pig manure, and the collection and cleaning of liquid manure can be achieved smoothly. It facilitates the quick and smooth cleaning of the flexible paving mechanism, while significantly reducing the difficulty of spreading the flexible paving mechanism. Example 4

[0026] Traditional manure removal equipment lacks a closed-loop feedback mechanism for environmental perception and decision-making, making it impossible to dynamically adjust operational strategies based on real-time manure conditions. To ensure thorough cleaning of pen manure, this embodiment introduces an intelligent detection technology capable of real-time identification of residual manure. Unlike the previous embodiment, this equipment also includes an intelligent target detection system and method for identifying residual manure, providing guidance for secondary cleaning and addressing the inefficiency of traditional "one-size-fits-all" operation methods. This method uses a camera to collect real-time images of the pigsty floor, quickly locating the position and shape of residual manure, providing data support for secondary cleaning. This method overcomes the limitations of traditional mechanical operations, achieving an intelligent closed loop of cleaning and re-inspection, and improving the thoroughness and efficiency of pigsty manure cleaning.

[0027] The intelligent residual feces detection system includes: an image acquisition module for real-time acquisition of image data of the pigsty floor via a camera to generate a raw image dataset; an image preprocessing module for performing image enhancement processing on the raw image dataset; an image feature extraction module for inputting the normalized image into a lightweight detection model, which is built based on the YOLOv11 architecture; a target detection module for performing multi-scale target recognition operations on the fused feature map, outputting the location coordinates and category probability distribution of the feces target, and deleting redundant boxes through a non-maximum suppression algorithm to obtain the final residual feces detection result; and a cleaning guidance module for sending the target location data and feces category data from the detection result to the cleaning control module to drive the manure cleaning equipment to perform targeted secondary cleaning operations.

[0028] The target detection implementation method is as follows: Data collection: The dataset was used to train the target detection model. In the early stage, a total of 16,825 images of feces in pigsties were collected through photography, the Internet and other methods, including 7,434 images of lumpy feces and 9,391 images of fluid feces. Image preprocessing: To avoid overfitting during model training, improve model robustness, and enhance model generalization ability, this method introduces data augmentation techniques to simulate different lighting conditions, viewing angle differences, and shooting interference scenarios in pigsties. This expands the semantic diversity of images and the breadth of input distribution, enabling the model to maintain good detection accuracy on unseen samples. Image data augmentation: Geometric deformation: Elastic deformation simulates feces of different shapes, and random rotation enhances posture robustness; Photometric perturbation: Independent offset of RGB channels to simulate different lighting color temperatures, and calculation of the offset channel values. as follows: R new=clamp(R original +ΔR,0,255) G new =clamp(G original +ΔG,0,255) B new =clamp(B original +ΔB,0,255) Among them, R new G new B new R represents the channel value after offset. original G original B original The original channel value is represented by ΔR, ΔG, and ΔB, which represent the offset of each channel. clamp(x, a, b) restricts x to the range [a, b]. The above photometric perturbation design is based on the following logic: First, for each pixel channel, controllable brightness offsets ΔR, ΔG, and ΔB are added to make the pixel brightness distribution closer to the real image under different times, different lighting conditions, or occlusion scenarios; Second, the clamp function is introduced to perform upper and lower boundary clipping to prevent pixel values ​​from overflowing into the illegal range of 255 or below 0. The perturbation image constructed in this way does not change the semantic structure of the original image, while expanding the illumination domain distribution of the image and improving the robustness of the model in recognizing "residual feces in unevenly illuminated areas"; Adding noise: Random noise is added to the image to enhance the model's robustness in noisy environments and improve its ability to handle image noise in practical applications. Types of noise to add include Gaussian noise and salt-and-pepper noise. Among them, Gaussian noise simulates the electrical signal fluctuation characteristics of the imaging sensor under low illumination, while salt-and-pepper noise is used to simulate interference such as camera aging and lens stains, aiming to increase the detection model's tolerance to blurred and edge-distorted fecal areas. Image fusion: By combining two or more images, new samples are generated, which increases the diversity of samples and improves the generalization ability of the model. Among them, Gaussian noise simulates the electrical signal fluctuation characteristics of the imaging sensor under low illumination, while salt-and-pepper noise is used to simulate interference such as camera aging and lens stains, aiming to increase the detection model's tolerance to blurred and edge-distorted fecal areas. Image normalization To eliminate data discrepancies, accelerate training, and enhance model feature extraction, pixel normalization is performed on the image. Specifically, the pixel values ​​in the image are adjusted to [0,1] using the following formula. Where, x normis the normalized pixel value, x is the original pixel value, X is the set of pixels in the entire image, min(X) is the minimum pixel value in the image, and max(X) is the maximum pixel value in the image. The normalization operation follows these steps: First, extract the minimum pixel min(X) and maximum pixel max(X) from the entire image to determine the dynamic range of the image pixels. Second, perform a linear stretching transformation on each pixel value, mapping it to the [0,1] interval, thereby unifying the brightness scale of all image inputs. This normalization method not only eliminates differences in image brightness but also improves the stability of gradient propagation in the neural network, accelerating model convergence. The algorithm model is applied to a mobile manure cleaning robot (belonging to pigsty floor manure cleaning equipment). Based on the application scenario of the equipment, this method constructs a lightweight object detection model based on the YOLOv11 architecture. In this method, Ghost convolution is used to replace Conv convolution to achieve network lightweighting, and SKNet attention mechanism is added. Among them, the Ghost convolution module generates some feature maps by standard convolution and the remaining feature maps by inexpensive linear transformation, effectively reducing the amount of computation and parameters; the SKNet module introduces a selective convolution kernel weighting mechanism, which dynamically selects the most suitable convolution channel for the current target shape according to the input features, thereby enhancing the ability to express the differential features of "lump-fluid state" in fecal images. Among them, the network structure is as follows Figure 9 As shown, this network structure can effectively fuse multi-scale feature information, improve the feature representation ability, and the selection module in the SKNet attention mechanism only selects a portion of the convolutional kernels for computation, thereby reducing computational cost and maintaining the network's efficiency. The input image (640×640×3) first enters the backbone. The backbone network uses GhostConv convolution to achieve lightweight feature extraction, significantly reducing the number of parameters by using "linear transformation + feature recombination" to balance the deployment efficiency of edge devices and the preservation of core fecal features. Subsequently, the image is passed through 4 sets of ChostConv+C3K2 modules to gradually reduce the resolution from 640×640 to deep semantic feature maps. SPPF spatial pyramid pooling expands the receptive field through multi-scale max pooling to capture the global morphology of feces. The SKAttention attention mechanism dynamically learns the weights of different convolution kernels to enhance the differentiated features of lumpy and fluid states, while suppressing background interference. In particular, SPPF’s three-layer pooling windows (such as 5×5, 9×9, and 13×13) correspond to the feature extraction of fecal targets of different physical sizes, ensuring that both small-volume and diffuse stains can be detected, thus building a scale-insensitive detection foundation. The C2PSA module at the end performs collaborative purification of channel and spatial attention, improving the visual distinguishability of "feces-background" and providing high semantic features for subsequent fusion. After the feature maps enter the Neck network, multi-scale feature fusion is performed; high-level semantic features are first upsampled through Unsample to increase resolution, and then fused with shallow detail features of the same scale as the Backbone through Concat concatenation, achieving complementarity between "deep semantics" and "shallow details"; the fused features are then compressed through C3K2 or Conv modules to enhance information integration and construct a hierarchical feature pyramid. By upsampling deep features and concatenating them with shallow features, and then downsampling shallow features and fusing them with deep features, sufficient interaction between the details of small targets and the semantics of large targets is ensured. The final output consists of three sets of fused feature maps with decreasing resolutions: 80×80, 40×40, and 20×20. The Head detection module receives multi-scale feature maps from the Neck network and uses a multi-scale hierarchical recognition mechanism to detect feces of different shapes and sizes. It then outputs the spatial coordinates and category probability distribution of the fecal targets, and eliminates overlapping detection boxes using a non-maximum suppression (NMS) algorithm, ultimately generating optimized end-to-end detection results. This network, GhostConv, features a lightweight design and incorporates the SKAttention mechanism, enabling rapid real-time detection of both lumpy and fluid feces on septic tank cleaning equipment, providing data support for the operation of such equipment.

Claims

1. A pigsty floor manure cleaning device, comprising a robot body (10), characterized in that: The robot body (10) is equipped with a squeezing collection mechanism for collecting clumps of feces on the ground into a container, and a flexible mopping mechanism for cleaning up non-clump feces on the ground.

2. The pigsty floor manure cleaning equipment according to claim 1, characterized in that: The flexible pier paving mechanism includes a vertically arranged connecting rod (11), the upper end of the connecting rod (11) is connected to a connecting rod lifting mechanism (15), and the lower end of the connecting rod (11) is connected to several independent flexible strip materials (12). A sleeve (13) is sleeved on the connecting rod (11). The sleeve (13) is fixedly installed. When the connecting rod (11) moves to the upper target position, all the flexible strip materials (12) are located in the sleeve (13) in a compacted state. When the connecting rod (11) moves to the lower target position, all the flexible strip materials (12) are radially dispersed.

3. The pigsty floor manure cleaning equipment according to claim 2, characterized in that: The flexible strip material (12) is made of bundled cotton or bundled fiber with a diameter of no more than 5 mm.

4. The pigsty floor manure cleaning equipment according to claim 3, characterized in that: Each flexible strip (12) has a beryllium copper wire with a diameter of 1~2mm arranged axially inside.

5. The pigsty floor manure cleaning equipment according to claim 2, characterized in that: The flexible strip material (12) is made of sponge with a diameter of no more than 15 mm. Each flexible strip material (12) is axially provided with a beryllium copper wire spring (14) with a diameter of 3~8 mm. The beryllium copper wire spring (14) is provided with sponge material (16) inside and outside.

6. The pigsty floor manure cleaning equipment according to any one of claims 4-5, characterized in that: When the flexible strip material (12) is in its initial state without being subjected to external force, the beryllium copper wire or beryllium copper wire spring (14) has an arched structure, and the top of the arched structure is connected to the bottom of the connecting rod (11).

7. The pigsty floor manure cleaning equipment according to claim 6, characterized in that: The angle between the inclined plane of the arch structure and the horizontal plane is 30~50°.

8. The pigsty floor manure cleaning equipment according to claim 6, characterized in that: The compression collection mechanism includes a vertical telescopic mechanism (1), the lower end of the telescopic rod of the vertical telescopic mechanism (1) is connected to the top of the cylinder (2), the cylinder (2) is arranged vertically, and the bottom wall of the cylinder (2) is provided with several arrayed through holes (3). The cylinder (2) serves as a container for collecting and storing pig manure, and the through holes (3) also serve as manure outlets. A piston (4) is provided in the inner cavity of the cylinder (2), and the piston (4) is connected to a telescopic device (5). The piston (4) is driven to move axially in the inner cavity of the cylinder (2) through the telescopic device (5). The through holes (3) have a structure that is thick at both ends and thin in the middle. The cylinder (2) is made of polytetrafluoroethylene material.

9. The pigsty floor manure cleaning equipment according to claim 8, characterized in that: During the movement of the pigsty floor manure cleaning equipment, the flexible pier mechanism is located behind the cylinder (2).

10. The pigsty floor manure cleaning equipment according to claim 6, characterized in that, The device also includes a target detection system that uses intelligent identification of residual feces, which includes: The image acquisition module is used to acquire image data of the pigsty floor in real time through a camera and generate a raw image dataset; The image preprocessing module is used to perform image enhancement processing on the original image dataset; An image feature extraction module is used to input the normalized image into a lightweight detection model, which is built on the YOLOv11 architecture. The target detection module is used to perform multi-scale target recognition operations on the fused feature map set, output the location coordinates and category probability distribution of the fecal target, and delete redundant boxes through a non-maximum suppression algorithm to obtain the final residual fecal detection result. It also includes a cleaning guidance module, which sends the target location data and fecal type data from the detection results to the cleaning control module to drive the fecal cleaning equipment to perform targeted secondary cleaning operations.

Images