Feeding system capable of automatically crushing feed and monitoring physical signs

Abstract

The invention relates to the technical field of breeding, in particular to a feeding system capable of automatically crushing feed and monitoring physical signs, comprising a feed storage device used for storing and carrying feed to a set feeding point; the feeding device comprises a shaftless spiral blade and is used for pushing the foodstuff to advance in the feeding channel; the material crushing device comprises a material crushing shaft, and a material crushing spiral is arranged on the surface of the material crushing shaft; the weighing device is used for weighing the received foodstuff and pouring the foodstuff to the material receiving device; the material receiving device is used for conveying foodstuff into the feeding trough; and the monitoring device is used for monitoring the animal state and residual foodstuff at the feeding cage. By adjusting and improving the feeding system, the foodstuff is crushed again in the conveying process, the uniformity of foodstuff particles is guaranteed, and the foodstuff is prevented from being left and clamped in the feeding device. The monitoring device can judge the survival signs and the remaining amount of the foodstuff of the animals by acquiring the image information and remind abnormal conditions, and can help to improve the automation degree and the reliability degree of the feeding system.

Description

Technical Field

[0001] The invention relates to the technical field of breeding, and in particular to a feeding system capable of automatically crushing materials and performing vital sign monitoring. Background Art

[0002] Automated machinery technology continues to develop and can undertake more and more work content. In recent years, it has also been widely used in the fields of agriculture and animal husbandry.

[0003] In animal husbandry, accurate feeding is crucial to ensuring animal growth. However, existing feeding devices, which mostly use pipelines for transportation, have many drawbacks: they are expensive, and the pipelines cannot crush and clean the feed, resulting in a certain amount of feed residue in the pipeline. This residual feed easily becomes spoiled residue, generating mold and bacteria, etc., which pollutes the entire pipeline, induces animal diseases, and endangers animal health. At the same time, the feed delivery accuracy of existing feeding equipment is low, and it is easy to under-deliver or over-deliver, resulting in insufficient feeding or feed waste, increasing overall costs. There is also the risk of excess feed being contaminated, which increases the pressure on site cleaning.

[0004] Furthermore, current automated farming systems suffer from monitoring gaps. Due to the large scale of operations, it's difficult for farmers to monitor the health of all animals, such as miscarriages and pup dropouts. These conditions are often not discovered in a timely manner, resulting in animals not receiving timely treatment and causing significant financial losses to the farm.

[0005] It can be seen that the current automated farming solutions still have problems that need to be solved. They should be optimized to improve the accuracy of feed delivery during the automated farming process. At the same time, they should be able to optimize the feed delivery structure, reduce feed residue, and avoid feed spoilage, thereby ensuring the health of the animal diet. In addition, they should be able to automatically monitor the animal's condition and issue prompt alerts when an animal is abnormal, so that the animal can be treated in time and unnecessary losses can be reduced. Therefore, it is necessary to propose a more reasonable technical solution to solve the technical problems existing in the existing technology. Summary of the Invention

[0006] In order to overcome at least one of the above-mentioned defects, the present invention proposes a feeding system with automatic crushing and vital sign monitoring, aiming to optimize the structure of the feeding device, so that the feed can be transported and crushed during the feeding process, so as to better control the feeding amount, reduce feed residue, and avoid food spoilage that affects the health of animals; at the same time, by real-time monitoring of the animal's condition and issuing reminders, abnormal situations can be discovered and handled in time to reduce losses.

[0007] In order to achieve the above objectives, the feeding system disclosed in the present invention can adopt the following technical solutions: An automatic crushing and vital signs monitoring feeding system, comprising: A storage device for storing and carrying food to a set feeding point; the storage device includes a storage trolley for storing food, which moves along a track and reaches the feeding point; A feeding device is used to transport the food in the storage trolley outward. The feeding device includes a feeding channel and a shaftless spiral blade arranged in the feeding channel. When the shaftless spiral blade rotates, it pushes the food in the feeding channel. When the food is pushed to the discharge point, it falls to the weighing device; The crushing device includes a crushing shaft coaxially matched with the shaftless spiral blade, a crushing spiral is provided on the surface of the crushing shaft for cooperating with the inner side of the shaftless spiral blade to crush the material, and the rotation direction of the crushing spiral is opposite to that of the shaftless spiral blade; A weighing device, used to weigh the received food, includes a weighing plate and a flipping mechanism. When the food on the weighing plate reaches a set weight, the flipping mechanism drives the weighing plate to flip and dump the food into the receiving device below; The material receiving device includes a material receiving hopper, and a conveying pipe is provided below the material receiving hopper, and the conveying pipe is used to convey the food in the material receiving hopper to the food trough; The monitoring device includes an animal monitoring device for monitoring the status of animals in feeding cages, and a food monitoring device for monitoring the feeding device and the residual food in the trough.

[0008] The above-disclosed feeding system uses a track to drive a storage trolley to travel between multiple feeding cages. The food in the storage device is transported to the weighing device by the feeding device. During the transportation process, it can be crushed by the crushing device to prevent the feeding device from clogging. When the food on the weighing device reaches the set weight, the food at the weighing device is dumped into the receiving device and reaches the corresponding feeding cage for the animals to eat. This improves the intelligence level of automatic feeding and the accuracy of feeding control, facilitating better feeding. At the same time, the vital signs of the animals are monitored by the monitoring device, including identifying and processing images of excrement to determine physical health, and identifying and processing images of cubs to determine survival status. At the same time, the amount of food remaining in the feeding device is monitored in real time to prevent the residual food from spoiling and causing dietary hazards. The above monitoring can issue an alarm in time to avoid causing greater losses.

[0009] Furthermore, the storage trolley serves as a load-bearing structure, driving the synchronous movement of other devices. This can be achieved through a variety of solutions, and its structure is not limited to a single one. Here, we optimize and propose one feasible option: the storage trolley is provided with a first bracket, on which a storage hopper is provided. The storage hopper is connected to the feeding device through a discharge pipe; the storage trolley is also provided with a second bracket, and the feeding device, crushing device, weighing device, and monitoring device are all connected and matched to the second bracket and move synchronously with the trolley. When adopting this solution, the first bracket can be configured as a fence structure for placing the storage hopper; the second bracket includes a cantilever for connecting to the storage trolley and a connecting plate provided on the cantilever, and the feeding device, crushing device, and weighing device are fixed to the connecting plate.

[0010] Furthermore, the feed channel for conveying food can be constructed in various forms, and its structure is not limited to a single one. Here, we optimize and propose one feasible option: the feed channel includes a lower tube body and an upper tube body that interlock with each other. The upper and lower tube bodies interlock to form a circular feed channel. The shaftless spiral blade is disposed within the feed channel. The discharge point includes a discharge port disposed in the lower tube body. When adopting the above solution, the upper and lower tube bodies can be semicircular tubes.

[0011] Furthermore, the shaftless spiral blades convey food during rotation, and their drive can be achieved through a variety of solutions. Here, we optimize and propose one feasible option: the ends of the shaftless spiral blades are cooperatively connected to a synchronous wheel, which is connected and driven by a first drive component. When the first drive component is activated, it drives the synchronous wheel to rotate, and the shaftless spiral blades rotate synchronously. When adopting this solution, the first drive component includes a servo, the output shaft of the servo is connected to the drive wheel, and a synchronous belt is connected between the drive wheel and the synchronous wheel, thereby achieving transmission and driving the shaftless spiral blades.

[0012] Furthermore, the crushing shaft cooperates with the shaftless spiral blade to shear and crush the food during the rotation process, thereby preventing the food particles from being too large, facilitating the control of feeding accuracy, and avoiding food residue. The driving of the crushing shaft can be achieved in a variety of ways. Here, we optimize and propose one of the feasible options: the end of the crushing shaft is connected to a second driving component, which drives the crushing shaft to rotate. When the crushing shaft and the shaftless spiral blade rotate coaxially, the crushing spiral and the shaftless spiral blade shear and crush the food. When the above solution is adopted, the second driving component can be a servo, which drives the crushing shaft to achieve continuous rotation. A certain gap is formed between the crushing spiral and the shaftless spiral blade, and when the food is crushed to a specified particle size, it can pass through the gap.

[0013] Furthermore, under the weighing device, when the food reaches a set weight, it is dumped into the receiving device below. This dumping action can be achieved by a flipping action. There are multiple specific solutions, and the structure is not limited to a single one. Here, we optimize and propose one feasible option: the flipping mechanism includes a flipping drive component and a flipping shaft. The flipping drive component drives the flipping shaft to rotate, and the weighing plate is connected to the flipping shaft and flips with the flipping shaft. When adopting this solution, the flipping drive component can use a servo, and under the drive of the servo, the weighing plate can achieve a 90° flip.

[0014] Furthermore, when pouring food from the weighing pan, it is necessary to maintain accuracy and avoid spillage. This effect can be achieved through improvements, and the structure is not limited to a single one. Here, we optimize and propose one feasible option: the edge of the weighing pan is provided with a pouring spout and a guide and anti-spill area extending to the pouring spout. When the weighing pan is turned over and the pouring spout is directed toward the material receiving device, the food enters the pouring spout along the guide and anti-spill area and flows into the material receiving device. When adopting this solution, the guide and anti-spill area is an arc-shaped surface and is provided on the side of the pouring spout.

[0015] Furthermore, the feeding cages are arranged in a continuous path along the track, with multiple separate feeding compartments spaced apart. Therefore, a receiving device is provided at each compartment. This structure is not strictly limited. Here, an optimization is proposed as a feasible option: the receiving devices are spaced along the track, with each feeding point equipped with a receiving device. When adopting this solution, if the feeding cages are stacked, separate receiving devices are provided for each upper and lower compartment.

[0016] Furthermore, the monitoring device is used to observe and provide feedback on the feeding situation and issue early warnings for abnormal situations. The animal monitoring device includes a first mounting frame, on which a first camera is installed. The first camera is used to obtain image data from the feeding cage and send it to the server. The image data includes images of excrement and pup images. When adopting the above solution, the first camera can be directed toward the feeding cage to obtain data. When the feeding cages are stacked, the first camera obtains images from below the feeding cage as judgment data. For example, the excrement of animals on this layer will fall below this layer. Monitoring and obtaining images here can determine the dietary health of the animals on this layer. When a pup dies or falls, it will also fall below this layer. The images here can detect abnormalities in the pup, thereby issuing a warning and taking quick measures.

[0017] Furthermore, the monitoring device also monitors the feeding area, including monitoring food delivery. The food monitoring device includes a second mounting bracket, mounted with a second camera. The second camera is configured to capture image data of the feeding device and the trough and transmit it to a server. The image data includes images of food residue. In this embodiment, as food is conveyed within the feeding device, the second camera captures and detects the amount of food residue. The camera then determines the amount of food residue based on the image information. If the amount of food residue reaches an alarm level, a prompt is issued.

[0018] Compared with the prior art, some of the beneficial effects of the technical solution disclosed in the present invention include: By adjusting and improving the feeding system, the present invention can re-crush the feed during transportation, ensuring the uniformity of the feed particles and preventing food residue and jamming within the conveyor. This also facilitates subsequent accurate weighing and feeding of the feed, thereby improving feeding accuracy. During system operation, the monitoring device can determine the animal's vital signs by capturing image information. It can also monitor the feed in the feeding device and trough, providing alerts and feedback when anomalies are detected in the image information, thereby helping to improve the automation and reliability of the feeding system. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only represent some embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without paying any creative work.

[0020] Figure 1 Schematic diagram of the overall structure of the feeding system.

[0021] Figure 2 This is a schematic diagram of the structure of the storage device in conjunction with the track.

[0022] Figure 3 This is a schematic diagram of the matching connections at the storage trolley.

[0023] Figure 4 It is a schematic diagram of the coordinated structure of the feeding device, crushing device and weighing device.

[0024] Figure 5 It is a structural diagram of the combination of shaftless spiral blades and crushing shaft.

[0025] Figure 6 Schematic diagram of the decomposition structure of the feeding channel.

[0026] Figure 7 Schematic diagram of the system's multi-task workflow.

[0027] Figure 8 This is the system architecture diagram of YOLOv8n.

[0028] Figure 9 Images of feed pellets resulting from an abortion event.

[0029] In the above drawings, the meanings of the symbols are as follows: 1. Track; 2. Storage trolley; 201. First bracket; 202. Storage hopper; 203. Cantilever; 204. Connecting plate; 205. First mounting frame; 3. Receiving hopper; 4. Conveying pipe; 5. Feeding trough; 6. Feeding cage; 7. Feeding device; 701. Shaftless spiral blade; 702. Crushing shaft; 703. Crushing spiral; 704. Lower tube body; 705. Upper tube body; 706. Feeding port; 8. Weighing plate; 801. Pour port; 802. Guide anti-spill zone; 9. Pressure measuring element; 10. First driving component; 1001. Driving wheel; 1002. Synchronous belt; 1003. Synchronous wheel; 1004. Reinforcement rib; 11. Second driving component; 12. Flipping driving component. DETAILED DESCRIPTION

[0030] This embodiment will be further explained below with reference to the accompanying drawings and specific examples.

[0031] In view of the shortcomings of the feeding equipment in the prior art, such as inaccurate feeding amount control and inability to monitor and provide feedback on animal vital signs, the following embodiments are optimized and overcome the defects in the prior art.

[0032] Example 1 like Figures 1 to 6 As shown, this embodiment provides a feeding system that automatically crushes materials and performs vital sign monitoring, aiming to improve the degree of automation and intelligence of the feeding system, improve the control accuracy during the feeding process, and can timely monitor and feedback abnormal situations, so as to facilitate timely measures to reduce losses.

[0033] like Figure 3 As shown, as a feeding system provided in this embodiment, one of its structures includes: The storage device is used to store and carry food to a set feeding point; the storage device includes a storage trolley 2 for storing food, and the storage trolley 2 travels along the track 1 and reaches the feeding point.

[0034] The storage trolley 2 serves as a load-bearing structure, driving other devices to move synchronously. This can be achieved through a variety of solutions, and its structure is not limited to a single one. This embodiment optimizes and adopts one of the feasible options: the storage trolley 2 is provided with a first bracket 201, and a storage hopper 202 is provided on the first bracket 201. The storage hopper 202 is connected and coordinated with the feeding device 7 through a discharge pipe; the storage trolley 2 is also provided with a second bracket, and the feeding device 7, crushing device, weighing device and monitoring device are all connected and coordinated to the second bracket and move synchronously with the trolley. When adopting the above solution, the first bracket 201 can be set as a fence structure for placing the storage hopper 202; the second bracket includes a cantilever 203 for connecting to the storage trolley 2, and a connecting plate 204 provided on the cantilever 203. The feeding device 7, crushing device, weighing device and connecting plate 204 are fixed together.

[0035] like Figure 4 、 Figure 5 、 Figure 6 As shown, the feeding system provided in this embodiment includes the following structure: The feeding device 7 is used to transport the food in the storage cart outward. The feeding device 7 includes a feeding channel and a shaftless spiral blade 701 arranged in the feeding channel. When the shaftless spiral blade 701 rotates, it pushes the food to move in the feeding channel. When the food is pushed to the unloading point, it falls to the weighing device.

[0036] The feeding channel is used to transport food and can be constructed in various forms. Its structure is not limited to a single one. This embodiment optimizes and adopts one feasible option: the feeding channel includes a lower tube body 704 and an upper tube body 705 that interlock with each other. The upper tube body 705 and the lower tube body 704 interlock to form a circular feeding channel. The shaftless spiral blade 701 is disposed in the feeding channel; the feeding point includes a feeding port 706 disposed in the lower tube body 704. When adopting the above solution, the upper tube body 705 and the lower tube body 704 can be semicircular tubes.

[0037] The shaftless spiral blade 701 conveys food material during rotation. This drive can be achieved through a variety of schemes. Here, we optimize and propose one feasible option: the end of the shaftless spiral blade 701 is cooperatively connected to a synchronous wheel 1003, which is connected and driven by the first drive component 10. When the first drive component 10 is activated, it drives the synchronous wheel 1003 to rotate, and the shaftless spiral blade 701 rotates synchronously. When adopting this scheme, the first drive component 10 includes a servo, the output shaft of which is connected to the drive wheel 1001. A synchronous belt 1002 is connected between the drive wheel 1001 and the synchronous wheel 1003, thereby achieving transmission and driving the shaftless spiral blade 701.

[0038] Preferably, in this embodiment, the shaftless spiral blade 701 is connected and fixed to the synchronous wheel 1003 via a reinforcing rib 1004 .

[0039] like Figure 5 As shown, the feeding system provided in this embodiment includes the following structure: The crushing device includes a crushing shaft 702 coaxially matched with the shaftless spiral blade 701. The surface of the crushing shaft 702 is provided with a crushing spiral 703 that cooperates with the inner side of the shaftless spiral blade 701 to crush the material. The rotation direction of the crushing spiral 703 is opposite to that of the shaftless spiral blade 701.

[0040] The crushing shaft 702 cooperates with the shaftless spiral blade 701 to shear and crush the food during rotation, thereby preventing the food particles from being too large, facilitating feeding accuracy control, and preventing food residue. The crushing shaft 702 can be driven in a variety of ways. This embodiment optimizes and adopts one feasible option: the end of the crushing shaft 702 is connected to a second drive component 11, which drives the crushing shaft 702 to rotate. When the crushing shaft 702 and the shaftless spiral blade 701 rotate coaxially, the crushing screw 703 and the shaftless spiral blade 701 shear and crush the food. When adopting this solution, the second drive component 11 can be a steering gear, which drives the crushing shaft 702 to achieve continuous rotation. A certain gap is formed between the crushing screw 703 and the shaftless spiral blade 701. After the food is crushed to a specified particle size, it can pass through this gap.

[0041] like Figure 4 As shown, the feeding system provided in this embodiment includes the following structure: The weighing device is used to weigh the received food, and includes a weighing plate 8 and a turning mechanism. When the food on the weighing plate 8 reaches a set weight, the turning mechanism drives the weighing plate 8 to turn over to dump the food into the receiving device below.

[0042] Preferably, a pressure measuring element 9 is provided below the weighing pan 8 for accurately measuring the weight of the food.

[0043] When the food reaches a set weight under the weighing device, it is dumped into the receiving device below. This dumping action can be achieved by a flipping action. A variety of specific solutions are possible, and the structure is not limited to a single one. This embodiment optimizes and adopts one feasible option: the flipping mechanism includes a flipping drive component 12 and a flip shaft. The flipping drive component 12 drives the flip shaft to rotate, and the weighing plate 8 is connected to the flip shaft and flips with the flip shaft. When adopting this solution, the flipping drive component 12 can be a steering gear, and the weighing plate 8 can be flipped 90 degrees under the drive of the steering gear.

[0044] When pouring food from the weighing pan 8, it is necessary to maintain accuracy and avoid spillage. This effect can be achieved through improvements, and its structure is not limited to a single one. This embodiment optimizes and adopts one feasible option: the edge of the weighing pan 8 is provided with a pouring port 801, and a guide anti-spill area 802 extending to the pouring port 801 is also provided. When the weighing pan 8 is turned over and the pouring port 801 is directed toward the material receiving device, the food enters the pouring port 801 along the guide anti-spill area 802 and is discharged to the material receiving device. When adopting the above solution, the guide anti-spill area 802 is an arc-shaped surface and is provided on the side of the pouring port 801.

[0045] like Figure 1 As shown, the feeding system provided in this embodiment includes the following structure: The material receiving device includes a material receiving hopper 3 , and a conveying pipe 4 is provided below the material receiving hopper 3 . The conveying pipe 4 is used to convey the food in the material receiving hopper 3 to the food trough 5 .

[0046] The feeding cage 6 is arranged in a continuous manner with the track 1, with multiple separate feeding compartments spaced apart between them. Therefore, a feeding device is provided at each compartment. This structure is not strictly limited. This embodiment optimizes and adopts one feasible option: the feeding devices are spaced along the extension direction of the track 1, with each feeding point equipped with a feeding device. In this embodiment, when the feeding cage 6 adopts a stacked structure, separate feeding devices are provided in the upper and lower compartments.

[0047] like Figure 3 As shown, the feeding system provided in this embodiment includes the following structure six: The monitoring device includes an animal monitoring device for monitoring the status of animals in the feeding cage 6 and a food monitoring device for monitoring the residual food in the feeding device 7 and the feeding trough 5.

[0048] The monitoring device is used to observe and provide feedback on feeding conditions and issue warnings for abnormalities. The animal monitoring device includes a first mounting bracket 205, on which a first camera is mounted. The first camera is used to capture image data from the feeding cage 6 and transmit it to a server. The image data includes images of excrement and pups. When using this solution, the first camera can face the feeding cage 6 to capture data. When the feeding cages 6 are stacked, the first camera captures images from below the feeding cage 6 as judgment data. For example, if animal excrement on a given layer falls below the layer, monitoring and capturing images from this layer can determine the dietary health of the animals on that layer. If a pup dies or falls, it will also fall below the layer. Images from this layer can detect any abnormalities in the pup, thereby issuing a warning and enabling quick action.

[0049] The monitoring device also monitors the feeding area, including food delivery. This food monitoring device includes a second mounting bracket, mounted with a second camera. This second camera is used to capture image data from the feeding device 7 and transmit it to a server. This image data includes images of food residue. In this embodiment, as food is conveyed within the feeding device 7, the second camera captures the amount of food residue. The camera then determines the amount of food residue based on the image information. If the amount of food residue reaches the alarm level, a prompt alert is issued.

[0050] The feeding system disclosed in this embodiment is provided with a track 1 to drive a storage trolley 2 to move between multiple feeding cages 6. The food in the storage device is transported to the weighing device by the feeding device 7. During the transportation process, it can be crushed by the crushing device to prevent the feeding device 7 from being blocked. When the food on the weighing device reaches the set weight, the food at the weighing device is poured into the receiving device and reaches the corresponding feeding cage 6 for the animals to eat. This improves the intelligence level of automatic feeding and the accuracy of feeding control, making it easier to complete feeding. At the same time, the vital signs of the animals are monitored by the monitoring device, including identifying and processing the images of excrement to determine the health of the body, and identifying and processing the images of the cubs to determine the survival status. At the same time, the residual amount of food in the feeding device 7 is monitored in real time to prevent the residual food from spoiling and causing dietary hazards. The above monitoring can issue an alarm in time to avoid causing greater losses.

[0051] Example 2 The above embodiment discloses a feeding system. This embodiment provides a feeding method when the above system is in operation, comprising the following steps: S01: The storage trolley 2 arrives above the designated feeding cage 6 along the track 1.

[0052] S02: The first camera of the monitoring device takes a picture of the feeding cage 6 to identify abnormal conditions such as miscarriage and pup falling; the second camera takes a picture of the feeding device 7 and the trough 5 to identify the amount of remaining food in the feeding device 7 and the trough 5.

[0053] S03: Calculate the appropriate feeding amount based on the remaining food in the trough 5 and the current animal information. The first driving component 10 rotates to drive the synchronous wheel 1003 to rotate, and the shaftless spiral blade 701 rotates to push the food in the storage hopper 202 to the weighing device below.

[0054] S04: The second driving component 11 rotates in the opposite direction, driving the crushing shaft 702 to rotate and crush the stuck food.

[0055] S05: The weighing plate 8 detects whether the weight of the food reaches a preset value.

[0056] S06: When the preset value is reached, the first driving component 10 and the second driving component 11 stop rotating, and the flip driving component 12 drives the weighing plate 8 to rotate, so that the food falls into the receiving device.

[0057] S07: The weighing plate 8 returns to its original position and the feeding process ends.

[0058] Example 3 The above embodiments disclose a feeding system and a feeding method. In this embodiment, a method for identifying, acquiring and feeding back image data is disclosed.

[0059] During system operation, the first and second cameras operate independently. The second camera periodically captures images and analyzes the feed status. When the feed percentage falls below a set value, a feed shortage warning is triggered. The first camera periodically captures images and detects bloodstains, paw paws, and fallen pups. When the system detects these anomalies, it triggers an abortion warning, a cleanup reminder, and a high-priority alarm, respectively.

[0060] The following system components are involved in image acquisition, recognition, and conversion into alarm signals: An animal monitoring device for capturing image data of a feeding cage; A food monitoring device for capturing image data from the feeding device; The server is used to receive and process image data, identify and read pixel information on the image data, and compare it with data stored in the server, thereby determining the corresponding image data as the situation represented by the corresponding data and marking it as normal or abnormal. When an abnormal result is marked, the server generates an alarm signal and sends it to the alarm device; The alarm device is used to receive the alarm signal from the server and provide alarm action.

[0061] Specifically, using the above system components, the following method is proposed to generate corresponding alarm signals through image data: The captured image data is sent to the server, which processes and parses it to extract parameters from the image, including but not limited to pixel values; Compare the extracted parameter values ​​with the data stored in the data model to find the closest data; The actual situation represented by the closest data is used as the actual situation faced by the extracted parameter value; If the actual situation is a negative one, such as miscarriage, death, food residue, etc., an alarm signal is generated and sent through the server, and the alarm signal is received by the alarm device; When the alarm device receives the alarm signal, it starts the alarm reminder.

[0062] Specifically, some specific processes in the above method are described in detail below.

[0063] The process of image parsing through a model includes establishing a data model and running the data model.

[0064] The process of building a data model includes: Step 1: Data preprocessing. This includes feature selection, dataset creation, and data standardization.

[0065] Step 2: Design the network structure and initialize the parameters.

[0066] Step 3: Train the model using the dataset.

[0067] Step 4: Obtain the prediction model. After completing the model training, the prediction model can be obtained.

[0068] The process of running the data model to make judgments includes: Step 1: Load the data model and successfully load the trained prediction model.

[0069] Step 2: Data preparation: organize the collected data and prepare them for model prediction.

[0070] Step 3: Model prediction: Make predictions on the collected data by training the model.

[0071] Step 4: Analyze the model, obtain the predicted results, and return the results.

[0072] According to the above process, the image data captured by the first camera and the second camera can be processed and abnormality judgment can be achieved.

[0073] Specifically, the second camera is used to monitor the amount of food residue, and data processing and abnormality judgment can be performed through the following process.

[0074] Image acquisition must account for varying lighting conditions, shooting angles, and background complexity to improve the model's generalization capabilities. Data annotation can be performed using, but is not limited to, the Labelme tool to precisely label the feed trough area. The labeled feed is assigned the label "feed." The annotation results are converted to COCO format and then to MindRecord format for easy input into the prediction model for training.

[0075] In image processing, the gradient of the image is calculated using the Sobel operator to extract the edge information of the feed area:

[0076]

[0077] In the above formula, Indicates that the image is at coordinates The pixel brightness value output; Represents the horizontal gradient, which is calculated by the brightness difference between the left and right adjacent pixels; Represents the vertical gradient, which is calculated by the brightness difference between upper and lower adjacent pixels.

[0078] After synthesizing the gradient, the edge detection result is obtained:

[0079] When the gradient value exceeds the set threshold, the edge point in the image is marked as the feed area boundary. Then, by calculating the feed area ratio in the image, we can determine whether the feed is sufficient:

[0080] In the above formula, represents the area of ​​the food region detected in the image (pixel integral), Indicates the area of ​​the total image region.

[0081] The threshold TfT_fTf is generally set to 1 / 5. When the calculated feed remaining ratio is lower than 1 / 5, the feed shortage warning is triggered.

[0082] In addition, the first camera is used to monitor vital signs, and data processing and abnormality judgment can be performed through the following process.

[0083] Animal monitoring devices are primarily used to detect abnormalities in feeding cages, including fecal accumulation, blood, shavings (feed mixed with feces), and fallen pups. To accurately label abnormal areas in images, use tools such as Labelme, but not limited to, to annotate images with labels such as "blood," "manure," "mixed," and "kit."

[0084] In bloodstain detection, the HIS color space is used to extract the light pink bloodstain area. The general light pink HIS value range is Hue ∈ [0, 20], Saturation: 0.05 — 0.35Intensity: 0.45 — 0.85. Bloodstains are effectively distinguished by calculating hue HH, saturation SS and brightness II:

[0085]

[0086]

[0087] In the above formula, R, G, and B represent the pixel values ​​of the red, green, and blue channels in the RGB color space; H represents hue, which is used to distinguish color categories, such as light pink bloodstains; S represents saturation, which indicates color purity; and I represents intensity, which indicates the brightness of the color.

[0088] The bloodstain area is extracted through binarization and further noise is removed through morphological operations.

[0089] Shavings detection uses texture analysis to distinguish the degree of mixing of feed and feces, and uses binocular stereo vision to calculate the volume of the mixed area:

[0090] In the above formula, Represents the height information of each pixel in binocular stereo vision; Indicates the area of ​​the mixing region (e.g., mixing of food and discrimination); Indicates the volume of the mixed area. This method allows the system to estimate the volume of the mixed area and determine whether it needs to be cleaned.

[0091] For the detection of falling puppies, the system is first trained on n manually annotated images containing puppies in the offline stage: the pink skin area is accurately marked along the outline of the rabbit in each image, Hue(HH) ∈[0,20], Saturation(SS): 0.05 — 0.35Intensity(II): 0.45 — 0.85. Bloodstains are effectively distinguished by calculating hue (HH), saturation (SS), and brightness (II), and their proportion in the entire image is calculated. The model then automatically learns an optimal threshold (around 20%) based on these proportion labels. When the proportion P of pixels in the image identified as "light pink skin / bloodstains" using the H,H / S,S / I,I (i.e., HH, SS, II) rule exceeds 0.20, the system identifies a "fallen cub." During online detection, the system converts the real-time image to the HIS color space, extracts the light pink area, and calculates its proportion P. When P exceeds the pre-trained threshold Th (0.20), it identifies a fallen cub and issues an alarm:

[0092] In the above formula, the image is taken at a fixed angle and at a fixed position, so the total area remains unchanged, the selection is not random, and will not interfere with model recognition. Indicates the area of ​​the pup's skin visible in the image; Indicates the total area of ​​the image.

[0093] Specifically, such as Figures 7 to 9 As shown, actual cases are presented here, in which the time of abortion is identified by detecting bloodstains and fetuses, and the behavior of feed spillage is identified by detecting spilled feed particles.

[0094] Case 1. Miscarriage Detection Based on YOLOv8n The YOLO (You Only Look Once) series of object detection algorithms are widely used in tasks such as livestock monitoring, automatic feeding, and health assessment. YOLOv8, in particular, excels in processing images to efficiently identify key targets. YOLOv8P2 is more accurate in detecting small targets and can avoid information loss.

[0095] like Figure 7 As shown in Figure 1, YOLOv8n (the lightest variant in the YOLOv8 family) is used to detect miscarriages in a rabbit breeding environment. The model takes an RGB image as input and outputs the location of relevant visual indicators such as bloodstains and fetuses.

[0096] like Figure 8 As shown, the YOLOv8n backbone network adopts a Cross-Stage Partial (CSP) architecture, starting with several convolutional layers and then passing through a series of C2f modules. These modules divide and merge feature channels through bottleneck blocks. At the deepest layer, the Spatial Pyramid Fast Pooling (SPPF) module aggregates contextual information from different receptive field sizes, improving the model's ability to capture fine textures and high-level semantics.

[0097] The neck adopts a bottom-up path, in which the deep feature maps from the backbone network are upsampled and concatenated with the shallow feature maps, while the top-down path reintroduces the fused output into higher layers through downsampling and further concatenation. This bidirectional flow ensures that each output feature map incorporates the detailed spatial cues of the shallow layers and the strong contextual information of the deep layers.

[0098] The detection head operates on three fused feature maps (H3, H4, and H5). For each scale, parallel convolutional branches generate classification scores and bounding box predictions. This design enables accurate localization and classification across a wide range of object sizes, such as those associated with bloodstains and fetuses in abortions.

[0099] 2. Feed spillage detection based on YOLOv8n-P2 Due to the extremely small size of scattered feed particles and the cluttered environment in which rabbits are raised, detecting them presents a significant challenge. To address this issue, the YOLOv8n-P2 model, a variant of the YOLOv8 family specifically designed for ultra-small object detection, was used to detect scattered feed particles.

[0100] like Figure 9As shown, YOLOv8n-P2 builds on the original YOLOv8n architecture by introducing an additional detection branch in the neck, focusing on upsampling the output of the shallowest C2f layer to match the spatial resolution of higher layers. These refined features are then concatenated in a similar bottom-up and top-down pathway in the neck to compute an information-rich feature map for further analysis by the detection head. This modification produces a higher-resolution feature representation that better preserves the fine details required for small object detection. As a result, the detection head in YOLOv8n-P2 operates at four scales: P2, P3, P4, and P5, corresponding to small objects, small to medium objects, medium to large objects, and large objects, respectively. By explicitly integrating P2 features in the neck through additional upsampling, YOLOv8n-P2 significantly improves its sensitivity to small objects, enabling more accurate localization of feed pellets in the second task.

[0101] 3. Result fusion through conditional filtering The detection of bloodstains or fetuses is combined with the detection of feed particles through conditional filtering. First, the two sets of detection results of the two tasks are defined as follows:

[0102] in A detection frame indicating blood or a fetus, A detection frame representing feed pellets, and are their corresponding confidence values ​​respectively.

[0103] Next, conditional filtering first filters the detection frames of bloodstains or fetuses in the miscarriage detection task to ensure that the results have a high confidence level. Then, the intersection over union (IoU) between each filtered bloodstain or fetus frame and each feed particle frame is calculated, and the feed particle frame with the largest IoU value is selected as the final detection result, as shown below:

[0104] in and are the detection boxes selected from the two tasks, respectively. The Intersection over Union (IoU) value quantifies the degree of overlap between the two boxes, with 0 indicating no overlap and 1 indicating complete overlap. In this case, only feed particles that spatially coexist with the abortion indicator are selected, resulting in an accurate and non-redundant feed spill detection output.

[0105] 4. Specific experimental content A. Data This study collected 4,985 high-resolution RGB images (3024×4032 pixels) from closed-environment rabbit farms in Sichuan Province, China. These images were captured from multiple angles, at a distance of 10–40 cm from the manure trays beneath the rabbit cages, under both natural and artificial lighting conditions to ensure data diversity. All images were annotated using the YOLOTXT format, with precise bounding boxes drawn for three object classes: bloodstains, fetuses, and feed pellets. To accommodate the multi-task design, two subsets were constructed, each containing approximately 2,500 images. One subset was used for detecting bloodstains and fetuses, and the other for detecting feed pellets. Each subset was split into training, validation, and test sets with a ratio of 80%, 10%, and 10%.

[0106] B. Implementation Details In this study, each 3024×4032 RGB image was resized to 640×640 and transformed by min-max normalization. A series of data augmentation methods, including image stitching, mixing, HSV jittering, random scaling, translation, and horizontal flipping, were used to promote the model to learn robust representations for detecting bloodstains, fetuses, and feed particles in various rabbit farming environments. The models in both detection tasks were trained using the standard loss formula of YOLOv8 as described in

[16] . During training, the maximum number of epochs was set to 1000 and the batch size was 16. In addition, the SGD optimizer was used with a momentum factor of 0.937, an initial learning rate set to 0.01, and gradually decayed to 1×10-4 with a decay factor of 5×10-4 through a cosine annealing schedule. FP16 mixed precision training was adopted to accelerate convergence and reduce memory usage. In our conditional filtering algorithm, the confidence threshold α was set to 0.3, which determined the minimum confidence level required for the detection of signs of miscarriage to participate in the fusion process. The overlap threshold β is set to 0.5 to limit the minimum IoU required between the aborted box and the feed particle box to be considered co-located.

[0107] C. Performance 1. Abortion Detection: As shown in Table 1, the YOLOv8n model achieved an overall mAP@0.5 of 95.7%, with excellent performance in the bloodstain category (98.4%) and accurate detection of the fetus category (94.0%). These results demonstrate that YOLOv8n not only accurately locates signs associated with abortion but also maintains consistency across diverse lighting and background conditions. Furthermore, YOLOv8n achieved a precision of 97.2%, a recall of 95.6%, and an F1 score of 96.4%, indicating low false positive and false negative rates. These metrics confirm that YOLOv8n provides accurate detection results for real-time monitoring of rabbit abortions and provides a reliable source of results for another task, feed spill detection.

[0108] Table 1 Performance of YOLOv8n in detecting miscarriage events

[0109] 2. Detection of feed spillage caused by abortion: The goal is to detect feed spillage caused by abortion events. The final detection result is calculated by fusion of the results after conditional filtering, as shown in Table 1.

[0110] Table 2 Performance of YOLOv8n and YOLOv8n-P2 in detecting feed spillage caused by abortion events

[0111] The YOLOv8n-P2 model accurately detected feed pellets, with an overall mAP@0.5 of 81%, a precision of 87%, and a recall of 79%. This indicates that the model has a low false positive rate, but relatively poor performance in accurately locating all positive boxes. To demonstrate the improved ability of YOLOv8n-P2 in detecting small feed pellets, Table 2 also lists the feed pellet detection performance using the original YOLOv8n. In comparison, the model variant YOLOv8n-P2 achieved more accurate results than the original YOLOv8n. Specifically, mAP@0.5, precision, and recall improved by 9%, 12%, and 9%, respectively. This confirms the effectiveness of YOLOv8n-P2 in capturing fine details. Figure 9 An example of our detection of feed pellets is shown, showing accurate box results for detecting feed spillage caused by an abortion event.

[0112] D. Computational efficiency Computational efficiency is crucial in detection tasks. In this work, the YOLOv8n and YOLOv8n-P2 detectors for both tasks were run on an NVIDIA A6000 GPU, occupying 4.2GB and 5.6GB of GPU memory, respectively, while achieving running speeds of 82.0FPS and 76.0FPS, enabling real-time inference for feed spill detection.

[0113] The above are the implementation methods listed in this embodiment, but this embodiment is not limited to the above optional implementation methods. Those skilled in the art can arbitrarily combine the above methods to obtain other various implementation methods. Anyone can derive other various implementation methods based on the inspiration of this embodiment. The above specific implementation methods should not be understood as limiting the scope of protection of this embodiment. The scope of protection of this embodiment should be based on the definition in the claims.

Claims

1. A feeding system for automatic material crushing and vital sign monitoring, characterized in that: include: A storage device for storing and carrying food to a set feeding point; the storage device includes a storage trolley (2) for storing food, the storage trolley (2) travels along a track (1) and reaches the feeding point; A feeding device (7) is used to transport food materials in the storage trolley outward, and the feeding device (7) includes a feeding channel and a shaftless spiral blade (701) arranged in the feeding channel. When the shaftless spiral blade (701) rotates, it pushes the food materials to move in the feeding channel. When the food materials are pushed to the discharge point, they fall to the weighing device; The crushing device comprises a crushing shaft (702) coaxially matched with the shaftless spiral blade (701), a crushing spiral (703) is provided on the surface of the crushing shaft (702) and is matched with the inner side of the shaftless spiral blade (701) to crush the material, and the rotation direction of the crushing spiral (703) is opposite to that of the shaftless spiral blade (701); A weighing device for weighing received food, comprising a weighing plate (8) and a turning mechanism. When the food on the weighing plate (8) reaches a set weight, the turning mechanism drives the weighing plate (8) to turn over and dump the food into a receiving device below. The material receiving device comprises a material receiving hopper (3), a conveying pipe (4) is provided below the material receiving hopper (3), and the conveying pipe (4) is used to convey the food in the material receiving hopper (3) to the food trough (5); The monitoring device comprises an animal monitoring device for monitoring the state of animals in a feeding cage (6), and a food monitoring device for monitoring the residual food in a feeding device (7) and a food trough (5).

2. The feeding system for automatic material crushing and vital sign monitoring according to claim 1, characterized in that: The storage trolley (2) is provided with a first bracket (201), a storage hopper (202) is provided on the first bracket (201), and the storage hopper (202) is connected and matched with the feeding device (7) through a discharge pipe; the storage trolley (2) is also provided with a second bracket, and the feeding device (7), crushing device, weighing device and monitoring device are all connected and matched to the second bracket and move synchronously with the trolley.

3. The feeding system for automatic material crushing and vital sign monitoring according to claim 1, characterized in that: The feeding channel comprises a lower tube body (704) and an upper tube body (705) that are interlocked. The upper tube body (705) and the lower tube body (704) are interlocked to form a circular feeding channel. The shaftless spiral blade (701) is arranged in the feeding channel; the discharge point comprises a discharge port (706) arranged in the lower tube body (704).

4. The feeding system for automatic material crushing and vital sign monitoring according to claim 1 or 3, characterized in that: The end of the shaftless spiral blade (701) is cooperatively connected to a synchronous wheel (1003), and the synchronous wheel (1003) is connected and driven by the first driving component (10). When the first driving component (10) is started, the synchronous wheel (1003) is driven to rotate, and the shaftless spiral blade (701) rotates synchronously.

5. The feeding system for automatic material crushing and vital sign monitoring according to claim 1, characterized in that: The end of the crushing shaft (702) is connected to a second driving component (11), and the second driving component (11) drives the crushing shaft (702) to rotate. When the crushing shaft (702) and the shaftless spiral blade (701) rotate coaxially, the crushing spiral (703) and the shaftless spiral blade (701) shear and crush the food.

6. The feeding system for automatic material crushing and vital sign monitoring according to claim 1, characterized in that: The flip mechanism comprises a flip drive component (12) and a flip shaft. The flip drive component (12) drives the flip shaft to rotate. The weighing plate (8) is connected to the flip shaft and flips along with the flip shaft.

7. The feeding system for automatic material crushing and vital sign monitoring according to claim 1 or 6, characterized in that: The edge of the weighing plate (8) is provided with a pouring port (801), and a guide anti-spill area (802) extending to the pouring port (801); when the weighing plate (8) is turned over and the pouring port (801) is directed toward the material receiving device, the food enters the pouring port (801) along the guide anti-spill area (802) and is discharged to the material receiving device.

8. The feeding system for automatic material crushing and vital sign monitoring according to claim 1, characterized in that: The material receiving devices are arranged at intervals along the extension direction of the track (1), and each feeding point is provided with a material receiving device.

9. The feeding system for automatic material crushing and vital sign monitoring according to claim 1, characterized in that: The animal monitoring device comprises a first mounting frame (205), on which a first camera is provided. The first camera is used to obtain image data at a feeding cage (6) and transmit the image data to a server, wherein the image data comprises images of excrement and images of cubs.

10. The feeding system for automatic material crushing and vital sign monitoring according to claim 1, characterized in that: The food monitoring device comprises a second mounting frame, on which a second camera is provided. The second camera is used to obtain image data at the feeding device (7) and send it to a server, wherein the image data includes an image of food residue.

Images