A feeding system that automatically grinds and performs vital sign monitoring

Abstract

The present application relates to the technical field of breeding, in particular to a kind of automatic feed and carry out sign monitoring feeding system, including storage device, to store and carry food to the set feeding point;Feeding device, including shaftless spiral blade, to push food to travel in feeding channel;Crushed material device, including crushed material shaft, the surface of crushed material shaft is provided with crushed material spiral;Weighing device, to weigh the food received, and pour food to receiving device;Receiving device, to transport food to trough;Monitoring device, to monitor the state of animals and residual food in feeding cage. Through the adjustment improvement of feeding system, the uniformity of food particles is guaranteed, and the residual and jamming of food in the feeding device are avoided. The monitoring device can judge the survival signs of animals and the amount of food by obtaining image information, and remind the abnormal situation, which can help to improve the automation and reliability of the feeding system.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of breeding, in particular to an automatic feed system for crushing feed and monitoring animal signs. BACKGROUND

[0002] With the development of automatic mechanical technology, more and more work can be undertaken, and in recent years, it has been widely used in the field of farming and animal husbandry.

[0003] In the process of animal breeding, accurate feed feeding is crucial to ensure the growth of animals. However, the existing feeding device mostly uses pipeline transportation for delivery, which has many drawbacks: high price, the pipeline cannot crush and clean the feed, resulting in a certain amount of residual feed in the pipeline, which is prone to become spoiled residue, generating mold, bacteria, etc., causing pollution to the entire delivery pipeline, inducing animal diseases, and harming animal health. At the same time, the existing feeding equipment has low feed feeding accuracy, which is prone to insufficient or excessive feeding, resulting in insufficient feeding or waste of feed, increasing the overall cost, and also causing excess feed to be contaminated, increasing the pressure on site cleaning.

[0004] In addition, the current automated breeding has a monitoring gap. Due to the large scale of breeding, it is difficult for breeders to pay attention to the health status of all animals, such as abortion, and fallen young animals. The discovery of such conditions is often not timely enough, resulting in animals not receiving timely treatment and causing economic losses to the breeding farm.

[0005] Therefore, the current automated breeding scheme still has problems to be solved, and should be optimized to improve the accuracy of feed feeding in the process of automated breeding, while optimizing the structure of feed feeding to reduce feed residue and avoid feed spoilage, thereby ensuring the health of animal diet, and automatically monitoring the state of animals to timely remind when there is an abnormal animal, timely treatment of the animal to reduce unnecessary losses. Therefore, a more reasonable technical scheme is needed to solve the technical problems in the prior art. SUMMARY

[0006] To overcome at least one of the above-mentioned defects, the present application provides an automatic feed system for crushing feed and monitoring animal signs, which aims to optimize the structure of the feeding device, crush the feed during feeding, better control the feeding amount, reduce the residual feed, and avoid food spoilage affecting animal health; at the same time, by monitoring the animal status in real time and giving a reminder, abnormal conditions can be found and handled in time to reduce losses.

[0007] To achieve the above-mentioned purpose, the feeding system disclosed by the present application can adopt the following technical scheme:

[0008] An automatic feed system for breaking and monitoring signs, comprising:

[0009] A storage device for storing and carrying feed to a set feeding point; the storage device comprises a storage trolley for storing feed, the storage trolley travels along a track and reaches the feeding point;

[0010] A feeding device for conveying the feed in the storage trolley outward, the feeding device comprises a feeding channel and a shaftless spiral blade arranged in the feeding channel, the shaftless spiral blade pushes the feed to travel in the feeding channel when it rotates, and the feed falls to a weighing device when it is pushed to a discharge point;

[0011] A breaking device comprising a breaking shaft coaxially matched with the shaftless spiral blade, the surface of the breaking shaft is provided with a breaking spiral for breaking the feed matched with the inner side of the shaftless spiral blade, and the rotation direction of the breaking spiral is opposite to that of the shaftless spiral blade;

[0012] A weighing device for weighing the received feed, comprising a weighing disc and a turnover mechanism, the turnover mechanism drives the weighing disc to turn over to pour the feed to a receiving device below when the feed on the weighing disc reaches a set weight;

[0013] A receiving device comprising a receiving hopper, a conveying pipeline is arranged below the receiving hopper, and the conveying pipeline is used to convey the feed in the receiving hopper to a feeding trough;

[0014] A monitoring device comprising an animal monitoring device for monitoring the state of animals at the feeding cage and a feed monitoring device for monitoring the residual feed in the feeding device and the feeding trough.

[0015] The above disclosed feeding system drives the storage trolley to travel between multiple feeding cages through the track, the feed in the storage device is conveyed to the weighing device by the feeding device, and the feed can be supplemented and broken by the breaking device during the conveying process to avoid blockage of the feeding device, when the feed on the weighing device reaches the set weight, the feed at the weighing device is poured to the receiving device and reaches the corresponding feeding cage for the animals to eat. In this way, the intelligent degree of automatic feeding and the precision of feeding control are improved, which facilitates better feeding. At the same time, the monitoring device is used to monitor the vital signs of the animals, including judging the physical health degree by recognizing and processing the excrement image, judging the survival condition by recognizing and processing the image of the pups, and monitoring the residual amount of the feed in the feeding device in real time to avoid deterioration of the residual feed and cause food safety hazards. The above monitoring can timely issue an alarm to avoid greater losses.

[0016] Further, the storage trolley is used as a bearing structure to drive other devices to move synchronously, which can be achieved by various schemes, and the structure is not uniquely limited. Here, one feasible option is optimized and proposed: the first support is arranged on the storage trolley, the storage hopper is arranged on the first support, and the storage hopper is communicated and matched with the feeding device through the discharge pipeline. The second support is also arranged on the storage trolley, and the feeding device, the crushing device, the weighing device and the monitoring device are all connected and matched to the second support and move synchronously with the trolley. When the above scheme is adopted, the first support can be arranged as a fence structure to place the storage hopper; the second support includes a cantilever connected to the storage trolley and a connecting plate arranged on the cantilever, and the feeding device, the crushing device, the weighing device and the connecting plate are matched and fixed.

[0017] Further, the feeding channel is used to transport food, which can be constructed in various forms, and the structure is not uniquely limited. Here, one feasible option is optimized and proposed: the feeding channel includes a lower pipe body and an upper pipe body that are buckled with each other, and the upper pipe body and the lower pipe body form a circular feeding channel after buckling, and the shaftless spiral blade is arranged in the feeding channel; the discharging point includes a discharging port arranged on the lower pipe body. When the above scheme is adopted, the upper pipe body and the lower pipe body can adopt a semicircular pipe.

[0018] Further, the shaftless spiral blade realizes the transportation of food during rotation, and the driving thereof can be achieved by various schemes. Here, one feasible option is optimized and proposed: the end of the shaftless spiral blade is matched and connected with a synchronous wheel, the synchronous wheel is connected and driven by a first driving component, and when the first driving component is started, the synchronous wheel rotates to drive the shaftless spiral blade to rotate synchronously. When the above scheme is adopted, the first driving component includes a rudder, the output shaft of the rudder is connected to a driving wheel, and a synchronous belt is connected between the driving wheel and the synchronous wheel, so as to realize transmission and drive the shaftless spiral blade.

[0019] Further, the crushing shaft cooperates with the shaftless spiral blade to shear and crush the food during rotation, so as to avoid that the particle size of the food is too large, facilitate the control of the feeding precision, and avoid food residues. The driving of the crushing shaft can be achieved by various ways. Here, one feasible option is optimized and proposed: the end of the crushing shaft is connected with a second driving component, the second driving component drives the crushing shaft to rotate, and the crushing screw and the shaftless spiral blade shear and crush the food when the crushing shaft and the shaftless spiral blade rotate coaxially. When the above scheme is adopted, the second driving component can adopt a rudder, and the rudder drives the crushing shaft to realize continuous rotation. The crushing screw and the shaftless spiral blade form a certain gap, and the food can pass through the gap after being crushed to a specified particle size.

[0020] Further, the food reaches the set weight under the weighing of the weighing device, and is poured into the lower receiving device after reaching the set weight. The pouring action can be achieved by a turnover action, and various schemes can be adopted. The structure is not uniquely limited, and one of the feasible options is optimized and proposed herein. The turnover mechanism includes a turnover driving part and a turnover shaft. The turnover driving part drives the turnover shaft to rotate. The weighing disc is connected with the turnover shaft and turns with the turnover shaft. When the above scheme is adopted, the rudder machine can be used as the turnover driving part. Under the driving of the rudder machine, the weighing disc can be turned by 90°.

[0021] Further, when the weighing disc pours the food, the accuracy of the pouring of the food needs to be maintained to avoid spilling. The effect can be achieved by improvement, and the structure is not uniquely limited. One of the feasible options is optimized and proposed herein. The edge of the weighing disc is provided with a pouring port, and a guide anti-spilling area extending to the pouring port is further provided. When the weighing disc is turned and the pouring port faces the receiving device, the food enters the pouring port along the guide anti-spilling area and is discharged to the receiving device. When the above scheme is adopted, the guide anti-spilling area is an arc surface, and is arranged beside the pouring port.

[0022] Further, the feeding cage is arranged in the track, and a plurality of separate feeding compartments are arranged in the middle. Therefore, the receiving device is arranged at each compartment, and the structure is not uniquely limited. One of the feasible options is optimized and proposed herein. The receiving device is arranged at intervals along the extension direction of the track, and each feeding point is provided with a receiving device. When the above scheme is adopted, when the feeding cage adopts a stacked structure, separate receiving devices are arranged in the upper and lower compartments.

[0023] Further, the monitoring device is used for observing and feeding back the feeding situation, and giving a warning for abnormal situation. The animal monitoring device includes a first mounting frame, and a first camera is arranged on the first mounting frame. The first camera is used to acquire image data of the feeding cage and send the image data to a server. The image data includes excrement image and pup image. When the above scheme is adopted, the first camera can acquire data towards the feeding cage. When the feeding cage adopts a stacked structure, the first camera acquires the image below the feeding cage as judgment data. For example, the excrement of the animals in the layer falls to the layer below. The image acquired by monitoring this place can determine the diet health degree of the animals in the layer. When a pup dies or falls, it also falls to the layer below. The image acquired by this place can find the abnormality of the pup, so as to give a reminder and take measures quickly.

[0024] Further, the monitoring device also includes monitoring the feeding place, including monitoring the food delivery, the food monitoring device includes a second mounting frame, the second mounting frame is provided with a second camera, the second camera is used to acquire image data of the feeding device and the trough and send to the server, the image data includes food residue image. When the food is transported in the feeding device, the residual amount is captured by the second camera, and the residual food is judged according to the image information, and when the residual food amount reaches the alarm amount, timely reminding is sent.

[0025] Compared with the prior art, some beneficial effects of the technical scheme of the present application include:

[0026] The present application can further improve the feeding system, and the food can be crushed again during transportation, so as to ensure the uniformity of the food particles, avoid the residual and jamming of the food in the conveying device, and facilitate the subsequent accurate weighing and feeding of the food, thereby improving the accuracy of feeding. During the operation of the system, the monitoring device can judge the vital signs of the animals by acquiring image information, and can also monitor the food in the feeding device and the trough. When there is an abnormality in the image information, a feedback can be given, so as to help improve the automation and reliability of the feeding system. BRIEF DESCRIPTION OF DRAWINGS

[0027] In order to more clearly illustrate the technical scheme of the embodiments of the present application, the following will briefly introduce the drawings needed in the embodiments. It should be understood that the following drawings only represent some embodiments of the present application, and should not be regarded as limiting the scope. For those skilled in the art, other related drawings can also be obtained without creative labor.

[0028] Figure 1 It is a schematic diagram of the overall structure of the feeding system.

[0029] Figure 2 It is a schematic diagram of the structure of the storage device cooperating with the track.

[0030] Figure 3 It is a schematic diagram of the cooperation connection at the storage trolley.

[0031] Figure 4 It is a schematic diagram of the cooperation structure of the feeding device, the crushing device and the weighing device.

[0032] Figure 5 It is a schematic diagram of the combination of the shaftless helical blade and the crushing shaft.

[0033] Figure 6 It is a schematic diagram of the exploded structure of the feeding channel.

[0034] Figure 7A schematic diagram of a system multi-task workflow.

[0035] Figure 8 A system architecture diagram of YOLOv8n.

[0036] Figure 9 An image of feed pellets caused by abortion events.

[0037] In the above drawings, the meanings of various marks are as follows:

[0038] 1, track; 2, storage trolley; 201, first support; 202, storage hopper; 203, cantilever; 204, connecting plate; 205, first mounting bracket; 3, receiving hopper; 4, conveying pipeline; 5, trough; 6, feeding cage; 7, feeding device; 701, shaftless spiral blade; 702, crushed material shaft; 703, crushed material screw; 704, lower pipe body; 705, upper pipe body; 706, discharge port; 8, weighing disc; 801, pouring port; 802, guide anti-spill area; 9, pressure measuring element; 10, first driving part; 1001, driving wheel; 1002, synchronous belt; 1003, synchronous wheel; 1004, reinforcing rib column; 11, second driving part; 12, overturning driving part. DETAILED DESCRIPTION

[0039] The present embodiment will be further explained in conjunction with the drawings and specific embodiments.

[0040] In view of the deficiencies in the prior art feeding equipment, such as inaccurate feeding amount control and inability to monitor and feedback animal physical signs, the following embodiments optimize and overcome the defects in the prior art.

[0041] Embodiment 1

[0042] As shown in Figures 1-6 , the present embodiment provides a feeding system for automatic crushed material and physical sign monitoring, aiming to improve the automation and intelligence of the feeding system, improve the control accuracy during the feeding process, and timely monitor and feedback abnormal conditions, so as to take timely measures and reduce losses.

[0043] As shown in Figure 3 , one of the structures of the feeding system provided in the present embodiment includes:

[0044] The storage device is used to store and carry food to the designated 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.

[0045] The storage trolley 2 serves as a bearing structure to drive other devices to move synchronously. The structure can be implemented in various schemes, and is not uniquely limited. In the embodiment, one feasible scheme is adopted: the first support 201 is arranged on the storage trolley 2, the storage hopper 202 is arranged on the first support 201, and the storage hopper 202 is connected and matched with the feeding device 7 through the discharge pipeline. The feeding device 7, the crushing device, the weighing device and the monitoring device are all connected and matched to the second support and move synchronously with the trolley. When the above scheme is adopted, the first support 201 can be arranged as a fence structure to place the storage hopper 202. The second support includes a cantilever 203 connected to the storage trolley 2 and a connecting plate 204 arranged on the cantilever 203. The feeding device 7, the crushing device, the weighing device and the connecting plate 204 are matched and fixed.

[0046] As shown in Figure 4 、 Figure 5 、 Figure 6 , the second structure of the feeding system provided in the embodiment includes:

[0047] The feeding device 7 is used to transport the food in the storage trolley to the outside. The feeding device 7 includes a feeding channel and an axisless spiral blade 701 arranged in the feeding channel. When the axisless spiral blade 701 rotates, the food is pushed to move in the feeding channel. When the food is pushed to the discharge point, it falls to the weighing device.

[0048] The feeding channel is used to transport the food and can be constructed in various forms. The structure is not uniquely limited. In the embodiment, one feasible scheme is adopted: the feeding channel includes a lower pipe body 704 and an upper pipe body 705 that are buckled with each other. After the upper pipe body 705 and the lower pipe body 704 are buckled, a circular feeding channel is formed, and the axisless spiral blade 701 is arranged in the feeding channel. The discharge point includes a discharge port 706 arranged on the lower pipe body 704. When the above scheme is adopted, the upper pipe body 705 and the lower pipe body 704 can adopt a semicircular pipe.

[0049] The axisless spiral blade 701 realizes the transportation of the food when it rotates. The driving can be realized in various schemes. Here, one feasible scheme is proposed: the end of the axisless spiral blade 701 is matched and connected with a synchronous wheel 1003. The synchronous wheel 1003 is connected and driven by the first driving part 10. When the first driving part 10 is started, the synchronous wheel 1003 is driven to rotate, and the axisless spiral blade 701 is synchronously rotated. When the above scheme is adopted, the first driving part 10 includes a rudder machine. The output shaft of the rudder machine is connected to drive the wheel 1001. The synchronous belt 1002 is connected between the drive wheel 1001 and the synchronous wheel 1003, so as to realize transmission and drive the axisless spiral blade 701.

[0050] Preferably, in the embodiment, the shaftless spiral blade 701 is connected and fixed with the synchronous wheel 1003 through the reinforcing rib column 1004.

[0051] As shown in Figure 5 , the feeding system provided in the embodiment has the following structure:

[0052] The crushing device comprises a crushing shaft 702 coaxially matched with the shaftless spiral blade 701, and the surface of the crushing shaft 702 is provided with a crushing spiral 703 matched with the inside of the shaftless spiral blade 701 to crush the food, and the rotation direction of the crushing spiral 703 is opposite to that of the shaftless spiral blade 701.

[0053] The crushing shaft 702 is matched with the shaftless spiral blade 701 to shear and crush the food in the process of rotation, so as to avoid that the food is too large in granularity, facilitate the control of the feeding precision, and avoid the food residues. The driving of the crushing shaft 702 can be realized in various ways, and the embodiment is optimized and one of the feasible choices is adopted: the end of the crushing shaft 702 is connected with a second driving component 11, the second driving component 11 drives the crushing shaft 702 to rotate, and the crushing spiral 703 and the shaftless spiral blade 701 shear and crush the food when the crushing shaft 702 and the shaftless spiral blade 701 rotate coaxially. When the above scheme is adopted, the second driving component 11 can adopt a rudder machine to realize continuous rotation of the crushing shaft 702. A certain gap is formed between the crushing spiral 703 and the shaftless spiral blade 701, and the food can pass through the gap when it is crushed to a specified granularity.

[0054] As shown in Figure 4 , the feeding system provided in the embodiment has the following structure:

[0055] The weighing device is used to weigh the received food, and comprises a weighing disc 8 and a turnover mechanism. When the food on the weighing disc 8 reaches a set weight, the turnover mechanism drives the weighing disc 8 to turn over to pour the food to the receiving device below.

[0056] Preferably, a pressure measuring element 9 is arranged below the weighing disc 8 to accurately measure the weight of the food.

[0057] After the food reaches the set weight under the weighing of the weighing device, it is poured to the receiving device below, and the pouring action can be realized by a turnover action, and various schemes can be adopted, and the structure is not uniquely limited. The embodiment is optimized and one of the feasible choices is adopted: the turnover mechanism comprises a turnover driving component 12 and a turnover shaft, the turnover driving component 12 drives the turnover shaft to rotate, and the weighing disc 8 is connected with the turnover shaft and turns over with the turnover shaft. When the above scheme is adopted, the turnover driving component 12 can adopt a rudder machine, and the weighing disc 8 can realize 90° turnover under the driving of the rudder machine.

[0058] In order to keep the food accurate and avoid spilling when the food is poured into the weighing disc 8, the pouring edge 801 of the weighing disc 8 is provided with a guide spilling prevention area 802 extending to the pouring edge 801, and when the weighing disc 8 is turned over and the pouring edge 801 faces the receiving device, the food is poured into the pouring edge 801 along the guide spilling prevention area 802 and discharged to the receiving device.

[0059] As shown in the figure, the feeding system provided in the embodiment includes the following structures: Figure 1 As shown in the figure, the feeding system provided in the embodiment includes the following structures:

[0060] The receiving device includes a receiving hopper 3, and a conveying pipeline 4 is arranged below the receiving hopper 3, which is used to convey the food in the receiving hopper 3 to the feeding trough 5.

[0061] The feeding cage 6 is arranged in the track 1 and is spaced apart into multiple separate feeding compartments, and the receiving device is arranged at each compartment, and the structure is not uniquely limited, and the embodiment is optimized and one of the feasible options is adopted: the receiving device is arranged at each feeding point along the extension direction of the track 1. When the feeding cage 6 adopts a stacked structure, the upper and lower compartments are respectively provided with separate receiving devices.

[0062] As shown in the figure, the feeding system provided in the embodiment includes the following structures: Figure 3 As shown in the figure, the feeding system provided in the embodiment includes the following structures:

[0063] The monitoring device includes an animal monitoring device used to monitor the state of the animals in the feeding cage 6 and a food monitoring device used to monitor the residual food in the feeding trough 5 and the feeding device 7.

[0064] The monitoring device is used to observe and feed back the feeding situation and give an early warning for abnormal situations, wherein the animal monitoring device includes a first mounting frame 205, and a first camera is arranged on the first mounting frame 205, which is used to obtain image data of the feeding cage 6 and send it to the server, and the image data includes excrement images and pup images. When the first camera obtains data towards the feeding cage 6, when the feeding cage 6 adopts a stacked arrangement, the first camera obtains the image below the feeding cage 6 as the judgment data, for example, the excrement of the animals in the layer will fall to the layer below, and monitoring and obtaining the image at this position can determine the diet health degree of the animals in the layer; when a pup dies or falls, it will also fall to the layer below, and the image at this position can find the abnormality of the pup, so as to give a prompt and take quick measures.

[0065] The monitoring device also includes monitoring the feeding station, including monitoring the food delivery, and the food monitoring device includes a second mounting bracket, and a second camera is arranged on the second mounting bracket, and the second camera is used to obtain image data at the feeding device 7 and send it to the server, and the image data includes food residue images. When the food is transported in the feeding device 7, the residual amount is captured by the second camera, and the residual food is judged according to the image information, and when the residual food amount reaches the alarm amount, a prompt is sent in time.

[0066] The feeding system disclosed in the embodiment drives the storage trolley 2 to walk between the multiple feeding cages 6 through the track 1, and the food in the storage device is transported to the weighing device by the feeding device 7. In the transportation process, the food can be supplemented and crushed by the crushing device to avoid blockage of the feeding device 7. 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. In this way, the intelligent degree of automatic feeding and the precision of feeding control are improved, which facilitates better feeding. At the same time, the monitoring device is used to monitor the vital signs of the animals, including identifying and processing the excrement images to judge the body health degree, identifying and processing the images of the pups to judge the survival situation, and monitoring the residual amount of food in the feeding device 7 in real time to avoid spoilage of the residual food and cause food safety hazards. The above monitoring can send an alarm prompt in time to avoid greater losses.

[0067] Embodiment 2

[0068] The above embodiment discloses a feeding system, and the present embodiment provides a feeding method for the operation of the above system, including the following steps:

[0069] S01: The storage trolley 2 reaches the designated feeding cage 6 above along the track 1.

[0070] S02: The first camera of the monitoring device takes pictures of the feeding cage 6 to identify abnormal situations such as miscarriage and pup falling; the second camera takes pictures of the feeding device 7 and the feeding trough 5 to identify the residual food amount in the feeding device 7 and the feeding trough 5.

[0071] S03: According to the residual food amount in the feeding trough 5 and the current animal information, the appropriate feeding amount is calculated. The first driving part 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.

[0072] S04: The second driving part 11 rotates in the opposite direction to drive the crushing shaft 702 to rotate to crush the food stuck.

[0073] S05: The weighing disc 8 detects whether the food weight reaches the preset value.

[0074] S06: When the preset value is reached, the first driving component 10 and the second driving component 11 stop rotating, and the overturning driving component 12 drives the weighing disc 8 to rotate, so that the food falls into the receiving device.

[0075] S07: The weighing disc 8 is homed, and the feeding process is completed.

[0076] Example 3

[0077] The above examples disclose a feeding system and a feeding method. In this example, a method for recognizing, obtaining and feeding back image data is disclosed.

[0078] During system operation, the first camera and the second camera work independently. The second camera takes images regularly, analyzes the food state, and triggers a food shortage warning when the food proportion is lower than the set value. The first camera takes images regularly, detects bloodstains, food and cub fall, and triggers an abortion warning, a cleaning reminder and a high-priority alarm when the system detects related abnormal conditions.

[0079] For image acquisition and recognition and conversion into alarm signals, the following system components are involved:

[0080] Animal monitoring device, used to capture image data at the feeding cage;

[0081] Food monitoring device, used to capture image data at the feeding device;

[0082] Server, used to receive and process image data, read pixel information on the image data by recognition, and compare with data stored in the server, so as to determine the corresponding image data as the situation represented by the corresponding data, and mark the determination result as normal or abnormal. When the abnormal result is marked, the server generates an alarm signal and sends it to the alarm device;

[0083] Alarm device, used to receive the alarm signal of the server and provide alarm action.

[0084] Specifically, by using the above system components, the corresponding alarm signal is generated through image data, and the following method is proposed:

[0085] The captured image data is sent to the server for processing and analysis, and the parameters in the image are extracted, including but not limited to pixel value;

[0086] Compare the extracted parameter value with the data stored in the data model to find the closest data;

[0087] The closest data represents the actual situation, as the extracted parameter value faces the actual situation;

[0088] If the actual situation is negative, such as miscarriage, death, food residue, etc., an alarm signal is generated and sent by the server, and the alarm signal is received by the alarm device;

[0089] When the alarm device receives the alarm signal, the alarm reminder is started.

[0090] Specifically, some specific processes in the above method are described in detail below.

[0091] The process of image analysis by the model includes establishing a data model and running the data model.

[0092] The process of establishing a data model includes:

[0093] Step one: data preprocessing, including feature selection, data set preparation, and data standardization.

[0094] Step two: design network structure, initialize parameters.

[0095] Step three: train the model, use the data set to train the model.

[0096] Step four: get the prediction model, after completing the training of the model, the prediction model can be obtained.

[0097] The process of running the data model for judgment includes:

[0098] Step one: load the data model, successfully load the trained prediction model.

[0099] Step two: data preparation, organize the collected data for model prediction.

[0100] Step three: model prediction: predict the collected data by the trained model.

[0101] Step four: model analysis, get the prediction result and return the result.

[0102] 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 realized.

[0103] Specifically, the second camera is used to monitor the residual amount of food, and the data can be processed and abnormality judgment can be realized through the following process.

[0104] Image acquisition needs to cope with different light conditions, shooting angles and background complexity to improve the generalization ability of the model. Data annotation can use but not limited to Labelme tool to accurately mark the trough area, and the marked feed part is assigned a "feed" label. The annotation result is converted into COCO format and then into MindRecord format, which is convenient for input into the prediction model for training.

[0105] In image processing, the gradient of the image is calculated by using the Sobel operator to extract the edge information of the feed area:

[0106]

[0107]

[0108] In the above formula, represents the image in the coordinate The pixel brightness value is 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 the upper and lower adjacent pixels.

[0109] After synthesizing the gradient, the edge detection result is obtained:

[0110]

[0111] When the gradient value exceeds the set threshold value, the edge point in the image is marked as the boundary of the feed area. Subsequently, by calculating the proportion of the feed area in the image Feed_RatioFeed_Ratio, it is judged whether the feed is sufficient:

[0112]

[0113] In the above formula, represents the area (pixel integral) of the feed area monitored in the image, represents the area of the total image area.

[0114] The threshold value TfT_fTf is generally set to 1 / 5. When the calculated proportion of feed remaining is less than 1 / 5, the feed shortage warning is triggered.

[0115] In addition, the first camera is used to monitor vital signs, which can be processed through the following process and judged for abnormalities.

[0116] The animal monitoring device is mainly used to detect abnormal situations on the feeding cage, including feces accumulation, bloodstains, feed mixing (mixing of feed and feces), and kit falling, etc. To accurately label the abnormal areas in the image, the image can be labeled using but not limited to the Labelme tool, and labeled as "blood" (bloodstains), "manure" (feces), "mixed" (mixed area), and "kit" (kits) respectively.

[0117] 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: 0.05-0.35, Intensity: 0.45-0.85. By calculating the hue HH, saturation SS, and intensity II, the bloodstains can be effectively distinguished:

[0118]

[0119]

[0120]

[0121] 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 the hue (Hue), which is used to distinguish color categories such as light pink bloodstains; S represents the saturation (Saturation), which represents the color purity; and I represents the intensity (Intensity), which represents the color brightness.

[0122] Through binarization processing, the bloodstain area is extracted and further processed by morphological operation to remove noise.

[0123] Feed mixing detection distinguishes the mixing degree of feed and feces through texture analysis, and uses binocular stereo vision to calculate the volume of the mixed area:

[0124]

[0125] In the above formula, represents the height information of each pixel point in binocular stereo vision; represents the area of the mixed area (such as feed and feces mixing); represents the volume of the mixed area. Through this method, the system can estimate the volume of the mixed area and determine whether it needs to be cleaned.

[0126] For the detection of the fall of the kits, the system first trains n artificial labeled images containing kits in the offline stage: in each image, the pink skin area is accurately marked along the rabbit outline, Hue (HH) ∈ [0, 20], Saturation (SS): 0.05 — 0.35 Intensity (II): 0.45 — 0.85. By calculating the hue HH, saturation SS and brightness II, the bloodstains are effectively distinguished, and their proportion in the whole image is calculated; then the model automatically learns an optimal threshold (about 20%) based on these proportion labels; when the proportion P of pixels in the image determined by the H, H / S, S / I (i.e. HH, SS, II) rule as “pale pink skin / bloodstains” is greater than or equal to 0.20, the system determines that it is “fall of the kits” and issues an alarm:

[0127]

[0128] In the above formula, the image in the above formula is taken at a fixed angle and in a fixed position, so the total area is constant, and there is no randomness in selection, which will not interfere with model recognition. represents the area of the kits skin region found in the image; represents the total area of the image.

[0129] Specifically, as shown in Figures 7-9 , actual cases are proposed here, which respectively detect bloodstains and fetuses to identify the abortion time, and detect spilled feed particles to identify the feed spillage behavior.

[0130] Case

[0131] I. Abortion detection based on YOLOv8n

[0132] The YOLO (You Only Look Once) series of target detection algorithms are widely used in livestock monitoring, automatic feeding and health assessment tasks, especially YOLOv8, which performs well in processing images to efficiently identify key targets; YOLOv8P2 is more accurate in monitoring small targets and can avoid information loss.

[0133] As shown in Figure 7 , YOLOv8n (the lightest variant in the YOLOv8 series) is used to detect abortion events in rabbit breeding environments. The model takes RGB images as input and outputs the positions of related visual indicators such as bloodstains and fetuses.

[0134] AsFigure 8 As shown, the backbone network of YOLOv8n adopts a cross-stage local (CSP) structure, starting from 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, a spatial pyramid fast pooling (SPPF) module aggregates context information of different receptive field sizes, improving the model's ability to capture fine textures and high-level semantics.

[0135] The neck adopts a bottom-up path, where deep feature maps from the backbone network are upsampled and concatenated with shallow feature maps, while the top-down path reintroduces the fused output to higher layers through downsampling and further concatenation. This bidirectional flow ensures that each output feature map is fused with detailed spatial cues from shallow layers and strong contextual information from deep layers.

[0136] The detection head operates on the three fused feature maps (H3, H4, and H5). For each scale, parallel convolutional branches generate classification scores and bounding box predictions, respectively. This design enables accurate localization and classification across a wide range of object sizes related to bloodstains and fetuses in abortion events.

[0137] II. Feed spill detection based on YOLOv8n-P2

[0138] Due to the extremely small size of scattered feed particles and the cluttered environment of rabbit breeding, detecting them poses a significant challenge. To address this issue, the YOLOv8n-P2 model is used to detect scattered feed particles. This model is a variant of the YOLOv8 series specifically designed for ultra-small target detection.

[0139] As shown in Figure 9 YOLOv8n-P2 introduces an additional detection branch in the neck based on the original YOLOv8n architecture, focusing on upsampling the C2f output of the shallowest layer to match the spatial resolution of higher layers. These refined features are then connected in the neck using similar bottom-up and top-down paths 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 target detection. As a result, the detection head in YOLOv8n-P2 operates on four scales: P2, P3, P4, and P5, corresponding to small targets, small-to-medium targets, medium-to-large targets, and large targets, respectively. By explicitly integrating P2 features through additional upsampling in the neck, YOLOv8n-P2 significantly improves its sensitivity to small targets, enabling more accurate localization of feed particles in the second task.

[0140] III. Result fusion through conditional filtering

[0141] The detection of blood or fetus and the detection of feed pellets are fused by a conditional filtering method. First, the two sets of detection results of the two tasks are defined as follows:

[0142]

[0143] wherein represents the detection frame of blood or fetus, represents the detection frame of feed pellets, and are their corresponding confidence values, respectively.

[0144] Next, the conditional filtering first filters the detection frame of blood or fetus in the abortion detection task to ensure that the result has a high confidence. Then, the intersection over union (IoU) between each filtered blood or fetus frame and each feed pellet frame is calculated, and then the feed pellet frame with a larger IoU value is selected as the final detection result, as follows:

[0145]

[0146] wherein and are the detection frames selected from the two tasks. The IoU value quantifies the degree of overlap of the two frames, 0 indicating no overlap and 1 indicating complete overlap. In this case, only the feed pellets coexisting with the abortion indicators in space are selected, resulting in accurate and non-redundant feed spill detection output.

[0147] IV. Specific experimental content

[0148] A. Data

[0149] In this study, 4985 high-resolution RGB images were collected from closed environment rabbit farms in Sichuan Province, China, with an image size of 3024x4032. These images were taken at a distance of 10-40 cm from the fecal plate below the rabbit cage, from multiple angles, under natural and artificial lighting conditions, to ensure the diversity of the data. All images were annotated in YOLO TXT format, with accurate bounding boxes drawn for three target classes (blood, fetus, and feed pellets). To adapt to the multi-task design, we constructed two subsets, each containing about 2500 images. One subset was used to detect blood and fetus, and the other subset was used to detect feed pellets. Each subset was divided into training set, validation set and test set in the ratio of 80%, 10% and 10%.

[0150] B. Implementation details

[0151] In this study, each RGB image of size 3024x4032 was resized to 640x640 and converted by min-max normalization. A series of data augmentation methods were adopted, including image stitching, mixing, HSV jitter, random scaling, translation, and horizontal flipping, to facilitate the model to learn robust representations for detecting bloodstains, fetuses, and feed pellets under 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 of 0.01, and a gradual decay to 1x10-4 through a cosine annealing scheduling strategy with a decay factor of 5x10-4. To accelerate convergence and reduce memory usage, FP16 mixed precision training was adopted. In our conditional filtering algorithm, the confidence threshold a was set to 0.3, which determined the minimum confidence level required for the abortion sign detection to participate in the fusion process. The overlap threshold b was set to 0.5 to limit the minimum IoU required for the abortion box and the feed pellet box to be considered coexisting.

[0152] C. Performance

[0153] 1. Abortion event detection: As shown in Table 1, the overall mAP@0.5 of the YOLOv8n model reached 95.7%, with excellent performance in the bloodstain class (98.4%) and accurate detection in the fetus class (94.0%). These results indicate that YOLOv8n not only accurately locates the signs related to abortion but also maintains consistency under different lighting and background conditions. In addition, the precision of YOLOv8n was 97.2%, the recall was 95.6%, and the F1 score was 96.4%, indicating that both the false positive rate and the false negative rate were low. These indicators confirm that YOLOv8n provides accurate detection results for real-time monitoring of rabbit abortion events and provides a reliable source of results for another feed spill detection task.

[0154] Table 1 Performance of YOLOv8n in detecting abortion events

[0155]

[0156] 2. Feed spill detection caused by abortion: The goal was to detect feed spills caused by abortion events. The final detection results were calculated by the results of the conditional filtering and fusion, as shown in Table 1.

[0157] Table 2 Performance of YOLOv8n and YOLOv8n-P2 in detecting feed spills caused by abortion events

[0158]

[0159] The YOLOv8n-P2 model accurately detects 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 detection performance of the original YOLOv8n on feed pellets. In comparison, the model variant YOLOv8n-P2 achieves more accurate results than the original YOLOv8n. Specifically, the mAP@0.5, precision, and recall are improved by 9%, 12%, and 9%, respectively. This confirms the effectiveness of YOLOv8n-P2 in capturing fine details. Figure 9 We demonstrate our detection example on feed pellets, showing accurate box results for detecting feed spillage caused by abortion events.

[0160] D. Computational Efficiency

[0161] Computational efficiency is crucial in detection tasks. In this work, both YOLOv8n and YOLOv8n-P2 detectors in the two tasks run on an NVIDIA A6000 GPU, occupying 4.2 GB and 5.6 GB of GPU memory, respectively, while achieving running speeds of 82.0 FPS and 76.0 FPS, enabling real-time inference for feed spillage detection.

[0162] The above is the implementation mode listed in the embodiment, but the embodiment is not limited to the above optional implementation mode, and those skilled in the art can obtain other various implementation modes by arbitrarily combining the above modes with each other. Any person can obtain other various forms of implementation modes under the inspiration of the embodiment. The above specific implementation mode should not be understood as a limitation on the protection scope of the embodiment, and the protection scope of the embodiment should be defined by the claims.

Claims

1. A feeding system that automatically crushes feed and monitors vital signs, characterized in that, include: A storage device for storing and carrying food to a designated feeding point; the storage device includes a storage trolley (2) for storing food, the storage trolley (2) travels along a track (1) and arrives at the feeding point; The feeding device (7) is used to transport the food in the storage trolley outward. The feeding device (7) includes a feeding channel and a shaftless spiral blade (701) set 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 discharge point, it falls to the weighing device. The crushing device includes a crushing shaft (702) coaxially engaged with the shaftless helical blade (701), and a crushing spiral (703) engaged with the inner side of the shaftless helical blade (701) for crushing materials on the surface of the crushing shaft (702). The direction of rotation of the crushing spiral (703) is opposite to that of the shaftless helical blade (701). The weighing device is used to weigh the received food, including a weighing pan (8) and a flipping mechanism. When the food on the weighing pan (8) reaches the set weight, the flipping mechanism drives the weighing pan (8) to flip so that the food is poured into the receiving device below. The receiving device includes a receiving hopper (3), and a conveying pipe (4) is provided below the receiving hopper (3). The conveying pipe (4) is used to convey the food in the receiving hopper (3) to the feeding trough (5). The monitoring device includes an animal monitoring device for monitoring the status of animals in the feeding cage (6) and a feed monitoring device for monitoring the feed device (7) and the feed trough (5) for monitoring the remaining feed.

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

3. The automatic feed shredding and vital sign monitoring system according to claim 1, characterized in that: The feeding channel includes a lower tube (704) and an upper tube (705) that interlock with each other. The upper tube (705) and the lower tube (704) interlock to form a circular feeding channel. The shaftless spiral blade (701) is disposed in the feeding channel. The discharge point includes a discharge port (706) disposed in the lower tube (704).

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

5. The automatic feed shredding and vital sign monitoring system according to claim 1, characterized in that: The end of the crushing shaft (702) is connected to a second driving component (11). 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 automatic feeding system for crushing and monitoring vital signs according to claim 1, characterized in that: The flipping mechanism includes a flipping drive component (12) and a flipping shaft. The flipping drive component (12) drives the flipping shaft to rotate. The weighing pan (8) is connected to the flipping shaft and flips with the flipping shaft.

7. The automatic feeding system for crushing and monitoring vital signs according to claim 1 or 6, characterized in that: The weighing pan (8) is provided with a pouring port (801) on its edge, and a guide anti-spillage zone (802) extending to the pouring port (801) is also provided; when the weighing pan (8) is flipped and the pouring port (801) faces the receiving device, the food enters the pouring port (801) along the guide anti-spillage zone (802) and is discharged to the receiving device.

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

9. The automatic feed shredding and vital sign monitoring system according to claim 1, characterized in that: The animal monitoring device includes a first mounting frame (205), on which a first camera is mounted. The first camera is used to acquire image data at the feeding cage (6) and send it to the server. The image data includes images of excrement and images of cubs.

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

Images