Pre-slaughter pig carbon dioxide stunning algorithm individual adjustment system

Abstract

This application relates to the fields of animal husbandry engineering and animal welfare technology. A pre-slaughter pig carbon dioxide stunning algorithm individual adjustment system includes: an identification device installed in the passageway of the waiting pen, used to identify the passing pigs by time and ID, forming a timestamp and identification ID; an inspection robot installed at the passageway and capable of moving along the length of the passageway, used to collect data from the pigs, forming static data; a stunning device with a gas release valve, used to receive instruction packets and adjust the opening of the gas release valve according to the instruction packets to stunning the pigs; and a control system used to receive the timestamp, identification ID, and static data, and to send instruction packets to the gas release valve, configured to: extract features from the static data to form static features, input the static features into a first AI algorithm model to form instructions, bind the instructions, identification ID, and timestamp to form an instruction packet, and send the instruction packet to the gas release valve.

Description

Technical Field

[0001] This application relates to the fields of animal husbandry engineering and animal welfare technology, specifically to the automation of slaughtering and processing and humane slaughter, and particularly to an individual adjustment system for pre-slaughter carbon dioxide stun dosage in pigs. Background Technology

[0002] Currently, in the large-scale pig slaughtering industry, carbon dioxide stunning is the mainstream pre-slaughter stunning method. This technology involves putting pigs into a closed environment containing 80%-90% high concentration of CO2, using hypoxia and blood acidosis to achieve stunning. Although it can avoid problems such as muscle spasms and fractures caused by electric shock stunning, it is widely used in slaughterhouses, but it still has many inherent technical defects.

[0003] Current carbon dioxide stunning technology employs a batch processing model, simultaneously placing multiple pigs in a confined space with a fixed CO2 concentration and an exposure time of 60-90 seconds. This completely ignores individual differences in weight, breed, genotype, and physiological state, easily leading to incomplete stunning of more tolerant pigs and excessive stunning of weaker pigs. High CO2 concentrations can cause severe respiratory discomfort in pigs, causing them to struggle violently during their conscious period before losing consciousness, violating humane slaughter principles. Furthermore, operators often deliberately increase the concentration and prolong the exposure time to ensure stunning effects, exacerbating physiological stress in pigs and leading to meat quality deterioration such as PSE (Polyester Sedimentary Extraction) and DFD (Distilled Dry Discharge) meat. Batch processing also easily causes pigs to trample each other and damage equipment, restricting slaughter efficiency. These problems have become critical technical pain points that the industry urgently needs to address. Summary of the Invention

[0004] The main objective of this application is to provide an individual adjustment system for the pre-slaughter carbon dioxide coma dosage of pigs to solve the technical problems in the background art.

[0005] To achieve the above objectives, this application proposes an individual adjustment system for the pre-slaughter carbon dioxide coma dose algorithm for pigs, comprising: An identification device is installed in the passageway of the slaughter pen to identify the passing pigs by time and ID, generating a timestamp and identification ID; An inspection robot is installed in the passageway and can move along the length of the passageway to collect data on the pigs and form static data. The stunning device has a gas release valve for receiving command packets and adjusting the opening of the gas release valve according to the command packets to stunning the pigs; The control system, connected to both the identification device and the inspection robot, receives the timestamp, the identification ID, and the static data, and sends the command packet to the gas release valve. The control system is configured to: The static data is subjected to feature extraction to form static features. The static features are input into the first AI algorithm model to form an instruction. The instruction, the identification ID and the timestamp are bound together to form an instruction packet. The instruction packet is sent to the gas release valve. The first AI algorithm model represents an AI algorithm model trained on historical data.

[0006] In some feasible embodiments, the inspection robot has a 3D depth camera used to collect data on the weight and breed of the pigs, forming static data.

[0007] In some feasible embodiments, the identification device includes an RFID scanning gate arranged within the passageway; wherein, When the pig passes through the passage, if the pig has an ear tag, the RFID scanning gate scans the ear tag of the pig to form the timestamp and the identification ID; If the pig does not have an ear tag, the RFID scanning gate will scan empty data, resulting in abnormal data. The control system is also configured to: The system receives the abnormal data and performs visual feature recognition based on the visible light visual image data acquired by the inspection robot at the same timestamp. If no ear tag feature is identified, the timestamp is bound to the static data to construct a virtual identification ID for the current pig. The visible light visual image data is the image data acquired synchronously when the 3D depth camera acquires the static data.

[0008] In some possible implementations, the identification device may also include an ear tag located on the pig's ear.

[0009] Among the feasible methods are: A visual data acquisition device is installed in the passageway or at the scale of the channel to collect data on the weight and breed of the pigs passing by, forming supplementary static data, wherein the supplementary static data are supplementary features of the static data. The control system is connected to the visual data acquisition device and is also used to receive the supplementary static data; wherein, The control system is also configured to: Feature extraction is performed on the supplementary static data to form supplementary static features. The supplementary static features are then fused with the static features to form target static features. The target static features are then input into the first AI algorithm model to form instructions.

[0010] In some feasible embodiments, the visual data acquisition device is a 3D vision module consisting of at least two cameras. The two cameras are separately set at different angles in the channel to identify different angles of the pig and generate first visual static data and second visual static data accordingly. The first visual static data and the second visual static data constitute the supplementary static data.

[0011] Among the feasible methods are: A body temperature monitoring device is installed on the inspection robot to monitor the temperature of the pigs passing by and generate a thermal imaging temperature map. The control system is connected to the body temperature monitoring device and is also used to receive the thermal imaging temperature map; wherein, The control system is also configured to: Based on the static features and their mirror features, a local three-dimensional model of the pig's ear base, eyes, and sides is formed. The thermal imaging temperature map is aligned with the local three-dimensional model, and the temperature features of the thermal imaging temperature map corresponding to the local three-dimensional model are extracted. The temperature features and the static features are fused and input into the second AI algorithm model to form the instruction. The second AI algorithm model represents an AI algorithm model formed by adding temperature parameter variables on the basis of the first AI algorithm model.

[0012] Among the feasible methods are: A microphone array, mounted on the inspection robot, is used to collect the sounds of the passing pigs to form sound information, wherein the sound information includes at least the frequency and pitch of the calls; The control system is connected to the microphone array and is also used to receive the sound information; wherein, The control system is also configured to: Feature extraction is performed on the sound information to form the sound features. Based on the time difference and phase difference of the sound features, the target sound features of the pig are located. The target sound features, the static features and the temperature features are fused and input into a third AI algorithm model to form an instruction. The third AI algorithm model is an AI algorithm model formed by adding sound parameter variables on the basis of the second AI algorithm model.

[0013] In some possible implementations, the stun-inducing device further includes a fan connected to the control system for receiving instruction packets sent to it by the control system and adjusting its operating time and power in response to the instruction packets.

[0014] In some possible implementations, the stunning device further includes a conveyor belt connected to the control system for receiving instruction packets sent to it by the control system and adjusting the transmission speed in response to the instruction packets.

[0015] The technical solutions provided by the embodiments of this application may include the following beneficial effects: This application presents an individual adjustment system for carbon dioxide coma induction dosage in pre-slaughter pigs, effectively addressing a series of industry pain points in existing large-scale pig slaughtering technologies. The system abandons the traditional batch processing model, using an inspection robot to accurately collect individual static data such as weight and breed from pigs in the aisle. This data is combined with a first AI algorithm model trained on historical data to calculate personalized coma parameters. Furthermore, by binding the timestamp and identification ID of an identification device, a unique instruction package is generated for each pig, controlling the opening of the gas release valve of the coma induction device. This achieves precise, individualized control of pig coma, completely resolving the problems of incomplete coma in pigs with strong tolerance and excessive coma in weak pigs under traditional fixed CO2 concentration and exposure time conditions, due to individual differences. The system precisely adjusts CO2 release based on individual pig characteristics, avoiding the blind use of 80%-90% high CO2 concentrations in traditional techniques. This effectively reduces respiratory discomfort in pigs and minimizes their violent struggles during the pre-consciousness period, aligning with humane slaughter principles. Furthermore, the system's precise stunning parameters, output by an AI algorithm model, replace the crude operation of operators deliberately increasing concentrations and extending exposure time to ensure stunning effects. This reduces physiological stress in pigs from the source, significantly decreasing the occurrence of PSE (Polyester Sedimentary Oxide) and DFD (Distilled Dioxide) meat quality deterioration, thus ensuring the quality of meat products after slaughter. Attached Figure Description

[0016] The accompanying drawings, which form part of this application, are used to provide a further understanding of the application and to make other features, objects, and advantages of the application more apparent. The illustrative embodiments and descriptions of this application are used to explain the application and do not constitute an undue limitation of the application. In the drawings: Figure 1 This is a schematic diagram of the structure of an individual adjustment system for the pre-slaughter carbon dioxide stun dosage algorithm for pigs provided in this application. Detailed Implementation

[0017] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0018] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0019] In this application, the terms "upper," "lower," "left," "right," "front," "rear," "top," "bottom," "inner," "outer," "middle," "vertical," "horizontal," "lateral," and "longitudinal" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These terms are primarily for the purpose of better describing this application and its embodiments, and are not intended to limit the indicated device, element, or component to having a specific orientation, or to be constructed and operated in a specific orientation.

[0020] Furthermore, in addition to indicating location or positional relationship, some of the aforementioned terms may also have other meanings. For example, the term "above" may also be used in some cases to indicate a certain dependency or connection relationship. Those skilled in the art can understand the specific meaning of these terms in this application based on the specific circumstances.

[0021] Furthermore, the terms "installation," "setup," "equipped with," "connection," "linked," and "socketing" should be interpreted broadly. For example, "connection" can be a fixed connection, a detachable connection, or an integral structure; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, or an internal connection between two devices, components, or parts. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.

[0022] like Figure 1 As shown, this application provides an individual adjustment system for pre-slaughter pig carbon dioxide stunning dosage algorithm, including an identification device, an inspection robot, stunning equipment, and a control system.

[0023] The inspection robot is located in the passage and can move along the length of the passage to collect data on the pigs and form static data.

[0024] Specifically, the inspection robot is equipped with a 3D depth camera, which is used to collect data on the weight and breed of the pigs to form static data.

[0025] It should be noted that a track is arranged on one side of the passage, and the inspection robot can move along the track. The specific movement method can be driven by a motor, and the movement distance can be set and controlled as needed. This application does not limit this.

[0026] The 3D depth camera is a conventional camera capable of capturing point cloud data and visible light image data. This application does not limit the model of the 3D depth camera.

[0027] An identification device is installed in the passageway of the slaughter pen to identify the time and ID of the passing pigs, generating a timestamp and identification ID.

[0028] Specifically, the identification device includes an RFID scanning gate, which is arranged inside the passage.

[0029] When a pig passes through the passage, if the pig has an ear tag, the RFID scanning gate scans the ear tag to generate a timestamp and an identification ID; if the pig does not have an ear tag, the RFID scanning gate scans empty data, generating abnormal data.

[0030] The control system is further configured to: receive the abnormal data and perform visual feature recognition based on the visible light visual image data acquired by the inspection robot at the same timestamp; if no ear tag feature is identified, the timestamp and the static data are bound together to construct a virtual identification ID for the current pig, wherein the visible light visual image data is the image data acquired synchronously when the 3D depth camera acquires the static data.

[0031] In addition, the identification device also includes an ear tag, which is located on the pig's ear.

[0032] It should be noted that the identification device can be fixedly installed at the entrance of the passage from the slaughtering pen to the stunning workshop. It is a checkpoint-style identification structure that all pigs must pass through. Its core function is to synchronously collect and bind the identity information and time information of each passing pig, forming a unique timestamp and identification ID. This serves as the core identification identifier for subsequent individualized stunning parameter matching of the pig. It works in conjunction with the inspection robot and control system to achieve full-scene identification of pigs with and without ear tags. The specific structure and working principle are as follows: Specifically, the identification device includes an RFID scanning gate, which is a channel-type identification structure that is adapted to the width of the slaughtering pen channel and is fixedly arranged in the channel at the necessary passage for pigs. The RFID scanning gate has a built-in radio frequency signal transmitting component, a radio frequency signal receiving component, and a time synchronization component (the structure of the RFID scanning gate is a conventional structure and has not been improved in this application). The radio frequency signal transmitting component can continuously transmit radio frequency detection signals to the entire cross-section of the channel, achieving blind-spot-free coverage of the channel identification area. The time synchronization component keeps the clock of the control system in precise synchronization, ensuring that the generated timestamp is consistent with the time base of the entire process data.

[0033] The identification device also includes an ear tag, which is an identification mark that can be fixed to the outside of the pig's ear. Each ear tag is pre-stored with a unique pig identification ID, which is a unique digital code that can be adapted to the radio frequency signal identification of RFID scanning gates.

[0034] When a pig passes normally through the identification area of ​​the RFID scanning gate along the passage, the RFID scanning gate receives the radio frequency signal fed back by the ear tag through the radio frequency signal receiving component, completing the synchronous identification of identity and time: if the pig is wearing an ear tag, the RFID scanning gate can quickly read the unique identification ID pre-stored in the ear tag, and at the same time collect the current identification time through the time synchronization component, generate a timestamp, bind the identification ID and the timestamp, and transmit them synchronously to the control system; if the pig is not wearing an ear tag, the radio frequency signal receiving component of the RFID scanning gate cannot receive a valid ear tag radio frequency feedback signal. At this time, the RFID scanning gate will determine that the scanning data is empty, and then generate abnormal data containing the current identification timestamp, and transmit the abnormal data to the control system in real time, triggering the visual verification process for pigs without ear tags.

[0035] The control system is also configured to: receive abnormal data transmitted by the RFID scanning gate in real time, and perform visual feature recognition of the pig's ears based on the visible light visual image data acquired by the inspection robot at the same timestamp; wherein, the visible light visual image data is image data generated simultaneously by the 3D depth camera while collecting static data of the pig. When the 3D depth camera collects static data related to the pig's weight and breed, it will simultaneously capture clear visual images of the pig's ears and head. This image data is bound to the same timestamp as the static data and transmitted to the control system in real time to ensure accurate matching of the time dimension.

[0036] Based on the aforementioned visible light visual image data, the control system identifies the visual morphological features of the pig's ear area. By distinguishing whether the physical outline features of the ear tag are present in the ear area, a secondary verification of the presence or absence of the ear tag is completed. If no ear tag features are identified, it is determined that the pig is indeed not wearing an ear tag. At this time, the control system uniquely binds the timestamp and the static data collected by the inspection robot at the same timestamp. Through the exclusive coding of the timestamp and static data, a globally unique virtual identification ID is constructed for the current pig. This virtual identification ID has the same identity identification function as the identification ID corresponding to the ear tag. It can serve as the core basis for subsequent matching of the pig's stunning parameters and full-process data traceability, ensuring that even pigs without ear tags can achieve accurate identification and binding of individual identities.

[0037] It should be noted that the existing image recognition methods for identifying the visual morphological features of the ear region of pigs are based on distinguishing the physical outline features of the ear tag. This application does not improve the image recognition method.

[0038] The stunning device has a gas release valve for receiving command packets and adjusting the opening of the gas release valve according to the command packets to stunning the pigs; Furthermore, the stun-inducing device also includes a fan, which is connected to the control system and is used to receive instruction packets sent to it by the control system, and to adjust the working time and power in response to the instruction packets.

[0039] Furthermore, the stunning device also includes a conveyor belt connected to the control system, which receives instruction packets sent to it by the control system and adjusts the transmission speed in response to the instruction packets.

[0040] Specifically, the stunning equipment is deployed in the stunning operation area at the end of the slaughter pen passage. To accommodate the continuous flow of pigs in a streamlined, sealed stunning chamber structure, it seamlessly connects with the slaughter pen passage, ensuring that pigs can enter the chamber in an orderly manner to complete the stunning process. The core of this stunning equipment is equipped with a gas release valve. Additionally, fans and conveyor belts can be added as needed. All three components are connected to the control system via industrial communication, allowing them to independently receive and analyze the same command packet from the control system. Each component adjusts its actions according to the corresponding control parameters within the command packet. Through the coordinated work of all components, personalized carbon dioxide stunning of pigs is achieved. The specific structure, connection relationships, and working principles of each component are as follows: The stunning device features a gas release valve, which is an electrically proportional regulating valve. One end of the valve is sealed and connected to the carbon dioxide storage device in the slaughterhouse, while the other end extends into the sealed cavity of the stunning device, enabling the directional release of carbon dioxide gas. The gas release valve has a built-in signal receiving module and an opening adjustment execution module (the gas release valve is an existing valve). It can receive command packets from the control system in real time and analyze the control parameters related to the gas release valve within the command packets through the execution module. Its core function is to adjust its opening degree according to the command packets, controlling the release flow rate of carbon dioxide gas into the cavity through stepless adjustment of the opening degree. This precisely controls the carbon dioxide concentration within the stunning cavity to meet the stunning concentration requirements of different pigs, ultimately achieving carbon dioxide stunning of pigs entering the cavity. Furthermore, the valve's opening adjustment precision perfectly matches the parameters in the command packets, ensuring accurate concentration control.

[0041] Furthermore, the stunning device also includes a fan, which is fixedly installed at the air inlet and exhaust outlet of the sealed cavity of the stunning device and communicates with the inside of the cavity. The power supply and control end of the fan are both connected to the control system, forming a two-way communication and control link, which can receive command packets sent to it by the control system in real time. The fan has a built-in frequency converter control and timing module, which can accurately respond to the control parameters related to the fan in the command packet. By adjusting the working power, the ventilation volume of the fan can be changed. The higher the power, the greater the ventilation volume. At the same time, the continuous ventilation duration of the fan can be controlled by adjusting the working time. Its core function is to start in time after the pigs have completed the stunning operation and work at the set power and duration to quickly expel the residual carbon dioxide gas in the stunning cavity and introduce fresh air into the cavity to achieve gas replacement inside the cavity. This avoids the current carbon dioxide concentration being incompatible with the carbon dioxide concentration required by the pigs entering later, or causing premature stress, and ensures the stability of the working environment of the stunning cavity.

[0042] Furthermore, the stunning device also includes a conveyor belt, which is laid entirely inside the sealed cavity of the stunning device and at the inlet and outlet ends of the cavity. It seamlessly connects to the end of the waiting pen channel, ensuring that pigs can smoothly enter the stunning cavity from the channel and move with the conveyor belt. The drive end of the conveyor belt is equipped with a variable frequency speed control module, which establishes a communication connection with the control system. This module can receive command packets sent by the control system in real time and accurately respond to the control parameters related to the conveyor belt within the command packets. The conveyor belt adjusts its transmission speed through the variable frequency speed control module, achieving stepless adjustment. Its core function is to change the movement speed of the pigs within the sealed cavity of the stunning device by adjusting the transmission speed, thereby controlling the residence time of the pigs within the cavity, i.e., the exposure time of the pigs to carbon dioxide gas. The slower the transmission speed, the longer the exposure time of the pigs within the cavity; the faster the transmission speed, the shorter the exposure time. In conjunction with the concentration control of the gas release valve, dual personalized control of carbon dioxide concentration and exposure time is achieved, ensuring precise adaptation of the stunning effect for each pig.

[0043] In summary, the gas release valve, fan, and conveyor belt of the stunning equipment share the same instruction package issued by the control system. Each component independently analyzes its own control parameters within the package and executes them synchronously. The gas release valve is responsible for controlling the stunning concentration, the conveyor belt is responsible for controlling the exposure time, and the fan is responsible for completing the gas replacement of the chamber after stunning. The three work together to achieve fully automated operation from precise execution of stunning parameters to restoration of the working environment, which is suitable for the rhythm of assembly line pig slaughtering operations.

[0044] The control system is connected to the identification device and the inspection robot respectively, and is used to receive the timestamp, the identification ID and the static data, and send the instruction packet to the gas release valve.

[0045] The control system is configured to: extract features from the static data to form static features, input the static features into a first AI algorithm model to form an instruction, bind the instruction, the identification ID and the timestamp to form an instruction packet, and send the instruction packet to the gas release valve. The first AI algorithm model represents an AI algorithm model trained on historical data.

[0046] Specifically, the control system can be deployed in a collaborative manner using an edge computing gateway and a cloud management platform. The edge computing gateway is deployed near the slaughterhouse's on-site operating area, while the cloud management platform is deployed at the slaughterhouse's remote control center. The two systems achieve real-time data synchronization via a network, balancing the real-time nature of on-site data processing with the professionalism of cloud data storage and model training. The control system establishes a stable two-way communication connection with the identification device and inspection robot via a wireless communication link. Simultaneously, it achieves precise command transmission with the gas release valve of the stunning equipment via an industrial control bus. Its core function is to uniformly receive various data uploaded by the identification device and inspection robot, process them according to standardization, generate personalized stunning command packages, and send them to the gas release valve to achieve individualized control of stunning parameters. The specific connection relationships and operational configurations are as follows: The control system establishes dedicated communication links with the identification device and the inspection robot, and can receive the unique identification ID of the pig (including the identification ID corresponding to the ear tag and the virtual identification ID of the pig without the ear tag) and timestamp transmitted by the identification device in real time. At the same time, it receives the static data of the pig transmitted by the inspection robot. All of the above data will be stored in real time by the control system and classified and archived according to the identification ID, ensuring that the identification information of each pig corresponds one-to-one with the collected data, and providing a complete and accurate data source for subsequent instruction generation.

[0047] The control system is configured to complete the entire process of static data processing, stunning command generation, command package construction and distribution. The working principles and details of each step are as follows: After receiving the static data of the pigs uploaded by the inspection robot, the control system first identifies this static data as unilateral body shape data collected by the robot's 3D depth camera along one side of the passage, reflecting only the body shape information of the pig facing the camera. Then, based on the physiological characteristic of bilateral symmetry of the pig's body, it generates matching mirror features using the unilateral static data as a benchmark. These mirror features complete the body shape information of the other side of the pig, forming the basic data for the pig's complete body shape. This lays the data dimensional foundation for accurate matching with supplementary static features collected from multiple perspectives. After completing the mirror feature completion, the integrated complete body shape basic data undergoes standardized preprocessing, retaining core quantitative information that truly reflects the inherent attributes of each individual pig. Subsequently, based on the parameter matching requirements for carbon dioxide stunning, core features related to the stunning effect are extracted from the preprocessed complete body shape data to form static features. These static features are quantitative information of a unified dimension, accurately characterizing key attributes such as the pig's weight and breed, and are the core data basis for subsequent matching of personalized stunning parameters.

[0048] The control system extracts the static features of the pigs and inputs them into a pre-trained first AI algorithm model. The first AI algorithm model then matches and generates a personalized stunning command suitable for the pig. This first AI algorithm model is trained based on historical data of carbon dioxide stunning in large-scale pig slaughter. This historical data covers the carbon dioxide gas release valve control parameters corresponding to effective and moderate stunning of pigs of different weights and breeds. The first AI algorithm model has summarized the matching rules between individual pig characteristics and stunning valve control parameters from the historical data. Based on the input static features, it can directly match a stunning command suitable for the current pig. The command content consists of specific control parameters that the gas release valve can directly execute, including the target valve opening, opening adjustment rate, and target opening duration. The same applies to the fan and conveyor belt, which will not be elaborated further.

[0049] The control system binds the personalized stunning commands generated by the first AI algorithm model to the unique identification ID and data collection timestamp of each pig, constructing a standardized command package. This command package uses a unified data format that the stunning equipment can directly parse. It explicitly includes core information such as the pig's identification, data collection time, and specific control parameters of the gas release valve, ensuring the stunning equipment can accurately identify the individual pig and the execution requirements corresponding to the command. After the command package is constructed, the control system, based on the pig's movement rhythm within the channel, sends the command package in real-time to the gas release valve of the stunning equipment via the industrial control bus. This ensures that the gas release valve can promptly receive and execute the command when a pig enters the stunning operation area, achieving matching between the individual pig and the stunning parameters.

[0050] If the system is equipped with a visual data acquisition device, after the control system completes the above-mentioned mirror feature completion and static feature extraction, it will perform data dimension calibration on the static feature after mirror feature completion and the multi-view supplemented static feature uploaded by the visual data acquisition device, so as to achieve the matching of single-sided acquisition + mirror feature completion data and multi-view acquisition data, and then perform feature fusion to form the target static feature.

[0051] In one embodiment, an individual adjustment system for pre-slaughter carbon dioxide coma dosage in pigs further includes: A visual data acquisition device is installed in the passageway or at the scale of the channel to collect data on the weight and breed of the pigs passing by, forming supplementary static data.

[0052] The supplementary static data refers to the supplementary features of the static data.

[0053] The control system is connected to the visual data acquisition device and is also used to receive the supplementary static data.

[0054] The control system is further configured to: extract features from the supplementary static data to form supplementary static features, fuse the supplementary static features with the static features to form target static features, and input the target static features into a first AI algorithm model to form instructions.

[0055] Furthermore, the visual data acquisition device is a 3D vision module composed of at least two cameras. The two cameras are separately set at different angles in the channel to identify different angles of the pig and form first visual static data and second visual static data accordingly. The first visual static data and the second visual static data constitute the supplementary static data.

[0056] It should be noted that the visual data acquisition device is a fixed data acquisition structure, set up in the straight section of the passageway in the slaughtering pen or at the weighing scale. These locations are all areas that pigs must pass through on the pig assembly line, and the pigs move at a slow speed in these areas without crowding or obstruction, which can ensure the effectiveness of data acquisition. Its core function is to collect secondary data related to weight and breed of each passing pig, forming supplementary static data. This supplementary static data serves as a supplementary feature to the static data collected by the inspection robot, enriching the dimensions of individual pig characteristic data, correcting the deviation of a single collection, and improving the accuracy of subsequent stunning parameter matching.

[0057] The visual data acquisition device establishes a communication connection with the control system. The control system can receive supplementary static data transmitted by the visual data acquisition device in real time. This supplementary static data will be bound to the identification ID and timestamp of the corresponding pig to ensure that it belongs to the same pig as the static data collected by the inspection robot, thus providing a data foundation for subsequent feature fusion.

[0058] The control system is further configured to process supplementary static data and fuse it with existing static features. The specific workflow is as follows: After receiving the supplementary static data, the control system performs standardized preprocessing. Then, it extracts core features highly correlated with carbon dioxide stunning parameters from the supplementary static data to form supplementary static features. Subsequently, based on the identification ID and timestamp of the same pig, the control system fuses the supplementary static features with the static features formed by the inspection robot's data collection to form a more comprehensive target static feature. This target static feature integrates effective data from both collections, compensating for potential perspective bias and data errors from a single collection, and more realistically and completely representing the actual individual attributes of the pig. Finally, the control system inputs this target static feature into the first AI algorithm model, which generates a suitable stunning command based on more accurate individual feature matching.

[0059] Furthermore, the visual data acquisition device is a 3D vision module composed of at least two cameras. Both cameras are professional 3D vision acquisition cameras, possessing the same body shape data acquisition principle as the 3D depth camera of the inspection robot, ensuring uniformity in the dimensions and quantification standards of the acquired data, facilitating subsequent feature fusion processing. The two cameras are separately installed at different angles in the passage, specifically symmetrically arranged on the fence supports on both sides of the passage, or staggered at different heights on the same side of the passage, forming a three-dimensional acquisition perspective of the pigs, minimizing incomplete acquisition problems caused by the pigs turning to the side or partially obstructing the view.

[0060] Two 3D vision cameras work synchronously. When a pig passes through the collection area, each camera performs a full-body scan and depth imaging of the pig from its own angle. Based on the principle of 3D vision imaging, the cameras acquire core information such as the pig's body outline and dimensions from different angles. One camera collects and generates first visual static data, while the other camera collects and generates second visual static data. Both the first and second visual static data are quantitative data containing information about the pig's weight and breed. The two data complement and verify each other, forming the supplementary static data, thus ensuring the comprehensiveness and accuracy of the secondary data collection from the source.

[0061] In one embodiment, an individual adjustment system for pre-slaughter carbon dioxide coma dosage in pigs further includes: A body temperature monitoring device is installed on the inspection robot to monitor the temperature of the pigs passing by and generate a thermal imaging temperature map. The control system is connected to the body temperature monitoring device and is also used to receive the thermal imaging temperature map; wherein, The control system is also configured to: Based on the static features and their mirror features, a local three-dimensional model of the pig's ear base, eyes, and sides is formed. The thermal imaging temperature map is aligned with the local three-dimensional model, and the temperature features of the thermal imaging temperature map corresponding to the local three-dimensional model are extracted. The temperature features and the static features are fused and input into the second AI algorithm model to form the instruction. The second AI algorithm model represents an AI algorithm model formed by adding temperature parameter variables on the basis of the first AI algorithm model.

[0062] Specifically, the body temperature monitoring device is fixedly mounted on the collection end of the inspection robot, maintaining the same collection angle and working rhythm as the inspection robot's 3D depth camera, and the collection range is matched with each other. It can move along the channel with the inspection robot to complete the temperature monitoring of each pig. The body temperature monitoring device is used to perform non-contact full-range scanning of the body surface temperature of passing pigs, capturing the infrared radiation signal of the pig's body surface and converting it into a thermal imaging temperature map. This thermal imaging temperature map is a pixelated temperature distribution image, in which each pixel corresponds to a unique temperature value, which can completely reflect the temperature distribution state of the pig's body surface and provide core basis for judging the physiological stress state of the pig.

[0063] The body temperature monitoring device and the control system establish a wired or wireless communication connection, and the two maintain the real-time and synchronous transmission of data. The control system can receive the thermal imaging temperature map uploaded by the body temperature monitoring device in real time, and the thermal imaging temperature map will be bound with the identification ID, timestamp, static features and mirror features of the corresponding pig to ensure that all data belong to the same pig, providing a complete and matching data source for subsequent local 3D model construction and temperature feature extraction.

[0064] The control system is configured to complete the entire process of local 3D model construction, thermal imaging temperature map alignment, temperature feature extraction, feature fusion, and dimming command generation. Each step relies on equipment calibration and spatial mapping principles to achieve precise connection. The specific working principle and details are as follows: Based on static and mirror features, a local 3D model is constructed. The control system uses the extracted static features of the pig as a foundation, combined with mirror features generated to complete the unilateral body shape information. According to the body shape contour rules corresponding to the pig's weight and breed, the system extracts the body shape contour features and spatial position information of three key parts: the base of the ear, the eyes, and the sides of the body, and constructs local 3D models of these three parts. This local 3D model is not a full-body model of the pig, but focuses only on the key areas for body temperature monitoring. It includes the spatial contour shape, relative position relationship, and spatial coordinate range of the base of the ear, the eyes, and the sides of the body. The coordinate system of the local 3D model is consistent with the acquisition coordinate system of the inspection robot's 3D depth camera and the body temperature monitoring device, laying a spatial foundation for the accurate alignment of the subsequent thermal imaging temperature map.

[0065] The thermal imaging temperature map and the local 3D model are coarsely aligned. The control system, based on the factory calibration and simple on-site calibration results of the inspection robot, performs a coarse spatial coordinate mapping alignment between the thermal imaging temperature map and the local 3D model. The approximate spatial coordinate range of the ear base, eye, and body side in the local 3D model is mapped to the pixel coordinate system of the thermal imaging temperature map, so as to achieve a rough match between the spatial positions of the two data. This ensures that the three key temperature measurement areas defined by the local 3D model can correspond to the approximate areas of the pig's ear base, eye, and body side on the thermal imaging temperature map. This spatially locks the target range of temperature data to be extracted from the thermal imaging temperature map, thus meeting the basic requirements for temperature extraction.

[0066] After extracting temperature features from the corresponding areas of the local 3D model and performing coarse alignment, the control system roughly selects the temperature acquisition areas corresponding to the pig's ear base, eyes, and sides in the thermal imaging temperature map based on the pixel coordinate range mapped from the local 3D model. The temperature values ​​of all pixels within each area are standardized to remove abnormal temperature points caused by environmental infrared interference or dirt on the pig's body surface. Then, the average temperature value of each area is calculated. Based on the physiological characteristics of the pig and the differences in the reference value of on-site temperature measurement, differentiated fixed weights are assigned to the average temperatures of the three areas, with the ear base having the highest weight, followed by the eyes, and the sides of the body having the lowest. The average temperature values ​​of each area are multiplied by their corresponding weight coefficients and then summed to form a temperature feature that accurately represents the pig's real-time physiological body temperature state. This temperature feature is a unified-dimensional quantitative numerical information and is the core basis for reflecting whether the pig is in a state of stress.

[0067] It should be noted that the ear root has the highest reference value. Its skin is extremely thin and rich in subcutaneous arteries and veins, making it most synchronous with the pig's core body temperature. The systemic temperature rise caused by pre-slaughter stress will be reflected first and most stably in the ear root. Furthermore, this area is less likely to be obstructed by light or equipment in the passageway, resulting in the most accurate temperature data with minimal environmental interference. The eye area has the next highest reference value. The skin around the eye is thin and contains conjunctival blood vessels, reflecting core body temperature. However, pigs may close their eyes in the passageway, or dust may adhere to the eyes, occasionally leading to slight data collection deviations, making its accuracy slightly lower than that of the ear root. The side of the body has the lowest reference value. Its skin is relatively thick, and the side of the body has a large surface area, making it easily affected by pig friction, hot air in the passageway, and direct sunlight. While it can reflect body temperature changes, it is most susceptible to environmental interference and has the lowest accuracy, serving only as a supplementary reference.

[0068] In the feature fusion and stunning command generation process, the control system integrates and fuses the extracted temperature features with the pig's static features to form a comprehensive feature set that includes the pig's inherent body shape attributes (weight, breed) and real-time physiological body temperature. This comprehensive feature set is then input into a second AI algorithm model, which generates a personalized stunning command tailored to the pig. The second AI algorithm model is an optimized model based on the first AI algorithm model, with the addition of temperature parameter variables. Building upon the existing matching rules between pig body shape and stunning parameters, the second AI algorithm model incorporates the adaptation logic between the pig's body temperature and stunning parameters. It can adjust the stunning control parameters according to the pig's body temperature characteristics, achieving more individualized stunning that better matches the pig's actual physiological state, ensuring the stunning effect while minimizing stress response in the pig.

[0069] It should be noted that the second AI algorithm model can also be a brand-new algorithm model that incorporates body temperature parameters and is constructed based on the static and temperature characteristics of pigs. The core of the new algorithm model is that the matching logic between the pig's body size and the stunning parameters is incorporated into the matching rules between the pig's body size and the stunning parameters. The stunning control parameters can be adjusted according to the pig's body temperature characteristics.

[0070] In one embodiment, an individual adjustment system for pre-slaughter carbon dioxide coma dosage in pigs further includes: A microphone array, mounted on the inspection robot, is used to collect the sounds of the passing pigs and generate sound information.

[0071] The sound information includes at least the frequency and pitch of the call.

[0072] The control system is connected to the microphone array and is also used to receive the sound information.

[0073] The control system is further configured to: extract features from the sound information to form sound features; locate the target sound features of the pig based on the time difference and phase difference of the sound features; and input the target sound features, the static features, and the temperature features into a third AI algorithm model to form an instruction. The third AI algorithm model is an AI algorithm model formed by adding sound parameter variables to the second AI algorithm model.

[0074] Specifically, the microphone array can be embedded and fixedly installed in the front-end acquisition area of ​​the inspection robot, maintaining the same acquisition range and synchronous working sequence as the 3D depth camera and body temperature monitoring device on the device, and moving along the channel with the inspection robot. The microphone array is composed of multiple independent sound receiving units, which can synchronously collect sound field signals in the channel, distinguish sound sources based on the sound propagation law, and extract the unique calls of each passing pig. The collected sound information forms standardized sound information, which includes at least two basic sound field contents: the frequency of the pig's calls and the pitch of the calls. It can intuitively reflect the current emotional agitation and stress intensity of the pig, serving as a supplementary basis for judging the real-time physiological state of the pig.

[0075] The microphone array establishes a communication connection with the control system, enabling lossless transmission of sound information. After receiving the sound information uploaded by the microphone array, the control system binds and archives the sound information with the corresponding pig's identification ID, timestamp, static features, mirror features, and temperature features, ensuring that the sound field data, body size data, and body temperature data all belong to the same pig, thus avoiding sound signal confusion when multiple pigs pass through.

[0076] The control system is also configured to perform the following processes: sound preprocessing, sound feature extraction, sound source localization and screening, multi-feature fusion, and final generation of a stunning command. It distinguishes sound sources based on the time and phase differences of sound propagation. The specific working principle and details are as follows: The basic preprocessing and feature extraction of the raw sound information: The control system first performs conventional noise reduction on the received raw sound information, such as filtering out the background noise of fixed-frequency equipment such as fan operation and conveyor belt drive, while retaining all biological vocal signals in the channel; then, it extracts two core contents, namely the frequency and pitch of the calls, from the noise-reduced sound field signal to initially form general sound features, without removing the vocal signals of other pigs in the surrounding area.

[0077] By accurately locking onto the target sound features using time and phase differences, when multiple sets of microphones simultaneously record sound, the same call will generate a natural reception time difference and sound wave phase difference when transmitted to different microphones. The control system uses the fixed installation spacing of the microphone array as a unified calculation benchmark. It first uses the reception time difference to reverse-calculate the approximate spatial range of the sound source to complete the coarse localization of the sound source. Then, it uses the sound wave phase difference to perform fine point calibration on the approximate spatial range to tighten the judgment range of the sound source. The real-time position of the pig in the channel is obtained by superimposing the walking positioning of the inspection robot at the same time stamp with the relative coordinates of the pig measured by the 3D depth camera. The control system relies on the inherent laws of the sound field, combined with the real-time position of the pig in the channel, to calculate and match the sound field parameters of the sound source specific to the pig. It locates the vocal signal of the currently identified pig from the mixed sound field, separates irrelevant sound sources such as distant pig calls and scattered environmental noises, and finally purifies and filters out the target sound features that correspond only to this pig.

[0078] The control system integrates the purified target sound features with the static features that were previously completed and the added weighted temperature features into a unified dimension, thus completing the fusion and summarization of the three core features and forming a comprehensive feature that takes into account the pig's inherent body shape attributes, real-time body temperature stress state, and on-site vocal emotional state.

[0079] The control system inputs the integrated features into the third AI algorithm model, which then generates a personalized final stunning command tailored to the pig. The third AI algorithm model can be optimized by adding sound parameter variables to the second AI algorithm model, or it can be directly constructed as a brand-new algorithm model that incorporates sound parameters and adapts to the collaborative judgment of three types of features. The new algorithm model, in addition to the original matching logic of body size and body temperature, further incorporates the stress emotions reflected by the pig's vocalizations. It can fine-tune the stunning concentration and passage rhythm according to the pig's vocal agitation level, minimizing the pig's stress response and taking into account both humane slaughter standards and a stable stunning effect.

[0080] It should be noted that, in this application, fusion refers to the interrelation and superposition of multiple sets of aggregated features to form a comprehensive feature that can be input into a single model.

[0081] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0082] Obviously, those skilled in the art should understand that the various units or steps of this application described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Optionally, they can be implemented using computer-executable program code, thereby storing them in a storage device for execution by a computing device, or fabricating them separately as individual integrated circuit modules, or fabricating multiple modules or steps into a single integrated circuit module. Thus, this application is not limited to any particular combination of hardware and software.

[0083] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A pre-slaughter carbon dioxide coma-inducing dosage algorithm individual adjustment system for pigs, characterized in that, include: An identification device is installed in the passageway of the slaughter pen to identify the passing pigs by time and ID, generating a timestamp and identification ID; An inspection robot is installed in the passageway and can move along the length of the passageway to collect data on the pigs and form static data. The stunning device has a gas release valve for receiving command packets and adjusting the opening of the gas release valve according to the command packets to stunning the pigs; The control system, connected to both the identification device and the inspection robot, receives the timestamp, the identification ID, and the static data, and sends the command packet to the gas release valve. The control system is configured to: The static data is subjected to feature extraction to form static features. The static features are input into the first AI algorithm model to form an instruction. The instruction, the identification ID and the timestamp are bound together to form an instruction packet. The instruction packet is sent to the gas release valve. The first AI algorithm model represents an AI algorithm model trained on historical data.

2. The individual adjustment system for pre-slaughter carbon dioxide coma dosage in pigs as described in claim 1, characterized in that, The inspection robot is equipped with a 3D depth camera, which is used to collect data on the weight and breed of the pigs to form static data.

3. The individual adjustment system for pre-slaughter carbon dioxide coma dosage in pigs as described in claim 2, characterized in that, The identification device includes an RFID scanning gate, which is arranged within the passageway; wherein... When the pig passes through the passage, if the pig has an ear tag, the RFID scanning gate scans the ear tag of the pig to form the timestamp and the identification ID; If the pig does not have an ear tag, the RFID scanning gate will scan empty data, resulting in abnormal data. The control system is also configured to: The system receives the abnormal data and performs visual feature recognition based on the visible light visual image data acquired by the inspection robot at the same timestamp. If no ear tag feature is identified, the timestamp is bound to the static data to construct a virtual identification ID for the current pig. The visible light visual image data is the image data acquired synchronously when the 3D depth camera acquires the static data.

4. The individual adjustment system for pre-slaughter carbon dioxide coma dosage in pigs as described in claim 3, characterized in that, The identification device also includes an ear tag, which is located on the pig's ear.

5. The individual adjustment system for pre-slaughter carbon dioxide coma dosage in pigs as described in claim 1, characterized in that, Also includes: A visual data acquisition device is installed in the passageway or at the scale of the channel to collect data on the weight and breed of the pigs passing by, forming supplementary static data, wherein the supplementary static data are supplementary features of the static data. The control system is connected to the visual data acquisition device and is also used to receive the supplementary static data; wherein, The control system is also configured to: Feature extraction is performed on the supplementary static data to form supplementary static features. The supplementary static features are then fused with the static features to form target static features. The target static features are then input into the first AI algorithm model to form instructions.

6. The individual adjustment system for pre-slaughter carbon dioxide coma dosage in pigs as described in claim 5, characterized in that, The visual data acquisition device is a 3D vision module consisting of at least two cameras. The two cameras are set separately at different angles in the channel to identify the pigs from different angles and generate first visual static data and second visual static data accordingly. The first visual static data and the second visual static data constitute the supplementary static data.

7. The individual adjustment system for pre-slaughter carbon dioxide coma dosage in pigs as described in claim 1, characterized in that, Also includes: A body temperature monitoring device is installed on the inspection robot to monitor the temperature of the pigs passing by and generate a thermal imaging temperature map. The control system is connected to the body temperature monitoring device and is also used to receive the thermal imaging temperature map; wherein, The control system is also configured to: Based on the static features and their mirror features, a local three-dimensional model of the pig's ear base, eyes, and sides is formed. The thermal imaging temperature map is aligned with the local three-dimensional model, and the temperature features of the thermal imaging temperature map corresponding to the local three-dimensional model are extracted. The temperature features and the static features are fused and input into the second AI algorithm model to form the instruction. The second AI algorithm model represents an AI algorithm model formed by adding temperature parameter variables on the basis of the first AI algorithm model.

8. The individual adjustment system for pre-slaughter carbon dioxide coma dosage in pigs as described in claim 7, characterized in that, Also includes: A microphone array, mounted on the inspection robot, is used to collect the sounds of the passing pigs to form sound information, wherein the sound information includes at least the frequency and pitch of the calls; The control system is connected to the microphone array and is also used to receive the sound information; wherein, The control system is also configured to: Feature extraction is performed on the sound information to form the sound features. Based on the time difference and phase difference of the sound features, the target sound features of the pig are located. The target sound features, the static features and the temperature features are fused and input into a third AI algorithm model to form an instruction. The third AI algorithm model is an AI algorithm model formed by adding sound parameter variables on the basis of the second AI algorithm model.

9. The individual adjustment system for pre-slaughter carbon dioxide coma dosage in pigs according to any one of claims 1-8, characterized in that, The stun-inducing device also includes a fan connected to the control system, which receives instruction packets sent by the control system and adjusts its operating time and power in response to the instruction packets.

10. The individual adjustment system for pre-slaughter carbon dioxide coma dosage algorithm for pigs as described in any one of claims 1-8, characterized in that, The stunning device also includes a conveyor belt connected to the control system, which is used to receive instruction packets sent to it by the control system and adjust the transmission speed in response to the instruction packets.

Images