Automatic conveying system and method for broiler feeding

Abstract

The invention provides an automatic transmission system and method for broiler feeding, and the method comprises the steps: determining the behavior activeness of a broiler group at different positions of a breeding region, and determining a pecking frequency distribution diagram of the broiler group at different positions of the breeding region; according to all the behavior activeness and a preset group density threshold value, determining a plurality of dynamic feed transmission subareas of the breeding area in combination with the multi-dimensional environmental factors of the breeding area; determining growth state tensors of the broiler groups in different dynamic feed transmission zones by combining the frequency distribution diagram and synchronous indexes of broiler group development in each dynamic feed transmission zone; and according to all the growth state tensors, controlling an electromagnetic distribution valve group to convey the feed required by each dynamic feed conveying partition to a movable feed conveying vehicle, and controlling the movable feed conveying vehicle to carry out dynamic feed conveying. According to the scheme, the breeding area can be dynamically partitioned and feed can be transported based on the group space layout and the group activity track of the broiler chickens in the breeding area.

Description

Technical Field

[0001] The present application relates to the field of visual intelligence technology, and more specifically, to an automatic transmission system and method for broiler feeding. Background Art

[0002] With the continued growth in global demand for animal-based food, precision farming has become a key direction for improving production efficiency and animal welfare. Precision animal husbandry relies on continuous real-time monitoring and analysis of individual growth and behavior to support refined management and resource optimization. Its core technologies include multi-channel sensor data fusion, computer vision, and three-dimensional point cloud reconstruction.

[0003] In existing technologies, most farms are equipped with Internet of Things (IoT) environmental monitoring systems to collect environmental parameters such as temperature, humidity, and ammonia concentration, and to implement automated feeding in combination with fixed-time feeding scheduling. However, these systems can usually only distribute feed based on static zoning and preset feeding plans, making it difficult to cope with the dynamic changes in the activity range and growth status of broiler groups, resulting in feed waste and nutritional imbalance. In addition, although existing automatic feeding devices (such as chain feeding systems and screw conveyors) can achieve centralized feeding, they lack the ability to respond to group behavior and environmental changes in real time, making it difficult to optimize individualized or regionalized feed distribution strategies. Therefore, how to dynamically partition the breeding area and transport feed based on the spatial layout of broiler groups and the group activity trajectories in the breeding area has become a difficult problem facing the industry. Summary of the Invention

[0004] The present application provides an automatic transmission system and method for broiler feeding, which can dynamically partition the breeding area and transport feed based on the spatial layout and group activity trajectory of broiler chickens in the breeding area.

[0005] In a first aspect, the present application provides a method for dynamic transportation of broiler feed based on visual intelligence, which is applied to an automatic transmission system for broiler feeding, wherein the system includes an electromagnetic dispensing valve group and a mobile feed transmission vehicle. The method includes: Collect dynamic distribution images of broiler chickens in the breeding area and then construct a 3D point cloud model of the breeding area; Performing instance segmentation on the three-dimensional point cloud model based on skeleton key point regression to extract the behavioral activity of the broiler flock at different locations in the breeding area, and determining the frequency distribution map of the broiler flock pecking at different locations in the breeding area; Based on the behavioral activity of broiler flocks at different locations in the breeding area and the preset population density threshold combined with the multi-dimensional environmental factors of the breeding area, the feed transmission nodes of the broiler flocks in the breeding area are dynamically divided into multiple areas to obtain multiple dynamic feed transmission zones; Determining the growth state tensors of the broiler flocks in different dynamic feed delivery zones using a long short-term memory network growth prediction model in combination with the frequency distribution diagram and the synchronization index of broiler flock development in each dynamic feed delivery zone; According to all growth state tensors, the electromagnetic dispensing valve group is controlled to deliver the feed required by each dynamic feed transmission partition to the mobile feed transmission vehicle, and the mobile feed transmission vehicle is controlled to perform dynamic feed transmission.

[0006] In some embodiments, performing instance segmentation on the three-dimensional point cloud model based on skeleton key point regression to extract the behavioral activity of the broiler flock at different locations in the breeding area specifically includes: Performing key point regression on the three-dimensional point cloud model to obtain three-dimensional offset vectors of all chicken skeleton key points; Clustering adjacent point clouds in the three-dimensional point cloud model based on the three-dimensional offset vectors of all chicken skeleton key points, thereby separating all independent chicken instances; Determine the feeding posture of each chicken instance in the breeding area; Determine the pecking density map of broiler chickens at different locations in the breeding area based on the feeding postures of all chicken instances; The behavioral activity of the broiler flock at different locations in the breeding area is extracted from the pecking density map.

[0007] In some embodiments, the feed transmission nodes of the broiler flocks in the breeding area are dynamically divided into multiple zones based on the behavioral activity of the broiler flocks at different locations in the breeding area and a preset population density threshold combined with the multi-dimensional environmental factors of the breeding area. The multiple dynamic feed transmission zones specifically include: According to all behavioral activities and the preset population density threshold, multiple high-density active areas and multiple low-density active areas of the broiler population in the breeding area are identified; Obtain multidimensional environmental factors of the breeding area; determining the environmental suitability of each high-density active area and each low-density active area based on the multidimensional environmental factors; The feed transmission nodes in the breeding area are dynamically divided according to all environmental suitabilities to obtain multiple dynamic feed transmission partitions.

[0008] In some embodiments, determining the growth state tensor of the broiler flocks in different dynamic feed delivery zones using the long short-term memory network growth prediction model in combination with the frequency distribution diagram and the synchronization index of the broiler flock development in each dynamic feed delivery zone specifically includes: Determine the synchronization indicators of broiler flock development in each dynamic feed delivery zone; For each dynamic feed transmission partition, using a long short-term memory network growth prediction model to generate a growth prediction curve of the broiler flock in the dynamic feed transmission partition in different future time periods based on the frequency distribution diagram and the synchronization index of the broiler flock development in the dynamic feed transmission partition; The growth state tensor of the broiler chicken group in the dynamic feed transmission zone is determined based on all the growth prediction curves, and then the growth state tensors of the broiler chicken group in different dynamic feed transmission zones are obtained.

[0009] In some embodiments, controlling the electromagnetic dispensing valve group according to all growth state tensors to deliver the feed required by each dynamic feed transmission zone to the mobile feed transmission vehicle specifically includes: For each dynamic feed delivery partition, determining a feed requirement of the dynamic feed delivery partition according to the growth state tensor; Mapping the feed demand of the dynamic feed transmission partition to the corresponding solenoid valve group number and generating the control instructions of the dynamic feed transmission partition; According to the control instructions, the feed required by the dynamic feed transmission partition is delivered to the mobile feed transmission vehicle, and then the feed required by each dynamic feed transmission partition is delivered to the mobile feed transmission vehicle.

[0010] In some embodiments, determining the feed requirement of the dynamic feed delivery partition according to the growth state tensor specifically includes: determining regional growth characteristics of the dynamic feed delivery partition based on the growth state tensor; The feed requirements of the broiler chicken population in the dynamic feed delivery zone are predicted based on the regional growth characteristics.

[0011] In some embodiments, controlling the mobile feed transport vehicle to perform dynamic feed delivery specifically includes: Extracting the movement trajectory of the broiler chicken group from the three-dimensional point cloud model; Obtain the spatial distribution of all dynamic feed transport zones in the farming area; Planning a feed transportation path covering all feeding points in the breeding area based on the spatial distribution and the movement trajectory of the broiler chicken group; The mobile feed transport vehicle is controlled by the feed transport path to feed the feed at each feeding point.

[0012] In a second aspect, the present application provides an automatic transmission system for broiler feeding, the system comprising an electromagnetic dispensing valve group and a mobile feed transmission vehicle, the system also comprising a broiler feed dynamic transport unit, the broiler feed dynamic transport unit comprising: The acquisition module is used to collect dynamic distribution images of broiler chickens in the breeding area and then construct a three-dimensional point cloud model of the breeding area; a processing module for performing instance segmentation on the three-dimensional point cloud model based on skeleton key point regression, thereby extracting the behavioral activity of the broiler flock at different locations in the breeding area, and determining a frequency distribution diagram of the broiler flock pecking at different locations in the breeding area; The processing module is used to dynamically divide the feed transmission nodes of the broiler flocks in the breeding area into multiple zones based on the behavioral activity of the broiler flocks at different locations in the breeding area and a preset population density threshold in combination with the multi-dimensional environmental factors of the breeding area, thereby obtaining multiple dynamic feed transmission zones; The processing module is used to determine the growth state tensor of the broiler chicken groups in different dynamic feed transmission partitions by using the long short-term memory network growth prediction model in combination with the frequency distribution map and the synchronization index of the broiler chicken group development in each dynamic feed transmission partition; The execution module is used to control the electromagnetic feeding valve group according to all growth state tensors to deliver the feed required by each dynamic feed transmission zone to the mobile feed transmission vehicle, and control the mobile feed transmission vehicle to perform dynamic feed transmission.

[0013] In a third aspect, the present application provides a computer device comprising a memory and a processor, wherein the memory stores a code, and the processor is configured to obtain the code and execute the above-mentioned visual intelligence-based dynamic transportation method for broiler feed.

[0014] In a fourth aspect, the present application provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the above-mentioned dynamic transportation method for broiler feed based on visual intelligence.

[0015] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects: In the automatic transmission system and method for broiler feeding provided in the present application, first, a dynamic distribution image of a broiler flock in a breeding area is collected, and then a three-dimensional point cloud model of the breeding area is constructed; the three-dimensional point cloud model is instance segmented based on skeletal key point regression, and then the behavioral activity of the broiler flock at different locations in the breeding area is extracted, and a frequency distribution diagram of the broiler flock pecking at different locations in the breeding area is determined; according to the behavioral activity of the broiler flock at different locations in the breeding area and a preset group density threshold combined with the multidimensional environmental factors of the breeding area, the feed transmission nodes of the broiler flock in the breeding area are dynamically divided into regions to obtain a plurality of dynamic feed transmission partitions; the growth state tensors of the broiler flock in different dynamic feed transmission partitions are determined by using a long short-term memory network growth prediction model combined with the frequency distribution diagram and the synchronization indicators of the broiler flock development in each dynamic feed transmission partition; according to all the growth state tensors, the electromagnetic distributing valve group is controlled to transport the feed required for each dynamic feed transmission partition to the mobile feed transmission vehicle, and the mobile feed transmission vehicle is controlled to perform dynamic feed transportation.

[0016] Thus, it can be seen that the present application controls the electromagnetic distributing valve group according to all growth state tensors to deliver the feed required by each dynamic feed transmission zone to the mobile feed transmission vehicle, and controls the mobile feed transmission vehicle to perform dynamic feed transmission; first, determining the behavioral activity can obtain the activity level of broilers in eating at different positions in the breeding area, and the determination of behavioral activity helps to evaluate the feeding activity of broilers in different spatial positions through the pitch amplitude and feeding depth of broilers' pecking, thereby achieving accurate feed demand prediction and dynamic area division, and improving breeding efficiency; then, determining the dynamic feed transmission zone can obtain feed distribution areas adapted to different broiler activity levels and environmental conditions, and the determination of the dynamic feed transmission zone enables the feed supply of each dynamic feed transmission zone to be adjusted according to actual needs, thereby avoiding overfeeding or feed waste, and at the same time making the feed transportation path more reasonable, reducing transportation time and energy consumption, and improving the timeliness and accuracy of feed delivery, It helps to achieve balanced growth of broiler chickens and improve overall breeding efficiency. In addition, the zoning strategy can be adjusted in real time as the breeding environment and broiler group behavior change, thereby enhancing the system's adaptability to external changes and ensuring the stability and sustainability of the breeding process. Finally, determining the growth state tensor can obtain a multidimensional data structure that integrates the growth prediction curves of broiler groups at different times and spaces. The determination of the growth state tensor helps to integrate the group pecking behavior time series, individual synchronous development level and environmental monitoring data into the same high-dimensional structure, effectively overcoming the limitation that a single indicator cannot reflect complex breeding dynamics, thereby improving the system's sensitivity and accuracy to changes in feed demand, and realizing accurate growth prediction of broiler groups in each dynamic feed transmission zone, thereby optimizing feed supply strategies and improving breeding efficiency and broiler health levels. In summary, based on the above scheme, the breeding area can be dynamically partitioned and feed can be transported based on the spatial layout of broiler groups and the group activity trajectory of broilers in the breeding area. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is an exemplary flow chart of a method for dynamic transportation of broiler feed based on visual intelligence according to some embodiments of the present application; Figure 2 is an exemplary flow chart of determining dynamic feed transmission zones according to some embodiments of the present application; Figure 3 is an operational flow chart for determining a growth state tensor of a broiler flock within a dynamic feed delivery zone according to some embodiments of the present application; Figure 4 is a schematic structural diagram of a broiler feed dynamic transport unit according to some embodiments of the present application; Figure 5This is a diagram of the internal structure of a computer device for implementing a dynamic transportation method for broiler feed based on visual intelligence as shown in some embodiments of the present application. DETAILED DESCRIPTION

[0018] In order to better understand the technical solution of the present application, the technical solution of the present application will be described in detail below with reference to the accompanying drawings and specific implementation methods.

[0019] refer to Figure 1 , which is an exemplary flow chart of a method for dynamic transportation of broiler feed based on visual intelligence according to some embodiments of the present application. The method 100 for dynamic transportation of broiler feed based on visual intelligence mainly includes the following steps: In step 101, dynamic distribution images of broiler chickens in a breeding area are collected, and then a three-dimensional point cloud model of the breeding area is constructed.

[0020] It should be noted that, in this application, the three-dimensional point cloud model is a three-dimensional data model representing the spatial distribution of the breeding area and the distribution of the movement trajectory of the broiler group in the breeding area. In this solution, the three-dimensional point cloud model can provide a reliable geometric basis for dynamic zoning and path planning, and significantly improve the feeding accuracy and efficiency. In specific implementation, the dynamic distribution image of the broiler group in the breeding area is collected, and then the three-dimensional point cloud model of the breeding area is constructed. This can be achieved in the following way, namely: first, multiple industrial cameras (such as Hikvision Starlight 4K smart ball cameras) arranged at different positions in the breeding area are used to continuously record the breeding area in real time within a preset collection period (such as one day) to obtain time-series video data covering the entire field, and then a time synchronization protocol (such as The time-series video data is synchronized at the sub-microsecond level using a precise time synchronization protocol for network measurement and control systems. Existing video processing techniques (for example, checkerboard camera calibration and optical distortion correction algorithms, and background modeling and foreground segmentation algorithms based on Gaussian mixture models) are used to perform distortion correction, background modeling, and foreground segmentation on the captured video stream to obtain a dynamic distribution image of the broiler flock. Finally, a three-dimensional reconstruction tool (such as a multi-view stereo 3D reconstruction system) is used to solve the internal and external parameters of the camera for this dynamic distribution image through checkerboard calibration, and a dense point cloud is generated based on an incremental process of structured beam adjustment to obtain a three-dimensional point cloud model of the breeding area. The dynamic distribution image is an image that reflects the distribution status of the broiler flock in space and time.

[0021] In step 102, the three-dimensional point cloud model is instance segmented based on skeleton key point regression, and the behavioral activity of the broiler group at different locations in the breeding area is extracted, and the frequency distribution diagram of the broiler group pecking at different locations in the breeding area is determined.

[0022] In some embodiments, instance segmentation based on skeleton key point regression is performed on the three-dimensional point cloud model to extract the behavioral activity of broiler chickens at different locations in the breeding area. The following steps can be used: Performing key point regression on the three-dimensional point cloud model to obtain three-dimensional offset vectors of all chicken skeleton key points; Clustering adjacent point clouds in the three-dimensional point cloud model based on the three-dimensional offset vectors of all chicken skeleton key points, thereby separating all independent chicken instances; Determine the feeding posture of each chicken instance in the breeding area; Determine the pecking density map of broiler chickens at different locations in the breeding area based on the feeding postures of all chicken instances; The behavioral activity of the broiler flock at different locations in the breeding area is extracted from the pecking density map.

[0023] It should be noted that in this application, the three-dimensional offset vector refers to the spatial offset of all points in the three-dimensional point cloud data relative to the key points of the chicken skeleton (such as the head, neck, chest, etc.). The three-dimensional offset vector can be used to guide the points in the point cloud to gather toward their corresponding bone key points, thereby realizing the segmentation of individual instances and the reconstruction of the skeleton structure; in this scheme, the application of the three-dimensional offset vector enables the point cloud data of each broiler to be accurately aggregated to the center of its skeleton in a dense and overlapping group of broilers, significantly improving the accuracy of individual recognition and posture analysis, and providing a reliable data basis for subsequent behavior monitoring and intelligent feeding.

[0024] In specific implementation, key point regression is performed on the three-dimensional point cloud model to obtain the three-dimensional offset vectors of all chicken bone key points. This can be achieved in the following way, namely: a pre-trained deep learning model (such as a point cloud deep feature extraction model) can be used to perform key point regression on the point cloud data in the three-dimensional point cloud model to obtain the three-dimensional offset vectors of all chicken bone key points of broilers; wherein the chicken bone key points are specific anatomical landmark points used to describe the individual skeletal structure of broilers, such as: the head vertex, the neck pivot point, the chest center point, etc.

[0025] It should be noted that in this application, the point cloud deep feature extraction model can effectively capture the local and global information of the point cloud by introducing multi-scale feature processing, and is suitable for three-dimensional point cloud analysis in complex scenes. In the pre-training stage, point cloud data with manually labeled chicken bone key points can be used as a training set to train the point cloud deep feature extraction model to learn to predict the offset vector from each point in the point cloud to the chicken bone key point and output the offset vector from each point to the chicken bone key point to form a three-dimensional offset vector.

[0026] In specific implementation, the adjacent point clouds in the three-dimensional point cloud model are clustered based on the three-dimensional offset vectors of all the key points of the chicken bones, and then all independent chicken instances are separated. This can be achieved in the following way, namely: an existing clustering algorithm (such as a density-based spatial clustering algorithm) can be used to assign a cluster label to each point in the three-dimensional point cloud model according to the three-dimensional offset vectors of all the key points of the chicken bones, indicating the chicken instance to which it belongs, and the points in the original point cloud are grouped according to the cluster labels, so that all independent chicken instances can be separated in the three-dimensional point cloud model; wherein, the chicken instance refers to the three-dimensional structural representation of a single broiler separated from the entire chicken flock. In this solution, identifying and extracting chicken instances can realize independent behavioral analysis of each broiler, especially accurate monitoring of pecking posture, thereby providing data support for individual growth status assessment and precise feed delivery, thereby improving breeding efficiency.

[0027] In specific implementation, the following method can be used to determine the feeding posture of each chicken instance in the breeding area, namely: for each chicken instance, first, the key points of the chicken bones can be connected according to the natural bone topology (such as head→neck→chest) to construct a three-dimensional skeleton model of the chicken body, and then the rate of change of the angle between the vector between the head and the neck and the horizontal plane in the three-dimensional skeleton model within a preset time window (such as 5 seconds) is calculated as the pitch change rate of the chicken instance when eating, and the rate of change of the vertical distance of the head relative to the chest in the three-dimensional skeleton model within a preset time window (such as 5 seconds) is calculated as the chicken instance eating. Finally, the vector composed of the pitch change rate, the depth change rate and the spatial position coordinates of the chicken instance is used as the feeding posture of the chicken instance when pecking in the breeding area. The above steps can be used to obtain the feeding posture of each chicken instance when pecking in the breeding area; wherein, the feeding posture refers to the three-dimensional posture composed of the spatial relative positions and angles of key bone points such as the head, neck and chest of broilers during the feeding process, which can be used to quantify the action pattern of its pecking behavior. In this solution, accurate identification and analysis of feeding posture helps to monitor the feeding behavior and health status of broilers in real time, optimize feeding strategies, and improve breeding efficiency.

[0028] It should be noted that in this application, the pecking posture characteristics are characteristics that quantify the feeding activity and behavioral rhythm of broilers in different breeding areas at different spatial positions. In this scheme, the pecking posture characteristics can be determined to evaluate the feeding activity and behavioral rhythm of broilers in different spatial positions through the pitch amplitude and feeding depth of the broilers' pecking, thereby achieving accurate feed demand prediction and dynamic area division, and improving breeding efficiency.

[0029] In specific implementation, the pecking density map of broiler chickens at different locations in the breeding area can be determined based on the feeding postures of all chicken instances. This can be achieved in the following way: first, the breeding area can be projected onto a plane and divided into spatial grids of equal size, and each spatial grid can be used as a feed transmission node. The feeding posture of the chicken instance is then mapped to the spatial position of the chicken instance in the breeding area. The broiler pecking events in each feed transmission node are identified through the feeding postures of all chicken instances, the broiler pecking frequency is calculated, and a pecking density map is generated (for example, a chicken instance with a pitch change rate and a depth change rate in the feeding posture that are both greater than 0.6 can be identified as a chicken instance that is eating); wherein, the pecking density map is an image showing the foraging activity of the broiler chicken group at various locations in the breeding area. The pecking density map can be used to identify high-density foraging areas, thereby optimizing feed delivery strategies and improving breeding efficiency.

[0030] In specific implementation, the behavioral activity of broiler chickens at different locations in the breeding area can be extracted from the pecking density map in the following manner, namely: the pecking frequency of broilers in each feed transmission node in the pecking density map can be used as the behavioral activity of the broiler chickens corresponding to each feed transmission node in the breeding area, thereby obtaining the behavioral activity of the broiler chickens at different locations in the breeding area; wherein, the behavioral activity refers to the activity level of broilers in eating at different locations in the breeding area. In this solution, the behavioral activity can be used to accurately locate high-demand feeding areas, thereby achieving precise zoning feeding, improving feed utilization efficiency and reducing waste.

[0031] In some embodiments, determining the frequency distribution of pecking at different locations in the breeding area by the broiler flock can be achieved by the following steps: Obtain the eating posture of each chicken instance in the breeding area; The frequency distribution diagram of broiler flocks pecking at different positions in the breeding area is determined based on the feeding postures of all chicken instances.

[0032] It should be noted that, in the present application, the frequency distribution diagram is a distribution image reflecting the changes in the foraging frequency of broiler chicken groups in various locations in the breeding area at different time points. The frequency distribution diagram can provide the temporal characteristics of the synchronization and pecking rhythm changes of the broiler chicken group, and can provide data support for the growth status prediction of the broiler chicken group in the subsequent steps; in specific implementation, the frequency distribution diagram of the broiler chicken group pecking at different locations in the breeding area can be determined based on the feeding posture of all chicken instances. This can be achieved in the following way, namely: the posture parameter changes of the feeding posture of all chicken instances in each feed transmission node in the breeding area in the continuous time frame of the three-dimensional point cloud model can be analyzed by existing time-frequency analysis methods (such as short-time Fourier transform), the frequency of the pecking rhythm can be extracted and a frequency distribution diagram can be generated.

[0033] In step 103, the feed transmission nodes of the broiler flocks in the breeding area are dynamically divided into regions according to the behavioral activity of the broiler flocks at different locations in the breeding area and the preset group density threshold combined with the multidimensional environmental factors of the breeding area to obtain multiple dynamic feed transmission partitions.

[0034] In some embodiments, reference Figure 2 This figure is an exemplary flow chart for determining dynamic feed transmission zones according to some embodiments of the present application. In this application, feed transmission nodes for broiler flocks in a breeding area are dynamically divided into zones based on the behavioral activity of broiler flocks at different locations in the breeding area and a preset population density threshold, combined with multi-dimensional environmental factors of the breeding area. The following steps can be used to obtain multiple dynamic feed transmission zones: In step 1031, multiple high-density active areas and multiple low-density active areas of the broiler chicken population in the breeding area are identified based on all behavioral activities and a preset population density threshold; In step 1032, multi-dimensional environmental factors of the breeding area are obtained; In step 1033, the environmental suitability of each high-density active area and each low-density active area is determined based on the multi-dimensional environmental factors; In step 1034, the feed transmission nodes in the breeding area are dynamically divided according to all environmental suitabilities to obtain multiple dynamic feed transmission partitions.

[0035] In specific implementation, the following method can be used to identify multiple high-density active areas and multiple low-density active areas of the broiler group in the breeding area based on all behavioral activity and the preset group density threshold, namely: for each feed transmission node divided in the breeding area, if the behavioral activity of the broiler group at the feed transmission node is greater than the preset group density threshold, the feed transmission node is marked as a high-density active area; if the behavioral activity of the broiler group at the feed transmission node is less than the preset group density threshold, the feed transmission node is marked as a low-density active area. Through the above steps, each feed transmission node divided in the breeding area can be marked, thereby obtaining multiple high-density active areas and multiple low-density active areas of the broiler group in the breeding area; wherein, the group density threshold is a key parameter for distinguishing between the "hot spot" area and the "cold spot" area where the broiler pecking behavior is active in the breeding area. Reasonable setting of the group density threshold can accurately identify the high-density active areas and low-density active areas of the broiler group in the breeding area, thereby realizing dynamic division of the breeding area, optimizing the feed transportation path, and improving resource utilization efficiency.

[0036] It should be noted that, in this application, high-density active areas refer to feed transmission nodes in breeding areas where the pecking frequency of broilers is significantly higher than that of surrounding areas. Such high-density active areas represent "hotspot" areas where broiler groups are highly concentrated and pecking behavior is active. Such areas can be used to achieve targeted feed delivery and transportation route optimization, thereby significantly improving breeding management efficiency and resource utilization. Low-density active areas refer to feed transmission nodes in breeding areas where the pecking frequency of broilers is significantly lower than that of surrounding areas. Such low-density active areas represent "cold spot" areas where broiler groups are relatively sparse and inactive. Such areas can be used to exclude low-priority areas when optimizing feed distribution, thereby concentrating resources on high-demand areas that require more attention, improving transportation efficiency and reducing energy consumption.

[0037] In specific implementation, the multidimensional environmental factors of the breeding area can be obtained in the following way, namely: multi-channel environmental sensor nodes (such as temperature and humidity sensors, ammonia sensors and carbon dioxide sensors) can be deployed at the center of each feed transmission node projected on the ground of the breeding area to collect the air temperature, relative humidity, ammonia concentration and carbon dioxide concentration in the feed transmission node, and the set of air temperature, relative humidity, ammonia concentration and carbon dioxide concentration of all feed transmission nodes is used as the multidimensional environmental factor of the breeding area; wherein, the multidimensional environmental factor refers to the set of environmental indicators synchronously collected by multiple sensors such as temperature and humidity, ammonia concentration, light intensity, carbon dioxide, etc. in the breeding area. The multidimensional environmental factor can be used to provide a comprehensive environmental suitability assessment for dynamic zoning, thereby accurately guiding feed transportation and resource allocation, and improving broiler health and production efficiency.

[0038] In specific implementation, the environmental suitability of each high-density active area and each low-density active area is determined based on the multidimensional environmental factors, which can be achieved in the following manner: first, for each high-density active area, the air temperature, relative humidity, ammonia concentration and carbon dioxide concentration of the high-density active area are first obtained within the multidimensional environmental factors, and the air temperature, the relative humidity, the ammonia concentration and the carbon dioxide concentration are normalized using an existing normalization method (such as standard deviation normalization) to eliminate the influence of different dimensions, and then the air temperature, the relative humidity, the ammonia concentration and the carbon dioxide concentration are weighted and summed according to a preset weight vector using an existing weighted summation method (such as a linear weighted method), and the sum is used as the environmental suitability of the high-density active area. The environmental suitability of each high-density active area can be obtained through the above steps, and then, for each low-density active area, the air temperature, relative humidity, ammonia concentration and carbon dioxide concentration of the low-density active area are first obtained within the multidimensional environmental factors. concentration and carbon dioxide concentration, and use the existing normalization method (such as standard deviation normalization) to normalize the air temperature, the relative humidity, the ammonia concentration and the carbon dioxide concentration to eliminate the influence of different dimensions, and then use the existing weighted summation method (such as linear weighted method) to perform weighted summation on the air temperature, the relative humidity, the ammonia concentration and the carbon dioxide concentration according to the preset weight vector, and the sum value obtained is used as the environmental suitability of the low-density active area. The environmental suitability of each high-density active area can be obtained through the above steps; wherein, the environmental suitability is an indicator that quantifies the degree of match between the ecological conditions of the breeding area and the growth requirements of broiler chickens. In this solution, the environmental suitability can be used to ensure that the feed transportation zone covers both high-activity areas and gives priority to the most suitable environmental areas, thereby improving feed utilization efficiency; wherein, the weight vector can be set according to historical feeding data. In other embodiments, other methods can also be used to set the weight vector, which is not specifically limited here.

[0039] It should be noted that, in this application, dynamic feed transmission partition refers to a feed distribution area that adapts to different broiler activity and environmental conditions. The dynamic feed transmission partition can be used to accurately guide feed supply and improve distribution efficiency and breeding balance. In specific implementation, the feed transmission nodes in the breeding area are dynamically divided according to all environmental suitabilities, and multiple dynamic feed transmission partitions can be obtained. The following method can be used, namely: all high-density active areas and all low-density active areas can be clustered according to all environmental suitabilities in the breeding area through the existing spatiotemporal density clustering algorithm (such as a density clustering algorithm based on space-time), and high-density active areas with similar environmental suitability (such as the difference in environmental suitability is less than or equal to 0.5) are automatically aggregated into a dynamic feed transmission partition, and low-density active areas with similar environmental suitability (such as the difference in environmental suitability is less than or equal to 0.5) are automatically aggregated into a dynamic feed transmission partition, thereby obtaining multiple dynamic feed transmission partitions.

[0040] In step 104, the growth state tensors of the broiler flocks in different dynamic feed transmission partitions are determined using the long short-term memory network growth prediction model in combination with the frequency distribution diagram and the synchronization index of the broiler flock development in each dynamic feed transmission partition.

[0041] In some embodiments, reference Figure 3 This figure is an operational flow chart for determining the growth status tensor of a broiler flock within a dynamic feed transmission zone according to some embodiments of the present application. In the present application, the long short-term memory network growth prediction model is combined with the frequency distribution diagram and the synchronization index of broiler flock development in each dynamic feed transmission zone to determine the growth status tensor of the broiler flock within different dynamic feed transmission zones. The following steps can be used: Determine the synchronization indicators of broiler flock development in each dynamic feed delivery zone; For each dynamic feed transmission partition, using a long short-term memory network growth prediction model to generate a growth prediction curve of the broiler flock in the dynamic feed transmission partition in different future time periods based on the frequency distribution diagram and the synchronization index of the broiler flock development in the dynamic feed transmission partition; The growth state tensor of the broiler chicken group in the dynamic feed transmission zone is determined based on all the growth prediction curves, and then the growth state tensors of the broiler chicken group in different dynamic feed transmission zones are obtained.

[0042] In specific implementation, the following method can be used to determine the synchronization indicators of the development of the broiler group in each dynamic feed transmission partition, namely: for each dynamic feed transmission partition, first, the weight of the broilers in the point cloud data of the dynamic feed transmission partition in the three-dimensional point cloud model can be estimated by a non-invasive weight estimation algorithm (such as a point cloud deep learning model based on deep features), and the average weight growth rate of the broilers in the dynamic feed transmission partition can be automatically output according to a preset time window (such as every hour) as the growth indicator of the broilers in the dynamic feed transmission partition. Then, the frequency of pecking events in each feed transmission node in the dynamic feed transmission partition is obtained in the pecking density map, and all feed transmission nodes are recorded. The standard deviation of the frequency of pecking events in the feed transmission node is used as the group behavior synchronization of broiler chickens in the dynamic feed transmission partition. Finally, the vector composed of the growth index and the group behavior synchronization is used as the synchronization index of the broiler group development in the dynamic feed transmission partition. The synchronization index of the broiler group development in each dynamic feed transmission partition can be obtained through the above steps; wherein, the synchronous development index is a comprehensive index that quantifies the multi-dimensional growth status of the weight growth rate and behavioral synchronization of the broiler group in the dynamic feed transmission partition. The synchronous development index can accurately capture the growth differences of different partitions by evaluating the developmental consistency of the broiler group, thereby optimizing feed demand prediction and allocation decisions.

[0043] In specific implementation, the long short-term memory network growth prediction model is used to generate a growth prediction curve of the broiler group in the dynamic feed transmission partition in different future time periods based on the frequency distribution diagram and the synchronization index of the development of the broiler group in the dynamic feed transmission partition. This can be achieved in the following way, namely: a pre-trained growth prediction model (such as a growth prediction model based on a long short-term memory network) can be loaded, and the frequency distribution diagram and the synchronization index of the development of the broiler group in the dynamic feed transmission partition are used as inputs of the growth prediction model. The growth prediction model is run, and the weight growth prediction curve of the broiler group in different future time periods (such as 1-4 hours, 4-8 hours, 8-12 hours, etc.) output by the growth prediction model is used as the growth prediction curve of the broiler group, thereby obtaining the growth prediction curve of the broiler group in the dynamic feed transmission partition in different future time periods; wherein, the growth prediction curve is a curve reflecting the changing trend of the growth index of the broiler group in the dynamic feed transmission partition, and the growth prediction curve can be used to accurately control the timing and quantity of feed feeding, thereby optimizing breeding efficiency and timely warning of growth abnormalities.

[0044] It should be noted that in this application, the long short-term memory network growth prediction model is used to integrate the spatiotemporal behavior and developmental indicators of the broiler group to achieve accurate prediction of the short-term weight growth trend of each dynamic feed transmission partition. The frequency distribution map and the synchronous development indicators are combined to form a multi-channel time series feature sequence, which is input into the long short-term memory network growth prediction model for sequence-to-sequence prediction. The model output is the weight growth prediction curve of each dynamic feed transmission partition in different time periods in the future (such as 1-4 hours, 4-8 hours, 8-12 hours, etc.), providing a quantitative basis for feed distribution and health management of broiler groups.

[0045] It should be noted that in the present application, the growth state tensor is a multidimensional data structure that integrates the growth prediction curves of the broiler group at different times and spaces. The growth state tensor can be used to comprehensively characterize the dynamic growth state of the broilers in the dynamic feed transmission partition, and realize accurate growth prediction of the broiler group in the dynamic feed transmission partition, thereby optimizing the feed supply strategy and improving the breeding efficiency and the health level of the broilers; in specific implementation, the growth state tensor of the broiler group in the dynamic feed transmission partition based on all growth prediction curves can be determined in the following way, namely: the growth prediction curves of the dynamic feed transmission partition in all future time periods can be arranged in chronological order and organized into a prediction matrix, and the prediction matrix is ​​used as the growth state tensor of the broiler group in the dynamic feed transmission partition.

[0046] In step 105, the electromagnetic distributing valve group is controlled according to all growth state tensors to deliver the feed required by each dynamic feed transmission zone to the mobile feed transmission vehicle, and the mobile feed transmission vehicle is controlled to perform dynamic feed transmission.

[0047] In some embodiments, the following steps may be used to control the electromagnetic dispensing valve group according to all growth state tensors to deliver the feed required by each dynamic feed transfer zone to the mobile feed transfer vehicle: For each dynamic feed delivery partition, determining a feed requirement of the dynamic feed delivery partition according to the growth state tensor; Mapping the feed demand of the dynamic feed transmission partition to the corresponding solenoid valve group number and generating the control instructions of the dynamic feed transmission partition; According to the control instructions, the feed required by the dynamic feed transmission partition is delivered to the mobile feed transmission vehicle, and then the feed required by each dynamic feed transmission partition is delivered to the mobile feed transmission vehicle.

[0048] Preferably, in some embodiments, determining the feed requirement of the dynamic feed delivery partition according to the growth state tensor can be achieved by using the following steps: determining regional growth characteristics of the dynamic feed delivery partition based on the growth state tensor; The feed requirements of the broiler chicken population in the dynamic feed delivery zone are predicted based on the regional growth characteristics.

[0049] In specific implementation, determining the regional growth characteristics of the dynamic feed transmission partition based on the growth status tensor can be achieved in the following manner, namely: existing feature extraction technology (such as principal component analysis) can be used to perform comprehensive feature extraction on the growth prediction curve of the dynamic feed transmission partition in all future time periods in the growth status tensor to obtain the regional growth characteristics of the dynamic feed transmission partition; wherein, the regional growth characteristics are compact numerical representations of the multi-dimensional growth status data of the broiler group in the dynamic feed transmission partition. The regional growth characteristics can accurately summarize the growth status of the broilers in the dynamic feed transmission partition, and can be used as an efficient input for feed demand prediction and optimal allocation, thereby significantly improving the allocation accuracy and system response speed.

[0050] In specific implementation, the feed requirement of the broiler group in the dynamic feed transmission partition is predicted based on the regional growth characteristics, and the feed requirement of the broiler group in each dynamic feed transmission partition is obtained. This can be achieved in the following way, namely: a trained long short-term memory network model can be loaded, the regional growth characteristics are used as the input of the long short-term memory network model, the long short-term memory network model is executed, and the next moment feed requirement prediction value in the output of the long short-term memory network model is used as the feed requirement of the broiler group in the dynamic feed transmission partition; wherein, the feed requirement refers to the total amount of feed required to meet the normal growth, development and health of the broiler group in the dynamic feed transmission partition. In this scheme, accurately predicting the feed requirement of each dynamic feed transmission partition helps to achieve precise allocation of feed supply, optimize resource utilization, reduce feeding costs, and improve the growth efficiency and health level of broilers, thereby improving the production efficiency of the entire breeding system.

[0051] It should be noted that in this application, historical breeding data can be used to train the long short-term memory network model, and the model parameters can be optimized by minimizing the error between the predicted value and the actual feed consumption. After the training is completed, the long short-term memory network model can receive new regional growth feature vectors in real time and predict the feed demand of each dynamic feed transmission partition, thereby providing a basis for feed distribution.

[0052] It should be noted that the solenoid valve group is a group of execution units that control the opening and closing of multiple valve cores through the on-off electrical signals of the electromagnetic coil, and is used to accurately adjust the distribution path of fluids (such as feed suspensions or pneumatic conveying media). In this solution, the solenoid valve group can accurately deliver feed to each dynamic feed transmission zone, thereby improving the efficiency and accuracy of feed distribution and ensuring that the nutritional needs of broiler groups in different areas are met.

[0053] In specific implementation, mapping the feed demand of the dynamic feed transmission partition to the corresponding solenoid valve group number and generating the control instruction of the dynamic feed transmission partition can be achieved in the following manner, namely: first, a mapping relationship table between the dynamic feed transmission partition and the solenoid valve group number can be obtained, and the mapping relationship table can be searched to determine the solenoid valve group number corresponding to the dynamic feed transmission partition, and then, the feed demand of the dynamic feed transmission partition is mapped to the solenoid valve group number corresponding to the dynamic feed transmission partition and a control instruction is generated; wherein, the control instruction is an instruction sent to the solenoid valve controller via wireless communication to achieve precise control of the solenoid valve, and the control instruction can ensure that the feed is accurately delivered to each dynamic feed transmission partition, and has the advantages of low cost, easy installation, and reliable operation; wherein, the mapping relationship table needs to be established in advance, that is: the unique identifier of each dynamic feed transmission partition is associated with its corresponding solenoid valve group number.

[0054] In specific implementation, the feed required by the dynamic feed transmission partition can be delivered to the mobile feed transmission vehicle according to the control instructions in the following manner, namely: the solenoid valve group can quickly respond to the control instructions and switch to the feeding channel of the dynamic feed transmission partition and deliver the feed required by the dynamic feed transmission partition to the mobile feed transmission vehicle through the feeding channel.

[0055] It should be noted that the mobile feed transport vehicle is an autonomous mobile platform that integrates automatic navigation, feed storage and quantitative feeding functions. It is often used to achieve unmanned management in large-scale farms. The vehicle is usually equipped with lidar, visual sensors or inertial measurement units, which can build environmental maps in real time and perform path planning; in this solution, the mobile feed transport vehicle can accurately reach each dynamic feed transmission zone for quantitative feed delivery based on the optimal path planned by the three-dimensional point cloud model, significantly reducing the number of manual inspections and labor intensity, improving feed distribution efficiency, and continuously optimizing feeding strategies through real-time sensor feedback.

[0056] In some embodiments, controlling the mobile feed transport vehicle to dynamically transport feed may be achieved by using the following steps: Extracting the movement trajectory of the broiler chicken group from the three-dimensional point cloud model; Obtain the spatial distribution of all dynamic feed transport zones in the farming area; Planning a feed transportation path covering all feeding points in the breeding area based on the spatial distribution and the movement trajectory of the broiler chicken group; The mobile feed transport vehicle is controlled by the feed transport path to feed the feed at each feeding point.

[0057] In specific implementation, the spatial distribution of all dynamic feed transmission partitions in the breeding area can be obtained in the following way, namely: the spatial distribution map of the breeding area after the dynamic feed transmission partitions are divided can be obtained in the three-dimensional point cloud data of the breeding area in the three-dimensional point cloud model; wherein, the spatial distribution refers to an image showing the geographical location and layout status of all dynamic feed transmission partitions in the breeding area. In this solution, accurately determining the spatial distribution of all dynamic feed transmission partitions in the breeding area can achieve accurate path planning and efficient feed distribution, ensuring that the mobile feed transmission vehicle can reasonably arrange the order and route of feed feeding according to the activity trajectory of the broiler group, thereby improving breeding efficiency and resource utilization.

[0058] In specific implementation, the motion trajectory of the broiler chicken group can be extracted from the three-dimensional point cloud model in the following manner, namely: the existing multi-target tracking algorithm (such as an end-to-end deep neural network model) can be used to estimate the point motion in the three-dimensional point cloud model by learning the scene flow to identify and track the motion trajectory of the chicken body instance, and the collection of the motion trajectories of all chicken body instances is used as the motion trajectory of the broiler chicken group; wherein, the motion trajectory of the broiler chicken group refers to the motion path of the broiler chicken group in the breeding area that changes with time in the three-dimensional space.

[0059] In specific implementation, the feed transportation path covering all feeding points in the breeding area based on the spatial distribution and the movement trajectory of the broiler group can be planned in the following manner, namely: a path planning algorithm (such as an improved bidirectional rapid expansion random tree algorithm) can be used to calculate the path according to the spatial distribution and the movement trajectory of the broiler group to obtain a feed transportation path covering all feeding points in the breeding area; wherein, the feed transportation path refers to the optimal driving route of a mobile feed transport vehicle starting from a feed source point (such as a storage bin), passing through the feeding points of each dynamic feed transmission partition in the breeding area, and finally returning to the starting point. The feed transportation path takes into account the spatial distribution of the breeding area, the movement trajectory of the broiler group, the location of obstacles, and environmental factors, thereby achieving high efficiency and accuracy in feeding feed. The path planning algorithm can quickly and smoothly plan the optimal obstacle avoidance path in a point cloud environment, and take into account the actual situation of the breeding area during the path planning process, such as channel width, obstacle location, and broiler activity area, to ensure the feasibility and efficiency of the feed transportation path.

[0060] In specific implementation, the feed transport path can be used to control the mobile feed transport vehicle to feed at each feeding point in the following way, that is, the planned feed transport path can be input into the control system of the mobile feed transport vehicle to guide it to travel along the predetermined path and complete the precise feed feeding task.

[0061] In addition, another aspect of the present application, in some embodiments, the present application provides an automatic transmission system for broiler feeding, the system comprising an electromagnetic dispensing valve group and a mobile feed transmission vehicle, the system also comprising a broiler feed dynamic transport unit, reference Figure 4 This figure is a schematic structural diagram of a broiler feed dynamic transport unit according to some embodiments of the present application. The broiler feed dynamic transport unit 400 includes: a collection module 401, a processing module 402, and an execution module 403, which are described as follows: Acquisition module 401, in this application, acquisition module 401 is mainly used to collect dynamic distribution images of broiler chickens in the breeding area, and then construct a three-dimensional point cloud model of the breeding area; Processing module 402, in this application, is mainly used to perform instance segmentation of the three-dimensional point cloud model based on skeleton key point regression, thereby extracting the behavioral activity of the broiler group at different locations in the breeding area, and determining the frequency distribution map of the broiler group pecking at different locations in the breeding area; It should be noted that the processing module 402 in the present application is also used to dynamically divide the feed transmission nodes of the broiler flocks in the breeding area according to the behavioral activity of the broiler flocks at different locations in the breeding area and the preset group density threshold combined with the multi-dimensional environmental factors of the breeding area, thereby obtaining multiple dynamic feed transmission zones; In addition, it should be noted that the processing module 402 in the present application is also used to determine the growth state tensor of the broiler flocks in different dynamic feed transmission partitions by using the long short-term memory network growth prediction model in combination with the frequency distribution diagram and the synchronization index of the broiler flock development in each dynamic feed transmission partition; Execution module 403, in this application, execution module 403 is mainly used to control the electromagnetic distribution valve group according to all growth state tensors to deliver the feed required by each dynamic feed transmission partition to the mobile feed transmission vehicle, and control the mobile feed transmission vehicle to perform dynamic feed delivery.

[0062] Each module in the aforementioned automatic conveying system for broiler feeding can be implemented in whole or in part through software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in hardware form, or can be stored in a computer device memory in software form, so that the processor can call and execute the corresponding operations of each module.

[0063] In addition, in one embodiment, the present application provides a computer device, which may be a server, and its internal structure diagram may be as follows: Figure 5As shown. The computer device includes a processor, a memory and a network interface connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data on a dynamic transportation method for broiler feed based on visual intelligence. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a dynamic transportation method for broiler feed based on visual intelligence is implemented.

[0064] Those skilled in the art will understand that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0065] In one embodiment, a computer device is also provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the steps in the above-mentioned embodiment of the dynamic transportation method for broiler feed based on visual intelligence are implemented.

[0066] In one embodiment, a computer-readable storage medium is provided, storing a computer program, which, when executed by a processor, implements the steps in the above-mentioned embodiment of the method for dynamic transportation of broiler feed based on visual intelligence.

[0067] In one embodiment, a computer program product or computer program is provided, comprising computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the steps of the embodiment of the method for dynamic transportation of broiler feed based on visual intelligence.

[0068] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory or optical memory, etc. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0069] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0070] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art could make various modifications and improvements without departing from the spirit of the present application, all of which fall within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be determined by the appended claims.

Claims

1. A dynamic transportation method for broiler feed based on visual intelligence, applied to an automatic transmission system for broiler feeding, the system includes an electromagnetic dispensing valve group and a mobile feed transmission vehicle, characterized in that: The method comprises the following steps: Collect dynamic distribution images of broiler chickens in the breeding area and then construct a 3D point cloud model of the breeding area; Performing instance segmentation on the three-dimensional point cloud model based on skeleton key point regression to extract the behavioral activity of the broiler flock at different locations in the breeding area, and determining the frequency distribution map of the broiler flock pecking at different locations in the breeding area; Based on the behavioral activity of broiler flocks at different locations in the breeding area and the preset population density threshold combined with the multi-dimensional environmental factors of the breeding area, the feed transmission nodes of the broiler flocks in the breeding area are dynamically divided into multiple areas to obtain multiple dynamic feed transmission zones; Determining the growth state tensors of the broiler flocks in different dynamic feed delivery zones using a long short-term memory network growth prediction model in combination with the frequency distribution diagram and the synchronization index of broiler flock development in each dynamic feed delivery zone; According to all growth state tensors, the electromagnetic dispensing valve group is controlled to deliver the feed required by each dynamic feed transmission partition to the mobile feed transmission vehicle, and the mobile feed transmission vehicle is controlled to perform dynamic feed transmission.

2. The method according to claim 1, wherein The three-dimensional point cloud model is segmented based on skeleton key point regression to extract the behavioral activity of the broiler chickens at different locations in the breeding area. Specifically, the following information is obtained: Performing key point regression on the three-dimensional point cloud model to obtain three-dimensional offset vectors of all chicken skeleton key points; Clustering adjacent point clouds in the three-dimensional point cloud model based on the three-dimensional offset vectors of all chicken skeleton key points, thereby separating all independent chicken instances; Determine the feeding posture of each chicken instance in the breeding area; Determine the pecking density map of broiler chickens at different locations in the breeding area based on the feeding postures of all chicken instances; The behavioral activity of the broiler flock at different locations in the breeding area is extracted from the pecking density map.

3. The method according to claim 1, wherein Based on the behavioral activity of broiler flocks at different locations in the breeding area and the preset population density threshold, combined with the multi-dimensional environmental factors of the breeding area, the feed transmission nodes of the broiler flocks in the breeding area are dynamically divided into multiple areas, and multiple dynamic feed transmission zones are obtained, including: According to all behavioral activities and the preset population density threshold, multiple high-density active areas and multiple low-density active areas of the broiler population in the breeding area are identified; Obtain multidimensional environmental factors of the breeding area; determining the environmental suitability of each high-density active area and each low-density active area based on the multidimensional environmental factors; The feed transmission nodes in the breeding area are dynamically divided according to all environmental suitabilities to obtain multiple dynamic feed transmission partitions.

4. The method according to claim 1, wherein The growth state tensor of the broiler chicken groups in different dynamic feed transmission partitions is determined by using the long short-term memory network growth prediction model in combination with the frequency distribution diagram and the synchronization index of the broiler chicken group development in each dynamic feed transmission partition, specifically including: Determine the synchronization indicators of broiler flock development in each dynamic feed delivery zone; For each dynamic feed transmission partition, using a long short-term memory network growth prediction model to generate a growth prediction curve of the broiler flock in the dynamic feed transmission partition in different future time periods based on the frequency distribution diagram and the synchronization index of the broiler flock development in the dynamic feed transmission partition; The growth state tensor of the broiler chicken group in the dynamic feed transmission zone is determined based on all the growth prediction curves, and then the growth state tensors of the broiler chicken group in different dynamic feed transmission zones are obtained.

5. The method according to claim 1, wherein According to all growth status tensors, the electromagnetic dispensing valve group is controlled to deliver the feed required by each dynamic feed transmission zone to the mobile feed transmission vehicle, specifically including: For each dynamic feed delivery partition, determining a feed requirement of the dynamic feed delivery partition according to the growth state tensor; Mapping the feed demand of the dynamic feed transmission partition to the corresponding solenoid valve group number and generating the control instructions of the dynamic feed transmission partition; According to the control instructions, the feed required by the dynamic feed transmission partition is delivered to the mobile feed transmission vehicle, and then the feed required by each dynamic feed transmission partition is delivered to the mobile feed transmission vehicle.

6. The method according to claim 5, wherein Determining the feed demand of the dynamic feed transmission partition according to the growth state tensor specifically includes: determining regional growth characteristics of the dynamic feed delivery partition based on the growth state tensor; The feed requirements of the broiler chicken population in the dynamic feed delivery zone are predicted based on the regional growth characteristics.

7. The method according to claim 1, wherein Controlling the mobile feed transport vehicle for dynamic feed delivery specifically includes: Extracting the movement trajectory of the broiler chicken group from the three-dimensional point cloud model; Obtain the spatial distribution of all dynamic feed transport zones in the farming area; Planning a feed transportation path covering all feeding points in the breeding area based on the spatial distribution and the movement trajectory of the broiler chicken group; The mobile feed transport vehicle is controlled by the feed transport path to feed the feed at each feeding point.

8. An automatic feeding system for broiler chicken feeding, comprising an electromagnetic feeding valve assembly and a mobile feed feeding vehicle, characterized in that: The system also includes a broiler feed dynamic transport unit, which includes: The acquisition module is used to collect dynamic distribution images of broiler chickens in the breeding area and then construct a three-dimensional point cloud model of the breeding area; a processing module for performing instance segmentation on the three-dimensional point cloud model based on skeleton key point regression, thereby extracting the behavioral activity of the broiler flock at different locations in the breeding area, and determining a frequency distribution diagram of the broiler flock pecking at different locations in the breeding area; The processing module is used to dynamically divide the feed transmission nodes of the broiler flocks in the breeding area into multiple zones based on the behavioral activity of the broiler flocks at different locations in the breeding area and a preset population density threshold in combination with the multi-dimensional environmental factors of the breeding area, thereby obtaining multiple dynamic feed transmission zones; The processing module is used to determine the growth state tensor of the broiler chicken groups in different dynamic feed transmission partitions by using the long short-term memory network growth prediction model in combination with the frequency distribution map and the synchronization index of the broiler chicken group development in each dynamic feed transmission partition; The execution module is used to control the electromagnetic feeding valve group according to all growth state tensors to deliver the feed required by each dynamic feed transmission zone to the mobile feed transmission vehicle, and control the mobile feed transmission vehicle to perform dynamic feed transmission.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the broiler feed dynamic transportation method based on visual intelligence described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method for dynamic transportation of broiler feed based on visual intelligence as described in any one of claims 1 to 7 are implemented.

Images