Animal feeding plan generation method and system based on ingestion monitoring data

Abstract

The invention provides an animal feeding plan generation method and system based on ingestion monitoring data. The method comprises the following steps: obtaining ingestion time sequence data of each animal in a feeding space for a feed storage box; acquiring health index sampling data of each animal; nutritional ingredients corresponding to the feed types of the feed storage boxes are obtained, so that a feed nutrition relation graph is constructed; processing the feed nutrition relation graph on the basis of a graph neural network to obtain updated node features of each feed type graph node; and defining an input state of a feeding plan generation model by using the acquired data, so as to determine an animal feeding plan corresponding to the target action through reinforcement learning. Therefore, through data driving and in combination with an advanced graph neural network and reinforcement learning technology, the feeding amount and the feeding time of the feed can be accurately calculated, and intelligent management of the animal feeding process is realized.

Description

Technical Field

[0001] The present application relates to the field of intelligent feeding technology, and in particular to a method and system for generating an animal feeding plan based on food intake monitoring data. Background Art

[0002] In animal husbandry and experimental research, scientific and reasonable feeding management has an important impact on the growth, health and metabolic status of animals. Taking crab-eating macaques as an example, as experimental animals commonly used in metabolic disease research (such as obesity, diabetes and cardiovascular disease), accurate monitoring of their food intake and nutritional requirements is the key to ensuring the reliability of experimental data.

[0003] Traditional feeding plans mostly rely on static settings, that is, setting a fixed feeding amount based on basic parameters such as the animal's weight, age, and gender. For example, suppose an adult male crab-eating macaque weighing 6 kg is set to consume a certain amount of standard feed every day in the study. Although this method can guarantee the basic nutritional needs of animals to a certain extent, it ignores the individual differences in animal feeding behavior. Specifically, even two crab-eating macaques of the same weight may have differences in metabolic rate, activity level, and feeding preferences.

[0004] In addition, as primates, crab-eating macaques have very complex feeding behaviors. Different monkeys may choose different types of food at different times, and even have significant differences in the speed, frequency, and method of food intake. Traditional static feeding programs usually fail to take these individual behavioral differences into account and can only set fixed feeding times and amounts, which makes it impossible to fully adapt to the natural feeding habits and health needs of crab-eating macaques.

[0005] In response to the above problems, the industry has not yet proposed a better technical solution. Summary of the invention

[0006] The present application provides a method, system, storage medium, computer program product and electronic device for generating an animal feeding plan based on food intake monitoring data, which are used to at least solve the problem of ignoring individual differences and insufficient adaptability of feeding plans in traditional feeding plans.

[0007] In a first aspect, an embodiment of the present application provides a method for generating an animal feeding plan based on feeding monitoring data, comprising: obtaining feeding time series data of each animal in the feeding space for each feed storage box in the feeding space; each feed storage box is used to put a corresponding type of feed, and the feeding time series data includes feeding time and feeding amount for the corresponding feed type; obtaining health indicator sampling data of each of the animals; obtaining the nutritional components corresponding to the feed type of each feed storage box to construct a feed nutritional relationship diagram; the feed nutritional relationship diagram includes feed graph nodes, nutritional graph nodes and edge connections, the feed graph nodes are used to indicate the corresponding feed type, the nutritional graph nodes are used to indicate the nutritional components of the corresponding type, and the feed-nutrient relationship diagram includes feed graph nodes, nutritional graph nodes and edge connections, the feed graph nodes are used to indicate the corresponding feed type, and the nutritional graph nodes are used to indicate the nutritional components of the corresponding type. The edge weight of the edge connection is defined by the strength of the association between the feed type and the nutritional components, and the edge weight of the nutrient-nutrient edge connection is defined by the dependency relationship between the nutritional components; based on the graph neural network, the node features of each feed type graph node and the feature information of the adjacent nodes in the feed nutritional relationship graph are aggregated and updated to obtain the updated node features of each feed type graph node; the input state of the feeding plan generation model is defined by combining the sampling data of each health indicator with the feeding timing data corresponding to each feed storage box, the updated node features and the current feed balance, so as to determine the animal feeding plan corresponding to the target action through reinforcement learning; the animal feeding plan includes the planned feed delivery time and the corresponding planned feed delivery amount of each feed storage box.

[0008] In a second aspect, an embodiment of the present application provides an animal feeding plan generation system based on feeding monitoring data, including: a feeding data acquisition unit, for acquiring feeding time series data of each animal in the feeding space for each feed storage box in the feeding space; each of the feed storage boxes is used to put a corresponding type of feed, and the feeding time series data includes feeding time and feeding amount for the corresponding feed type; a health data acquisition unit, for acquiring health indicator sampling data of each of the animals; a nutrition map construction unit, for acquiring the nutritional components corresponding to the feed types of each of the feed storage boxes, so as to construct a feed nutrition relationship diagram; the feed nutrition relationship diagram includes feed map nodes, nutrition map nodes and edge connections, the feed map nodes are used to indicate the corresponding feed types, and the nutrition map nodes are used to indicate the nutritional components of the corresponding types, The edge weight of the feed-nutrient edge connection is defined by the strength of the association between the feed type and the nutrient component, and the edge weight of the nutrient-nutrient edge connection is defined by the dependency relationship between the nutrients; a graph structure updating unit is used to aggregate and update the node features of each feed type graph node and the feature information of the adjacent nodes in the feed nutrient relationship graph based on the graph neural network to obtain the updated node features of each feed type graph node; a feeding plan generating unit is used to define the input state of the feeding plan generating model by combining each of the health indicator sampling data with the feeding timing data corresponding to each of the feed storage boxes, the updated node features and the current feed remaining amount, so as to determine the animal feeding plan corresponding to the target action through reinforcement learning; the animal feeding plan includes the planned feed delivery time and the corresponding planned feed delivery amount of each feed storage box.

[0009] According to a third aspect, an electronic device is provided, comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the steps of the method for generating an animal feeding plan based on food intake monitoring data of any embodiment of the present application.

[0010] In a fourth aspect, an embodiment of the present application provides a storage medium having a computer program stored thereon, characterized in that when the program is executed by a processor, the steps of the method for generating an animal feeding plan based on food intake monitoring data of any embodiment of the present application are implemented.

[0011] In a fifth aspect, an embodiment of the present application provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the steps of the method for generating an animal feeding plan based on food intake monitoring data of any embodiment of the present application.

[0012] The method and system for generating an animal feeding plan based on feeding monitoring data provided by the present application can produce at least the following technical effects: (1) By obtaining the feeding time series data and health indicators of each animal and constructing the feed nutrition relationship diagram of each feed storage box, the complex relationship between feed and nutrients is modeled and analyzed using graph neural networks, and the intelligent agent optimization calculation is performed using the reinforcement learning model. The deep learning capabilities of graph neural networks and reinforcement learning are comprehensively applied to achieve data-driven intelligent decision-making, which can dynamically optimize the feed delivery time and amount. As a result, the feeding plan can respond to the behavioral changes and health status of animals in real time, significantly improve the flexibility and adaptability of feeding management, and avoid feed waste or animal malnutrition caused by fixed feeding time and delivery amount.

[0013] (2) By obtaining real-time feeding time series data, it can reflect the feeding habits of animals in different time periods and dynamically optimize the feeding plan based on the health indicators of the animals. This not only adapts to the changes in the feeding habits of individual animals, but also can timely adjust the feeding strategy through real-time monitoring results, providing a more personalized feeding plan to ensure that each animal can receive feed that meets its health needs.

[0014] (3) For each feed storage box arranged in the feed space, the relationship between different feed types and their nutrients is systematically modeled by constructing a feed nutrition relationship diagram. The features of feed type nodes and nutrient component nodes are aggregated and updated through graph neural networks, thereby achieving a deep understanding of the complex dependencies between feed and nutrients. As a result, the dependencies between feed nutrients can be comprehensively considered when making feed placement decisions, and the most appropriate feed type and nutrient ratio can be intelligently selected according to the health status and feeding behavior of the animals, avoiding the negative impact of unbalanced feed nutrients on animal health. In addition, combined with the decision-making mechanism of reinforcement learning, the system can intelligently adjust its strategy in the process of continuously acquiring new feeding and health data to adapt to the growth changes and health status of the animals and automatically generate the optimal feeding plan.

[0015] Through this technical solution, through data-driven and combined with advanced graph neural network and reinforcement learning technology, the feed delivery amount and delivery time are accurately calculated, realizing the intelligent management of the animal feeding process, which can effectively meet the personalized nutritional needs of animals and improve the efficiency and accuracy of feeding management. At the same time, by real-time monitoring of the animal's feeding behavior and health indicators, the system can respond and adjust the feeding plan in a timely manner, effectively avoiding the limitation of traditional static feeding plans that cannot adapt to changes in animal health in a timely manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following is a brief introduction to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0017] Figure 1 A flowchart showing an example of a method for generating an animal feeding plan based on food intake monitoring data according to an embodiment of the present application; Figure 2 A schematic diagram of the exterior structure design of a feed storage box according to an embodiment of the present application is shown; Figure 3 A connection schematic diagram of an example of an electronic structure inside a food trough according to an embodiment of the present application is shown; Figure 4 An operation flow chart of an example of obtaining food intake time series data according to an embodiment of the present application is shown; Figure 5 An operation flow chart of an example of constructing a feed nutrition relationship diagram according to an embodiment of the present application is shown; Figure 6 A schematic diagram showing a structural connection of an example of a feeding plan generation model according to an embodiment of the present application is shown; Figure 7 A structural block diagram of an example of a system for generating an animal feeding plan based on food intake monitoring data according to an embodiment of the present application is shown; Figure 8 It is a schematic structural diagram of an embodiment of an electronic device of the present application. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0019] In the technical solution of this application, the collection, storage, use, processing, transmission, provision and disclosure of user personal information involved shall comply with the provisions of relevant laws and regulations and shall not violate public order and good morals.

[0020] Figure 1 A flowchart showing an example of a method for generating an animal feeding plan based on food intake monitoring data according to an embodiment of the present application.

[0021] Regarding the executor of the method of the embodiment of the present application, it can be any controller or processor with computing or processing capabilities. By integrating advanced data acquisition, graph neural network analysis and reinforcement learning optimization technology, automatic feeding plan generation can be achieved, which reduces human intervention and operational errors, and significantly improves the intelligence level and scientific nature of animal feeding management.

[0022] In some examples, it may be integrated and configured in an electronic device, terminal or server through software, hardware or a combination of software and hardware, and the types of the terminal, electronic device or server may be diverse, such as a mobile phone, tablet computer or desktop computer, etc.

[0023] like Figure 1 As shown, in step S110, for each feed storage box in the breeding space, the feeding time series data of each animal in the breeding space for the feed storage box is obtained.

[0024] Here, each feed storage box is used to put a corresponding type of feed, and the feeding time series data includes the feeding time and feeding amount for the corresponding feed type.

[0025] In some embodiments, in the feeding space, a feeding monitoring device (such as a feeding image sensing system, an RFID tag, an NFC tag, a weight sensor, etc.) is installed for each feed storage box to record the feeding behavior of each animal in different feed storage boxes in real time. In addition, each feed storage box corresponds to a specific type of feed, for example, a storage box is specifically used for placing high-protein feed, and another storage box is used for placing high-fiber feed. In this way, each time the animal eats, the sensor can accurately record the timestamp of the feeding, the length of time for each feeding, the type of feed ingested, the amount of feed ingested, and so on.

[0026] In step S120, the health index sampling data of each animal is obtained.

[0027] In one example of the embodiment of the present application, the breeder can collect the health index data of the animals by regularly checking the animals, such as measuring the blood sugar and blood lipids of the animals, and enter the collected data into the system platform. In another example of the embodiment of the present application, the health index data can be collected by non-invasive wearable sensor devices (e.g., smart collars) to achieve real-time sampling and recording of animal health indicators, such as real-time monitoring of heart rate through photoplethysmography (PPG) technology, real-time monitoring of animal blood sugar and blood lipids through near-infrared spectral analysis of spectral and optical sensors, monitoring of animal weight through smart weight scales installed at the bottom of the feeding area, etc., to ensure that the feeding plan can be dynamically adjusted according to the health status of the animals.

[0028] In some embodiments, the feeding behavior data includes at least one of the following: feeding time, feeding frequency, and feeding duration. In addition, the health index sampling data is used to indicate at least one of the following health indicators: body weight, heart rate, blood sugar, and blood lipids. By incorporating multi-dimensional feeding behavior data such as feeding time, feeding frequency, and feeding duration into the model, rich behavior pattern information is provided. In addition, by incorporating multiple health indicators such as body weight, heart rate, blood sugar, and blood lipids during data analysis, the health status of the animal can be more comprehensively reflected.

[0029] In step S130, the nutritional components corresponding to the feed types in each feed storage box are obtained to construct a feed nutritional relationship diagram.

[0030] Here, the feed nutritional relationship graph includes feed graph nodes, nutritional graph nodes and edge connections. The feed graph nodes are used to indicate the corresponding feed types, and the nutritional graph nodes are used to indicate the corresponding types of nutrients. The edge weights of the feed-nutrient edge connections are defined by the strength of the association between the feed type and the nutrients, and the edge weights of the nutrient-nutrient edge connections are defined by the dependency between the nutrients.

[0031] In some embodiments, the nutritional table of the feed type is parsed to obtain a corresponding set of nutritional component data, such as protein, fat, carbohydrates, vitamins and minerals, etc. Exemplarily, the user obtains the nutritional table by uploading it to the system platform for parsing, or querying the nutritional database based on the feed type.

[0032] Specifically, by constructing a heterogeneous graph structure containing multiple types of graph nodes, feed graph nodes are used to express different feed types, and nutrition graph nodes are used to express different nutrients. There may be edge connections between feed graph nodes and nutrition graph nodes, which can reflect the correlation between feed and nutrients, such as the proportion of a certain nutrient in the feed, so as to define the edge weight of the feed-nutrient edge connection. In addition, there may also be edge connections between different nutrition graph nodes, which can reflect the correlation between nutrients, such as synergistic or antagonistic relationships. For example, there is a synergistic relationship between protein and amino acids, a certain synergistic ratio relationship between calcium and phosphorus, vitamin D affects calcium absorption, iron and zinc have competitive absorption phenomena, and so on. Furthermore, by referring to nutrition research knowledge or sample experimental data, such synergistic or antagonistic relationships are quantified into corresponding nutrient-nutrient edge weights.

[0033] In step S140, the node features of each feed type graph node and the feature information of the adjacent nodes in the feed nutrition relationship graph are aggregated and updated based on the graph neural network to obtain the updated node features of each feed type graph node.

[0034] Here, the Graph Neural Network (GNN) uses a message passing mechanism to aggregate and update the features of each graph node (feed type and nutrient content) based on the information of its adjacent nodes to achieve node feature updates. For example, the features of the feed type node will consider the feature information of the adjacent nutrient component nodes to update its own features, so that these updated node features can reflect the overall nutritional value of the feed.

[0035] In step S150, the input state of the feeding plan generation model is defined by combining the sampling data of each health indicator with the feeding time series data corresponding to each feed storage box, the updated node features and the current feed remaining, so as to determine the animal feeding plan corresponding to the target action through reinforcement learning.

[0036] Here, the animal feeding plan includes the planned feed delivery time and the corresponding planned feed delivery amount of each feed storage box.

[0037] It should be noted that the feeding plan generation model can adopt various types of reinforcement learning models, such as Q-learning, deep Q network, etc., which can generate the optimal target action (feed plan delivery time and corresponding feed plan delivery amount) according to the current input status (such as animal health indicators, feeding time series data corresponding to each feed storage box, updated node features and current feed balance).

[0038] In the embodiment of the present application, the agent is responsible for generating a feeding plan, which can make decisions (i.e., animal feeding plan) based on the input state. The agent operates in the environment and obtains feedback from the environment. Here, the environment contains a state space defined by information such as health indicator sampling data, feeding time series data, node characteristics, and feed remaining.

[0039] It should be understood that the design of the reward function can be diverse and can be set or adjusted according to business needs. In some examples, the reward function can be set comprehensively based on factors such as the health status of the animal or the effective utilization rate of the feed. For example, if the health indicators of the animal are improved after the feeding plan is implemented, the agent will receive a positive reward or a larger reward feedback; if the feed is wasted or the health status of the animal is reduced due to an unreasonable feeding plan, the agent will receive a negative reward or a smaller reward feedback.

[0040] In some embodiments, as factors such as the health status, feeding behavior, or feed balance of the animal change in real time, the system can adjust the output of the reinforcement learning model and automatically update the feeding time and corresponding feeding amount of different feed storage boxes, thereby adaptively meeting the personalized feeding needs of the animal.

[0041] Figure 2 A schematic diagram of the exterior structural design of a feed storage box according to an embodiment of the present application is shown.

[0042] like Figure 2 As shown, the feed storage box adopts a multi-layer box structure, and the feed storage box includes an upper box 21 and a lower box 22, wherein the upper box 21 is a feeding area provided with a feeding trough 211, and the lower box is used to store animal feces from the feeding area through the grid through holes at the bottom of the upper box. In this way, the animal's food and urine and feces can be effectively separated, and the food intake can be calculated more accurately while ensuring the animal's dietary hygiene. In addition, the breeder generally only needs to clean the feces in the lower box 22, which can also reduce the workload burden on the breeder to a certain extent. Preferably, as Figure 2 Multiple feed storage boxes can also be stacked to further improve the space utilization of the breeding space.

[0043] Figure 3 A connection diagram of an example of the electronic structure inside the trough according to an embodiment of the present application is shown.

[0044] like Figure 3 As shown, the electronic structure inside the trough 300 includes an RFID reader 310, a weighing sensor 320, and a signal transmitter 330. In some embodiments, the RFID reader 310 can be designed on the side close to the animal activity area to sense the unique RFID tag worn by each animal and identify the target animal in the feeding area in real time. In addition, each RFID reader equipped with the trough has a unique corresponding reader device number, thereby realizing the induction identification of the type of feed ingested by the animal.

[0045] The weighing sensor 320 is designed as a tray at the bottom of the trough, which is used to sense the weight change of the contained feed in real time and record the corresponding feed consumption and consumption time.

[0046] The signal transmitter 330 is used to upload the data collected by the RFID reader 310 and the weighing sensor 320 to the animal feeding management platform. For example, the MQTT protocol or other lightweight protocols suitable for IoT devices can be used to achieve efficient data transmission.

[0047] Figure 4 An operational flow chart of an example of obtaining food intake timing data according to an embodiment of the present application is shown.

[0048] like Figure 4 As shown, in step S410, the target animals in the feeding area are monitored in real time based on the RFID reader disposed inside the feeding trough sensing the RFID tags of the animals.

[0049] Specifically, taking crab-eating macaques as an example, each crab-eating macaque is equipped with a unique RFID tag, which contains the animal's unique identification data (UID) and related basic information (such as number, gender, age, health status, etc.). The RFID tag is securely fixed to the collar or other parts of each crab-eating macaque to ensure that the tag is not easily removed or damaged during the animal's feeding process. When the crab-eating macaque is in the feeding area near the trough, the RFID reader detects and records the UID, entry time and exit time of the animal entering the feeding area in real time. By matching the read UID with the animal information in the database, the accuracy of the data and the reliability of individual identification are ensured.

[0050] In step S420, the feed consumption amount and feed consumption time of the corresponding feed type are monitored in real time based on a weighing sensor disposed inside the feeding trough to determine the feeding timing data of the corresponding target animal.

[0051] Here, electronic weighing sensors that support high precision and fast response, such as strain gauge sensors, can be selected, and the weighing sensors can be installed on the bottom tray of the trough to ensure that the feed consumption can be accurately measured.

[0052] Therefore, individualized feeding behavior monitoring can be achieved through RFID sensing technology to ensure that the feeding data of each animal is independently collected and analyzed. High-precision weighing sensors can ensure the high accuracy and reliability of feeding amount and feeding time data, providing a reliable data basis for personalized feeding management.

[0053] Figure 5 An operational flow chart of an example of constructing a feed nutritional relationship diagram according to an embodiment of the present application is shown.

[0054] like Figure 5 As shown, in step S510, based on the feed nutrient database, the nutrient content of the feed type in each feed storage box and the bioavailability and nutrient metabolism experimental data corresponding to each nutrient are determined.

[0055] Here, the feed nutrient database can be a self-built database or a third-party database, and can be constructed by integrating scientific literature, laboratory test results, data provided by feed manufacturers, etc. By collecting nutrient data of various types of feed, including specific contents of protein, fat, carbohydrates, vitamins, minerals, etc. Investigate the bioavailability data of each nutrient, that is, the proportion of actual absorption and utilization in the animal body. Collect nutritional metabolism experimental data on animal health and metabolic functions related to nutrition, for example, through animal metabolism experiments, determine the bioavailability of each nutrient, which can also include indicators such as metabolic rate, energy consumption, and nutrient absorption efficiency. In some examples, nutritional metabolism experimental data can also include experimental samples that consider the impact of nutrient intake on various animal health indicators, which helps to reveal the relationship between nutrient intake and animal health.

[0056] In step S520, a feed nutrition relationship diagram is constructed based on the nutrient content of each feed type and the bioavailability and nutrient metabolism experimental data corresponding to each nutrient.

[0057] Here, the node characteristics of the feed map node are defined by the feed name, feed classification and feed nutrient content of the corresponding feed type, and the node characteristics of the nutrition map node are defined by the nutritional indicator name, nutritional classification, nutritional metabolism experimental data and bioavailability of the corresponding nutrients.

[0058] Specifically, a dual graph structure is constructed, including feed graph nodes and nutrition graph nodes. The graph nodes are connected by edges, which represent the relationship between feed and nutrients or the relationship between nutrients. Each feed type corresponds to a node, and the node features include feed name, feed classification, content of each nutrient, etc. Each nutrient corresponds to a node, and the node features include nutritional index name (such as protein, fat, etc.), nutritional classification, related nutritional metabolism experimental data (such as the impact on metabolic rate or health indicators), bioavailability, etc.

[0059] In some examples of the embodiments of the present application, the feed classification includes at least one of the following: plant feed, animal feed, energy feed, fiber feed and protein feed. By introducing more sophisticated feed classification and nutrition classification in the graph node definition, the characteristics of each feed type node not only reflect the nutritional content, but also characterize its nutritional function positioning and biological role through the category identification. Nutritional classification includes at least one of the following: macronutrients, micronutrients, electrolytes and essential amino acids. Macronutrients provide a large amount of energy required by animals on a daily basis, including protein, fat and carbohydrates. Animals have a small demand for micronutrients, but they are essential for health and physiological functions, including vitamins and minerals. Electrolytes such as sodium, potassium, calcium, and magnesium maintain fluid balance and nerve conduction. Essential amino acids such as lysine and tryptophan cannot be synthesized by animals themselves and must be ingested through diet. In this way, the nutritional node features are more systematic and explanatory, so that the feature representation of the graph node is no longer a simple nutritional content vector, but a multidimensional feature set that is organically integrated with categories, functional attributes and physiological meanings, which greatly enriches the model's available information for processing nodes of the same type in heterogeneous graphs.

[0060] Furthermore, the edge weight of the feed-nutrient edge connection is calculated as follows: , formula (1) , formula (2) , formula (3) , formula (4) In the formula, Indicates i The feed type corresponding to each feed graph node, Indicates j The nutrients corresponding to each nutrition map node, Indicates that it is used to connect and The edge weights of the edge connections of the corresponding graph nodes; express middle The standardized content of express Important factors for animal health and metabolic function, express The bioavailability factor, represents the nonlinear parameter used to control the sensitivity of nutrient content, is the first metric global scaling factor; Indicates that it contains m An array of animal health indicators; Indicates aA health indicator, whose value is normalized to [0,1], and the larger the value, the better the health status; express The intake level and The correlation measure between them is in the range of [-1,1]. A positive value indicates that the intake is positively correlated with the health index, and a negative value indicates that the intake is negatively correlated with the health index. is the total number of experimental samples in the nutritional metabolism experimental data, including In the The intake measured in the samples was ,against In the The performance value measured in the samples is , For all experimental samples The mean of For all experimental samples The mean of Represents the correlation measure Mapped to the [0,1] interval corresponding to the positive score; Health indicators The weight of the indicator indicates the relative importance of the indicator to overall health.

[0061] It should be noted that the importance factor of a nutrient reflects the degree of correlation between changes in the intake of that nutrient and multidimensional health indicators (such as weight stability, healthy heart rate range, blood sugar stability, normal blood lipid levels, etc.). Specifically, for each nutrient By analyzing the experimental sample data, we calculated the correlation between the data and each health index. , and mapped to a positive score. Then, using the weight Perform weighted summation to obtain the corresponding importance factor .if It has a positive impact on multiple key health indicators (the correlation coefficient is positive and large), then If there is a negative correlation, the value will be lower.

[0062] Here, the Pearson Correlation Coefficient is used to measure the correlation, which can measure the strength and direction of the linear relationship between two variables. When the value is close to 1, it means that the nutrient intake is significantly positively correlated with the health indicator (the health indicator tends to improve when the nutrient intake increases). Close to -1, indicating a significant negative correlation (health indicators tend to deteriorate when nutrient intake increases). When it is close to 0, it means there is no significant linear relationship between the two.

[0063] Furthermore, in order to avoid negative correlation from having too great an inhibitory effect on the calculated importance factor, The function maps the correlation to the interval [0,1]. By analyzing historical sample experimental data and calculating the linear correlation between nutrient intake levels and health indicators, the model can comprehensively weigh the response of multiple health dimensions to the intake of a single nutrient. The system can dynamically adjust feed ingredients according to different health priorities, which helps to generate more targeted and personalized nutrition plans.

[0064] The edge weight of the nutrient-nutrient edge connection is calculated as follows: , formula (5) , formula (6) , formula (7) In the formula, Indicates k The nutrients corresponding to each nutrition map node, Indicates that it is used to connect and The edge weights of the edge connections of the corresponding graph nodes; express and The metabolic correlation between for The bioavailability factor, To control the parameters of the sensitivity of nutrient relationships to bioavailability, For the The nonlinear mapping parameters to the positive value space, is the second global scaling factor, represents the exponential function, express and Quantification of the synergistic or antagonistic effects of nutrients; Consider two nutrients and Combined intake levels and health indicators The correlation measure of and Indicates single intake and and health indicators The correlation measure of Health indicators Weight in determining relationships between nutrients.

[0065] It should be noted that the relationship between nutrients can be reflected as synergistic or antagonistic effects, which are determined by analyzing historical experimental data to determine the differences in the effects on health indicators when two nutrients are consumed simultaneously and when consumed separately.

[0066] Specifically, the correlation between the health indicators of two nutrients taken at the same time is calculated. , then calculate the relative The average contribution of .like If it is significantly higher than the average value, it indicates synergy; if it is lower than the average value, it indicates antagonism. Since the value may be negative (antagonism) or positive (synergy), it is necessary to map it to the [-1,1] interval and distinguish between positive and negative relationships, and then use the hyperbolic tangent function (tanh) to smoothly map the value. ,but Close to 1, indicating strong synergy; if ,but Close to -1, indicating strong antagonism.

[0067] Therefore, when the two nutrients are taken together, they are significantly better than when they are taken alone in terms of multidimensional health indicators. It is positive and large, and is close to 1 after mapping with the hyperbolic tangent function, indicating obvious synergy. When the performance of two nutrients ingested simultaneously is significantly worse than that of taking them alone, the value is negative and large, and is close to -1 after mapping, indicating strong antagonism. If the relationship between the two is not significant, the value will hover around 0. Therefore, the ability to identify and quantify the synergistic or antagonistic effects between nutrients from historical experimental data allows the system to more accurately consider the complex interactions between nutrients when formulating feed plans, thereby avoiding misjudgments caused by simple linear addition, and ultimately improving animal health and feeding efficiency.

[0068] Through the embodiment of the present application, the constructed feed nutrition relationship diagram and the corresponding parameter calculation formula (such as feed-nutrient edge weight, nutrition-nutrient relationship measurement, nutrient importance factor, etc.) are used to transform the original static, experience-driven feeding strategy into a data-driven decision-making process. By aggregating and updating the information extracted from the feeding time series data, physiological indicator sampling data and feed nutrient database, the ability to automatically extract core nutrition relationship features from large-scale complex data is technically realized, thereby providing a basis for subsequent feeding plan decision analysis.

[0069] In some examples of the embodiments of the present application, the graph neural network adopts a graph attention network (GAT), which dynamically assigns weights of neighbor nodes through an attention mechanism and incorporates category information and edge weights to aggregate and update node features.

[0070] Specifically, for each pair of adjacent graph nodes, the attention score is calculated: , formula (8) In the formula, Represents adjacent graph nodes and , and denote the learnable linear transformation matrix and attention parameter vector, respectively. represents a vector concatenation operation, and Represents graph nodes and graph nodes Node characteristics; is the information vector corresponding to the feed classification or nutrition classification, which is encoded using one-hot encoding; Represents the LeakyReLU activation function.

[0071] Here, the feed classification or nutritional classification information of the graph nodes is integrated into the attention calculation in the form of one-hot encoding, so that the relationship between nodes of different classifications can be distinguished when the model calculates the attention weight, thereby enhancing the expressiveness of node features, thereby improving the pertinence and effectiveness of feature aggregation, and helping the system to better identify and utilize the characteristics of different feeds.

[0072] Incorporate the edge weight into the attention score to form a weighted attention score: , formula (9) In the formula, represents the weighted attention score, Representing a graph node With graph nodes The edge weights of the edge connections between them.

[0073] Here, the edge weights of the precomputed edge connections are Incorporating it into the attention score calculation formula allows the feature aggregation process to not only consider the similarity of node features, but also reflect the association strength between feed and nutrients and the metabolic correlation between nutrients.

[0074] Normalize the attention scores to get the weighted class attention weights: , formula (10) In the formula, Representing a graph node With graph nodes The weighted category attention weight between Representing a graph node The set of neighboring graph nodes Any graph node in , Representing a graph node With graph nodes The weighted category attention weights between .

[0075] The node features of neighboring nodes are weighted aggregated based on the weighted category attention weights to update the node features: , formula (11) In the formula, Represents a nonlinear activation function (such as RELU activation function, etc.), Representing a graph node The updated node features.

[0076] As a preferred implementation of the embodiment of the present application, multiple layers of GAT having the above structure may be stacked to capture deeper node relationships and feature representations.

[0077] Through the embodiment of the present application, the weights of neighbor nodes are adaptively allocated based on the attention mechanism, and the category information and pre-calculated edge weights are integrated to organically integrate the multi-dimensional features of feed and nutrition, thereby improving the expressiveness and scientificity of node features, generating personalized features for different feed types, and supporting the formulation of individualized feeding strategies. In addition, based on the attention weight mechanism, the explainability of the feed-nutrition relationship is provided, helping managers understand the feature update process and enhancing the transparency and trust of the system.

[0078] Figure 6 A schematic diagram of structural connections of an example of a feeding plan generation model according to an embodiment of the present application is shown.

[0079] like Figure 6 As shown, the feeding plan generation model 600 adopts a hierarchical reinforcement learning model, which includes an attention fusion layer 610, a high-level policy network 620 and a low-level policy network 630.

[0080] The attention fusion layer 610 is used to fuse each of the health indicator sampling data with the feeding time series data corresponding to each of the feed storage boxes, the updated node features and the current feed remaining amount through the self-attention mechanism to obtain the attention fusion feature.

[0081] , formula (12) In the formula, represents the attention fusion feature, represents the self-attention mechanism, Represents the time step health indicators, Represents the time step Each feeding time series data, Represents the time step Each updated node feature of Represents the time step The current feed balance of each Represents the multilayer perceptron processing function.

[0082] Here, a multilayer perceptron (MLP) is used to fuse features from different sources and extract high-dimensional feature representation. Specifically, the model will transform the health indicator data ( )、Eating time series data( ), graph node features ( )、Current feed balance( ) and other information are projected into a unified feature space through a multi-layer perceptron.

[0083] Then, the self-attention mechanism is used to automatically select more important information and give it a higher weight according to the relevance of each feature to the current decision, and then perform weighted fusion on them to obtain the weighted fusion feature. It is a comprehensive representation that adaptively weights and integrates multiple heterogeneous data to ensure that decisions are made based on the most relevant and useful information.

[0084] The input state is defined based on the attention fusion feature to trigger the high-level policy network 620 to determine the corresponding feed plan delivery time.

[0085] , formula (13) In the formula, represents the time step decision vector of feed delivery output by the high-level policy network, which is of length A vector where each element represents the probability of feeding at the corresponding time step; Represents the sofmax activation function; and Represent the weight matrix and bias vector of the high-level policy network respectively.

[0086] Here, the high-level policy network takes the attention fusion feature as input state and generates a policy output through a series of linear and nonlinear transformations , which expresses a probability distribution of “when to replenish feed to each feed storage box next”, which is used to decide when to perform the delivery action.

[0087] The attention fusion features and the feed plan delivery time are input into the low-level strategy network 630 to determine the corresponding feed plan delivery amount.

[0088] , formula (14) , formula (15) , formula (16) , formula (17) , formula (18) , formula (19) In the formula, represents the input feature vector of the low-level policy network, , and Represent the weight matrix, bias vector and output features of the first hidden layer of the low-level strategy network, respectively. , and Represent the weight matrix, bias vector and output features of the second hidden layer of the low-level strategy network, respectively. and Represent the weight matrix and bias vector of the output mean layer of the low-level strategy network, and Represent the weight matrix and bias vector of the output standard deviation layer of the low-level policy network respectively; represents the mean vector of the amount of feed stored in each feed storage box, represents the standard deviation vector of the amount of feed stored in each feed storage box, Represents the Softplus activation function; represents a normal distribution, used to sample continuous actions; Indicates that each feed storage box is in the time step The feed delivery amount vector.

[0089] Here, the model concatenates the temporal decision results of the high-level output with the attention fusion features of the input state to form the input of the low-level policy network, which is processed by nonlinear transformation and multi-layer processing. Specifically, through two layers of ReLU activated hidden layers ( , ), and finally output two distribution parameters for "feed delivery amount": and , and Respectively represent the mean and standard deviation of the amount of investment. Finally, from the normal distribution Sampling , use the normal distribution parameterization strategy to ensure the continuity and differentiability of the output action, and the low-level strategy network makes the final decision output, that is, it finally decides how much feed should be put into each feed storage box at this moment.

[0090] Through the embodiments of the present application, the high-level strategy network and the low-level strategy network hierarchically process decision-making tasks. The high-level strategy first determines when to release (i.e., time decision), and on this basis, the low-level strategy further refines and determines the amount of feed to be released. This hierarchical structure reduces the complexity of the decision space, allowing the model to maintain efficient learning and adaptability in high-dimensional decision-making problems (dual variables of time and release amount). In addition, the reinforcement learning model learns through continuous interaction and trial and error, and dynamically adapts decisions to the real-time behavior and health status changes of animals by updating strategy parameters. When external conditions (such as animal activity levels, fluctuations in health indicators, changes in feed ingredients) change, the model can automatically adjust the timing and amount of feeding, and has better environmental adaptability and real-time feedback capabilities.

[0091] In some examples, the feeding plan generation model uses the PPO (Proximal Policy Optimization) algorithm as the policy optimization algorithm. Specifically, the current policy network interacts with the environment, collects data on state, action, reward, and next state, and trains according to the training process of the PPO algorithm. The reward changes and the stability of the policy update during the training process are monitored to ensure model convergence.

[0092] Compared with the traditional policy gradient method, the PPO algorithm limits the difference between the new and old strategies when updating the policy parameters (i.e., "proximal" constraints), so that the policy will not deviate from the optimal direction during the training process due to a single overly aggressive update, thereby reducing the instability and risk of policy collapse during the training process. In addition, in the generation of feeding plans, the experimental and sampling costs are high. PPO can improve data usage efficiency by using the same batch of sample data for multiple updates, which helps to obtain better policy performance with fewer actual experimental resources.

[0093] In some examples of the embodiments of the present application, the reward function of the feeding plan generation model is: , formula (20) In the formula, Represents the time step The reward value is used to evaluate the reward at the time step Take Action The advantages and disadvantages of Represents the time step The degree of improvement of health indicators after Indicates The feed storage box is in time step The amount of feed provided, Indicates The feed storage box is in time step The current feed balance, Indicates the total number of feed storage boxes in the feeding space. , and They respectively represent the weight coefficient of the health index improvement item, the weight coefficient of the total feed input penalty item, and the weight coefficient of the feed surplus change penalty item.

[0094] Here, the reward function adopts a multi-objective design. Through multi-objective optimization, it ensures that the model can maintain good decision-making performance under various conditions. The model shows stronger robustness and stability when facing different breeding environments and data changes.

[0095] Specifically, through the health indicator reward term in the reward function ( ), by real-time monitoring and quantifying changes in health indicators, dynamically adjusting feeding strategies, encouraging the model to prioritize feeding plans that can significantly improve animal health indicators, and providing personalized nutrition plans for animals in different health conditions, ensuring that animals get the nutritional support they need and promote healthy growth.

[0096] Through the feed delivery reward item in the reward function By quantifying the impact of feed input on rewards, the model is guided to find the most economical feeding plan, encouraging the model to reduce feed input as much as possible while meeting health needs, effectively reducing feed costs and improving feed utilization efficiency.

[0097] Controlling the reward term by changing the amount of food left in the reward function , by controlling the fluctuation of feed balance, penalizing drastic changes in feed balance, encouraging the model to maintain stable feed inventory, supporting more accurate feed management, and ensuring the stability of feed inventory balance.

[0098] The reward function provided in the embodiment of the present application clearly reflects key factors such as health index improvement, feed input and feed balance changes through quantifiable multi-dimensional reward items, making the decision basis of the system clear and transparent, enhancing the user's trust and acceptance of the system, and helping users understand and trust the decision-making process of the system. In addition, , and It is used to balance the weight coefficient of different reward items. It can balance the importance of different reward items and can be adjusted according to the actual needs of feeding. For example, when paying more attention to animal health, it can be appropriately increased. The value can be appropriately increased when more emphasis is placed on feed cost. When it is more desirable to maintain a balanced feed reserve, it can be appropriately increased. Therefore, by adjusting the weight coefficient, the system can flexibly switch the focus according to actual feeding needs to meet the needs of diversified feeding scenarios.

[0099] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity of description, they are all expressed as a series of actions combined, but those skilled in the art should be aware that the present application is not limited by the described order of actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application. In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0100] Figure 7 A structural block diagram of an example of a system for generating an animal feeding plan based on food intake monitoring data according to an embodiment of the present application is shown.

[0101] like Figure 7 As shown, the animal feeding plan generation system 700 based on feeding monitoring data includes a feeding data acquisition unit 710, a health data acquisition unit 720, a nutrition map construction unit 730, a map structure update unit 740 and a feeding plan generation unit 750.

[0102] The feeding data acquisition unit 710 is used to obtain the feeding timing data of each animal in the feeding space for each feed storage box in the feeding space; each feed storage box is used to put a corresponding type of feed, and the feeding timing data includes the feeding time and amount for the corresponding feed type.

[0103] The health data acquisition unit 720 is used to acquire the health index sampling data of each of the animals.

[0104] The nutrition map construction unit 730 is used to obtain the nutrients corresponding to the feed types of each feed storage box to construct a feed nutrition relationship map; the feed nutrition relationship map includes feed map nodes, nutrition map nodes and edge connections, the feed map nodes are used to indicate the corresponding feed types, the nutrition map nodes are used to indicate the corresponding types of nutrients, the edge weight of the feed-nutrient edge connection is defined by the association strength between the feed type and the nutrients, and the edge weight of the nutrient-nutrient edge connection is defined by the dependency relationship between the nutrients.

[0105] The graph structure updating unit 740 is used to aggregate and update the node features of each feed type graph node and the feature information of the adjacent nodes in the feed nutrition relationship graph based on the graph neural network to obtain the updated node features of each feed type graph node.

[0106] The feeding plan generation unit 750 is used to define the input state of the feeding plan generation model by combining the sampling data of each of the health indicators with the feeding timing data corresponding to each of the feed storage boxes, the updated node features and the current feed remaining, so as to determine the animal feeding plan corresponding to the target action through reinforcement learning; the animal feeding plan includes the planned feed delivery time and the corresponding planned feed delivery amount of each feed storage box.

[0107] In some embodiments, an embodiment of the present application provides a non-volatile computer-readable storage medium, in which one or more programs including execution instructions are stored, and the execution instructions can be read and executed by an electronic device (including but not limited to a computer, a server, or a network device, etc.) to execute any of the steps of the above-mentioned method for generating an animal feeding plan based on food intake monitoring data in the present application.

[0108] In some embodiments, the embodiments of the present application also provide a computer program product, which includes a computer program stored on a non-volatile computer-readable storage medium, and the computer program includes program instructions. When the program instructions are executed by a computer, the computer performs any step of the above-mentioned method for generating an animal feeding plan based on food intake monitoring data.

[0109] In some embodiments, the embodiments of the present application also provide an electronic device, comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the steps of a method for generating an animal feeding plan based on food intake monitoring data.

[0110] Figure 8 FIG. 1 is a schematic diagram of the hardware structure of an electronic device for executing a method for generating an animal feeding plan based on food intake monitoring data provided by another embodiment of the present application. Figure 8 As shown, the device includes: One or more processors 810 and memory 820, Figure 8 A processor 810 is taken as an example.

[0111] The device for executing the method for generating an animal feeding plan based on food intake monitoring data may further include: an input device 830 and an output device 840 .

[0112] The processor 810, the memory 820, the input device 830 and the output device 840 may be connected via a bus or other means. Figure 8 The example of connecting through bus is taken in the following.

[0113] The memory 820, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs, non-volatile computer executable programs and modules, such as program instructions / modules corresponding to the method for generating an animal feeding plan based on food intake monitoring data in the embodiment of the present application. The processor 810 executes various functional applications and data processing of the server by running the non-volatile software programs, instructions and modules stored in the memory 820, that is, the method for generating an animal feeding plan based on food intake monitoring data in the above method embodiment is implemented.

[0114] The memory 820 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the electronic device, etc. In addition, the memory 820 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some embodiments, the memory 820 may optionally include a memory remotely arranged relative to the processor 810, and these remote memories may be connected to the electronic device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0115] The input device 830 may receive input digital or character information and generate signals related to user settings and function control of the electronic device. The output device 840 may include a display device such as a display screen.

[0116] The one or more modules are stored in the memory 820, and when executed by the one or more processors 810, perform the animal feeding plan generation method based on food intake monitoring data in any of the above method embodiments.

[0117] The above-mentioned product can execute the method provided in the embodiment of the present application, and has the functional modules and beneficial effects corresponding to the execution method. For technical details not fully described in this embodiment, please refer to the method provided in the embodiment of the present application.

[0118] The electronic devices of the embodiments of the present application exist in various forms, including but not limited to: (1) Mobile communication equipment: This type of equipment is characterized by having mobile communication functions and its main purpose is to provide voice and data communications. This type of terminal includes: smart phones, multimedia phones, functional phones, and low-end phones.

[0119] (2) Ultra-mobile personal computer devices: These devices belong to the category of personal computers, have computing and processing functions, and generally also have mobile Internet access features. These terminals include: PDA, MID and UMPC devices, etc.

[0120] (3) Portable entertainment devices: These devices can display and play multimedia content. They include audio and video players, handheld game consoles, e-books, smart toys, and portable car navigation devices.

[0121] (4) Other onboard electronic devices with data interaction functions, such as on-board devices installed in vehicles.

[0122] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0123] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a general hardware platform, and of course, by hardware. Based on this understanding, the above technical solution, in essence, or the part that contributes to the relevant technology, can be embodied in the form of a software product, which can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiment.

[0124] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for generating an animal feeding plan based on feeding monitoring data, comprising: For each feed storage box in the feeding space, obtaining the feeding time series data of each animal in the feeding space for the feed storage box; each feed storage box is used to put a corresponding type of feed, and the feeding time series data includes the feeding time and amount for the corresponding feed type; Obtaining sampling data of health indicators of each of the animals; Obtaining the nutritional components corresponding to the feed types in each of the feed storage boxes to construct a feed nutritional relationship diagram; The feed nutrition relationship graph includes feed graph nodes, nutrition graph nodes and edge connections, wherein the feed graph nodes are used to indicate corresponding feed types, the nutrition graph nodes are used to indicate corresponding types of nutrients, the edge weights of feed-nutrient edge connections are defined by the strength of association between feed types and nutrients, and the edge weights of nutrients-nutrient edge connections are defined by the dependency relationship between nutrients; Based on the graph neural network, node features of each feed type graph node and feature information of adjacent nodes in the feed nutrition relationship graph are aggregated and updated to obtain updated node features of each feed type graph node; The input state of the feeding plan generation model is defined by combining the sampling data of each of the health indicators with the feeding timing data corresponding to each of the feed storage boxes, the updated node features and the current feed remaining, so as to determine the animal feeding plan corresponding to the target action through reinforcement learning; the animal feeding plan includes the planned feed delivery time and the corresponding planned feed delivery amount of each feed storage box.

2. The method according to claim 1, wherein: The eating behavior data includes at least one of the following: eating time, eating frequency and eating duration; the health index sampling data is used to indicate at least one type of health index among the following: body weight, heart rate, blood sugar and blood lipids.

3. The method according to claim 1, wherein: The feed storage box comprises an upper box body and a lower box body, wherein the upper box body is a feeding area provided with a feeding trough, and the lower box body is used to store animal excrement from the feeding area through the grid through holes at the bottom of the upper box body.

4. The method according to claim 3, wherein: The step of obtaining the feeding time series data of each animal in the feeding space with respect to the feed storage box comprises: Based on the RFID reader disposed inside the feeding trough sensing the RFID tag of each animal, the target animal in the feeding area is monitored in real time; Based on the weighing sensor arranged inside the feeding trough, the feed consumption amount and feed consumption time of the corresponding feed type are monitored in real time to determine the feeding timing data corresponding to the target animal.

5. The method according to claim 1, wherein: The obtaining of the nutritional components corresponding to the feed types in each feed storage box to construct a feed nutritional relationship diagram includes: Based on the feed nutrient database, determine the nutrient content of the feed type in each feed storage box and the bioavailability and nutrient metabolism experimental data corresponding to each nutrient; the nutrient metabolism experimental data includes experimental data on animal health and metabolic function related to nutrition; Based on the nutrient content of each feed type and the bioavailability and nutrient metabolism experimental data corresponding to each nutrient, a feed nutrient relationship diagram is constructed; wherein the node characteristics of the feed diagram node are defined by the feed name, feed classification and feed nutrient content of the corresponding feed type, and the node characteristics of the nutrient diagram node are defined by the nutrient index name, nutrient classification, nutrient metabolism experimental data and bioavailability of the corresponding nutrient; The edge weight of the feed-nutrient edge connection is calculated by the following formula: ， ， ， ， In the formula, Indicates i The feed type corresponding to each feed graph node, Indicates j The nutrients corresponding to each nutrition map node, Indicates that it is used to connect and The edge weights of the edge connections of the corresponding graph nodes; express middle The standardized content of express Important factors for animal health and metabolic function, express The bioavailability factor, represents the nonlinear parameter used to control the sensitivity of nutrient content, is the first metric global scaling factor; Indicates that it contains m An array of animal health indicators; Indicates a A health indicator, whose value is normalized to [0,1], and the larger the value, the better the health status; express The intake level and The correlation measure between them is in the range of [-1,1]. A positive value indicates that the intake is positively correlated with the health index, and a negative value indicates that the intake is negatively correlated with the health index. is the total number of experimental samples in the nutritional metabolism experimental data, including In the The intake measured in the samples was ,against In the The performance value measured in the samples is , For all experimental samples The mean of For all experimental samples The mean of Represents the correlation measure Mapped to the [0,1] interval corresponding to the positive score; Health indicators The weight of the indicator indicates the relative importance of the indicator to overall health; The edge weight of the nutrient-nutrient edge connection is calculated by the following formula: ， ， ， In the formula, Indicates k The nutrients corresponding to each nutrition map node, Indicates that it is used to connect and The edge weights of the edge connections of the corresponding graph nodes; express and The metabolic correlation between for The bioavailability factor, To control the parameters of the sensitivity of nutrient relationships to bioavailability, For the The nonlinear mapping parameters to the positive value space, is the second global scaling factor, represents the exponential function, express and Quantification of the synergistic or antagonistic effects of nutrients; Consider two nutrients and Combined intake levels and health indicators The correlation measure of and Indicates single intake and and health indicators The correlation measure of Health indicators Weight in determining relationships between nutrients.

6. The method according to claim 5, wherein: The feed classification includes at least one of the following: plant feed, animal feed, energy feed, fiber feed and protein feed; the nutrition classification includes at least one of the following: macronutrients, micronutrients, electrolytes and essential amino acids.

7. The method according to claim 5, wherein: The graph neural network adopts a graph attention network, which dynamically assigns weights to neighbor nodes through an attention mechanism and incorporates category information and edge weights to aggregate and update node features: For each pair of adjacent graph nodes, calculate the attention score: ， In the formula, Represents adjacent graph nodes and , and denote the learnable linear transformation matrix and attention parameter vector, respectively. represents a vector concatenation operation, and Represents graph nodes and graph nodes Node characteristics of is the information vector corresponding to the feed classification or nutrition classification, which is encoded using one-hot encoding; Represents the LeakyReLU activation function; Incorporate the edge weight into the attention score to form a weighted attention score: ， In the formula, represents the weighted attention score, Representing a graph node With graph nodes The edge weights of the edge connections between them; Normalize the attention scores to get the weighted class attention weights: ， In the formula, Representing a graph node With graph nodes The weighted category attention weight between Representing a graph node The set of neighboring graph nodes Any graph node in , Representing a graph node With graph nodes The weighted category attention weights between ; The node features of neighboring nodes are weighted aggregated based on the weighted category attention weights to update the node features: ， In the formula, represents a nonlinear activation function, Representing a graph node The updated node features.

8. The method according to claim 7, wherein: The feeding plan generation model adopts a hierarchical reinforcement learning model, which includes an attention fusion layer, a high-level strategy network and a low-level strategy network; The attention fusion layer is used to fuse each of the health indicator sampling data with the feeding time series data corresponding to each of the feed storage boxes, the updated node features and the current feed remaining amount through the self-attention mechanism to obtain the attention fusion feature: ， In the formula, represents the attention fusion feature, represents the self-attention mechanism, Represents the time step health indicators, Represents the time step Each feeding time series data, Represents the time step Each updated node feature of Represents the time step The current feed balance of each Represents the multi-layer perceptron processing function; Defining an input state based on the attention fusion feature to trigger the high-level strategy network to determine a corresponding feed plan delivery time; ， In the formula, represents the time step decision vector of feed delivery output by the high-level policy network, which is of length A vector where each element represents the probability of feeding at the corresponding time step; Represents the sofmax activation function; and Represent the weight matrix and bias vector of the high-level policy network respectively; The attention fusion feature and the planned feed delivery time are input into the low-level strategy network to determine the corresponding planned feed delivery amount: ， ， ， ， ， ， In the formula, represents the input feature vector of the low-level policy network, , and Represent the weight matrix, bias vector and output features of the first hidden layer of the low-level strategy network, respectively. , and Represent the weight matrix, bias vector and output features of the second hidden layer of the low-level strategy network, respectively. and Represent the weight matrix and bias vector of the output mean layer of the low-level strategy network, and Represent the weight matrix and bias vector of the output standard deviation layer of the low-level policy network respectively; represents the mean vector of the amount of feed stored in each feed storage box, represents the standard deviation vector of the amount of feed stored in each feed storage box, Represents the Softplus activation function; represents a normal distribution, used to sample continuous actions; Indicates that each feed storage box is in the time step The feed delivery amount vector.

9. The method according to claim 8, wherein: The reward function of the feeding plan generation model is: ， In the formula, Represents the time step The reward value is used to evaluate the reward at the time step Take Action The advantages and disadvantages of Represents the time step The degree of improvement of health indicators after Indicates The feed storage box is in time step The amount of feed provided, Indicates The feed storage box is in time step The current feed balance, Indicates the total number of feed storage boxes in the feeding space. , and They respectively represent the weight coefficient of the health index improvement item, the weight coefficient of the total feed input penalty item, and the weight coefficient of the feed surplus change penalty item.

10. A system for generating an animal feeding plan based on feeding monitoring data, comprising: A feeding data acquisition unit is used to acquire feeding time series data of each animal in the feeding space for each feed storage box in the feeding space; each feed storage box is used to put a corresponding type of feed, and the feeding time series data includes feeding time and feeding amount for the corresponding feed type; A health data acquisition unit, used to acquire health index sampling data of each of the animals; A nutrition map construction unit is used to obtain the nutritional components corresponding to the feed types of each feed storage box to construct a feed nutrition relationship map; the feed nutrition relationship map includes feed map nodes, nutrition map nodes and edge connections, the feed map nodes are used to indicate the corresponding feed types, the nutrition map nodes are used to indicate the nutritional components of the corresponding types, the edge weights of the feed-nutrient edge connections are defined by the association strength between the feed types and the nutritional components, and the edge weights of the nutrient-nutrient edge connections are defined by the dependency relationship between the nutritional components; A graph structure updating unit, configured to aggregate and update the node features of each feed type graph node and the feature information of adjacent nodes in the feed nutrition relationship graph based on a graph neural network, so as to obtain updated node features of each feed type graph node; A feeding plan generation unit is used to define the input state of a feeding plan generation model by combining the health indicator sampling data with the feeding time series data corresponding to each feed storage box, the updated node characteristics and the current feed remaining amount, so as to determine the animal feeding plan corresponding to the target action through reinforcement learning; the animal feeding plan includes the planned feed delivery time and the corresponding planned feed delivery amount of each feed storage box.

Images