Intelligent rationing system for pig house feed supply

Abstract

The application relates to the technical field of artificial intelligence, and discloses a pig house feed supply intelligent allocation system, which comprises the following steps: the pigs in a feeding area are identified by radio frequency identification tags, and the body weight, growth cycle and feeding behavior characteristics are extracted; a dynamic nutritional demand model of the pigs is constructed based on historical feeding records, health data and environmental factors, and the current feed demand is predicted by combining a multivariate regression algorithm; the feeding amount and feeding frequency of the corresponding trough are adjusted in real time according to the feed demand result; the power of a conveying motor and the pressure difference of a pipeline section are dynamically adjusted by using a variable frequency drive control technology; the pig feeding reaction and the feed supply execution state are collected, the updated pig allocation result is output by combining online feedback and historical data differences, and the application has the advantage of improving the feeding stability.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, specifically to an intelligent feed distribution system for pigsties. Background Technology

[0002] In existing automated feeding systems for pigsties, feeding valves are typically opened at fixed time intervals to achieve timed feeding. While some systems have the ability to distribute feed according to age or stage, uneven feed distribution at the end of the feed delivery pipeline is a common problem in actual operation. Due to differences in pipeline length and pressure, feed inlets closer to the main feed bin receive feed first, while feed inlets at the end experience delays or even temporary lack of feed. This results in pigs in the last pen not eating synchronously at the start of each feeding. This persistent imbalance accumulates over a long feeding cycle, leading to weight differences among pigs, affecting slaughter consistency, and increasing the difficulty of pig farming management. Therefore, it is essential to design an intelligent feed distribution system for pigsties that improves feeding stability. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention provides an intelligent feed distribution system for pigsties, which has the advantage of improving feeding stability and solves the problems mentioned in the background technology.

[0004] To achieve the aforementioned goal of improving feeding stability, the present invention provides the following technical solution: an intelligent feed distribution system for pigsties, comprising: Pig identification module: Identifies pigs in the feeding area using radio frequency identification tags, extracts their weight, growth cycle and feeding behavior characteristics, determines whether the pigs need to be fed, and if so, proceeds to the feeding parameter modeling module. Feeding parameter modeling module: Based on historical feeding records, health data and environmental factors, construct a dynamic nutritional requirement model for pigs, and combine it with multivariate regression algorithm to predict current feed demand, determine whether there is any abnormality in feed demand, and if so, enter the feed supply control module. Feed control module: Based on the feed demand results, adjust the feeding amount and feeding frequency of the corresponding feed trough in real time, and use a multi-point pressure monitoring device to determine whether there is an abnormal flow at the end of the feed pipeline. If so, it enters the pipeline balance control module. Pipeline balancing control module: Utilizes variable frequency drive control technology to dynamically adjust the power of the conveyor motor and the pressure difference of the pipeline section. Based on the operating data of the conveyor motor and the feeding pipeline after each feeding, it determines whether the operating data needs to be optimized. If so, it enters the data adaptive optimization module. Data Adaptive Optimization Module: Collects pigs' feeding responses and feeding execution status, combines online feedback with differences in historical data, dynamically corrects feeding strategy parameters through an adaptive learning algorithm, and outputs updated pig allocation results.

[0005] Preferably, the process for determining whether pigs need to be fed is as follows: Read the RFID tag information of the pigs and match it with the pigs' health records; Collect data on the feeding behavior characteristics of pigs within the current time period, including feeding frequency, feeding duration, and changes in feeding posture; By combining the pigs' current weight, growth stage, and historical feeding records, the feeding probability value of the pigs is calculated through a behavioral pattern analysis model. When the probability value of feeding a pig is higher than the set behavioral response threshold, and the current time interval exceeds the minimum feeding interval threshold, it is determined that the pig needs to enter the feeding process.

[0006] Preferably, the process of constructing a dynamic nutritional requirement model for pigs based on historical feeding records, health data, and environmental factors is as follows: Physiological and behavioral characteristics of pigs at each growth stage were collected, including stage-specific feed intake, weight change trends, physiological development indicators, and daily behavioral activity. Based on time series recording, physiological and behavioral characteristics are structured and organized into historical samples of individual pig nutrition intake, forming a time feature sequence for constructing a dynamic nutrition demand model. Environmental monitoring data were collected in the pigsty, and the impact coefficient of environmental monitoring data on pigs' feeding behavior and nutrient absorption efficiency was calculated by statistical regression. A dynamic nutritional requirement model that adapts to changes in pig status was constructed using a multivariate regression algorithm.

[0007] Preferably, the process of predicting current feed demand using a multivariate regression algorithm is as follows: Collect current weight, growth cycle and health data of pigs as basic input factors for individual status; Environmental monitoring data and pigsty operation data are input into the dynamic nutritional requirement model. The internal weighting mechanism of the dynamic nutritional requirement model is used to differentiate the basic input factors, and regularization is used to suppress the interference of redundant variables. Finally, the feed requirement of the pigs at the current moment is output.

[0008] Preferably, the process for determining whether there is an abnormality in feed demand is as follows: Using the current physiological and behavioral characteristics of pigs and environmental monitoring data as input, the current predicted feed requirement value is output through a dynamic nutritional requirement model. Measured feed supply data were extracted from the historical feeding records of pigs, and the average feed requirement and standard deviation of feed requirement of pigs were calculated. The current predicted feed demand is compared with the average feed demand of pigs, the relative deviation ratio is calculated, and the abnormal deviation score is obtained by using the standard deviation of feed demand as the normalization scale. By combining the current health status indicators and activity status of pigs, the abnormality bias score is corrected to obtain the abnormality confidence factor; If the abnormal confidence factor is greater than the set tolerance threshold, the current feed demand is determined to be abnormal. If the anomaly confidence factor is less than or equal to the set tolerance threshold, then the current feed demand is determined to be non-abnormal.

[0009] Preferably, the process for determining whether there is an abnormal flow rate at the end of the feed pipeline is as follows: By deploying multiple micro-pressure sensor nodes at the end of the feeding pipeline, data on pressure changes over time during the feeding process are collected in real time. Based on the target pressure variation range preset during the design of the feeding system, and combined with the actual pressure data collected by the current micro-pressure sensor, a pressure difference index is constructed, and the flow rate fluctuation rate per unit time is calculated. If the flow rate fluctuation rate is greater than the set time threshold, it is determined that there is an abnormal flow rate at the end of the material supply. If the flow rate fluctuation rate is less than or equal to the set time threshold, it is determined that there is no flow abnormality at the feeding end.

[0010] Preferably, the process of dynamically adjusting the power of the conveyor motor and the pressure difference of the pipeline section is as follows: Real-time data collection of feed end pressure, pressure differential at each intermediate section, and flow rate at the end generates a multi-dimensional parameter set reflecting the dynamic state of the feed, which constitutes the feed status parameters for the current cycle. It determines whether the pressure differential of the pipeline section is within the target stable range. If it is not within the target stable range, it triggers an adjustment command. The adjustment command is executed through variable frequency drive control technology to adjust the working frequency of the conveyor motor in real time and regulate the motor output power and speed; After each round of material feeding, the duration of each section of material feeding, the motor power change curve, and the differential pressure fluctuation amplitude are recorded as the material feeding response data for this round, and summarized to form a material feeding response database. A control strategy is constructed based on the material supply response database. A self-learning gain adjustment algorithm is introduced into the control strategy. The material supply status parameters, motor output and differential pressure change data of historical high-efficiency cycles are extracted from the material supply response database. Combined with the parameters of the current cycle, the key adjustment coefficients in the control strategy are dynamically corrected. Compare the motor power consumption, feeding time, and differential pressure stability indicators before and after adjustment under the same feeding conditions, and calculate the stability score of the feeding process.

[0011] Preferably, the process of determining whether the running data needs optimization is as follows: After each feeding task is completed, record the key operating data during the feeding process, including the power change curve of the conveyor motor, the duration of each feeding section, the pressure fluctuation of the pipeline, and the real-time feeding amount fed back by the end flow sensor. Collect behavioral response information of the pigs being fed in the current batch and calculate the matching degree of feeding behavior response; The performance deviation factor was calculated by comparing the current round of material supply operation data with the material supply parameter samples that showed high stability and high matching degree of material supply behavior response in the historical records. When the performance deviation factor exceeds the preset threshold, it is determined that there is room for optimization in the current material supply control parameters, and the material supply control parameters are sent to the data adaptive optimization module.

[0012] Preferably, the process of outputting the updated pig allocation results is as follows: Based on key operational data recorded during the feeding process, the effectiveness of the current feeding strategy is evaluated, and a feeding feedback dataset is generated. The recommended feed supply parameters output by the dynamic nutrition requirement model are compared with the actual feeding performance of each pig in the feed feedback dataset after this round of feeding to identify the deviation between the recommended feed supply parameters and the actual performance. Based on the deviation results of recommended feed supply parameters, the weights of pig characteristics and environmental adaptation parameters in the dynamic nutritional requirement model are fine-tuned using the built-in online learning mechanism. The output includes the pig number, recommended feed type, recommended feed amount, and feeding time for the pig allocation.

[0013] Compared with the prior art, the present invention provides an intelligent feed distribution system for pigsties, which has the following beneficial effects: 1. By using radio frequency identification (RFID) technology, pigs in the feeding area can be quickly and accurately identified. This not only efficiently obtains key data such as individual weight, growth cycle, and behavioral characteristics, but also dynamically assesses whether pigs meet their current feeding needs by combining real-time behavior judgment mechanisms. This avoids overfeeding or underfeeding, and improves feed utilization efficiency and the health level of the pig herd.

[0014] 2. By integrating historical feeding data, physiological health information, and environmental change factors, a refined individual dynamic nutritional demand model is constructed. A multivariate regression algorithm is introduced to improve the accuracy of feed demand prediction and personalized response capability. Potential abnormal feed demand can be identified in a timely manner, thereby enhancing the scientific nature and adaptability of feeding decisions.

[0015] 3. By adjusting the feeding parameters of the trough in real time, precise feeding on demand can be achieved. At the same time, by using a multi-point pressure monitoring device to sense the flow status at the end of the feeding pipeline, potential feeding anomalies can be detected in time, effectively ensuring the stability and continuity of the feeding process.

[0016] 4. Through frequency conversion drive and dynamic feedback control mechanism, the power of the conveyor motor and the pressure difference of each section of the pipeline are intelligently adjusted to make the feeding process more energy-efficient and efficient. At the same time, the system operation status is evaluated by combining the operation data after the feeding is completed, so as to ensure that the system maintains good stability and adjustment sensitivity under different workloads.

[0017] 5. By integrating individual feeding behavior feedback and pipeline feeding execution data, and utilizing online learning algorithms to continuously optimize feeding strategies, the system dynamically outputs individualized feed type, feeding amount, and feeding time adjustment results, thereby achieving iterative upgrades of the intelligent system and improving the accuracy and resource utilization efficiency of long-term operation. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the structure of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Example 1: Please refer to Figure 1 As shown in the embodiment of the present invention, a smart feed distribution system for pigsties includes: Pig identification module: Identifies pigs in the feeding area using radio frequency identification tags, extracts their weight, growth cycle and feeding behavior characteristics, determines whether the pigs need to be fed, and if so, proceeds to the feeding parameter modeling module.

[0021] The process by which the pig identification module determines whether a pig needs to be fed is as follows: The system reads the RFID tag information of the pigs and matches it with their health records. It automatically obtains the electronic ear tag identification code of the pigs entering the feeding area through a high-frequency RFID reader deployed at the entrance of the feeding area. The identification code is compared with the health records in the local database at high speed to extract the corresponding pig record data, including date of birth, vaccination history, disease records, growth curves and history of abnormal feeding. Collect feeding behavior data of pigs within the current time period, including feeding frequency, feeding duration and feeding posture changes; record the time pigs stay in front of the feed trough, head movement trajectory and feeding posture changes in real time; use image recognition technology to analyze feeding action patterns, and combine pressure change data to identify the start and end times of feeding, and then calculate feeding frequency and duration. By combining the pig's current weight, growth stage, and historical feeding records, a behavioral pattern analysis model is used to calculate the feeding probability value of the pig; real-time weight data is obtained, and the current stage of the pig, such as suckling, nursery, fattening, or growing, is determined; historical feeding records and reference feeding models for the corresponding growth stages are called up, and the current feeding behavior characteristics are input into the behavioral pattern analysis model built based on deep decision trees or lightweight neural networks; the feeding probability value at the current moment is output, reflecting the degree of matching between the pig's willingness to accept feed and its physiological needs; When the pig's feeding probability value is higher than the set behavioral response threshold and the current time interval exceeds the minimum feeding interval threshold, it is determined that the pig needs to enter the feeding process; two threshold control variables are set: behavioral response threshold and minimum feeding interval threshold; when the feeding probability value output by the behavior model is higher than the set threshold and the time since the last feeding exceeds the minimum interval, it is determined that the pig has generated an effective feeding demand.

[0022] Feeding parameter modeling module: Based on historical feeding records, health data and environmental factors, a dynamic nutritional requirement model for pigs is constructed, and a multivariate regression algorithm is used to predict the current feed requirement and determine whether there is any abnormality in the feed requirement. If so, the feed supply control module is entered.

[0023] The process of constructing a dynamic nutritional requirement model for pigs based on historical feeding records, health data, and environmental factors is as follows: Physiological and behavioral characteristics of pigs at various growth stages were collected, including stage-specific feed intake, weight change trends, physiological development indicators, and daily behavioral activity. The data collected included stage-specific feed intake, weight change trends, and physiological development indicators. Behavioral activity data was acquired through a visual monitoring system or inertial sensors, including standing time, activity path, and feeding frequency. Each data point was associated with a timestamp and a unique pig number, forming a set of physiological and behavioral characteristics of the pig at a specific growth stage. Based on time series recording, physiological and behavioral characteristics are structured and organized into historical samples of individual pig nutrition intake, forming a time feature sequence for constructing a dynamic nutrition demand model. Data is aggregated using a fixed time window to construct continuous time series samples. Each time point sample includes structured features such as feed intake, weight change, and behavioral indicators. A two-dimensional sample matrix arranged in chronological order is formed to explore the correlation trend between pig nutrition intake and physiological and behavioral changes. Environmental monitoring data in pigsties were collected, and the impact coefficient of environmental monitoring data on pigs' feeding behavior and nutrient absorption efficiency was calculated by statistical regression. Daily temperature, humidity, ammonia concentration, light intensity and density of pigsties were obtained from environmental sensors in the pigsties. The sensitivity score of environmental factors to feeding behavior was obtained by calculating the correlation between environmental indicators and historical changes in feed intake. A dynamic nutritional requirement model that adapts to changes in pig status is constructed using a multivariate regression algorithm. An input variable matrix is ​​established, including the physiological characteristics and environmental indicators of pigs at the current time point. Using historical feed intake data as the target variable, the model is constructed using methods such as multiple linear regression or ridge regression, and the estimated nutritional requirement value at the current time point is output.

[0024] The process of predicting current feed demand using a multivariate regression algorithm in the feeding parameter modeling module is as follows: The system collects current weight, growth cycle, and health data of pigs as basic input factors for individual status; obtains pig identification information through a pig identification module and retrieves current weight from real-time weighing equipment or historical weight database; reads pig growth cycle information to reflect the pig's development stage; obtains key physiological health indicators through physiological sensing equipment or health monitoring system in real time; and summarizes and formats the above data to form a multi-dimensional input vector of individual pig status. Environmental monitoring data and pigsty operation data are input into a dynamic nutritional requirement model (NMR) to differentiate the basic input factors using the model's internal weighting mechanism. Regularization is then used to suppress redundant variables, ultimately outputting the feed requirement for each pig at the current moment. Environmental monitoring data and pigsty operation data are collected and combined with individual status inputs to form a complete set of input variables, which is then fed into a multi-factor regression model. Based on the weighting coefficients derived during training, different importance weights are assigned to different input variables within the model, highlighting the role of key factors, and ultimately outputting the feed requirement for each pig at the current moment. After calculation, the model generates a predicted feed requirement for the pig at the current time point, representing a reasonable feed supply. This predicted value serves as a reference standard for subsequent feed control modules, used to dynamically adjust the feed amount and frequency.

[0025] The process for determining whether there are abnormalities in feed demand in the feeding parameter modeling module is as follows: The model takes the current physiological and behavioral characteristics of pigs and environmental monitoring data as input, and outputs the current predicted feed requirement value through a dynamic nutrition requirement model. It uses the current weight, growth cycle, physiological health indicators of pigs, and data such as temperature, humidity and ammonia concentration collected by environmental sensors as input into the dynamic nutrition requirement model. After the model is calculated, it outputs the predicted theoretical feed requirement of the pig at the current time point, reflecting a reasonable feed supply standard. Measured feed supply data are extracted from the historical feeding records of pigs, and the average feed requirement and standard deviation of feed requirement are calculated. The historical feeding database of pigs is queried to select feeding records with similar growth stages and similar health indicators. The selected historical data are statistically analyzed to calculate the average feed requirement and standard deviation of the requirement, which serve as a reference for the normal requirement range of the pig. The current predicted feed demand is compared with the average feed demand of pigs to calculate the relative deviation ratio. The abnormal deviation score is obtained by normalizing the feed demand standard deviation. The difference between the current predicted feed demand and the historical average is calculated: Deviation = Current predicted value - Historical average value. The deviation is divided by the historical standard deviation to obtain the standardized deviation score, which reflects the degree to which the current demand deviates from the normal level. The larger the score, the more the current predicted demand deviates from the historical normal range, and the higher the potential abnormality. By combining the current health status indicators and activity status of pigs, the abnormal deviation score is corrected to obtain the abnormal confidence factor; the real-time health status indicators and behavioral activity indicators of pigs are obtained; the abnormal deviation score is weighted and adjusted according to the health and behavioral status, for example, the confidence of the abnormal score is increased when the health is abnormal; finally, the corrected abnormal confidence factor is output as a reference for judging abnormal feed demand. If the abnormal confidence factor is greater than the set tolerance threshold, the current feed demand is determined to be abnormal. If the anomaly confidence factor is less than or equal to the set tolerance threshold, then the current feed demand is determined to be non-abnormal.

[0026] Feed control module: Based on the feed demand results, adjust the feeding amount and feeding frequency of the corresponding feed trough in real time, and use a multi-point pressure monitoring device to determine whether there is an abnormal flow at the end of the feed pipeline. If so, it enters the pipeline balancing control module.

[0027] The process by which the material supply control module determines whether there is an abnormal flow rate at the end of the material supply pipeline is as follows: Multiple micro-pressure sensor nodes are deployed at the end of the feeding pipeline to collect data on pressure changes over time during the feeding process in real time; multiple micro-pressure sensors are evenly installed at the end of the feeding pipeline to ensure coverage of key monitoring points; the sensors collect the pressure values ​​in the pipeline in real time during the feeding process, and the sampling frequency meets the requirements for dynamic change detection. Based on the target pressure variation range preset in the design of the feeding system, and combined with the actual pressure data collected by the current micro-pressure sensor, a pressure difference index is constructed, and the flow rate fluctuation rate per unit time is calculated. According to the design specifications of the feeding system, a target pressure variation range is set as a reference range for the pressure during normal operation. By comparing the real-time collected actual pressure data with the target pressure range, the pressure difference at each time point is calculated to form a pressure difference index. The change amplitude of the pressure difference per unit time is statistically analyzed and converted into a flow rate fluctuation rate index to evaluate the stability of the feeding flow rate. If the flow rate fluctuation rate is greater than the set time threshold, it is determined that there is an abnormal flow rate at the end of the material supply. If the flow rate fluctuation rate is less than or equal to the set time threshold, it is determined that there is no flow abnormality at the feeding end.

[0028] Understandably, the purpose of determining whether there is an abnormal flow rate at the end of the material supply pipeline is: Function 1: To promptly detect abnormalities such as blockages, leaks, or air blockages during the feeding process, ensuring that feed can be smoothly delivered to the terminal feed trough according to the set flow rate, and ensuring continuous and stable feeding for pigs. Function 2: Provides data support for the dynamic control of the feeding system, making it easy for the system to quickly switch to the pipeline balancing control module after detecting anomalies, and adjust the motor power and valve opening to maintain the overall pressure balance and energy efficiency optimization of the feeding system.

[0029] The technical solution of this embodiment is as follows: By dynamically modeling the pigs' feeding behavior, weight changes, and nutritional needs, the feeding amount and frequency of the corresponding feed troughs are adjusted in real time. Simultaneously, multi-point pressure monitoring devices deployed at the end of the feeding pipeline can collect pressure change data in real time to determine if there are any flow anomalies. If an anomaly is detected, the system automatically switches to the pipeline balancing control module to perform closed-loop regulation of the conveyor motor power and pipeline valve opening. This achieves precise and efficient feeding management for pigs, significantly improving feed utilization and the health level of the pig herd. Furthermore, through multi-point pressure monitoring and adaptive flow control, problems such as feed blockage and flow rate imbalance are effectively prevented, ensuring the continuity and stability of the feeding process, thereby improving the energy efficiency and reliability of the entire intelligent feeding system.

[0030] Example 2: Figure 1 As shown, an intelligent feed distribution system for pigsties also includes the following modules: Pipeline balancing control module: Utilizing variable frequency drive control technology, it dynamically adjusts the power of the conveyor motor and the pressure difference of the pipeline section. Based on the operating data of the conveyor motor and the feeding pipeline after each feeding, it determines whether the operating data needs optimization. If so, it enters the data adaptive optimization module.

[0031] The process of dynamically adjusting the power of the conveyor motor and the pressure difference of the pipeline section in the pipeline balancing control module is as follows: Real-time data collection of feed-end pressure, differential pressure at each intermediate section, and flow rate at the end generates a multi-dimensional parameter set reflecting the dynamic state of the feeding process, forming the feeding status parameters for the current cycle. It determines whether the differential pressure at each pipeline section is within the target stable range; if not, it triggers an adjustment command. An embedded sensor network is used to collect real-time data from the feed-end pressure sensor, differential pressure sensors at each intermediate pipeline section, and the flow meter output during the feeding process. The data collected by each sensor is synchronously integrated and timestamped to form continuous multi-dimensional data points. After filtering and outlier removal of the collected data, a multi-dimensional feeding status parameter set for the current feeding cycle is constructed according to the time series. The adjustment command is executed through variable frequency drive control technology, which adjusts the working frequency of the conveyor motor in real time and regulates the motor output power and speed. During the material feeding execution phase, the variable frequency drive control module is invoked to dynamically adjust the working frequency of the conveyor motor according to the current material feeding status parameters. The motor frequency is adjusted in real time to synchronously control its speed and output power, ensuring that the material feeding speed matches the target flow rate. The execution unit in the parallel control valve module is used to synchronously adjust the valve opening of each pipe section according to the pressure difference change trend of each section. After each feeding cycle, the feeding duration, motor power change curve, and differential pressure fluctuation amplitude of each segment are recorded as the feeding response data for this cycle, and summarized to form a feeding response database. After each feeding cycle, the key adjustment indicators for that cycle are automatically recorded, including the feeding duration of each segment; the power change curve of the conveyor motor; and the differential pressure fluctuation amplitude of each segment. The feeding response efficiency of this cycle is compared with the adjustment effect samples of historical high-efficiency cycles, and feature classification is performed based on statistical learning. An adjustment response database containing a triplet of input state, control action, and feedback effect is constructed. A control strategy is constructed based on a material supply response database. A self-learning gain adjustment algorithm is introduced into the control strategy to extract historical high-efficiency cycle material supply status parameters, motor output, and differential pressure change data from the database. Combined with the current cycle parameters, key adjustment coefficients in the control strategy are dynamically corrected. A self-learning mechanism is introduced to compare the deviation of the current cycle's material supply response from the historical best response in the database. The gain adjustment algorithm automatically fine-tunes key parameters in the control strategy, such as the gain coefficient in the motor power regulation function and the adjustment ratio corresponding to the valve execution speed. Through continuous iteration and feedback correction, the accuracy of motor power output response and valve linkage stability are improved, optimizing the dynamic adaptability of the overall material supply process. Compare the motor power consumption, feeding time, and differential pressure stability indicators before and after adjustment under the same feeding conditions, and calculate the stability score of the feeding process.

[0032] The process by which the pipeline balancing module determines whether the operating data needs optimization is as follows: After each feeding task is completed, key operational data during the feeding process are recorded, including the power change curve of the conveyor motor, the feeding duration of each section, the pressure fluctuation amplitude of the pipeline, and the real-time feeding amount fed back by the end flow sensor; the power change of the conveyor motor is monitored and recorded in real time throughout the feeding process to form a power change curve; the feeding duration of each pipeline section is recorded to reflect the time distribution of material conveying; pressure fluctuation data of each node in the pipeline is collected, the pressure fluctuation amplitude is calculated, and the pressure stability during the conveying process is evaluated; real-time feeding amount feedback is obtained through the end flow sensor to verify whether the actual feeding has achieved the expected target. Collect behavioral response information of the pigs being fed in the current batch and calculate the matching degree of feed feeding behavior; use a pig behavior monitoring system to collect indicators such as feeding duration, feeding frequency, and post-feeding behavioral activity of pigs during the feeding process; analyze the collected behavioral data to evaluate the pigs' acceptance of the current feed and the effect of their response; compare the behavioral response indicators with the expected feeding model to calculate the matching degree of behavioral response, reflecting the suitability and effectiveness of the feed supply; The data from this round of material supply operation is compared with historical data on material supply parameters that demonstrate high stability and high matching degree of material supply behavior response. The performance deviation factor is calculated. Typical sample data with excellent material supply parameters, stable system operation, and high matching degree of behavior response are selected from the historical database. Key indicators such as the current round's material supply power curve, material supply duration, pressure fluctuation, and flow feedback are compared with typical samples item by item. The deviation between the current data and historical samples is calculated using statistical analysis methods to obtain the comprehensive performance deviation factor and quantify the degree of deviation of the current material supply parameters. When the performance deviation factor exceeds the preset threshold, it is determined that there is room for optimization in the current material supply control parameters, and the material supply control parameters are sent to the data adaptive optimization module.

[0033] Data adaptive optimization module: Collects pigs' feeding response and feeding execution status, combines online feedback with historical data differences, and outputs updated pig allocation results.

[0034] The process of outputting the updated pig allocation results in the data adaptive optimization module is as follows: Based on key operational indicators recorded during the feeding process, the effectiveness of the current feeding strategy is evaluated, and a feeding feedback dataset is generated. Key operational indicators for this round of feeding are automatically collected, including feeding time, feed amount, and conveying efficiency. The recommended feed supply parameters output by the dynamic nutritional requirement model are compared with the actual feeding performance of each pig in the feed feedback dataset after this round of feeding to identify the deviation between the recommended feed supply parameters and the actual performance; the feed requirement value predicted by the dynamic nutritional requirement model for the corresponding pig is retrieved; the difference between the actual feeding performance and the predicted value of the dynamic nutritional model is compared; the matching degree of the feeding parameters is evaluated to determine whether there is a deviation of insufficient or excessive feeding. Based on the deviation results of recommended feed supply parameters, the weights of pig characteristics and environmental adaptation parameters in the dynamic nutritional requirement model are fine-tuned using the built-in online learning mechanism; the model parameters are adjusted using an incremental learning algorithm based on feedback data and deviation results; and the weights of individual pig characteristics and environmental impact parameters are refined and updated. The output includes the pig number, recommended feed type, recommended feed amount, and feeding time period for pig allocation. The output information includes: Pig ID: A unique identifier for each pig; Recommended feed type: Feed category determined based on nutritional requirements and feed formulation; Recommended feed amount: The precise weight or volume of feed determined based on the current needs of the pigs; Feeding time: Recommended start and end times for feeding to ensure optimal matching between feed delivery and feeding behavior; The rationing results are sent to automatic feeding equipment or feeders through the system to achieve precise feeding.

[0035] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0036] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A smart feed distribution system for pigsties, characterized in that, include: Pig identification module: Identifies pigs in the feeding area using radio frequency identification tags, extracts their weight, growth cycle and feeding behavior characteristics, determines whether the pigs need to be fed, and if so, proceeds to the feeding parameter modeling module. Feeding parameter modeling module: Based on historical feeding records, health data and environmental factors, construct a dynamic nutritional requirement model for pigs, and combine it with multivariate regression algorithm to predict current feed demand, determine whether there is any abnormality in feed demand, and if so, enter the feed supply control module. Feed control module: Based on the feed demand results, adjust the feeding amount and feeding frequency of the corresponding feed trough in real time, and use a multi-point pressure monitoring device to determine whether there is an abnormal flow at the end of the feed pipeline. If so, it enters the pipeline balance control module. Pipeline balancing control module: Utilizes variable frequency drive control technology to dynamically adjust the power of the conveyor motor and the pressure difference of the pipeline section. Based on the operating data of the conveyor motor and the feeding pipeline after each feeding, it determines whether the operating data needs to be optimized. If so, it enters the data adaptive optimization module. Data adaptive optimization module: Collects pigs' feeding response and feeding execution status, combines online feedback with historical data differences, and outputs updated pig allocation results.

2. The intelligent feed distribution system for pigsties according to claim 1, characterized in that, The process of determining whether pigs need to be fed is as follows: Read the RFID tag information of the pigs and match it with the pigs' health records; Collect data on the feeding behavior characteristics of pigs within the current time period, including feeding frequency, feeding duration, and changes in feeding posture; By combining the pigs' current weight, growth stage, and historical feeding records, the feeding probability value of the pigs is calculated through a behavioral pattern analysis model. When the probability value of feeding a pig is higher than the set behavioral response threshold, and the current time interval exceeds the minimum feeding interval threshold, it is determined that the pig needs to enter the feeding process.

3. The intelligent feed distribution system for pigsties according to claim 2, characterized in that, The process of constructing a dynamic nutritional requirement model for pigs based on historical feeding records, health data, and environmental factors is as follows: Physiological and behavioral characteristics of pigs at each growth stage were collected, including stage-specific feed intake, weight change trends, physiological development indicators, and daily behavioral activity. Based on time series recording, physiological and behavioral characteristics are structured and organized into historical samples of individual pig nutrition intake, forming a time feature sequence for constructing a dynamic nutrition demand model. Environmental monitoring data were collected in the pigsty, and the impact coefficient of environmental monitoring data on pigs' feeding behavior and nutrient absorption efficiency was calculated by statistical regression. A dynamic nutritional requirement model that adapts to changes in pig status was constructed using a multivariate regression algorithm.

4. The intelligent feed distribution system for pigsties according to claim 3, characterized in that, The process of predicting current feed demand using multivariate regression algorithms is as follows: Collect current weight, growth cycle and health data of pigs as basic input factors for individual status; Environmental monitoring data and pigsty operation data are input into the dynamic nutritional requirement model. The internal weighting mechanism of the dynamic nutritional requirement model is used to differentiate the basic input factors, and regularization is used to suppress the interference of redundant variables. Finally, the feed requirement of the pigs at the current moment is output.

5. The intelligent feed distribution system for pigsties according to claim 4, characterized in that, The process for determining whether there are abnormalities in feed demand is as follows: Using the current physiological and behavioral characteristics of pigs and environmental monitoring data as input, the current predicted feed requirement value is output through a dynamic nutritional requirement model. Measured feed supply data were extracted from the historical feeding records of pigs, and the average feed requirement and standard deviation of feed requirement of pigs were calculated. The current predicted feed demand is compared with the average feed demand of pigs, the relative deviation ratio is calculated, and the abnormal deviation score is obtained by using the standard deviation of feed demand as the normalization scale. By combining the current health status indicators and activity status of pigs, the abnormality bias score is corrected to obtain the abnormality confidence factor; If the abnormal confidence factor is greater than the set tolerance threshold, the current feed demand is determined to be abnormal. If the anomaly confidence factor is less than or equal to the set tolerance threshold, then the current feed demand is determined to be non-abnormal.

6. The intelligent feed distribution system for pigsties according to claim 5, characterized in that, The process for determining whether there is an abnormal flow rate at the end of the feed pipeline is as follows: By deploying multiple micro-pressure sensor nodes at the end of the feeding pipeline, data on pressure changes over time during the feeding process are collected in real time. Based on the target pressure variation range preset during the design of the feeding system, and combined with the actual pressure data collected by the current micro-pressure sensor, a pressure difference index is constructed, and the flow rate fluctuation rate per unit time is calculated. If the flow rate fluctuation rate is greater than the set time threshold, it is determined that there is an abnormal flow rate at the end of the material supply. If the flow rate fluctuation rate is less than or equal to the set time threshold, it is determined that there is no flow abnormality at the feeding end.

7. The intelligent feed distribution system for pigsties according to claim 6, characterized in that, The process of dynamically adjusting the power of the conveyor motor and the pressure difference of the pipeline section is as follows: Real-time data collection of feed end pressure, pressure differential at each intermediate section, and flow rate at the end generates a multi-dimensional parameter set reflecting the dynamic state of the feed, which constitutes the feed status parameters for the current cycle. It determines whether the pressure differential of the pipeline section is within the target stable range. If it is not within the target stable range, it triggers an adjustment command. The adjustment command is executed through variable frequency drive control technology to adjust the working frequency of the conveyor motor in real time and regulate the motor output power and speed; After each round of material feeding, the duration of each section of material feeding, the motor power change curve, and the differential pressure fluctuation amplitude are recorded as the material feeding response data for this round, and summarized to form a material feeding response database. A control strategy is constructed based on the material supply response database. A self-learning gain adjustment algorithm is introduced into the control strategy. The material supply status parameters, motor output and differential pressure change data of historical high-efficiency cycles are extracted from the material supply response database. Combined with the parameters of the current cycle, the key adjustment coefficients in the control strategy are dynamically corrected. Compare the motor power consumption, feeding time, and differential pressure stability indicators before and after adjustment under the same feeding conditions, and calculate the stability score of the feeding process.

8. The intelligent feed distribution system for pigsties according to claim 7, characterized in that, The process of determining whether the running data needs optimization is as follows: After each feeding task is completed, record the key operating data during the feeding process, including the power change curve of the conveyor motor, the duration of each feeding section, the pressure fluctuation of the pipeline, and the real-time feeding amount fed back by the end flow sensor. Collect behavioral response information of the pigs being fed in the current batch and calculate the matching degree of feeding behavior response; The performance deviation factor was calculated by comparing the current round of material supply operation data with the material supply parameter samples that showed high stability and high matching degree of material supply behavior response in the historical records. When the performance deviation factor exceeds the preset threshold, it is determined that there is room for optimization in the current material feeding operation data, and the material feeding operation data is sent to the data adaptive optimization module.

9. The intelligent feed distribution system for pigsties according to claim 8, characterized in that, The process of outputting the updated pig allocation results is as follows: Based on key operational data recorded during the feeding process, the effectiveness of the current feeding strategy is evaluated, and a feeding feedback dataset is generated. The recommended feed supply parameters output by the dynamic nutrition requirement model are compared with the actual feeding performance of each pig in the feed feedback dataset after this round of feeding to identify the deviation between the recommended feed supply parameters and the actual performance. Based on the deviation results of recommended feed supply parameters, the weights of pig characteristics and environmental adaptation parameters in the dynamic nutritional requirement model are fine-tuned using the built-in online learning mechanism. The output includes the pig number, recommended feed type, recommended feed amount, and feeding time for the pig allocation.

Images