Veterinary drug feed accurate feeding and metering control and regulation system

Abstract

The invention discloses a veterinary drug feed accurate feeding metering control adjusting system, which relates to the technical field of livestock breeding and comprises seven modules, namely a material storage and conveying module, a multi-dimensional material characteristic dynamic calibration module and an accurate metering execution module. The core innovation lies in that the multi-dimensional material characteristic dynamic calibration module collects humidity, particle size distribution and fluidity parameters in real time, and a metering coefficient is corrected through a dynamic calibration model and iterative optimization, so that the problem of metering deviation caused by material characteristic fluctuation is solved; the feeding behavior-feeding amount dynamic coupling adjustment module calculates and couples material parameters in stages based on animal feeding speed and uniformity data to adjust feeding amount and speed, and matches real-time feeding requirements. The system achieves closed-loop control through module cooperation, improves the metering precision to + / -0.1 g, reduces the feed waste rate, guarantees the medication safety, optimizes the breeding benefits, and is suitable for a large-scale breeding scene.

Description

Technical Field

[0001] This invention relates to the field of animal husbandry, specifically to a precise metering and control system for the administration of veterinary drugs and feed. Background Technology

[0002] In large-scale livestock farming, the precise administration of veterinary drugs and feed is directly related to animal health, farming costs, and product quality. Existing veterinary drug and feed administration systems mostly employ fixed-weight metering or timed dispensing methods, relying on a single weight sensor to collect data and using a simple PID algorithm for adjustment.

[0003] However, in practical applications, changes in the moisture content of veterinary drug feed can cause the material to clump or become loose, and differences in particle size distribution can affect the flowability of the material. These fluctuations in characteristics directly reduce the accuracy of metering. At the same time, existing systems do not consider the differences in real-time feeding behavior of animals. Fixed dosages are prone to problems such as some animals not eating enough and others eating too much and wasting food. Furthermore, they cannot be dynamically adjusted according to the uniformity of feeding in the group, resulting in unstable drug efficacy and increased breeding costs. Therefore, there is an urgent need for a precise metering and control system that can adapt to fluctuations in material characteristics and match the real-time feeding needs of animals. Summary of the Invention

[0004] The purpose of this invention is to provide a precise metering and control system for veterinary drug and feed administration, in order to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a veterinary drug and feed precision dispensing and metering control system, comprising a material storage and conveying module, a multi-dimensional material characteristic dynamic calibration module, a precision metering execution module, an animal state sensing module, a feeding behavior-dispensing amount dynamic coupling adjustment module, a central control module, and a dispensing terminal module; the material storage and conveying module includes partitioned storage compartments (for storing veterinary drug premixes and basic feeds respectively) and a screw conveyor device. The inner wall of the storage compartment is provided with an anti-stick coating to prevent material clumping and residue. The material storage and conveying module receives the feeding instructions from the central control module and drives the screw conveyor device through a variable frequency motor to deliver the material to the precision metering execution module according to a preset ratio. The conveying speed is linked to the metering rhythm of the metering execution module. The multi-dimensional material property dynamic calibration module includes a humidity sensor, a laser particle size analyzer, a flowability testing unit, and a calibration model calculation unit. The multi-dimensional material property dynamic calibration module is used to collect material property parameters such as humidity, particle size distribution, and flowability of veterinary drug feed in real time, establish a dynamic calibration model and correct the measurement coefficients, and establish a dynamic mapping relationship between material property parameters and measurement coefficients. The precise metering execution module is a combination structure of a volumetric metering cylinder and an electromagnetic flow valve. A pressure sensor is installed on the inner wall of the metering cylinder to assist in verifying the material filling amount. The precise metering execution module receives the basic metering command and the calibrated metering correction coefficient issued by the central control module, which is the actual metering feedback value. It adjusts the volume setting of the metering cylinder through a stepper motor and combines the opening and closing speed of the electromagnetic flow valve to achieve precise metering. The animal status perception module includes an infrared thermal imaging camera, a weight-sensing feeding trough, and a sound sensor deployed in the breeding pens, covering all feeding areas. The infrared thermal imaging camera captures the animals' feeding posture in real time and identifies the number of individuals feeding. The weight-sensing feeding trough collects the weight change of the material in the trough every 200ms to calculate the feeding speed (g / s) of a single animal. The sound sensor collects the chewing sounds and squabbles of the animals while feeding to help determine the uniformity of feeding in the group (no squabbles indicate high uniformity, while frequent squabbles indicate low uniformity). The collected data on the number of individuals feeding, feeding speed, and group feeding uniformity, i.e., the animal status data, are uploaded to the central control module in real time.

[0006] The dynamic coupling adjustment module for feeding behavior and feeding amount includes a feeding behavior analysis unit, a feeding parameter coupling unit, and an adjustment instruction generation unit. The dynamic coupling adjustment module for feeding behavior and feeding amount is used to achieve dynamic matching between feeding amount and the animal's real-time feeding needs, establish the coupling relationship between feeding behavior parameters and feeding parameters, and dynamically adjust the feeding speed and single feeding amount through the coupling algorithm. The central control module receives data uploaded by each module, forwards material characteristic parameters to the multi-dimensional material characteristic dynamic calibration module, and forwards animal status data to the feeding behavior-feeding amount dynamic coupling adjustment module; it receives the output results of the multi-dimensional material characteristic dynamic calibration module and the feeding behavior-feeding amount dynamic coupling adjustment module, integrates them to generate metering instructions and feeding instructions, and sends them to the precision metering execution module and the feeding terminal module respectively; at the same time, it monitors the working status of each module and triggers an alarm when an abnormality occurs. The dispensing terminal module includes a rotatable dispensing nozzle and a position adjustment device, with each breeding pen corresponding to an independent dispensing terminal. The dispensing terminal module receives dispensing instructions from the central control module, aligns the position adjustment device with the feeding trough area, and evenly dispenses the precisely measured veterinary drug feed into the feeding trough according to the set dispensing speed, dispensing amount, and dispensing frequency. After dispensing, it sends a feedback signal to the central control module to complete one dispensing cycle.

[0007] Preferably, the multi-dimensional material property dynamic calibration module is implemented as follows: Step 1: Real-time Parameter Acquisition: Within the transfer silo along the material conveying path, a humidity sensor, laser particle size analyzer, and flowability testing unit are integrated to continuously acquire three core parameters. The humidity sensor employs a high-frequency capacitive sensor, acquiring the material humidity value every 50ms (accuracy ±0.5%RH). The laser particle size analyzer measures the material particle size distribution using a scattering method and outputs... Three characteristic particle size values ​​are collected in 100ms; the flowability test unit calculates the flowability coefficient (range 0-1, the larger the value, the better the flowability) by measuring the sliding speed and angle of repose of the material in the inclined pipe, with a collection period of 200ms. Step 2, Data Preprocessing: Outliers in the collected humidity, particle size distribution, and flowability coefficient data are removed using the 3σ criterion, and the data is smoothed using the moving average method (window size of 5 collection cycles) to avoid calibration deviations caused by instantaneous fluctuations, thus obtaining the preprocessed parameters; Step 3, Dynamic Calibration Model Construction and Iteration: Based on the preprocessed parameters, the preset initial calibration model is invoked (this model is established through a large number of orthogonal experiments, using the measurement coefficients corresponding to the standard material characteristic parameters as a benchmark); the three parameters collected in real time are input into the model, and the measurement correction coefficients corresponding to the current material are calculated through a multivariate nonlinear regression algorithm. ,in For the corrected measurement coefficients, For standard measurement coefficients, This is the humidity deviation value. For real-time feature particle size, Standard characteristic particle size, For liquidity coefficient, The calibration coefficient is used for the experiment. Step 4, Coefficient Output and Feedback: The calculated measurement correction coefficients are sent to the central control module in real time. Simultaneously, the calibration model is iteratively optimized every 5 data acquisition cycles. The deviation between the actual measurement results and the theoretical values ​​is input back into the model for correction. The coefficient improves long-term calibration accuracy.

[0008] Preferably, in step 3, the initial calibration model is specifically constructed through orthogonal experiments to establish a standardized mapping relationship, providing a reliable benchmark for real-time measurement coefficient correction. The steps are as follows: Step 3.1, Preliminary Establishment of the Initial Calibration Model: Determination of Experimental Factors: The three core characteristics affecting the accuracy of veterinary drug and feed metering are selected as experimental factors, namely humidity. Characteristic particle size Liquidity coefficient It covers the common fluctuation range of materials in actual breeding scenarios; Factor level settings: Each factor has 5 levels to ensure coverage of the entire fluctuation range. humidity : 40%RH is set as dry; 45%RH and 50%RH are standard values; 55%RH and 60%RH are high humidity and prone to agglomeration; characteristic particle size : 0.5mm is defined as fine powder; 0.65mm and 0.8mm are standard values; 0.95mm and 1.1mm are coarse particles; flowability coefficient : Set 0.2 as poor liquidity; 0.4 and 0.6 as standard values; 0.8 and 1.0 as excellent liquidity; Orthogonal experimental design: The L25 (5³) orthogonal experimental table was used to design a total of 25 groups of experiments, with each group of experiments repeated 3 times to avoid random errors; the experimental materials were commonly used compound veterinary drug premixes (main components: vitamins, minerals, antibiotic premixes) in animal husbandry. Standard measurement coefficient calibration: In each test group, the standard material characteristic parameters, i.e. The corresponding measurement coefficients Using this as a benchmark, a high-precision electronic balance (accuracy ±0.01g) was used to measure the deviation between the actual and theoretical measured values ​​under different combinations of characteristic parameters, and the actual measurement coefficient corresponding to each group of experiments was then calculated. ,in ; Model fitting and parameter calibration: The three-factor parameters of 25 experimental groups... "and actual measurement coefficient" "As a dataset, a multivariate nonlinear regression algorithm is used for fitting, and the initial calibration model formula is constructed as follows:" Solving the model coefficients using the least squares method Finally, the initial coefficient values ​​are obtained through calibration (e.g., after experimental verification). Meanwhile, the model fit was calculated to be R²≥0.98 to ensure the reliability of the model; Model storage and standardization: The fitted initial calibration model (including formulas, standard parameter values, and calibration coefficients) is stored and standardized. (Goodness-of-fit data) are stored in the local database of the central control module; Step 3.2, Real-time calling and parameter input of the initial calibration model: Triggering conditions: After the multi-dimensional material property dynamic calibration module is started, when the humidity sensor, laser particle size analyzer, and flowability test unit complete the first round of parameter acquisition, with acquisition periods of 50ms, 100ms, and 200ms respectively, the module automatically sends a model call request to the central control module with the latest completed flowability parameter acquisition as the trigger node. Standard parameter synchronous extraction: After responding to the request, the central control module extracts the standard parameters of the initial calibration model from the local database, i.e. The material characteristic parameters collected in real time are synchronously transmitted to the calibration model calculation unit; Real-time parameter preprocessing and transformation: Converting the preprocessed parameters into calculated values ​​that can be directly substituted into the initial calibration model. (If the real-time humidity is 55% RH, then) ); (If real-time D50 = 0.9mm, then the ratio = 0.9 / 0.8 ​​= 1.125); Flowability coefficient Use the preprocessed real-time values ​​directly; Step 3.3, Real-time calculation and output of measurement correction coefficients: Model calculation: The calibration model calculation unit will convert the model into a model. Particle size ratio Substitute the values ​​into the initial calibration model formula, and through hardware-accelerated calculation (ensuring a single calculation time ≤10ms), obtain the metering correction coefficient corresponding to the current material. ; Verification of calculation results: The calculated results The value is compared with the preset reasonable range (0.8-1.2, the effective correction interval determined based on orthogonal experimental data). If... If the value is out of range, an anomaly warning is triggered, and the previous valid value is retained. If the value is within a reasonable range, it is determined to be a valid correction coefficient. Step 3.4, Iterative Optimization of the Initial Calibration Model: Iteration Trigger Cycle: The model iterative optimization process is initiated after every 5 complete sets of steps 3.1-3.3, i.e., the "parameter acquisition - coefficient calculation - measurement execution" cycle; Iteration Data Source: Extracting the "material characteristic parameters" and "calculated correction coefficients" within this cycle. The measurement deviation rate for each set of data is calculated using the "actual measurement feedback value" and "theoretical measurement value" from the "precision measurement execution module". Model parameter correction: Using the deviation rate as a feedback indicator, the coefficients are fine-tuned using the gradient descent method in the initial calibration model. This makes the next round of calculations The values ​​are closer to actual measurement needs, ensuring that the model dynamically adapts to changes in material properties and equipment wear over long-term use, thereby continuously improving calibration accuracy.

[0009] Preferably, the specific working steps of the feeding behavior-feeding amount dynamic coupling adjustment module are as follows: Step S1: Feeding Behavior Data Analysis: The central control module receives feeding data forwarded from the animal status perception module. The feeding data includes the number of individuals feeding. Average feeding speed Maximum feeding speed and the evenness coefficient of group feeding ,in The value ranges from 0 to 1 and is calculated by combining sound sensor data and weight change data. The feeding behavior analysis unit performs time series analysis on the feeding data to analyze the current feeding stage, which is divided into the initial feeding stage, the stable feeding stage, and the end of the feeding stage. Step S2, Calculation of Demand: Based on the characteristics of different feeding stages, and combined with the pre-set daily veterinary drug intake standards for animal species and growth stages. Calculate real-time feeding demand; initial feeding phase, i.e., 3 minutes before feeding: feeding amount ,in This refers to the duration and speed of deployment during this phase. (Avoid material accumulation); Stable feeding phase, i.e., 3-15 minutes after feeding: Feeding amount Deployment speed (Matching stable feeding needs); Late feeding period, i.e., 15 minutes after feeding: Feeding amount Deployment speed (To avoid wasting leftover materials); Step S3, Parameter Coupling Adjustment: The parameter coupling unit couples the calculated dispensing amount and speed at each stage with the measurement correction coefficient output by the multi-dimensional material characteristic dynamic calibration module to correct the dispensing parameters; if the material flowability coefficient (Poor mobility) If the feeding speed is low, reduce it by 10%-20% to avoid blockages; if the uniformity of feeding in the group is high... If the uniformity is poor, the feeding amount should be divided into 2-3 feedings with a 30-second interval to improve the uniformity of feeding. Step S4, Adjustment command generation and issuance: The adjustment command generation unit sends the coupled and corrected instructions on the amount, speed and number of times of delivery to the central control module. At the same time, it receives the updated data from the animal status perception module in real time and recalculates the delivery parameters every 30 seconds to achieve dynamic adjustment.

[0010] Preferably, in step S1, the group feeding uniformity coefficient The specific calculation steps are as follows: A. Data Acquisition and Synchronization Alignment: Data collection equipment and objects: The animal status sensing module is set up in each breeding pen with weight-sensing feeding troughs deployed according to the standard of "4 feeding troughs for 50 animals". The weight-sensing feeding troughs have a data collection accuracy of ±0.1g and a data collection period of 200ms. At the same time, a sound sensor is deployed at the top center of the pen. The sound sensor has a sampling rate of 44.1kHz and a data collection period of 200ms to ensure coverage of weight changes and acoustic signals in all feeding areas. Data collection definition: Weight data: Real-time weight value of each feeding trough is collected every 200ms and recorded as . For each of the four feeding troughs, calculate the weight change between two consecutive feedings. , Negative values ​​indicate weight loss due to animal feeding, while positive values ​​indicate abnormal supplementation requiring further processing. Acoustic data: Audio signals of 200ms duration are collected every 200ms, focusing on capturing chewing sounds (100-500Hz) and fighting sounds (800-2000Hz, characterized by short pulses and high amplitude) during animal feeding. Data synchronization mechanism: Based on the system clock of the central control module, timestamps are added to each set of weight data and acoustic data, accurate to 1ms. The synchronization and alignment of weight data and acoustic data are achieved through timestamp matching, ensuring that the calculation is based on the feeding status of the same time window. B. Raw data preprocessing: Weight data preprocessing: Outlier removal: Using the 3σ criterion, weight variations in individual feeding troughs are removed. Extreme values ​​(such as sudden large fluctuations caused by animals colliding with the tank), if If the feed intake exceeds the range of -5g to 0g (determined based on actual measurements of single-feeding rates in fattening pigs), it is marked as abnormal, and the previous effective dose should be used. Replacement; Smoothing: Effective for 4 feeding troughs The average weight change rate of each feeding trough was obtained by smoothing the data using a moving average method (window size of 5 acquisition cycles, i.e., 1 second). Unit: g / s The absolute value reflects the feeding speed; Acoustic data preprocessing: Noise filtering: A bandpass filter (100-2000Hz) is used to filter out environmental noise (such as equipment operation noise and external interference noise) while retaining the relevant acoustic signals; Feature signal extraction: The audio signal is converted into a spectrum diagram through short-time Fourier transform to identify the characteristics of the competition sound, i.e., short pulse signals with amplitude ≥ 0.5V and duration ≤ 50ms, and the number of competition sound pulses in each acquisition cycle is counted. And the percentage of total time spent in the scrambling That is, the total duration of the fighting / 200ms; C. Core Feature Quantification: Weight Dimension Feature: Calculate the coefficient of variation of the average weight change rate of the four feeding troughs. , The formula reflecting the dispersion of feeding rates in each trough is: ,in, , ; The larger the value, the greater the difference in feeding speed among the feeding troughs, and the more uneven the feeding in the group. Acoustic Dimension Features: Constructing a Competition Intensity Index The formula, which comprehensively reflects the frequency of competing behavior, is as follows: ,in, The maximum number of fighting pulses was preset (the maximum number of pulses during intense fighting among fattening pigs was measured and set to 20 pulses / 200ms). Normalize the number of contests to the range of 0-1; Values ​​range from 0 to 1. The closer it is to 1, the more intense the competition and the lower the uniformity of feeding. D. Uniformity coefficient Coupling calculation and output: Weight allocation: Based on orthogonal experimental verification, weight dimension features Directly reflects differences in foraging resource allocation, with a weight set at 0.6; acoustic dimension features To aid in judging the degree of competition, a weight of 0.4 is set to ensure that the calculation results closely match actual foraging scenarios; Coupling calculation formula: ,in, To preset the maximum coefficient of variation, The coefficient of variation was normalized to the range of 0-1; where The closer it is to 1, the more uniform the feeding speed of each trough; The closer to 1, the less competition there is. The weighted sum of the two results in... (0-1 range); Result Verification and Output: The calculated results... The value is compared with the preset threshold (0.5). If... If, it is determined to be a "uniform feeding state"; if This was determined to be an "uneven feeding state"; The value is sent to the central control module in real time and simultaneously forwarded to the dynamic coupling adjustment module of feeding behavior and feeding amount, serving as the core basis for adjusting the feeding parameters; E. Dynamic adaptation and iteration: Coefficient calibration: every 24 hours based on the feed intake data of the livestock population (such as different time periods and different feed types). Based on the observed values ​​and actual feeding uniformity, the weighting coefficients are fine-tuned (0.6 and 0.4 can be optimized within ±0.1) to ensure... The degree of matching between the value and the actual feeding status; Real-time updates: The value is continuously calculated and updated at a frequency of 200ms / time, and is synchronously transmitted to the central control module along with the correction coefficient of the multi-dimensional material characteristic dynamic calibration module. This provides real-time data support for the dynamic adjustment of the feeding speed and the number of feedings, and realizes a seamless connection between "feeding status perception - uniformity measurement - feeding parameter optimization".

[0011] Compared with the prior art, the beneficial effects of the present invention are: Significantly improved measurement accuracy: The multi-dimensional material characteristic dynamic calibration module breaks through the limitations of traditional single weight measurement. By collecting three core parameters in real time—humidity, particle size distribution, and flowability—and combining them with a dynamic calibration model constructed through orthogonal experiments and iterative optimization using the gradient descent method, it dynamically corrects the measurement coefficients, effectively offsetting the measurement deviations caused by fluctuations in material characteristics, and ensuring the accuracy of veterinary drug dosage and the safety of medication.

[0012] Precise matching of feeding needs: The dynamic coupling adjustment module of feeding behavior and feeding amount is based on real-time animal feeding data. It dynamically calculates the feeding amount and feeding speed according to the feeding stage, and coupled with the material characteristic correction coefficient. It adaptively adjusts the parameters for scenarios with poor mobility or low feeding uniformity, avoiding material waste or insufficient animal feeding caused by fixed feeding mode, and reducing breeding costs.

[0013] The system is highly efficient and stable: Each module achieves seamless data interaction and closed-loop control through a central control module. The entire process, from material supply and characteristic calibration to precise metering, dynamic adjustment, and targeted delivery, is interconnected with a response latency of ≤5ms, adapting to the complex scenarios of large-scale farming. Simultaneously, the modules possess anomaly warning and adaptive fault-tolerance mechanisms, combined with model iterative optimization capabilities, to cope with variables encountered during long-term use, such as changes in material characteristics and equipment wear, ensuring the long-term stability and reliability of the system.

[0014] Comprehensive optimization of livestock farming efficiency: Through the synergistic effect of precise measurement and dynamic administration, the weight uniformity of the animal population is improved, the efficacy of veterinary drugs is fully realized, the risk of disease caused by uneven drug administration is reduced, and the quality of livestock products is improved. The system can complete the entire process of administration control without human intervention, reducing the intensity of manual labor and providing efficient and intelligent technical support for large-scale livestock farming. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the system structure of the present invention; Figure 2 This is a schematic diagram of the workflow of the multi-dimensional material property dynamic calibration module of the present invention; Figure 3 This is a schematic diagram of the workflow of the feeding behavior-feeding amount dynamic coupling adjustment module of the present invention. Detailed Implementation

[0016] 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.

[0017] Please see Figure 1-3 This invention provides a technical solution: a veterinary drug and feed precision dispensing and metering control system, comprising a material storage and conveying module, a multi-dimensional material characteristic dynamic calibration module, a precision metering execution module, an animal state sensing module, a feeding behavior-dispensing amount dynamic coupling adjustment module, a central control module, and a dispensing terminal module; the material storage and conveying module includes partitioned storage bins (for storing veterinary drug premixes and basic feeds respectively) and a screw conveyor device. The inner wall of the storage bins is coated with an anti-stick coating to prevent material clumping and residue. The material storage and conveying module receives feeding instructions from the central control module and drives the screw conveyor device via a variable frequency motor to deliver the material to the precision metering execution module according to a preset ratio. The conveying speed is linked to the metering rhythm of the metering execution module. The partitioned storage bins and the screw conveyor device are existing mature technologies, which will be briefly described below in conjunction with the invention: The partitioned storage area features a dual-compartment independent design, with one compartment for storing veterinary drug premix and the other for storing basic feed. The volume ratio of the two compartments is set according to the conventional needs of "small dose of veterinary drug + large dose of basic feed" in aquaculture scenarios (e.g., 50kg for premix and 500kg for basic feed) to avoid cross-contamination between different materials. All inner walls of the compartments are coated with a polytetrafluoroethylene (PTFE) non-stick coating. This coating has a high surface smoothness and a low coefficient of friction, which can effectively reduce the clumping of veterinary drug premix on the compartment walls due to moisture absorption, ensuring accurate feeding composition.

[0018] The operation of the modules is entirely scheduled by the central control module. Upon system startup, the central control module, combining the "total amount per feeding" output from the feeding behavior-feeding dynamic coupling adjustment module, calculates the required ratio of veterinary drug premix to basic feed (e.g., 1:100) and issues a feeding instruction containing "material type, conveying quantity, and conveying speed" to this module. Once the instruction is triggered, the electric valve at the outlet of the corresponding storage bin opens, and the variable frequency motor drives the screw conveyor to operate. The screw blades employ a variable pitch design; the larger pitch at the feeding end improves feeding efficiency, while the smaller pitch at the discharge end stabilizes material output, ensuring precise and controllable conveying volume.

[0019] To achieve coordination with the precision metering execution module, the conveying speed employs a "linkage adaptation" mechanism: the central control module receives the "metering progress signal" from the precision metering execution module in real time. If the metering cylinder is about to fill, the variable frequency motor speed is reduced to decrease the conveying speed; if the metering cylinder has finished unloading and is empty, the motor speed is increased to accelerate the conveying, avoiding problems such as "insufficient supply leading to metering interruption" or "excessive supply leading to material accumulation." Once the material is delivered to the inlet of the precision metering execution module according to the preset ratio, the module sends a "feeding completion signal" back to the central control module, then closes the discharge valve and the conveying device, completing a single feeding cycle and awaiting the next instruction.

[0020] The multi-dimensional material property dynamic calibration module includes a humidity sensor, a laser particle size analyzer, a flowability testing unit, and a calibration model calculation unit. The multi-dimensional material property dynamic calibration module is used to collect material property parameters such as humidity, particle size distribution, and flowability of veterinary drug feed in real time, establish a dynamic calibration model and correct the measurement coefficients, and establish a dynamic mapping relationship between material property parameters and measurement coefficients. The implementation of the multi-dimensional material property dynamic calibration module is as follows: Step 1: Real-time Parameter Acquisition: Within the transfer silo along the material conveying path, a humidity sensor, laser particle size analyzer, and flowability testing unit are integrated to continuously acquire three core parameters. The humidity sensor employs a high-frequency capacitive sensor, acquiring the material humidity value every 50ms (accuracy ±0.5%RH). The laser particle size analyzer measures the material particle size distribution using a scattering method and outputs... Three characteristic particle size values ​​are collected in 100ms; the flowability test unit calculates the flowability coefficient (range 0-1, the larger the value, the better the flowability) by measuring the sliding speed and angle of repose of the material in the inclined pipe, with a collection period of 200ms. Step 2, Data Preprocessing: Outliers in the collected humidity, particle size distribution, and flowability coefficient data are removed using the 3σ criterion, and the data is smoothed using the moving average method (window size of 5 collection cycles) to avoid calibration deviations caused by instantaneous fluctuations, thus obtaining the preprocessed parameters; Step 3, Dynamic Calibration Model Construction and Iteration: Based on the preprocessed parameters, the preset initial calibration model is invoked (this model is established through a large number of orthogonal experiments, using the measurement coefficients corresponding to the standard material characteristic parameters as a benchmark); the three parameters collected in real time are input into the model, and the measurement correction coefficients corresponding to the current material are calculated through a multivariate nonlinear regression algorithm. ,in For the corrected measurement coefficients, For standard measurement coefficients, This is the humidity deviation value. For real-time feature particle size, Standard characteristic particle size, For liquidity coefficient, The calibration coefficient is used for the experiment. Step 4, Coefficient Output and Feedback: The calculated measurement correction coefficients are sent to the central control module in real time. Simultaneously, the calibration model is iteratively optimized every 5 data acquisition cycles. The deviation between the actual measurement results and the theoretical values ​​is input back into the model for correction. The coefficient improves long-term calibration accuracy.

[0021] In step 3, the initial calibration model is specifically constructed through orthogonal experiments to establish a standardized mapping relationship, providing a reliable benchmark for real-time measurement coefficient correction. The steps are as follows: Step 3.1, Preliminary Establishment of the Initial Calibration Model: Determination of Experimental Factors: The three core characteristics affecting the accuracy of veterinary drug and feed metering are selected as experimental factors, namely humidity. Characteristic particle size Liquidity coefficient It covers the common fluctuation range of materials in actual breeding scenarios; Factor level settings: Each factor has 5 levels to ensure coverage of the entire fluctuation range. humidity : 40%RH is set as dry; 45%RH and 50%RH are standard values; 55%RH and 60%RH are high humidity and prone to agglomeration; characteristic particle size : 0.5mm is defined as fine powder; 0.65mm and 0.8mm are standard values; 0.95mm and 1.1mm are coarse particles; flowability coefficient : Set 0.2 as poor liquidity; 0.4 and 0.6 as standard values; 0.8 and 1.0 as excellent liquidity; Orthogonal experimental design: The L25 (5³) orthogonal experimental table was used to design a total of 25 groups of experiments, with each group of experiments repeated 3 times to avoid random errors; the experimental materials were commonly used compound veterinary drug premixes (main components: vitamins, minerals, antibiotic premixes) in animal husbandry. Standard measurement coefficient calibration: In each test group, the standard material characteristic parameters, i.e. The corresponding measurement coefficients Using this as a benchmark, a high-precision electronic balance (accuracy ±0.01g) was used to measure the deviation between the actual and theoretical measured values ​​under different combinations of characteristic parameters, and the actual measurement coefficient corresponding to each group of experiments was then calculated. ,in ; Model fitting and parameter calibration: The three-factor parameters of 25 experimental groups... "and actual measurement coefficient" "As a dataset, a multivariate nonlinear regression algorithm is used for fitting, and the initial calibration model formula is constructed as follows:" Solving the model coefficients using the least squares method Finally, the initial coefficient values ​​are obtained through calibration (e.g., after experimental verification). Meanwhile, the model fit was calculated to be R²≥0.98 to ensure the reliability of the model; Model storage and standardization: The fitted initial calibration model (including formulas, standard parameter values, and calibration coefficients) is stored and standardized. (Goodness-of-fit data) are stored in the local database of the central control module; Step 3.2, Real-time calling and parameter input of the initial calibration model: Triggering conditions: After the multi-dimensional material property dynamic calibration module is started, when the humidity sensor, laser particle size analyzer, and flowability test unit complete the first round of parameter acquisition, with acquisition periods of 50ms, 100ms, and 200ms respectively, the module automatically sends a model call request to the central control module with the latest completed flowability parameter acquisition as the trigger node. Standard parameter synchronous extraction: After responding to the request, the central control module extracts the standard parameters of the initial calibration model from the local database, i.e. The material characteristic parameters collected in real time are synchronously transmitted to the calibration model calculation unit; Real-time parameter preprocessing and transformation: Converting the preprocessed parameters into calculated values ​​that can be directly substituted into the initial calibration model. (If the real-time humidity is 55% RH, then) ); (If real-time D50 = 0.9mm, then the ratio = 0.9 / 0.8 ​​= 1.125); Flowability coefficient Use the preprocessed real-time values ​​directly; Step 3.3, Real-time calculation and output of measurement correction coefficients: Model calculation: The calibration model calculation unit will convert the model into a model. Particle size ratio Substitute the values ​​into the initial calibration model formula, and through hardware-accelerated calculation (ensuring a single calculation time ≤10ms), obtain the metering correction coefficient corresponding to the current material. ; Verification of calculation results: The calculated results The value is compared with the preset reasonable range (0.8-1.2, the effective correction interval determined based on orthogonal experimental data). If... If the value is out of range, an anomaly warning is triggered, and the previous valid value is retained. If the value is within a reasonable range, it is determined to be a valid correction coefficient. Step 3.4, Iterative Optimization of the Initial Calibration Model: Iteration Trigger Cycle: The model iterative optimization process is initiated after every 5 complete sets of steps 3.1-3.3, i.e., the "parameter acquisition - coefficient calculation - measurement execution" cycle; Iteration Data Source: Extracting the "material characteristic parameters" and "calculated correction coefficients" within this cycle. The measurement deviation rate for each set of data is calculated using the "actual measurement feedback value" and "theoretical measurement value" from the "precision measurement execution module". Model parameter correction: Using the deviation rate as a feedback indicator, the coefficients are fine-tuned using the gradient descent method in the initial calibration model. This makes the next round of calculations The values ​​are closer to actual measurement needs, ensuring that the model dynamically adapts to changes in material properties and equipment wear over long-term use, thereby continuously improving calibration accuracy.

[0022] The precise metering execution module is a combination of a volumetric metering cylinder and an electromagnetic flow valve. A pressure sensor is installed on the inner wall of the metering cylinder to assist in verifying the material filling amount. The precise metering execution module receives basic metering commands and calibrated metering correction coefficients from the central control module, i.e., the actual metering feedback value. It adjusts the volumetric setting of the metering cylinder via a stepper motor, and combined with the opening and closing speed of the electromagnetic flow valve, achieves precise metering. The combination of the volumetric metering cylinder and the electromagnetic flow valve is a mature existing technology; its specific structure will not be described in detail. Its function will be briefly explained using examples from this system: Specifically, this invention can adopt a composite structure design of "volume metering cylinder + electromagnetic flow valve + dual sensor verification". The core component, the volume metering cylinder, has a food-grade wear-resistant rubber lining on its inner wall, which avoids material residue and is adaptable to veterinary drug feeds with different particle sizes. The electromagnetic flow valve adopts a high-frequency response model (response time ≤10ms), which can accurately control the material output rate. At the same time, it integrates a pressure sensor and a displacement sensor. The former detects the filling pressure of the material in the metering cylinder to determine the density, and the latter monitors the piston displacement of the metering cylinder to assist in calculating the volume. Cross-verification of data from the dual sensors ensures the reliability of the measurement.

[0023] Its workflow fully responds to the system's closed-loop control logic: First, it receives a combined instruction from the central control module consisting of a "basic measurement value + measurement correction coefficient." The basic measurement value comes from the real-time demand calculation of the feeding behavior-feeding amount dynamic coupling adjustment module, while the measurement correction coefficient is provided by the multi-dimensional material characteristic dynamic calibration module. The two are combined to generate the actual measurement target (e.g., with a basic measurement of 250g and a correction coefficient of 1.08, the actual target is 270g). After the instruction is triggered, the module first controls the material to enter the metering cylinder from the conveying module through an electromagnetic flow valve. When the feedback data from the pressure sensor and displacement sensor both reach the threshold corresponding to the actual measurement target, the electromagnetic flow valve immediately closes, completing the material metering.

[0024] After metering is completed, the module performs two actions: First, it sends a "metering completion signal" and measured data from the dual sensors to the central control module, providing feedback for the model iteration of the multi-dimensional material characteristic dynamic calibration module; second, it receives the delivery instruction forwarded by the central control module and smoothly transports the metered material to the delivery terminal module through the pneumatic valve at the discharge port, ensuring that the material is not lost or subject to secondary metering deviation during the transfer process, laying a precise foundation for subsequent targeted delivery.

[0025] The animal status perception module includes infrared thermal imaging cameras, weight-sensing feeding troughs, and sound sensors deployed in the breeding pens, covering all feeding areas. The infrared thermal imaging cameras capture the animals' feeding postures in real time and identify the number of individuals feeding. The weight-sensing feeding troughs collect data on the weight change of the material in the trough every 200ms to calculate the feeding speed (g / s) of a single animal. The sound sensors collect chewing and fighting sounds of the animals while feeding to help determine the uniformity of group feeding (no fighting sounds indicate high uniformity, while frequent fighting sounds indicate low uniformity). The collected data on the number of individuals feeding, feeding speed, and group feeding uniformity, i.e., animal status data, are uploaded to the central control module in real time.

[0026] The dynamic coupling adjustment module for feeding behavior and feeding amount includes a feeding behavior analysis unit, a feeding parameter coupling unit, and an adjustment instruction generation unit. The dynamic coupling adjustment module for feeding behavior and feeding amount is used to achieve dynamic matching between feeding amount and the animal's real-time feeding needs, establish the coupling relationship between feeding behavior parameters and feeding parameters, and dynamically adjust the feeding speed and single feeding amount through the coupling algorithm. The specific working steps of the feeding behavior-feeding amount dynamic coupling adjustment module are as follows: feeding behavior analysis unit, feeding parameter coupling unit, and adjustment command generation unit. Step S1: Feeding Behavior Data Analysis: The central control module receives feeding data forwarded from the animal status perception module. The feeding data includes the number of individuals feeding. Average feeding speed Maximum feeding speed and the evenness coefficient of group feeding ,in The value ranges from 0 to 1 and is calculated by combining sound sensor data and weight change data. The feeding behavior analysis unit performs time series analysis on the feeding data to analyze the current feeding stage, which is divided into the initial feeding stage, the stable feeding stage, and the end of the feeding stage. Step S2, Calculation of Demand: Based on the characteristics of different feeding stages, and combined with the pre-set daily veterinary drug intake standards for animal species and growth stages. Calculate real-time feeding demand; initial feeding phase, i.e., 3 minutes before feeding: feeding amount ,in This refers to the duration and speed of deployment during this phase. (Avoid material accumulation); Stable feeding phase, i.e., 3-15 minutes after feeding: Feeding amount Deployment speed (Matching stable feeding needs); Late feeding period, i.e., 15 minutes after feeding: Feeding amount Deployment speed (To avoid wasting leftover materials); Step S3, Parameter Coupling Adjustment: The parameter coupling unit couples the calculated dispensing amount and speed at each stage with the measurement correction coefficient output by the multi-dimensional material characteristic dynamic calibration module to correct the dispensing parameters; if the material flowability coefficient (Poor mobility) If the feeding speed is low, reduce it by 10%-20% to avoid blockages; if the uniformity of feeding in the group is high... If the uniformity is poor, the feeding amount should be divided into 2-3 feedings with a 30-second interval to improve the uniformity of feeding. Step S4, Adjustment command generation and issuance: The adjustment command generation unit sends the coupled and corrected instructions on the amount, speed and number of times of delivery to the central control module. At the same time, it receives the updated data from the animal status perception module in real time and recalculates the delivery parameters every 30 seconds to achieve dynamic adjustment.

[0027] Group feeding evenness coefficient in step S1 The specific calculation steps are as follows: A. Data Acquisition and Synchronization Alignment: Data collection equipment and objects: The animal status sensing module is set up in each breeding pen with weight-sensing feeding troughs deployed according to the standard of "4 feeding troughs for 50 animals". The weight-sensing feeding troughs have a data collection accuracy of ±0.1g and a data collection period of 200ms. At the same time, a sound sensor is deployed at the top center of the pen. The sound sensor has a sampling rate of 44.1kHz and a data collection period of 200ms to ensure coverage of weight changes and acoustic signals in all feeding areas. Data collection definition: Weight data: Real-time weight value of each feeding trough is collected every 200ms and recorded as . For each of the four feeding troughs, calculate the weight change between two consecutive feedings. , Negative values ​​indicate weight loss due to animal feeding, while positive values ​​indicate abnormal supplementation requiring further processing. Acoustic data: Audio signals of 200ms duration are collected every 200ms, focusing on capturing chewing sounds (100-500Hz) and fighting sounds (800-2000Hz, characterized by short pulses and high amplitude) during animal feeding. Data synchronization mechanism: Based on the system clock of the central control module, timestamps are added to each set of weight data and acoustic data, accurate to 1ms. The synchronization and alignment of weight data and acoustic data are achieved through timestamp matching, ensuring that the calculation is based on the feeding status of the same time window. B. Raw data preprocessing: Weight data preprocessing: Outlier removal: Using the 3σ criterion, weight variations in individual feeding troughs are removed. Extreme values ​​(such as sudden large fluctuations caused by animals colliding with the tank), if If the feed intake exceeds the range of -5g to 0g (determined based on actual measurements of single-feeding rates in fattening pigs), it is marked as abnormal, and the previous effective dose should be used. Replacement; Smoothing: Effective for 4 feeding troughs The average weight change rate of each feeding trough was obtained by smoothing the data using a moving average method (window size of 5 acquisition cycles, i.e., 1 second). Unit: g / s The absolute value reflects the feeding speed; Acoustic data preprocessing: Noise filtering: A bandpass filter (100-2000Hz) is used to filter out environmental noise (such as equipment operation noise and external interference noise) while retaining the relevant acoustic signals; Feature signal extraction: The audio signal is converted into a spectrum diagram through short-time Fourier transform to identify the characteristics of the competition sound, i.e., short pulse signals with amplitude ≥ 0.5V and duration ≤ 50ms, and the number of competition sound pulses in each acquisition cycle is counted. And the percentage of total time spent in the scrambling That is, the total duration of the fighting / 200ms; C. Core Feature Quantification: Weight Dimension Feature: Calculate the coefficient of variation of the average weight change rate of the four feeding troughs. , The formula reflecting the dispersion of feeding rates in each trough is: ,in, , ; The larger the value, the greater the difference in feeding speed among the feeding troughs, and the more uneven the feeding in the group. Acoustic Dimension Features: Constructing a Competition Intensity Index The formula, which comprehensively reflects the frequency of competing behavior, is as follows: ,in, The maximum number of fighting pulses was preset (the maximum number of pulses during intense fighting among fattening pigs was measured and set to 20 pulses / 200ms). Normalize the number of contests to the range of 0-1; Values ​​range from 0 to 1. The closer it is to 1, the more intense the competition and the lower the uniformity of feeding. D. Uniformity coefficient Coupling calculation and output: Weight allocation: Based on orthogonal experimental verification, weight dimension features Directly reflects differences in foraging resource allocation, with a weight set at 0.6; acoustic dimension features To aid in judging the degree of competition, a weight of 0.4 is set to ensure that the calculation results closely match actual foraging scenarios; Coupling calculation formula: ,in, To preset the maximum coefficient of variation, The coefficient of variation was normalized to the range of 0-1; where The closer it is to 1, the more uniform the feeding speed of each trough; The closer to 1, the less competition there is. The weighted sum of the two results in... (0-1 range); Result Verification and Output: The calculated results... The value is compared with the preset threshold (0.5). If... If, it is determined to be a "uniform feeding state"; if This was determined to be an "uneven feeding state"; The value is sent to the central control module in real time and simultaneously forwarded to the dynamic coupling adjustment module of feeding behavior and feeding amount, serving as the core basis for adjusting the feeding parameters; E. Dynamic adaptation and iteration: Coefficient calibration: every 24 hours based on the feed intake data of the livestock population (such as different time periods and different feed types). Based on the observed values ​​and actual feeding uniformity, the weighting coefficients are fine-tuned (0.6 and 0.4 can be optimized within ±0.1) to ensure... The degree of matching between the value and the actual feeding status; Real-time updates: The value is continuously calculated and updated at a frequency of 200ms / time, and is synchronously transmitted to the central control module along with the correction coefficient of the multi-dimensional material characteristic dynamic calibration module. This provides real-time data support for the dynamic adjustment of the dispensing speed and frequency, achieving a seamless connection between "feeding status perception - uniformity measurement - dispensing parameter optimization". The central control module receives data uploaded by each module, forwards material characteristic parameters to the multi-dimensional material characteristic dynamic calibration module, and forwards animal status data to the feeding behavior-feeding amount dynamic coupling adjustment module; it receives the output results of the multi-dimensional material characteristic dynamic calibration module and the feeding behavior-feeding amount dynamic coupling adjustment module, integrates them to generate metering instructions and feeding instructions, and sends them to the precision metering execution module and the feeding terminal module respectively; at the same time, it monitors the working status of each module and triggers an alarm when an abnormality occurs. The dispensing terminal module includes a rotatable dispensing nozzle and a position adjustment device, with each breeding pen corresponding to an independent dispensing terminal. The dispensing terminal module receives dispensing instructions from the central control module, aligns itself with the feeding trough area via the position adjustment device, and evenly dispenses precisely measured veterinary feed into the feeding trough according to the set dispensing speed, amount, and number of dispensing cycles. After dispensing, it sends a feedback signal to the central control module, completing one dispensing cycle. The rotatable dispensing nozzle and position adjustment device are existing mature technologies, so their specific mechanical structure will not be described in detail. The following is a brief description of their functions: In this embodiment, the feeding terminal module can be configured as a "rotatable feeding nozzle + electric position adjustment device," deployed according to the principle of one terminal per pen, with each breeding pen corresponding to an independent feeding terminal to avoid interference from cross-pen feeding. The rotatable nozzle adopts a fan-shaped discharge design, and the discharge angle can be adjusted from 30° to 90° to adapt to the coverage needs of different sizes of feeding troughs; the position adjustment device is driven by a stepper motor, which can achieve horizontal ±15cm and vertical ±10cm displacement fine adjustment to ensure that the nozzle is accurately aligned with the feeding trough area.

[0028] Its workflow is deeply integrated with the system's closed-loop control: First, it receives a complete dispensing command from the central control module, including the dispensing amount after dynamic calibration based on multi-dimensional material characteristics, and the dispensing speed and number of dispensing times determined by the feeding behavior-dispensing amount dynamic coupling adjustment module. After the command is triggered, the position adjustment device first completes positioning calibration according to preset coordinates, and then the rotatable nozzle starts dispensing material at the set speed. If batch dispensing is required (such as when the uniformity of group feeding is low), the "dispensing-pause" action is executed cyclically at intervals to ensure that the material is evenly spread in the feeding trough and avoid local accumulation or spillage.

[0029] After dispensing is completed, the module immediately sends a "dispensing completion feedback signal" to the central control module, simultaneously uploading data such as the actual dispensing duration and nozzle working status, facilitating the central control module's monitoring of the entire process to ensure its normal operation. Upon receiving the feedback, the nozzle returns to its initial position, awaiting the next metering completion signal to begin a new dispensing cycle, ensuring seamless coordination with the precision metering execution module and the central control module.

[0030] This invention discloses a precise dosage and control system for veterinary drugs and feed, relating to the field of animal husbandry technology. The system includes a material storage and conveying module, a multi-dimensional material characteristic dynamic calibration module, a precise dosage execution module, an animal status sensing module, a feeding behavior-dosage dynamic coupling adjustment module, a central control module, and a dosing terminal module. The multi-dimensional material characteristic dynamic calibration module collects real-time data on the humidity, particle size distribution, and flowability parameters of the veterinary drug and feed to establish a dynamic calibration model and correct the dosage coefficients. The feeding behavior-dosage dynamic coupling adjustment module dynamically couples the dosing parameters based on real-time animal feeding speed and group feeding uniformity data. This invention solves the problems of low measurement accuracy and resource waste caused by material characteristic fluctuations and mismatches between dosage and animal feeding needs in traditional systems. Through the synergistic effect of the two core innovative modules, it achieves precise and dynamic adjustment of veterinary drug and feed dosing, improving breeding efficiency and medication safety.

[0031] 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 veterinary feed precision dosing metering control regulating system, characterized by, The application relates to a livestock feed dispensing system, which comprises a material storage and conveying module, a multi-dimensional material characteristic dynamic calibration module, a precise metering execution module, an animal state sensing module, a feeding behavior-dispensing amount dynamic coupling adjustment module, a central control module and a dispensing terminal module; the material storage and conveying module comprises a partitioned storage bin and a spiral conveying device, the inner wall of the storage bin is provided with an anti-sticking coating to avoid material caking and residual, and the material storage and conveying module receives a feeding instruction from the central control module, drives the spiral conveying device through a variable frequency motor, and conveys the material to the precise metering execution module according to a preset proportion, wherein the conveying speed is linked with the metering rhythm of the metering execution module; The multi-dimensional material characteristic dynamic calibration module comprises a humidity sensor, a laser particle size analyzer, a flowability test unit and a calibration model calculation unit; the multi-dimensional material characteristic dynamic calibration module is used for collecting the material characteristic parameters of the humidity, particle size distribution and flowability of the veterinary feed in real time, establishing a dynamic calibration model and correcting a metering coefficient, and establishing a dynamic mapping relationship between the material characteristic parameters and the metering coefficient; The precise metering execution module is a combined structure of a volumetric metering cylinder and an electromagnetic flow valve, the inner wall of the metering cylinder is provided with a pressure sensor to assist in verifying the material filling amount; and the precise metering execution module receives a basic metering instruction and a corrected metering coefficient from the central control module, that is, an actual metering feedback value, adjusts the volume gear position of the metering cylinder through a stepping motor, and realizes precise metering by combining the opening and closing speed of the electromagnetic flow valve; The animal state sensing module comprises infrared thermal imaging cameras, weight sensing type feeding troughs and sound sensors arranged in the breeding cage, and covers all feeding areas; the infrared thermal imaging cameras capture the feeding postures of animals in real time and identify the number of feeding individuals; the weight sensing type feeding troughs collect the weight change of the material in the trough every 200ms, and calculate the feeding speed of a single animal; the sound sensors collect the chewing sound and fighting sound of animals during feeding, and assist in judging the uniformity of group feeding; the collected number of feeding individuals, feeding speed and group feeding uniformity data, that is, the animal state data, are uploaded to the central control module in real time; The feeding behavior-dispensing amount dynamic coupling adjustment module comprises a feeding behavior analysis unit, a dispensing parameter coupling unit and a regulation instruction generation unit, and is used for realizing the dynamic matching of the dispensing amount and the real-time feeding demand of animals, establishing a coupling relationship between the feeding behavior parameters and the dispensing parameters, and dynamically adjusting the dispensing speed and the single dispensing amount through a coupling algorithm; The central control module receives the data uploaded by each module, forwards the material characteristic parameters to the multi-dimensional material characteristic dynamic calibration module, and forwards the animal state data to the feeding behavior-dispensing amount dynamic coupling adjustment module; the central control module receives the output results of the multi-dimensional material characteristic dynamic calibration module and the feeding behavior-dispensing amount dynamic coupling adjustment module, integrates the metering instruction and the dispensing instruction, and respectively issues the metering instruction and the dispensing instruction to the precise metering execution module and the dispensing terminal module; meanwhile, the central control module monitors the working states of each module and triggers an alarm when an abnormality occurs. The feeding terminal module comprises a rotatable feeding nozzle and a position adjusting device, and each breeding pen corresponds to an independent feeding terminal; the feeding terminal module receives the feeding instruction of the central control module, aligns the feeding trough area through the position adjusting device, and uniformly feeds the accurately measured veterinary feed into the feeding trough according to the set feeding speed, feeding amount and feeding frequency; after the feeding is completed, a feedback signal is sent to the central control module, and a feeding cycle is completed.

2. The system according to claim 1, wherein the system comprises a plurality of said feed dispensing units. The multi-dimensional material property dynamic calibration module is implemented as follows: Step 1: Real-time Parameter Acquisition: Within the transfer silo along the material conveying path, a humidity sensor, laser particle size analyzer, and flowability testing unit are integrated to continuously acquire three core parameters. The humidity sensor uses a high-frequency capacitive sensor, acquiring the material humidity value every 50ms. The laser particle size analyzer measures the material particle size distribution using a scattering method and outputs... Three characteristic particle size values, with a data acquisition period of 100ms; the flowability test unit calculates the flowability coefficient by measuring the sliding speed and angle of repose of the material in the inclined pipe, with a data acquisition period of 200ms; Step 2, data preprocessing: the collected humidity, particle size distribution and flowability coefficient data are subjected to 3σ criterion for outlier rejection, and the data are smoothed by the moving average method to avoid calibration deviation caused by instantaneous fluctuation, and the preprocessed parameters are obtained; Step 3, dynamic calibration model construction and iteration: based on the pre-processed parameters, an initial calibration model is called; three parameters collected in real time are input into the model, and the current material corresponding metering correction coefficient is calculated through multivariate nonlinear regression algorithm: wherein is the corrected metering coefficient, is the standard metering coefficient, is the humidity deviation value, is the real-time characteristic particle size, is the standard characteristic particle size, is the flowability coefficient, is the test calibration coefficient; Step 4, coefficient output and feedback: the calculated metrology correction coefficient is sent to the central control module in real time, and the calibration model is iteratively optimized every 5 acquisition cycles, the deviation of the actual metrology result from the theoretical value is input into the model in reverse, and the correction coefficient is improved to improve long-term calibration accuracy.

3. The system according to claim 1, wherein the system comprises a plurality of said dispensing units. The initial calibration model in step 3 is specifically constructed by orthogonal test to construct a standardized mapping relationship, which provides a reliable reference for real-time metering coefficient correction, and the steps are as follows: Step 3.1, Pre-establishment of initial calibration model: test factor determination: take three core characteristics affecting the feed metering accuracy of veterinary drugs as test factors, respectively, humidity , characteristic particle size , flowability coefficient , covering the common fluctuation range of materials in actual breeding scenes; Factor level setting: 5 gradient levels are set for each factor to ensure coverage of the full fluctuation interval: Humidity : Set 40% RH as dry; 45% RH, 50% RH as standard value; 55% RH, 60% RH as high humidity easy to caking; Characteristic particle size : Set 0.5 mm as fine powder; 0.65 mm, 0.8 mm as standard value; 0.95 mm, 1.1 mm as coarse particles; Flowability coefficient : Set 0.2 as poor flowability; 0.4, 0.6 are standard values; 0.8, 1.0 are excellent flowability; Orthogonal test scheme design: L25 (5³) orthogonal test table is used, 25 groups of tests are designed, each test is repeated 3 times to avoid accidental errors; Standard metering coefficient calibration: in each test group, the standard material characteristic parameters, i.e. , are used as the reference, the high-precision electronic balance is used to measure the deviation between the actual metering value and the theoretical metering value under different characteristic parameter combinations, and the actual metering coefficient corresponding to each test group is inversely deduced , wherein ;​ Model fitting and parameter calibration: 25 groups of test "three-factor parameters " and "actual measurement coefficient " as a data set, using multiple nonlinear regression algorithm fitting, to build the initial calibration model formula: ; through the least square method to solve the model coefficient , the initial coefficient value is finally calibrated, and the initial calibration model fitting degree R²≥0.98 is calculated, to ensure the reliability of the initial calibration model; Model storage and standardization: the fitted initial calibration model is stored in the local database of the central control module; Step 3.2, real-time calling and parameter substitution of the initial calibration model: Triggering condition for calling: after the multi-dimensional material property dynamic calibration module is started, when the humidity sensor, laser particle size analyzer and flowability test unit complete the first round of parameter collection, the collection period is 50ms, 100ms and 200ms respectively, and the latest completed flowability parameter collection is taken as the triggering node, and the module automatically sends a model calling request to the central control module; Standard parameter synchronous extraction: the central control module extracts the standard parameters of the initial calibration model from the local database in response to the request, that is synchronously transmitted to the calibration model calculation unit with the real-time collected material characteristic parameters; Real-time parameter pre-processing and conversion: converting the pre-processed parameters into calculated values that are directly substituted into the initial calibration model: ; ; flowability coefficient Directly using pre-processed real-time values; Step 3.3, real-time calculation and output of metering correction coefficient: Model operation: the calibrated model calculation unit substitutes the converted particle size ratio, value into the initial calibration model formula, and obtains the metering correction coefficient corresponding to the current material through hardware acceleration operation ; Verification of calculation results: The calculated results The value is compared with a preset reasonable range. If the value is out of range, an anomaly warning is triggered, and the previous valid value is retained. If the value is within a reasonable range, it is determined to be a valid correction coefficient. Step 3.4, Iterative optimization of initial calibration model: Iterative trigger period: every 5 complete sets of step 3.1-3.3, i.e. parameter collection-coefficient calculation-measurement execution cycle, start the model iterative optimization process; Iterative data source: extract the material characteristic parameters, calculated correction coefficients in this period , the actual measurement feedback value and the theoretical measurement value of the accurate measurement execution module, and calculate the measurement deviation rate of each group of data, ; Model parameter correction: Using the bias rate as a feedback indicator, the coefficients are fine-tuned using the gradient descent method in the initial calibration model. This makes the next round of calculations The value is closer to actual measurement needs.

4. The system according to claim 1, wherein the system comprises a plurality of said dispensing units. The feeding behavior-feeding amount dynamic coupling adjustment module specifically works as follows: Step S1, foraging behavior data analysis: receiving the feeding data forwarded from the animal state perception module by the central control module, the feeding data including the number of foraging individuals , average foraging speed , maximum foraging speed , and group foraging uniformity coefficient , wherein the value range is 0-1, and the value is calculated by the sound sensor and the weight change data; the feeding data is analyzed by the feeding behavior analysis unit in time sequence, and the current foraging stage is analyzed, wherein the foraging stage is divided into initial foraging stage, stable foraging stage and end foraging stage; Step S2, the calculation of the demand for dispensing: according to the characteristics of different feeding stages, combined with the animal breed, the growth stage and the preset single-day animal drug intake standard, the real-time dispensing demand is calculated ; the initial feeding stage, i.e. 3 minutes before eating: the dispensing amount , wherein is the duration of the stage, and the dispensing speed ; the stable feeding stage, i.e. 3-15 minutes of eating: the dispensing amount , and the dispensing speed ; the end of the feeding stage, i.e. after 15 minutes of eating: the dispensing amount , and the dispensing speed ; Step S3, coupling adjustment of feeding parameters: the coupling unit of feeding parameters couples the calculated feeding amount and feeding speed of each stage with the measurement correction coefficient output by the multi-dimensional material characteristic dynamic calibration module to correct the feeding parameters; if the material flowability coefficient is less than 0.8, the feeding speed is reduced by 10%-20% to avoid blockage; if the group feeding uniformity coefficient is greater than 0.9, the single feeding amount is split into 2-3 times with an interval of 30 seconds for feeding to improve the feeding uniformity; Step S4, adjustment instruction generation and issuance: the adjustment instruction generation unit sends the coupled and corrected feeding amount, feeding speed and feeding frequency instructions to the central control module, and simultaneously receives the updated data of the animal state sensing module in real time, and recalculates the feeding parameters every 30 seconds to realize dynamic adjustment.

5. The precise dosing metering control regulating system for veterinary feed according to claim 4, characterized in that: The coefficient of uniformity of intake of the population in step S1 The specific calculation steps are as follows: A, data acquisition and synchronous alignment: Acquisition equipment and object: the animal state sensing module is configured with 4 weight sensing feeding troughs in each breeding pen according to the standard deployment of 50 animals per feeding trough, the weight sensing feeding trough has an acquisition accuracy of ±0.1g and an acquisition period of 200ms, and a sound sensor is arranged at the top center of the pen, the sound sensor has a sampling rate of 44.1kHz and an acquisition period of 200ms, which ensures that the weight change and acoustic signals of all feeding areas are covered; Collection content definition: weight data: collect the real-time weight value of each feeding trough every 200 ms, denoted as , corresponding to 4 feeding troughs, calculate the weight change of adjacent two times , The negative value represents that the animal feeding causes the weight to decrease, and the positive value is an abnormal supplement, which needs subsequent processing; acoustic data: collect an audio signal with a length of 200 ms every 200 ms, focusing on capturing the chewing sound and fighting sound when the animal feeds; Data synchronization mechanism: taking the system clock of the central control module as the reference, a timestamp is added to each set of weight data and acoustic data with an accuracy of 1ms, the synchronous alignment of the weight data and acoustic data is realized through the timestamp matching, and the calculation is based on the same time window of the feeding state; B, original data preprocessing: Weight data preprocessing: Outlier removal: Using the 3σ criterion, weight variations in individual feeding troughs are removed. The extreme values ​​in, if If the value exceeds the range of -5g to 0g, it will be marked as abnormal, and the previous effective value will be used. Replacement; Smoothing: Effective for 4 feeding troughs The average weight change rate for each feeding trough was obtained by smoothing using the moving average method. , The absolute value reflects the feeding speed; Acoustic data preprocessing Noise filtering: band-pass filter is used to filter environmental noise and keep foraging-related acoustic signals; feature signal extraction: short-time Fourier transform is used to convert audio signals into frequency spectrum, identify the characteristics of the fighting sound, i.e. short pulse signals with amplitude ≥ 0.5V and duration ≤ 50ms, and count the number of fighting sound pulses in each collection period and the proportion of the total duration of the fighting sound , i.e. the total duration of the fighting sound / 200ms; C. Core feature quantification: weight dimension feature: coefficient of variation of the average weight change rate of 4 feeders , Reflect the dispersion degree of the feeding speed of each trough, the formula is: , , ; The larger the value, the greater the difference in feeding speed of each feeder, and the more uneven the group feeding; Acoustic dimension feature: build the contest intensity index , which comprehensively reflects the frequency of the contest behavior, and the formula is: , wherein, is the preset maximum number of contest pulses, normalize the number of contests to the range of 0-1; the value is 0-1, the closer to 1, the more intense the contest, the lower the uniformity of feeding; D, uniformity coefficient Coupling calculation and output: Weight distribution: based on orthogonal test verification, weight dimension characteristics Directly reflect the difference of foraging resource allocation, weight set to 0.6; acoustic dimension characteristics Auxiliary judgment of the degree of competition, weight set to 0.4, make the calculation result fit the actual foraging scene; Coupling calculation formula: wherein, is a preset maximum coefficient of variation, normalizing the coefficient of variation to a range of 0-1; wherein the closer to 1, the more uniform the feeding speed of each slot; the closer to 1, the milder the fighting, and the weighted sum of the two is ; Result Verification and Output: The calculated results... The value is compared with the preset threshold (0.5). If... If, it is determined to be a "uniform feeding state"; if This was determined to be an "uneven feeding state"; The value is sent to the central control module in real time and simultaneously forwarded to the dynamic coupling adjustment module for feeding behavior and feeding amount. E. Dynamic adaptation and iteration: Coefficient calibration: every 24 hours based on the feeding data of the farming population, fine-tune the weight coefficient to ensure that the value matches the actual feeding state. the value matches the actual feeding state. Real-time update: The value is continuously calculated and updated at a frequency of 200 ms / time, and the correction coefficient of the multi-dimensional material characteristic dynamic calibration module is transmitted to the central control module synchronously, thereby providing real-time data support for dynamic adjustment of the feeding speed and feeding times.

Images