Method and system for intelligently regulating and controlling feed feeding amount

Abstract

The invention relates to the technical field of intelligent animal husbandry, in particular to a feed feeding amount intelligent regulation and control method and system.The method comprises the following steps that weight data is collected, growth acceleration is calculated and matched with growth trend to correct configuration, environment parameters are monitored, heat dissipation power and environment enthalpy value compensation amount are calculated, and target feeding total weight is generated in combination with basic requirements; a swallowing phase silent window period is generated based on sound wave envelope analysis, and a trigger signal is responded to drive a motor to execute pulse release. According to the method, the correction configuration is generated by evaluating the growth inertia state to adapt to the actual development rate, the compensation configuration is generated by quantifying the environmental heat effect to offset the cold stress loss, the chewing and swallowing phases are accurately segmented based on the acoustic signals, and pulse driving is triggered by swallowing action; full-dimensional accurate feeding from a macroscopic growth trend to a microcosmic ingestion rhythm is realized, real-time synchronization of energy supply and biological metabolism requirements is ensured, and the feed utilization efficiency is remarkably improved while the growth potential of livestock is ensured.

Description

Technical Field

[0001] This invention relates to the field of intelligent animal husbandry technology, and in particular to a method and system for intelligent control of feed feeding amount. Background Technology

[0002] The field of intelligent livestock technology involves a comprehensive technical system for digitally monitoring and managing the entire livestock and poultry breeding process using IoT sensors, wireless communication networks, and automated execution equipment. Its core aspects include the control of breeding environment parameters, monitoring of livestock physiological characteristics, early warning of disease risks, and precise supply of feed and water. This field involves deploying temperature sensors, humidity sensors, harmful gas detectors, and video surveillance cameras within the breeding pens to collect real-time environmental data and the condition of the livestock themselves. The collected data is then transmitted to a central management platform via a network. The platform, based on a pre-set breeding process model, directs power to fans, water curtains, supplemental lighting, and other equipment. Automatic feeders issue control commands to maintain a suitable growth environment for livestock. The traditional method of intelligent control of feed feeding involves farm managers manually inputting fixed feeding times and the corresponding screw conveyor motor running time or feeding valve opening angle on the control terminal of the automatic feeder based on past breeding experience or standard feeding manuals. When the system time reaches the preset feeding time, the clock controller inside the feeder detects that the system time has reached the preset feeding time and closes the circuit to drive the motor to rotate or control the solenoid valve to open, transporting the feed from the storage tower to the feeding trough. When the motor running time or valve opening time reaches the preset duration, the power is cut off to stop feeding.

[0003] Traditional feed feeding methods rely on the past experience of farmers or the use of standard manuals to set fixed feeding parameters. They control the amount of feed by simply setting the motor running time or the valve opening angle, ignoring the dynamic differences in the growth rate of individual livestock and the real-time impact of environmental temperature and humidity changes on the organism's energy metabolism. This makes it impossible to accurately match the actual nutritional needs of livestock at different growth stages and under different climatic conditions. Furthermore, the lack of a mechanism for monitoring and responding to the real-time feeding behavior of livestock leads to overfeeding and waste, or underfeeding that limits growth potential, reduces feed conversion rate, and lowers the economic benefits of farming. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of existing technologies and to propose an intelligent control method and system for feed feeding.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: an intelligent control method for feed feeding amount, comprising the following steps:

[0006] S1: Collect time-series data on average weight of livestock populations, calculate population growth acceleration data using discrete difference, obtain theoretical growth acceleration references by combining livestock breeds and compare them, and match growth trend correction configurations.

[0007] S2: Based on the growth trend, the configuration is corrected, the surface area data and metabolic power of the organism are calculated, the air flow velocity data at the back height is monitored to calculate the convective heat transfer coefficient, the heat dissipation power is calculated by combining the surface area and ambient air temperature data, and the environmental enthalpy compensation amount is obtained based on the heat dissipation power and metabolic power.

[0008] S3: Obtain basic feeding requirement data based on livestock breed, weight and correct the basic feeding requirement data and the growth trend correction configuration, obtain the inertial correction amount, superimpose the inertial correction amount and the environmental enthalpy compensation amount, and generate the target total feeding weight.

[0009] S4: Collect acoustic signal flow in the feed trough area, perform Hilbert transform and generate acoustic amplitude envelope data, construct background noise basis parameters, segment chewing peak segments and intermittent trough segments, update average chewing cycle parameters and generate minimum swallowing pause limit, and generate swallowing phase silence window when the intermittent trough segment exceeds the minimum swallowing pause limit.

[0010] S5: In response to the trigger signal of the swallowing phase silent window period, drive the motor to release the single feed amount, deduct the single release amount from the target total feeding weight and update the remaining feed amount record, terminate the instruction and generate a feeding task completion record when the remaining feed amount record is cleared to zero.

[0011] As a further embodiment of the present invention, the growth trend correction configuration includes a growth rate gain coefficient, a developmental inertia factor, and a dynamic energy demand adjustment index; the environmental enthalpy compensation amount includes a convective heat loss compensation value, a cold stress energy compensation amount, and a feed heat conversion increment; the target total feeding weight includes the basal diet baseline mass, the growth inertia correction mass, and the environmental thermal additional mass; the swallowing phase quiescent window period specifically refers to the window start timestamp, the effective duration period identifier, and the swallowing action confidence level; and the feeding task completion record includes the cumulative total feeding mass, pulse execution count, and task termination time node.

[0012] As a further aspect of the present invention, the process of obtaining the growth trend correction configuration is specifically as follows:

[0013] S111: Collect time-series data on the average weight of livestock herds, perform second-order discrete difference operations on the time dimension of the average weight time-series data of livestock herds, calculate the rate of change of weight increment at adjacent sampling times and perform smoothing filtering on the results to remove observation noise and random fluctuations, quantify the dynamic net growth rate characteristics of the herd within the current monitoring time window, and generate herd growth acceleration data.

[0014] S112: Based on the livestock breed, obtain a pre-set theoretical growth curve model containing breed characteristics, read the age-time index parameters of the current breeding stage, deduce the standard metabolic growth acceleration under the current time index, construct a benchmark reference value, and establish a theoretical growth acceleration reference.

[0015] S113: Call the population growth acceleration data and the theoretical growth acceleration reference, calculate the algebraic difference between the two, determine the current growth inertia state, retrieve the preset multidimensional adjustment coefficient matrix, find and extract the matching control parameter combination row, and generate the growth trend correction configuration.

[0016] As a further aspect of the present invention, the process of obtaining the theoretical growth curve model is specifically as follows:

[0017] Data on the weight and age of corresponding livestock breeds throughout their life cycle under standardized feeding conditions were collected. The nonlinear least squares method was used to perform numerical fitting on the data to determine the asymptotic limit weight, maximum relative growth rate, and growth inflection point time parameters that characterize growth features. A continuous time function was generated and stored.

[0018] As a further aspect of the present invention, the process of establishing the multidimensional adjustment coefficient matrix is ​​specifically as follows:

[0019] A discrete deviation amplitude gradient covering the positive and negative deviation directions is set. For each gradient, a multi-level nutrient level compensation feeding test is performed. Growth rate recovery data under different combinations of feed energy and feeding amount are recorded. Based on the data, the growth rate is reverse-solved to return to the target control parameters of the theoretical growth acceleration reference. The corresponding energy density correction value and feeding amount ratio value are extracted and written into the storage unit according to the deviation direction and amplitude index mapping.

[0020] As a further aspect of the present invention, the process of obtaining the environmental enthalpy compensation amount is specifically as follows:

[0021] S211: Monitor the air flow velocity data and ambient air temperature data at the back height, calculate the convective heat transfer coefficient based on the air flow velocity data at the back height, extract the current average weight of the population associated with the growth trend correction configuration, calculate the biological body surface area data and basal metabolic heat production power data based on the current average weight of the population, and establish a set of biological thermal basic parameters.

[0022] S212: Call the biological thermal basic parameter set, combine the ambient air temperature data and convective heat transfer coefficient to perform convective heat transfer calculation, calculate the heat flux loss rate of the livestock body surface, multiply the heat flux loss rate by the biological body surface area data, and generate total heat dissipation power data.

[0023] S213: Call the total heat dissipation power data and the basic metabolic heat production power data in the biological thermal fundamental parameter set to calculate the environmental enthalpy compensation.

[0024] As a further aspect of the present invention, the process of obtaining the target total feeding weight is specifically as follows:

[0025] S311: Input the average weight of the livestock population and the age of the livestock as an index into the theoretical growth curve model, traverse the nutrient allocation logic embedded in the model, match the unit net energy value required to maintain basal metabolism and standard weight gain, combine the energy conversion density of the pre-set feed formula, quantify the theoretical feed consumption quality under the theoretical preset path, and generate basic feeding demand data.

[0026] S312: Based on the basic feeding requirement data, read the growth trend correction configuration, dynamically scale the benchmark feed amount according to the deviation of the actual growth acceleration from the theoretical value, determine the correction feed value that conforms to the current actual growth rate, and generate the inertia correction amount.

[0027] S313: Call the inertial correction amount, combine it with the environmental enthalpy compensation amount to perform linear superposition operation, and perform discretization and rounding and preset maximum feeding threshold truncation processing on the merged result according to the minimum feeding resolution and maximum single feeding limit of the automatic feeding equipment, construct the execution instruction of the single feeding task, and generate the target total feeding weight.

[0028] As a further aspect of the present invention, the process of obtaining the swallowing phase silent window period is specifically as follows:

[0029] S411: Collect acoustic signal stream in the trough area, perform Hilbert transform on the acoustic signal stream and generate acoustic amplitude envelope data, construct background noise floor parameters by performing histogram statistical analysis, segment peak morphology and trough segments according to background noise floor parameters, extract the time interval between adjacent peaks and generate chewing cycle time series set.

[0030] S412: Call the chewing cycle time series set, calculate the arithmetic mean and discrete standard deviation of the time intervals, and establish the minimum swallowing pause limit;

[0031] S413: Monitor the duration of the current trough segment in real time, compare the duration with the minimum swallowing pause limit in the time domain, and trigger the state latching logic when the duration exceeds the minimum swallowing pause limit to generate a swallowing phase silence window.

[0032] As a further aspect of the present invention, the process of obtaining the feeding task completion record is specifically as follows:

[0033] S511: In response to the trigger signal of the swallowing phase silent window period, call the motor drive timing parameters to construct a discrete pulse drive command, control the screw conveyor motor to perform a single quantitative rotation motion, collect the angular displacement signal of the motor shaft and verify the mechanical action, and generate a single pulse feeding execution record.

[0034] S512: Call the single pulse feeding execution record, deduct the single rotation feeding weight calibration value from the target total feeding weight according to the single rotation feeding weight calibration value, write the calculation result back to the memory to overwrite the original value, and establish real-time remaining feed quantity status data;

[0035] S513: Obtain the real-time remaining feed quantity status data, compare it with the zero value cutoff bit in real time, send a blocking command to the motor control port when the feed quantity reaches or falls below the zero value cutoff bit, summarize the start timestamp and cumulative release weight of each feeding cycle, and generate a feeding task completion record.

[0036] A feed feeding rate intelligent control system, the feed feeding rate intelligent control system being used to execute the above-mentioned feed feeding rate intelligent control method, the system comprising:

[0037] The growth trend analysis module collects time-series data on the average weight of livestock groups, calculates the growth acceleration data of the group using discrete difference, obtains theoretical growth acceleration references by combining them with livestock breeds and compares them, and matches the growth trend correction configuration.

[0038] The environmental compensation calculation module corrects the configuration according to the growth trend, calculates the surface area data and metabolic power of the organism, monitors the air flow velocity data at the back height to calculate the convective heat transfer coefficient, calculates the heat dissipation power by combining the body surface area and ambient air temperature data, and obtains the environmental enthalpy compensation amount based on the heat dissipation power and metabolic power.

[0039] The feeding amount decision module obtains basic feeding requirement data based on livestock breed, weights and corrects the basic feeding requirement data and the growth trend correction configuration, obtains the inertial correction amount, superimposes the inertial correction amount and the environmental enthalpy compensation amount, and generates the target total feeding weight.

[0040] The feeding behavior recognition module collects acoustic signal streams in the feed trough area, performs Hilbert transform and generates acoustic amplitude envelope data, constructs background noise baseline parameters, segments chewing peak segments and intermittent trough segments, updates the average chewing cycle parameters and generates the minimum swallowing pause limit, and generates a swallowing phase silence window when the intermittent trough segment exceeds the minimum swallowing pause limit.

[0041] The feeding execution control module responds to the trigger signal of the swallowing phase silent window period, drives the motor to release the single feed amount, deducts the single release amount from the target total feeding weight and updates the remaining feed amount record, terminates the command and generates a feeding task completion record when the remaining feed amount record is cleared to zero.

[0042] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0043] In this invention, livestock population growth data is collected and combined with theoretical models to assess the growth inertia state, generating a growth trend correction configuration to dynamically adapt to the actual development rate. Environmental wind speed and temperature data are monitored to calculate the convective heat transfer coefficient and the heat dissipation power of the organism, quantifying the environmental thermal effect and generating an enthalpy compensation configuration to offset the energy loss of cold stress. Based on the Hilbert transform of the sound wave signal and the background noise basis, the chewing and swallowing phases are accurately segmented and a silent window period is generated. The swallowing action triggers discrete pulse drive commands, realizing precise feeding from macroscopic growth trends to microscopic feeding rhythms, ensuring that energy supply and biological metabolic needs are synchronized in real time, and significantly improving feed utilization efficiency while protecting the growth potential of livestock. Attached Figure Description

[0044] Figure 1 This is a schematic diagram of the workflow of the present invention;

[0045] Figure 2 This is a flowchart illustrating the process of obtaining the growth trend correction configuration in this invention;

[0046] Figure 3 This is a flowchart for obtaining the environmental enthalpy compensation amount according to the present invention;

[0047] Figure 4 This is a flowchart illustrating how the total weight of the target feed is obtained according to the present invention.

[0048] Figure 5 This is a flowchart of the process for obtaining the silent window period of the swallowing phase in this invention;

[0049] Figure 6 This is a flowchart illustrating the process of obtaining the feeding task completion record according to the present invention. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0051] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0052] Please see Figure 1 This invention provides a technical solution, a method for intelligent control of feed feeding amount, comprising the following steps:

[0053] S1: Collect time-series data on average weight of livestock populations, calculate population growth acceleration data using discrete difference, obtain theoretical growth acceleration references by combining livestock breeds and compare them, and match growth trend correction configurations.

[0054] S2: Adjust the configuration according to the growth trend, calculate the surface area data and metabolic power of the organism, monitor the air flow velocity data at the back height to calculate the convective heat transfer coefficient, combine the surface area and ambient air temperature data to calculate the heat dissipation power, and obtain the environmental enthalpy compensation amount based on the heat dissipation power and metabolic power.

[0055] S3: Obtain basic feeding requirement data based on livestock breed, weight and correct the basic feeding requirement data and growth trend correction configuration, obtain inertial correction amount, superimpose inertial correction amount and environmental enthalpy compensation amount, and generate target total feeding weight.

[0056] S4: Collect acoustic signal flow in the feed trough area, perform Hilbert transform and generate acoustic amplitude envelope data, construct background noise basis parameters, segment chewing peak segments and intermittent trough segments, update average chewing cycle parameters and generate minimum swallowing pause limit, and generate swallowing phase silence window when the intermittent trough segment exceeds the minimum swallowing pause limit.

[0057] S5: Responding to the trigger signal during the swallowing phase silent window, drive the motor to release the single feed amount, deduct the single release amount from the target total feeding weight and update the remaining feed amount record, terminate the instruction and generate a feeding task completion record when the remaining feed amount record is cleared to zero.

[0058] The growth trend correction configuration includes growth rate gain coefficient, developmental inertia factor, and dynamic energy demand adjustment index. The environmental enthalpy compensation includes convective heat loss compensation value, cold stress energy compensation amount, and feed heat conversion increment. The target total feeding weight includes basal diet baseline mass, growth inertia correction mass, and environmental thermal added mass. The swallowing phase quiescent window period specifically refers to the window start timestamp, effective duration identifier, and swallowing action confidence. The feeding task completion record includes cumulative total feeding mass, pulse execution count, and task termination time node.

[0059] Please see Figure 2 The specific process for obtaining the growth trend correction configuration is as follows:

[0060] S111: Collect time-series data on the average weight of livestock herds, perform second-order discrete difference operations on the time dimension of the average weight time-series data of livestock herds, calculate the rate of change of weight increment at adjacent sampling times and perform smoothing filtering on the results to remove observation noise and random fluctuations, quantify the dynamic net growth rate characteristics of the herd within the current monitoring time window, and generate herd growth acceleration data.

[0061] Time-series data on the average weight of livestock herds were collected using four-point high-precision resistance strain gauge weighing sensors deployed at the bottom of the feeding station. The sensor sampling frequency was set to 1 Hz to capture the pressure signal when the livestock were standing in real time. Due to frequent hoof movements and center-of-gravity shifts during feeding, the raw signal contained a large amount of high-frequency mechanical vibration noise. First, a median filter with a sliding time window of 60 seconds was used for initial data cleaning. Then, a low-pass Butterworth digital filter with a cutoff frequency of 0.5 Hz was used to remove motion artifacts, thus obtaining an effective steady-state weight value with a signal-to-noise ratio higher than 20 dB. Second-order discrete difference operations were performed on the time-series data of the average weight of livestock herds. This operation logic was not a simple subtraction of adjacent points, but rather employed a five-point quadratic smoothing differential algorithm (Savitzky-Golay differential filter) to improve sensitivity to trend changes. First, the derivative of the daily average weight data is calculated to obtain the first derivative, i.e., the daily weight gain rate (kg / day). For example, at time t in the monitoring window, the average weight of the population is 85.50 kg, and at time t-1 it is 84.75 kg, thus calculating the instantaneous growth rate. Next, a difference operation is performed on the growth rate sequence to obtain the second derivative, i.e., the rate of change of weight gain, which directly characterizes the acceleration of growth. The calculation results need to be smoothed and filtered, using a dead-zone limiter with a threshold of 0.05 kg / day² to remove observation noise and random fluctuations caused by short-term mass mutations due to drinking or excretion. Finally, the dynamic net growth rate characteristics of the population within the current monitoring time window are quantified, generating population growth acceleration data. For example, in actual monitoring, the calculated growth acceleration of a certain batch of fattening pigs at 100 days of age was 0.025 kg / day². This value accurately reflects the release trend of the group's growth potential. The experimental data shows that, compared with the traditional daily weight gain index, the acceleration index based on second-order difference has an average lead time of 3.5 days when predicting the growth inflection point.

[0062] S112: Based on the livestock breed, obtain a pre-set theoretical growth curve model containing breed characteristics, read the age-time index parameters of the current breeding stage, deduce the standard metabolic growth acceleration under the current time index, construct a benchmark reference value, and establish a theoretical growth acceleration reference.

[0063] A pre-defined theoretical growth curve model matching the livestock breed characteristics is obtained. This model is stored in the non-relational database MongoDB and is built based on the Gompertz growth equation. Its mathematical expression includes three core biological parameters: the asymptotic value of mature body weight A (set to 120.5 kg), the age inflection point T at which the maximum growth rate occurs (set to 105 days), and the relative growth rate decay coefficient k (set to 0.018). The age-time index parameter of the current breeding stage is read (e.g., the current day is 98), and this time parameter is substituted into the second derivative form of the above model equation. The standard metabolic growth acceleration at the current time index is derived. This process is essentially taking the second derivative of the growth curve function with respect to time, that is, calculating the instantaneous rate of change of growth rate (daily weight gain) at the current moment, which is used to construct a benchmark reference value. For example, substituting t=98 into the equation, the theoretical growth acceleration reference value is calculated to be 0.032 kg / day². This reference value represents the standard growth acceleration capability achievable under ideal environmental and nutritional conditions, given the breed's genotype. To ensure the model's applicability, the model parameters A and k are periodically fine-tuned based on the farm's most recent five batches of historical slaughter data. Parameter updates utilize gradient descent, with the objective function being the minimization of the root mean square error of historical predictions. The learning rate is set to 0.001, and the number of iterations is 500, ensuring dynamic alignment between the theoretical baseline and the farm's current management level.

[0064] S113: Call the population growth acceleration data and theoretical growth acceleration reference, calculate the algebraic difference between the two, determine the current growth inertia state, retrieve the preset multidimensional adjustment coefficient matrix, find and extract the matching control parameter combination row, and generate the growth trend correction configuration.

[0065] The system retrieves population growth acceleration data (e.g., 0.025 kg / day²) and theoretical growth acceleration reference (e.g., 0.032 kg / day²), performs high-precision floating-point subtraction to calculate the numerical deviation between them, which is -0.007 kg / day². It then identifies the positive and negative deviation directions, with a negative value indicating that the actual growth kinetic energy lags behind the theoretical potential. The current growth inertia state is determined to be "Negative Lag Level II". A pre-set multidimensional adjustment coefficient matrix is ​​retrieved. This matrix is ​​a hash table stored in a Redis cache, where row indices correspond to the direction of the growth deviation (e.g., "Positive Lead", "Negative Lag"), and column indices correspond to the absolute value range of the deviation (e.g., [0, 0.005), [0.005, 0.010)). Matching control parameter combination rows are found and extracted. For the aforementioned deviation of -0.007, the data in the intersection cell of the "Negative Lag" row and the "[0.005, 0.010)" column is matched. The growth trend correction configuration stored in this unit includes an energy density correction factor of 1.05 and a feed ratio compensation coefficient of 1.08. The logic behind this process is that when growth shows a slowing trend, it is not only necessary to increase the feed amount, but also to simultaneously increase the energy concentration of the feed to compensate for the metabolic deficit.

[0066] Please see Figure 3 The process of obtaining the environmental enthalpy compensation is as follows:

[0067] S211: Monitor air flow velocity data and ambient air temperature data at the back elevation, calculate the convective heat transfer coefficient based on the air flow velocity data at the back elevation, extract the current average weight of the population associated with the growth trend correction configuration, calculate the biological body surface area data and basal metabolic heat production power data based on the current average weight of the population, and establish a set of basic biological thermal parameters.

[0068] Data on airflow velocity and ambient air temperature at the back height of the pigsty were monitored using a hot-wire anemometer and a PT1000 platinum resistance temperature sensor, suspended 0.8 meters above the pigpen (corresponding to the average back height of fattening livestock). The sampling period was 30 seconds. Based on the airflow velocity data at the back height (e.g., 0.8 m / s), the convective heat transfer coefficient was calculated using empirical formulas for horizontal turbulence in fluid mechanics. The Reynolds number Re was calculated to be approximately 45000. The corresponding Nusselt number Nu was obtained from a table, and the convective heat transfer coefficient h was then calculated to be approximately 12.5 W / (m²·Kelvin). The average weight of the current livestock population (e.g., 85.5 kg) was used. Based on this weight data, the body surface area was calculated using the Meeh formula (body surface area = k × weight^0.67, k = 0.09), resulting in approximately 1.76 m². Meanwhile, the basal metabolic rate (BMR) was calculated based on Kleiber's law (basal metabolic rate = 70 × body weight^0.75), and an activity-based metabolic correction factor (valued at 2.0) was introduced to cover both standing and food intake expenditures. The units were converted to International Units (SI). The calculation process is as follows: The result was approximately 195 watts. A set of basic biological thermal parameters was established, which included not only the instantaneous values ​​mentioned above but also the moving average over the past hour to eliminate the influence of instantaneous fluctuations in environmental parameters and ensure the stability of thermal calculations.

[0069] S212: Call the biological thermal basic parameter set, combine the ambient air temperature data and convective heat transfer coefficient to perform convective heat transfer calculation, calculate the heat flux loss rate of the livestock body surface, multiply the heat flux loss rate by the biological body surface area data, and generate total heat dissipation power data.

[0070] The biological thermal fundamental parameter set is invoked, and convective heat transfer calculations are performed using ambient air temperature data (e.g., 18 degrees Celsius) and the convective heat transfer coefficient (12.5 W / (m²·Kelvin)). First, the average body surface temperature of the livestock is measured or estimated (usually set to 34 degrees Celsius), and the temperature difference driving force (34-18=16 degrees Celsius) is calculated. The heat flux loss rate of the livestock's body surface is calculated, i.e., heat flux = convective heat transfer coefficient × temperature difference = 12.5 × 16 = 200 W / m². Based on the biological body surface area data (1.76 m²), the heat flux loss rate is integrally processed by area, i.e., 200 × 1.76, generating a total heat dissipation power of 352 W. In this step, if the ambient humidity exceeds 75%, a latent heat correction coefficient of 1.1 is introduced to compensate for the total heat dissipation power, reflecting the "wet cold" effect caused by increased air thermal conductivity and decreased insulation effectiveness of the livestock's fur in high humidity environments; that is, humid air accelerates the conduction and loss of heat from the body surface. In a specific example, if the current humidity is 80%, the corrected total heat dissipation power is 352 × 1.1 = 387.2 watts. This data accurately quantifies the actual rate of heat energy loss of an organism under the current physical environment.

[0071] S213: Use the total heat dissipation power data and the basal metabolic heat production power data from the biological thermal fundamental parameter set, employing the following formula:

[0072] ;

[0073] Calculate the environmental enthalpy compensation amount;

[0074] in, The environmental enthalpy compensation amount represents the additional feed mass required to offset the environmental heat effect within a single control cycle. The control cycle duration represents the time window length for a single feeding control, and is read from the preset timer configuration. This is a net energy parameter for feed, representing the available energy per unit mass of feed that can be used by organisms for production and maintenance. It is retrieved from a feed nutrient database. This represents the total heat dissipation power data, indicating the rate at which heat is lost from the livestock's body surface to the environment. This is a turbulence correction factor, representing an enhancement factor for convective heat transfer under non-laminar conditions. It is obtained by looking up the Reynolds number range based on the airflow velocity data at the back elevation. Basal metabolic heat production power data represents the endogenous heat generated by livestock to maintain vital functions. It is calculated based on body weight data using the allometric growth equation. The body heat volume factor represents the sensible heat capacity characteristic of an organism to resist temperature changes, and is preset based on livestock breed characteristic data. This is the temperature difference modulus, representing the absolute temperature difference between the body surface reference temperature and the ambient air temperature. It is obtained through calculation of the difference between sensor-monitored data. The stress time constant represents the response hysteresis characteristics of the body's thermal regulation system, and is obtained by fitting historical thermal stress experimental data.

[0075] The total heat dissipation power data (set to 387.2 watts) and the basal metabolic heat production power data (set to 195 watts) from the biological thermal fundamental parameter set are used, and these parameters are then substituted into the formula for calculating the environmental enthalpy compensation. In the formula, the duration of the regulation cycle is... Set to 14400 seconds (i.e., adjusted every 4 hours); Net energy parameters of feed The value retrieved from the database is 9800 joules / gram (i.e., 9.8 megajoules / kilogram). Turbulence correction factor. The value was obtained from a table based on the Reynolds number range corresponding to a measured wind speed of 0.8 m / s, and was set to 1.2 to correct for the enhanced heat transfer effect of non-laminar wind fields. (Body heat volume factor) The value is selected as 3.4 kJ / (kg·°C) to represent the heat storage capacity of the pig's body; the temperature difference modulus 16 degrees Celsius; stress time constant The time was set to 1800 seconds, based on measurements from a cold stress recovery experiment. The values ​​were then substituted into the formula for calculation:

[0076] explicit heat loss item watt;

[0077] Latent stress compensation items watt;

[0078] The sum of the two is 346.34 watts. The final calculation of the environmental enthalpy compensation is as follows:

[0079] The calculation results indicate that, in order to offset the heat loss caused by the current low temperature and high wind speed environment, an additional 509 grams of feed needs to be added to the basal diet to prevent livestock from using body fat to maintain body temperature. Table 1 lists the comparison of the compensation values ​​calculated under different environmental parameters.

[0080] Table 1. Comparison of Environmental Parameters and Enthalpy Compensation Configurations:

[0081]

[0082] As shown in Table 1, the calculated compensation amount increases significantly and nonlinearly as environmental conditions deteriorate.

[0083] Please see Figure 4 The specific process for obtaining the target total weight of feed is as follows:

[0084] S311: Input the average weight of the livestock population and the age of the livestock as an index into the theoretical growth curve model, traverse the nutrient allocation logic embedded in the model, match the unit net energy value required to maintain basal metabolism and standard weight gain, combine the energy conversion density of the pre-set feed formula, quantify the theoretical feed consumption quality under the theoretical preset path, and generate basic feeding demand data.

[0085] The average weight of the livestock population (85.5 kg) and the age at which they were raised (98 days) were used as indices to input into the theoretical growth curve model, traversing the nutrient allocation logic embedded in the model. This logic is based on the NRC (National Research Council) swine nutrition standards, matching the net energy required to maintain basal metabolism to 1.5 MJ / day, and the net energy required for production to achieve standard weight gain (assuming a target daily weight gain of 900 g) to 22.5 MJ / day, for a total energy requirement of 24.0 MJ / day. Combined with the energy conversion density of the pre-set feed formula (9.8 MJ / kg), the theoretical feed consumption mass under the theoretically preset path was quantified, calculated as 24.0 / 9.8 = 2.449 kg. This value is defined as the daily basal requirement. If the current feeding window is 4 hours, then the basal amount per feeding is 2.449 / 6 ≈ 0.408 kg, or 408 g. Basal feeding requirement data is generated. This step is based on an idealized model, not considering environmental and individual differences, and serves as the cornerstone for subsequent dynamic adjustments.

[0086] S312: Based on basic feeding demand data, read the growth trend correction configuration, dynamically scale the benchmark feed amount according to the deviation of the actual growth acceleration from the theoretical value, determine the correction feed value that matches the current actual growth rate, and generate the inertia correction amount.

[0087] Based on the basic feeding requirement data (408 grams), the growth trend correction configuration (feeding ratio compensation coefficient 1.08) is read. The baseline feed amount is dynamically scaled according to the deviation of the actual growth acceleration from the theoretical value, performing a multiplication operation: 408 grams × 1.08 = 440.64 grams. This result is judged to be a corrected feed value that matches the current actual growth rate, meaning that in order to catch up with the previously detected growth lag (the negative lag determined in S113), the feed amount needs to be artificially increased by 8%. An inertial correction amount (440.64 grams) is generated. A safety check is also performed here. If the correction coefficient exceeds 1.2 (i.e., an increase of more than 20%), an "abnormal overfeeding risk warning" is triggered, and the coefficient is forcibly clamped at 1.2 to prevent overeating-related digestive system diseases caused by data jumps. In this embodiment, 1.08 is within the safe range.

[0088] S313: Call the inertial correction amount, combine it with the environmental enthalpy compensation amount to perform linear superposition operation, and perform discretization and rounding and preset maximum feeding threshold truncation processing on the merged result according to the minimum feeding resolution and maximum single feeding limit of the automatic feeding equipment, construct the execution instruction of the single feeding task, and generate the target total feeding weight.

[0089] The inertial correction (440.64 grams) is combined with the environmental enthalpy compensation (509 grams) using a linear superposition operation, resulting in a preliminary sum of 440.64 + 509 = 949.64 grams. This value reflects the total demand to meet both the nutritional needs for catch-up growth and to offset the heat loss caused by severe cold stress. Subsequently, based on the minimum feeding resolution of the automatic feeding equipment (the physical limitation of the screw motor's rotation accuracy, e.g., 10 grams per pulse) and the maximum single feeding limit (e.g., no more than 1500 grams per feeding to prevent overflow), the combined result is discretized, rounded, and truncated using a preset maximum feeding threshold. 949.64 grams is rounded down to a multiple of 10 grams, i.e., 940 grams. The execution instruction for a single feeding task is constructed, and this value is encapsulated in the instruction package, generating a target total feeding weight of 940 grams. This process realizes the practical conversion from theoretical calculation to hardware execution instructions, ensuring that the final feeding amount meets both physiological needs and is within the mechanical execution capability.

[0090] Please see Figure 5 The process of obtaining the silent window period of the swallowing phase is as follows:

[0091] S411: Collect acoustic signal stream in the trough area, perform Hilbert transform on the acoustic signal stream and generate acoustic amplitude envelope data, construct background noise floor parameters by performing histogram statistical analysis, segment peak morphology and trough segments according to background noise floor parameters, extract the time interval between adjacent peaks and generate chewing cycle time series set.

[0092] Acoustic signal streams were collected from the feeding trough area using a MEMS microphone array mounted on the inner edge of the feeding trough, with a sampling depth of 16 bits and a sampling rate of 44.1 kHz. Hilbert transform was performed on the acoustic signal stream to convert the real-domain time-domain signal into an analytic signal, thereby extracting its instantaneous amplitude and generating acoustic amplitude envelope data. This processing effectively eliminated carrier frequency interference, highlighting the energy fluctuation characteristics of chewing motion. Histogram statistical analysis was performed on the acoustic amplitude envelope data to construct background noise floor parameters. The distribution of envelope amplitude over the past 10 seconds was statistically analyzed, and the interval with the highest frequency and lowest amplitude in the histogram (usually the left-hand peak pattern) was selected as ambient noise. The 95th percentile of this pattern (e.g., amplitude 0.05V) was set as the noise floor threshold. Based on the background noise floor parameters, peak and trough segments were segmented. Segments with amplitudes consistently above 0.05V for more than 100 milliseconds were marked as chewing peaks, and segments below this threshold were marked as troughs. Extract the time interval between adjacent peaks and generate a chewing cycle time series set, for example, a series of interval data were collected: [850ms, 820ms, 880ms, 840ms, ...].

[0093] S412: Call the chewing cycle time series set, calculate the arithmetic mean and discrete standard deviation of the time intervals, using the formula:

[0094] ;

[0095] Calculate and establish the minimum swallowing pause limit;

[0096] in, The minimum swallowing pause limit, The arithmetic mean of the time intervals represents the baseline average chewing cycle for the current eating phase. It is obtained by averaging historical data from a chewing cycle time series set. The fluctuation-weighted coefficient represents the amount of weight adjustment for the width of the impact of chewing rhythm instability, and is read from the preset algorithm configuration parameters. The standard deviation of the time interval represents the degree of dispersion or random fluctuation of the chewing rhythm cycle, obtained through statistical calculations. The system response delay time represents the inherent hardware lag time from the acoustic wave acquisition link to the control response end, obtained through hardware calibration testing. The physiological minimum swallowing time represents the theoretically shortest anatomical time required for this breed of livestock to complete one bolus swallowing action, retrieved from a biological characteristic database. The statistical confidence factor represents the scaling ratio used to adjust the physiological baseline value to control the false alarm rate, and is obtained by mapping based on a preset allowable false alarm rate index.

[0097] Use the chewing cycle time series set (such as the dataset above) to calculate the arithmetic mean of the time intervals. (e.g., 847.5 milliseconds) and discrete standard deviation (For example, 25.8 milliseconds). The minimum swallowing pause limit is calculated using the formula, where the parameters are selected as follows: fluctuation weighting coefficient. Set to 3.0, based on the normal distribution of 3. In principle, it covers 99.7% of chewing fluctuations; response latency time The time was measured to be 35 milliseconds through a high-speed camera and audio synchronization calibration experiment; this is the minimum physiological swallowing time. The shortest anatomical time for swallowing a food bolus in this breed of pig was determined by fluoroscopic swallowing smear (VFSS), set at 600 milliseconds; confidence factors were statistically analyzed. The value is set to 1.1 to moderately tighten the lower physiological limit and improve the specificity of the judgment. Substitute the value into the formula:

[0098] Composite deviation term = milliseconds; physiological constraints = Milliseconds. Ultimately. Milliseconds. A minimum swallowing pause limit (approximately 1450 milliseconds) was calculated and established. Table 2 shows a comparison of the effect of this dynamic limit with a traditional fixed threshold.

[0099] Table 2. Comparison of swallowing behavior recognition accuracy experimental data:

[0100]

[0101] As shown in Table 2, by introducing dynamic calculations of statistical fluctuations and physiological constraints, the probability of misjudging normal long chewing intervals as swallowing is reduced.

[0102] S413: Monitor the duration of the current trough segment in real time, compare the duration with the minimum swallowing pause limit in the time domain, and trigger the state latching logic when the duration exceeds the minimum swallowing pause limit to generate a swallowing phase silence window.

[0103] The duration of the current trough segment is monitored in real time, and the timer value currently at a low level (i.e., trough state) is continuously refreshed during millisecond-level clock interrupts. The duration is compared with the minimum swallowing pause threshold (1450 milliseconds) in the time domain. Assuming that a pause has been detected for 1451 milliseconds, the condition is met. When the duration exceeds the minimum swallowing pause threshold, the state latching logic is triggered, immediately marking the current state as "swallowing certainty state" and generating a swallowing phase silence window signal. This signal is a high-level pulse containing a timestamp, indicating that the livestock has just completed a real swallowing action, the esophagus is empty, and it is waiting for the next feeding, which is the optimal time window for feeding intervention.

[0104] Please see Figure 6 The specific process for obtaining the feeding task completion record is as follows:

[0105] S511: Responds to the trigger signal of the swallowing phase silent window period, calls the motor drive timing parameters to construct discrete pulse drive instructions, controls the screw conveyor motor to perform a single quantitative rotational motion, collects the angular displacement signal of the motor shaft and verifies the mechanical action, and generates a single pulse feeding execution record.

[0106] In response to the trigger signal during the swallowing phase silence window, the microcontroller (MCU) immediately wakes up the motor drive subroutine. It calls the motor drive timing parameters to construct discrete pulse drive instructions, setting the PWM carrier frequency to 20kHz and the duty cycle to 85% to provide sufficient starting torque. It controls the screw conveyor motor to perform a single quantitative rotation, for example, driving the motor to rotate 90 degrees (1 / 4 turn). After mechanical calibration, this rotation angle corresponds to the precise release of 10 grams of feed. During motor execution, a Hall encoder installed at the motor's tail collects the angular displacement signal of the motor shaft and verifies the mechanical action, ensuring that the motor actually rotates past 90 degrees and is not stuck due to feed jamming. If the encoder feedback angle is insufficient, a reverse feeding program is automatically executed. A single pulse feeding execution record is generated, including the execution timestamp (e.g., 12:00:01.050) and the actual feed amount (10 grams).

[0107] S512: Call the single pulse feeding execution record, deduct the single rotation feeding weight calibration value from the target total feeding weight according to the single rotation feeding weight calibration value, write the calculation result back to the memory to overwrite the original value, and establish real-time remaining feed quantity status data;

[0108] The system retrieves the single-pulse feeding execution record (10 grams) and deducts the single-pulse feeding weight calibration value from the target total feeding weight (940 grams) based on this value. The calculation logic is: Remaining amount = Current remaining amount - 10. Initially, the current remaining amount is 940 grams, which becomes 930 grams after the first deduction. The calculation result is written back to memory to overwrite the original value, establishing real-time remaining feed status data. This process is performed at high speed in SRAM to prevent data races in a multi-threaded environment.

[0109] S513: Obtain real-time data on the remaining feed quantity, compare it with the zero-value cutoff bit in real time, send a blocking command to the motor control port when the feed quantity reaches or falls below the zero-value cutoff bit, summarize the start timestamp and cumulative release weight of each feeding cycle, and generate a feeding task completion record.

[0110] Acquire real-time data on the remaining feed quantity (e.g., after 93 feedings, the remaining amount drops to 10 grams; after the 94th feeding, the remaining amount drops to 0 grams). Compare this data in real-time with the zero-value cutoff (0 grams), and determine when the data reaches or falls below the zero-value cutoff (i.e., the remaining amount). 0) Immediately send a blocking command to the motor control port to terminate the pulse drive command, pull the GPIO output low, cut off the enable signal of the motor driver, physically prohibit subsequent feeding actions, and prevent overfeeding. At this time, summarize the start timestamp (12:00:00), end timestamp (12:15:30), and cumulative released weight (940 grams) of this feeding cycle, pack these data into non-volatile Flash memory, generate a feeding task completion record, and send a "task completed" status code to the host computer.

[0111] A feed feeding rate intelligent control system is provided, which is used to execute the above-mentioned feed feeding rate intelligent control method. The system includes:

[0112] The growth trend analysis module collects time-series data on the average weight of livestock groups, calculates the growth acceleration data of the group using discrete difference, obtains theoretical growth acceleration references by combining them with livestock breeds and compares them, and matches the growth trend correction configuration.

[0113] The environmental compensation calculation module adjusts the configuration according to the growth trend, calculates the surface area data and metabolic power of the organism, monitors the air flow velocity data at the back height to calculate the convective heat transfer coefficient, calculates the heat dissipation power by combining the body surface area and ambient air temperature data, and obtains the environmental enthalpy compensation amount based on the heat dissipation power and metabolic power.

[0114] The feeding amount decision module obtains basic feeding requirement data based on livestock breed, weights and corrects the basic feeding requirement data and growth trend correction configuration, obtains inertial correction amount, superimposes inertial correction amount and environmental enthalpy compensation amount, and generates target total feeding weight.

[0115] The feeding behavior recognition module collects acoustic signal streams in the feed trough area, performs Hilbert transform and generates acoustic amplitude envelope data, constructs background noise baseline parameters, segments chewing peak segments and intermittent trough segments, updates the average chewing cycle parameters and generates the minimum swallowing pause limit, and generates a swallowing phase silence window when the intermittent trough segment exceeds the minimum swallowing pause limit.

[0116] The feeding execution control module responds to the trigger signal during the swallowing phase silent window, drives the motor to release the amount of feed in a single feeding, deducts the amount released in a single feeding from the target total feeding weight and updates the remaining amount to be fed record, terminates the command and generates a feeding task completion record when the remaining amount to be fed record is cleared to zero.

[0117] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A method for intelligent control of feed feeding amount, characterized in that, Includes the following steps: S1: Collect time-series data on average weight of livestock populations, calculate population growth acceleration data using discrete difference, obtain theoretical growth acceleration references by combining livestock breeds and compare them, and match growth trend correction configurations. S2: Based on the growth trend, the configuration is corrected, the surface area data and metabolic power of the organism are calculated, the air flow velocity data at the back height is monitored to calculate the convective heat transfer coefficient, the heat dissipation power is calculated by combining the surface area and ambient air temperature data, and the environmental enthalpy compensation amount is obtained based on the heat dissipation power and metabolic power. S3: Obtain basic feeding requirement data based on livestock breed, weight and correct the basic feeding requirement data and the growth trend correction configuration, obtain the inertial correction amount, superimpose the inertial correction amount and the environmental enthalpy compensation amount, and generate the target total feeding weight. S4: Collect acoustic signal flow in the feed trough area, perform Hilbert transform and generate acoustic amplitude envelope data, construct background noise basis parameters, segment chewing peak segments and intermittent trough segments, update average chewing cycle parameters and generate minimum swallowing pause limit, and generate swallowing phase silence window when the intermittent trough segment exceeds the minimum swallowing pause limit. S5: In response to the trigger signal of the swallowing phase silent window period, drive the motor to release the single feed amount, deduct the single release amount from the target total feeding weight and update the remaining feed amount record, terminate the instruction and generate a feeding task completion record when the remaining feed amount record is cleared to zero.

2. The intelligent feed feeding amount control method according to claim 1, characterized in that, The growth trend correction configuration includes a growth rate gain coefficient, a developmental inertia factor, and a dynamic energy demand adjustment index. The environmental enthalpy compensation includes a convective heat loss compensation value, a cold stress energy compensation amount, and a feed heat conversion increment. The target total feeding weight includes the basal diet baseline mass, the growth inertia correction mass, and the environmental thermal added mass. The swallowing phase quiescent window period specifically refers to the window start timestamp, the effective duration period identifier, and the swallowing action confidence level. The feeding task completion record includes the cumulative total feeding mass, pulse execution count, and task termination time node.

3. The intelligent feed feeding method according to claim 1, characterized in that, The process of obtaining the growth trend correction configuration is as follows: S111: Collect time-series data on the average weight of livestock herds, perform second-order discrete difference operations on the time dimension of the average weight time-series data of livestock herds, calculate the rate of change of weight increment at adjacent sampling times and perform smoothing filtering on the results to remove observation noise and random fluctuations, quantify the dynamic net growth rate characteristics of the herd within the current monitoring time window, and generate herd growth acceleration data. S112: Based on the livestock breed, obtain a pre-set theoretical growth curve model containing breed characteristics, read the age-time index parameters of the current breeding stage, deduce the standard metabolic growth acceleration under the current time index, construct a benchmark reference value, and establish a theoretical growth acceleration reference. S113: Call the population growth acceleration data and the theoretical growth acceleration reference, calculate the algebraic difference between the two, determine the current growth inertia state, retrieve the preset multidimensional adjustment coefficient matrix, find and extract the matching control parameter combination row, and generate the growth trend correction configuration.

4. The intelligent feed feeding method according to claim 3, characterized in that, The specific process for obtaining the theoretical growth curve model is as follows: Data on the weight and age of corresponding livestock breeds throughout their life cycle under standardized feeding conditions were collected. The nonlinear least squares method was used to perform numerical fitting on the data to determine the asymptotic limit weight, maximum relative growth rate, and growth inflection point time parameters that characterize growth features. A continuous time function was generated and stored.

5. The intelligent feed feeding method according to claim 3, characterized in that, The process of establishing the multidimensional adjustment coefficient matrix is ​​as follows: A discrete deviation amplitude gradient covering the positive and negative deviation directions is set. For each gradient, a multi-level nutrient level compensation feeding test is performed. Growth rate recovery data under different combinations of feed energy and feeding amount are recorded. Based on the data, the growth rate is reverse-solved to return to the target control parameters of the theoretical growth acceleration reference. The corresponding energy density correction value and feeding amount ratio value are extracted and written into the storage unit according to the deviation direction and amplitude index mapping.

6. The intelligent feed feeding method according to claim 3, characterized in that, The process of obtaining the environmental enthalpy compensation is as follows: S211: Monitor the air flow velocity data and ambient air temperature data at the back height, calculate the convective heat transfer coefficient based on the air flow velocity data at the back height, extract the current average weight of the population associated with the growth trend correction configuration, calculate the biological body surface area data and basal metabolic heat production power data based on the current average weight of the population, and establish a set of biological thermal basic parameters. S212: Call the biological thermal basic parameter set, combine the ambient air temperature data and convective heat transfer coefficient to perform convective heat transfer calculation, calculate the heat flux loss rate of the livestock body surface, multiply the heat flux loss rate by the biological body surface area data, and generate total heat dissipation power data. S213: Call the total heat dissipation power data and the basic metabolic heat production power data in the biological thermal fundamental parameter set to calculate the environmental enthalpy compensation.

7. The intelligent feed feeding amount control method according to claim 6, characterized in that, The process of obtaining the target total feeding weight is as follows: S311: Input the average weight of the livestock population and the age of the livestock as an index into the theoretical growth curve model, traverse the nutrient allocation logic embedded in the model, match the unit net energy value required to maintain basal metabolism and standard weight gain, combine the energy conversion density of the pre-set feed formula, quantify the theoretical feed consumption quality under the theoretical preset path, and generate basic feeding demand data. S312: Based on the basic feeding requirement data, read the growth trend correction configuration, dynamically scale the benchmark feed amount according to the deviation of the actual growth acceleration from the theoretical value, determine the correction feed value that conforms to the current actual growth rate, and generate the inertia correction amount. S313: Call the inertial correction amount, combine it with the environmental enthalpy compensation amount to perform linear superposition operation, and perform discretization and rounding and preset maximum feeding threshold truncation processing on the merged result according to the minimum feeding resolution and maximum single feeding limit of the automatic feeding equipment, construct the execution instruction of the single feeding task, and generate the target total feeding weight.

8. The intelligent feed feeding method according to claim 7, characterized in that, The process of obtaining the swallowing phase silent window is as follows: S411: Collect acoustic signal stream in the trough area, perform Hilbert transform on the acoustic signal stream and generate acoustic amplitude envelope data, construct background noise floor parameters by performing histogram statistical analysis, segment peak morphology and trough segments according to background noise floor parameters, extract the time interval between adjacent peaks and generate chewing cycle time series set. S412: Call the chewing cycle time series set, calculate the arithmetic mean and discrete standard deviation of the time intervals, and establish the minimum swallowing pause limit; S413: Monitor the duration of the current trough segment in real time, compare the duration with the minimum swallowing pause limit in the time domain, and trigger the state latching logic when the duration exceeds the minimum swallowing pause limit to generate a swallowing phase silence window.

9. The intelligent feed feeding method according to claim 8, characterized in that, The process of obtaining the feeding task completion record is as follows: S511: In response to the trigger signal of the swallowing phase silent window period, call the motor drive timing parameters to construct a discrete pulse drive command, control the screw conveyor motor to perform a single quantitative rotation motion, collect the angular displacement signal of the motor shaft and verify the mechanical action, and generate a single pulse feeding execution record. S512: Call the single pulse feeding execution record, deduct the single rotation feeding weight calibration value from the target total feeding weight according to the single rotation feeding weight calibration value, write the calculation result back to the memory to overwrite the original value, and establish real-time remaining feed quantity status data; S513: Obtain the real-time remaining feed quantity status data, compare it with the zero value cutoff bit in real time, send a blocking command to the motor control port when the feed quantity reaches or falls below the zero value cutoff bit, summarize the start timestamp and cumulative release weight of each feeding cycle, and generate a feeding task completion record.

10. A smart feed feeding rate control system, characterized in that, The system is used to implement the intelligent feed feeding amount control method according to any one of claims 1-9, the system comprising: The growth trend analysis module collects time-series data on the average weight of livestock groups, calculates the growth acceleration data of the group using discrete difference, obtains theoretical growth acceleration references by combining them with livestock breeds and compares them, and matches the growth trend correction configuration. The environmental compensation calculation module corrects the configuration according to the growth trend, calculates the surface area data and metabolic power of the organism, monitors the air flow velocity data at the back height to calculate the convective heat transfer coefficient, calculates the heat dissipation power by combining the body surface area and ambient air temperature data, and obtains the environmental enthalpy compensation amount based on the heat dissipation power and metabolic power. The feeding amount decision module obtains basic feeding requirement data based on livestock breed, weights and corrects the basic feeding requirement data and the growth trend correction configuration, obtains the inertial correction amount, superimposes the inertial correction amount and the environmental enthalpy compensation amount, and generates the target total feeding weight. The feeding behavior recognition module collects acoustic signal streams in the feed trough area, performs Hilbert transform and generates acoustic amplitude envelope data, constructs background noise baseline parameters, segments chewing peak segments and intermittent trough segments, updates the average chewing cycle parameters and generates the minimum swallowing pause limit, and generates a swallowing phase silence window when the intermittent trough segment exceeds the minimum swallowing pause limit. The feeding execution control module responds to the trigger signal of the swallowing phase silent window period, drives the motor to release the single feed amount, deducts the single release amount from the target total feeding weight and updates the remaining feed amount record, terminates the command and generates a feeding task completion record when the remaining feed amount record is cleared to zero.

Images