An intelligent heating control method and system for livestock farms

Abstract

The present invention relates to the field of intelligent heating control technology, and in particular to an intelligent heating control method and system for livestock farms. The method comprises the following steps: monitoring the behavior and activity status of livestock in the livestock farm through electronic monitoring equipment, and acquiring temperature, humidity, and greenhouse gas fluctuation data in real time in combination with temperature and humidity sensors and greenhouse gas sensors; generating livestock posture difference data by performing posture difference analysis on the monitoring video; further deconstructing spatial thermal energy vector loss through temperature, humidity, and gas data to obtain heat energy loss data, and calculating livestock body hypothermia data in combination with the posture difference data. Based on these data, the heating hot steam cycle demand is matched, and a heating adaptive regulation model is constructed. The present invention makes the intelligent heating control technology more perfect by optimizing the intelligent heating control technology.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent heating control, and in particular to an intelligent heating control method and system for livestock farms. Background Art

[0002] Previous heating methods for livestock farms often relied on manual adjustments, making it difficult to achieve precise control of environmental factors such as temperature, humidity, and gas concentrations, and unable to respond promptly to climate change and real-time changes in livestock health. Intelligent heating systems, through the introduction of advanced electronic monitoring, environmental sensing, and data analysis technologies, can monitor multiple key indicators, including livestock behavior, ambient temperature and humidity, and greenhouse gas concentrations, in real time. They can then combine artificial intelligence algorithms for precise analysis and dynamic adjustment. By monitoring and analyzing livestock behavior, combined with the spatial distribution of environmental heat loss, the system can intelligently match heating needs. However, traditional intelligent heating control methods suffer from low accuracy in analyzing spatial heat loss in livestock farms and the degree of hypothermia among livestock of different body shapes, resulting in large errors in heating adjustment. Summary of the Invention

[0003] Based on this, it is necessary to provide an intelligent heating control method and system for livestock farms to solve at least one of the above technical problems.

[0004] To achieve the above object, a method for intelligent heating control of a livestock farm is provided, the method comprising the following steps:

[0005] Step S1: Monitor the behavior and activity status of livestock in a livestock farm using electronic monitoring equipment to obtain a livestock behavior and activity status monitoring video; monitor the temperature and humidity fluctuations and greenhouse gas fluctuations in the livestock farm under extreme cold weather using the temperature and humidity sensors and greenhouse gas sensors in the environmental sensing module to obtain temperature and humidity fluctuation data and greenhouse gas fluctuation data respectively;

[0006] Step S2: Performing posture difference analysis on livestock behavior and activity status monitoring videos to generate livestock posture difference data; performing spatial thermal energy vector loss deconstruction on temperature and humidity fluctuation data and greenhouse gas fluctuation data to obtain spatial thermal energy vector loss deconstruction data; performing posture hypothermia measurement integration between livestock of different postures on the livestock posture difference data based on the spatial thermal energy vector loss deconstruction data to generate livestock posture hypothermia measurement data;

[0007] Step S3: Based on the livestock hypothermia learning data, the heating hot steam circulation demand between livestock of different body shapes is matched to obtain the heating hot steam circulation demand; a heating regulation model is constructed according to the heating hot steam circulation demand to obtain a heating adaptive regulation model, and the heating adaptive regulation model is sent to the cloud platform to implement intelligent heating control of the livestock farm.

[0008] Preferably, step S2 includes the following steps:

[0009] Step S21: performing posture difference analysis on the livestock behavior and activity status monitoring video to generate livestock posture difference data;

[0010] Step S22: performing group movement frequency analysis on the livestock behavior and activity status monitoring video to obtain livestock group movement frequency data;

[0011] Step S23: Acquire livestock farm structural design data; perform gas disturbance dispersion analysis on greenhouse gas fluctuation data based on livestock group movement frequency data and livestock farm structural design data to obtain greenhouse gas disturbance dispersion structure data;

[0012] Step S24: performing spatial thermal energy vector loss deconstruction on the temperature and humidity fluctuation data and the greenhouse gas disturbance dispersion structure data according to the livestock farm structural design data to obtain spatial thermal energy vector loss deconstruction data;

[0013] Step S25: performing a body hypothermia measurement integration between livestock with different body shapes on the livestock body shape difference data based on the spatial thermal energy vector loss deconstruction data to generate livestock body hypothermia measurement data.

[0014] Preferably, step S23 includes the following steps:

[0015] Step S231: Acquire livestock farm structural design data; perform gas circulation design path analysis on the livestock farm structural design data to obtain a gas circulation design path;

[0016] Step S232: performing group gait acceleration frequency analysis on the livestock group movement frequency data to obtain the group gait acceleration frequency;

[0017] Step S233: Calculating the continuous pressure difference of the peripheral airflow based on the group gait acceleration frequency to obtain the continuous pressure difference of the peripheral airflow;

[0018] Step S234: Deducing the change of azimuthal velocity difference of the gas flow section on the designed gas flow path according to the continuous pressure difference driven by the surrounding airflow, and obtaining the change data of azimuthal velocity difference of the gas flow section;

[0019] Step S235: performing mass / volume regional distribution difference analysis on the greenhouse gas fluctuation data to obtain the greenhouse gas mass / volume regional distribution difference;

[0020] Step S236: performing gas disturbance dispersion analysis on the greenhouse gas mass / volume regional distribution difference based on the flow section azimuthal velocity difference change data to obtain greenhouse gas disturbance dispersion structure data.

[0021] Preferably, step S24 includes the following steps:

[0022] Step S241: performing material thermal insulation performance degradation analysis on the livestock farm structural design data to obtain material thermal insulation performance degradation data;

[0023] Step S242: performing temperature and humidity spatial fluctuation distribution mapping on the temperature and humidity fluctuation data according to the livestock farm structural design data to obtain temperature and humidity spatial fluctuation distribution mapping data;

[0024] Step S243: Based on the material thermal insulation performance loss data, the spatial heat energy loss divergence per unit time is calculated for the temperature and humidity spatial fluctuation distribution mapping data and the greenhouse gas disturbance dispersion structure data to obtain spatial heat energy loss divergence data;

[0025] Step S244: performing spatial heat energy vector loss deconstruction based on the spatial heat energy loss divergence data to obtain spatial heat energy vector loss deconstruction data.

[0026] Preferably, step S243 includes the following steps:

[0027] The incremental coefficient of structural pore thermal conductivity is calculated based on the thermal insulation performance loss data of the material to obtain the incremental coefficient of structural pore thermal conductivity;

[0028] Based on the incremental thermal conductivity coefficient of structural pores, the interface heat transfer acceleration proportional identification is performed on the material insulation performance loss data to obtain the interface heat transfer acceleration proportional data;

[0029] The low temperature / low humidity relative change rate difference is calculated for the temperature and humidity spatial fluctuation distribution mapping data to generate the spatial distribution low temperature / low humidity relative change rate difference;

[0030] According to the relative change rate difference of low temperature / low humidity in spatial distribution, the spatial discrete heat loss density value increment of greenhouse gas disturbance dispersion structure data per unit time is integrated to obtain the spatial heat loss density increment value per unit time;

[0031] Based on the interface heat transfer acceleration proportional data and the spatial heat loss density increment value per unit time, the spatial heat energy loss divergence per unit time is calculated to obtain the spatial heat energy loss divergence data.

[0032] Preferably, step S25 includes the following steps:

[0033] Step S251: performing an extreme value analysis of low temperature tolerance among livestock with different body shapes on the livestock body shape difference data to obtain the extreme values of low temperature tolerance among livestock with different body shapes;

[0034] Step S252: performing Fourier thermal spectrum transformation on the spatial thermal energy vector loss deconstruction data to obtain a frequency domain distribution spectrum of thermal energy loss;

[0035] Step S253: performing a time-varying nonlinear dynamic evolution of body surface heat energy loss among livestock of different body shapes on the livestock body shape difference data based on the heat energy loss frequency domain distribution spectrum and the extreme values of livestock low temperature tolerance among livestock of different body shapes, to obtain nonlinear time-varying data of body surface heat energy loss;

[0036] Step S254: performing convergence-constrained power series expansion on the nonlinear time-varying data of body surface heat energy loss to obtain a convergent power series of body surface heat energy loss;

[0037] Step S255: performing a body hypothermia measurement integration between livestock with different body shapes on the livestock body shape difference data according to the convergence power series of body surface heat energy loss to generate livestock body hypothermia measurement data.

[0038] Preferably, step S254 includes the following steps:

[0039] The instantaneous growth index of heat energy loss is calculated for the nonlinear time-varying data of body surface heat energy loss to obtain the instantaneous growth index of heat energy loss;

[0040] Based on the instantaneous growth index of heat energy loss, the relative entropy difference of local mutation of heat energy loss is analyzed to obtain the relative entropy difference of local mutation of loss;

[0041] Based on the relative entropy difference of the local mutation of loss, the convergence of the mutation point of the nonlinear time-varying data of the body surface heat energy loss is analyzed to obtain the convergence data of the loss mutation point;

[0042] According to the convergence data of loss mutation point, the nonlinear time-varying data of body surface heat energy loss is expanded with convergence constraint power series to obtain the convergence power series of body surface heat energy loss.

[0043] Preferably, step S3 includes the following steps:

[0044] Step S31: normalizing the spatial thermal energy vector loss deconstruction data to obtain spatial thermal energy vector loss normalized data;

[0045] Step S32: performing logic learning processing on the livestock hypothermia measurement data to obtain livestock hypothermia learning data;

[0046] Step S33: performing spatial thermal energy replenishment rate matching according to the spatial thermal energy vector loss normalization data to obtain spatial thermal energy replenishment rate data;

[0047] Step S34: matching heating steam circulation requirements between livestock of different body shapes based on the livestock hypothermia learning data to obtain heating steam circulation requirements;

[0048] Step S35: Construct a heating regulation model based on the spatial heat energy replenishment rate data and the heating hot steam cycle demand to obtain a heating adaptive regulation model, and send the heating adaptive regulation model to the cloud platform to execute intelligent heating control of the livestock farm.

[0049] Preferably, the present invention further provides an intelligent heating control system for a livestock farm, which is used to execute the intelligent heating control method for a livestock farm as described above. The intelligent heating control system for a livestock farm comprises:

[0050] The data monitoring module is used to monitor the behavior and activity status of livestock in the livestock farm through electronic monitoring equipment to obtain livestock behavior and activity status monitoring videos; the temperature and humidity sensors and greenhouse gas sensors in the environmental sensing module are used to monitor the temperature and humidity fluctuations and greenhouse gas fluctuations in the livestock farm under extremely cold weather, and obtain temperature and humidity fluctuation data and greenhouse gas fluctuation data respectively;

[0051] The postural hypothermia measurement module is used to analyze postural differences in livestock behavior and activity status monitoring videos to generate livestock postural difference data; perform spatial thermal energy vector loss deconstruction on temperature and humidity fluctuation data and greenhouse gas fluctuation data to obtain spatial thermal energy vector loss deconstruction data; and perform postural hypothermia measurement integration between livestock of different postural shapes on livestock postural difference data based on the spatial thermal energy vector loss deconstruction data to generate livestock postural hypothermia measurement data;

[0052] The heating regulation model construction module is used to match the heating hot steam circulation demand between livestock of different body shapes based on livestock hypothermia learning data to obtain the heating hot steam circulation demand; the heating regulation model is constructed according to the heating hot steam circulation demand to obtain a heating adaptive regulation model, and the heating adaptive regulation model is sent to the cloud platform to implement intelligent heating control of livestock farms.

[0053] The beneficial effects of the present invention lie in monitoring the behavior and activity of livestock on livestock farms through electronic monitoring equipment, producing videos of their behavior and activity. Simultaneously, temperature and humidity sensors and greenhouse gas sensors monitor environmental changes in extreme cold weather. This process enables real-time acquisition of livestock activity status information, as well as ambient temperature, humidity, and gas concentration fluctuation data, providing accurate data support for subsequent analysis. Promptly understanding livestock behavior and environmental changes helps identify potential health issues and the impact of temperature fluctuations, providing fundamental data for intelligent control. By analyzing the posture differences in the livestock behavior and activity monitoring videos, the generated posture difference data helps understand the health and activity changes of livestock of different postures in specific environments. Simultaneously, the temperature, humidity, and greenhouse gas fluctuation data are subjected to spatial thermal energy vector decomposition to quantify the distribution of thermal energy losses. This process enables the system to accurately understand the impact of environmental factors on livestock posture, further calculate the hypothermia of livestock of different postures, and generate posture hypothermia measurement data, providing a quantitative basis for subsequent adjustment of heating demand. Based on the livestock hypothermia learning data, the heating steam circulation requirements for livestock of different body shapes are matched to determine the heat required for each type of livestock. This process ensures that the heating system can provide accurate hot steam supply according to the specific needs of livestock of different body shapes, while avoiding energy waste. By constructing a heating regulation model based on the heating steam circulation requirements, intelligent adaptive regulation of the heating system is achieved. This regulation model can automatically adjust the heating intensity according to different environmental and body conditions, thereby improving heating efficiency and ensuring the healthy growth of livestock. Therefore, the present invention is an optimization of a traditional intelligent heating control method for livestock farms. It solves the problem of low accuracy in analyzing the spatial heat energy loss of livestock farms and the degree of hypothermia among livestock of different body shapes, which leads to large errors in heating regulation. It improves the accuracy of analyzing the spatial heat energy loss of livestock farms and the degree of hypothermia among livestock of different body shapes, and reduces the error in heating regulation. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 A schematic flow chart of the steps of an intelligent heating control method for livestock farms;

[0055] Figure 2 for Figure 1 Detailed implementation steps of step S2 in FIG.

[0056] Figure 3 for Figure 1 Detailed implementation steps of step S3 in FIG. DETAILED DESCRIPTION

[0057] See also Figures 1 to 3, an intelligent heating control method for a livestock farm, the method comprising the following steps:

[0058] Step S1: Monitor the behavior and activity status of livestock in a livestock farm using electronic monitoring equipment to obtain a livestock behavior and activity status monitoring video; monitor the temperature and humidity fluctuations and greenhouse gas fluctuations in the livestock farm under extreme cold weather using the temperature and humidity sensors and greenhouse gas sensors in the environmental sensing module to obtain temperature and humidity fluctuation data and greenhouse gas fluctuation data respectively;

[0059] Step S2: Performing posture difference analysis on livestock behavior and activity status monitoring videos to generate livestock posture difference data; performing spatial thermal energy vector loss deconstruction on temperature and humidity fluctuation data and greenhouse gas fluctuation data to obtain spatial thermal energy vector loss deconstruction data; performing posture hypothermia measurement integration between livestock of different postures on the livestock posture difference data based on the spatial thermal energy vector loss deconstruction data to generate livestock posture hypothermia measurement data;

[0060] Step S3: Based on the livestock hypothermia learning data, the heating hot steam circulation demand between livestock of different body shapes is matched to obtain the heating hot steam circulation demand; a heating regulation model is constructed according to the heating hot steam circulation demand to obtain a heating adaptive regulation model, and the heating adaptive regulation model is sent to the cloud platform to implement intelligent heating control of the livestock farm.

[0061] In the embodiment of the present invention, reference Figure 1 The above is a schematic flow chart of the steps of an intelligent heating control method for a livestock farm according to the present invention. In this example, the intelligent heating control method for a livestock farm includes the following steps:

[0062] Step S1: Monitor the behavior and activity status of livestock in a livestock farm using electronic monitoring equipment to obtain a livestock behavior and activity status monitoring video; monitor the temperature and humidity fluctuations and greenhouse gas fluctuations in the livestock farm under extreme cold weather using the temperature and humidity sensors and greenhouse gas sensors in the environmental sensing module to obtain temperature and humidity fluctuation data and greenhouse gas fluctuation data respectively;

[0063] In an embodiment of the present invention, in order to realize intelligent heating control of livestock farms under extremely cold weather conditions, step S1 is first implemented. In this step, an all-weather high-definition infrared camera with a resolution of not less than 1920×1080 is installed in the center area and four corners of the ceiling of the livestock breeding house to form full-angle coverage and collect livestock behavior and activity status monitoring videos. The camera has infrared night vision and thermal imaging functions, and can realize 24-hour continuous data collection. The video frame rate is set to 25 frames per second to ensure the continuity of dynamic behavior collection. At the same time as the video is collected, environmental parameters are collected through environmental sensing modules arranged in different locations of the breeding house. The temperature and humidity sensor has an accuracy of ±0.1℃ and ±1.5%RH, and the sampling period is set to once every 10 seconds; the greenhouse gas sensor uses the NDIR principle The sensor model is GSS ExplorIR-W, with an accuracy of ±30ppm + 3% of reading and a sampling period of 10 seconds. Collection points are located at the entrance, in the middle, in areas with high livestock activity, and near exhaust ducts to ensure representative and comprehensive environmental data. All collected data is synchronously stored with timestamps in a local edge data concentration system and connected to the data processing module via a wired network.

[0064] Step S2: Performing posture difference analysis on livestock behavior and activity status monitoring videos to generate livestock posture difference data; performing spatial thermal energy vector loss deconstruction on temperature and humidity fluctuation data and greenhouse gas fluctuation data to obtain spatial thermal energy vector loss deconstruction data; performing posture hypothermia measurement integration between livestock of different postures on the livestock posture difference data based on the spatial thermal energy vector loss deconstruction data to generate livestock posture hypothermia measurement data;

[0065] In this embodiment of the present invention, the collected livestock behavior and activity monitoring video is first subjected to structured image segmentation. The outlines of individual livestock are extracted using edge recognition and contour extraction techniques. The spatial positions of the livestock's limbs, trunk, and head nodes are annotated using a human posture recognition method based on the OpenPose algorithm. Subsequently, the spatial position differences of the nodes at different time points are quantified using Euclidean distance and angle change comparison methods, thereby generating livestock posture difference data. Furthermore, the temperature and humidity fluctuation data and greenhouse gas fluctuation data acquired in step S1 are time-series aligned and spatially mapped on a per-minute basis. A three-dimensional spatial grid is created based on the physical structural parameters of the livestock shed (including length, width, height, and ventilation opening location and dimensions), with a unit side length of 0.5 meters. The principles of energy conservation and thermal diffusion are combined with air density, specific heat capacity, and wind speed data to analyze the direction and magnitude of heat transfer within each grid unit. A spatial heat energy vector transfer matrix is constructed using the finite difference method, and the directionality and amount of heat energy loss per unit time are decomposed to obtain spatial heat energy vector loss decomposition data. Finally, the surface area change rate and the degree of skin movement activity per unit time extracted from the livestock body shape difference data were used to integrate the exposure heat flux of livestock with different body shapes in each heat loss area. The trapezoidal integration method was used to sum them up to obtain the body hypothermia measurement integral between livestock with different body shapes, and generate livestock body hypothermia measurement data.

[0066] Step S3: Based on the livestock hypothermia learning data, the heating hot steam circulation demand between livestock of different body shapes is matched to obtain the heating hot steam circulation demand; a heating regulation model is constructed according to the heating hot steam circulation demand to obtain a heating adaptive regulation model, and the heating adaptive regulation model is sent to the cloud platform to implement intelligent heating control of the livestock farm.

[0067] In this embodiment of the present invention, the livestock hypothermia data obtained in step S2 is first normalized to unify the heat flux loss values of different livestock to the range of 0 to 1, eliminating order-of-magnitude shifts caused by body size differences. This data is then used as an input variable to match the corresponding livestock activity areas and estimate the distributed heat energy demand density. A kernel density estimation method is used to fit the heat energy demand variation density surface for each region. This surface is then compared and analyzed region by region with the spatial heat energy vector loss deconstruction data obtained in step S2. Based on the heat energy difference, a matching calculation is performed on the regional heating replenishment rate to determine the steam heat energy supply required for each spatial unit. Parameters for the steam circulation heating system include a maximum hourly heat output of 120 kW, a maximum steam pipe pressure of 1.6 MPa, allocation of main and branch pipes based on heat demand priority, and a setting of the activation threshold for each pipe based on the unit heat loss density. Using a multi-objective constrained programming algorithm, a heating regulation model is constructed as a control logic diagram, which includes parameters such as heating on-time, steam flow rate, and heater on-power, and outputs an adaptive heating regulation model. The model is encapsulated in the form of OPC-UA protocol and uploaded to the cloud platform scheduling system through a wired LAN connection to perform remote intelligent control and achieve differentiated heating regulation goals for different areas of the livestock farm.

[0068] Step S2 includes the following steps:

[0069] Step S21: performing posture difference analysis on the livestock behavior and activity status monitoring video to generate livestock posture difference data;

[0070] Step S22: performing group movement frequency analysis on the livestock behavior and activity status monitoring video to obtain livestock group movement frequency data;

[0071] Step S23: Acquire livestock farm structural design data; perform gas disturbance dispersion analysis on greenhouse gas fluctuation data based on livestock group movement frequency data and livestock farm structural design data to obtain greenhouse gas disturbance dispersion structure data;

[0072] Step S24: performing spatial thermal energy vector loss deconstruction on the temperature and humidity fluctuation data and the greenhouse gas disturbance dispersion structure data according to the livestock farm structural design data to obtain spatial thermal energy vector loss deconstruction data;

[0073] Step S25: performing a body hypothermia measurement integration between livestock with different body shapes on the livestock body shape difference data based on the spatial thermal energy vector loss deconstruction data to generate livestock body hypothermia measurement data.

[0074] As an example of the present invention, refer to Figure 2 As shown, in this example, step S2 includes:

[0075] Step S21: performing posture difference analysis on the livestock behavior and activity status monitoring video to generate livestock posture difference data;

[0076] In this embodiment of the present invention, in step S21, a high-definition infrared camera with a resolution of 1920×1080 and a frame rate of 30 frames per second is used to continuously capture video of livestock activities within a livestock farm. The captured video images enter the image processing process via an edge server. The YOLOv5 object detection algorithm is used to identify and number individual livestock, obtaining the bounding box position of each individual livestock in each frame. By analyzing the temporal changes in the bounding box positions in consecutive frames, the position information of key livestock body joints is extracted using OpenPose skeleton recognition technology. After feature point extraction, the relative position changes of each body joint are analyzed using a sliding window with a time window of 2 seconds. Body similarity is compared using the cosine similarity metric, and the K-means clustering algorithm is used to cluster and annotate body movement categories. The differences in posture angles and the degree of skeleton deformation between different body categories are quantified into feature vector groups, ultimately constructing livestock body difference data. The structure of the posture difference data is in the form of a two-dimensional matrix, where each row corresponds to an individual livestock, and each column is the numerical indicator of a certain type of posture difference vector. The total dimension is the number of livestock individuals × the number of posture feature dimensions. For example, in actual tests, for 30 beef cattle, 16 key point coordinates are extracted from each cow to form a feature vector, and finally a 30×16 posture difference data matrix is formed. This matrix serves as the basic parameter for subsequent precise control of heating.

[0077] Step S22: performing group movement frequency analysis on the livestock behavior and activity status monitoring video to obtain livestock group movement frequency data;

[0078] In this embodiment of the present invention, in step S22, based on the individual livestock numbers and bounding box trajectory sequences obtained in step S21, the optical flow method is used to track the motion trajectory of each livestock in consecutive frames. This method employs the dense optical flow Lucas-Kanade algorithm to track the pixel movement of the livestock center point in the image coordinate system. The total pixel displacement of each individual livestock movement path is calculated every second and then converted into actual spatial movement distance (based on camera installation parameters and field calibration data, with a conversion ratio of 1 pixel to 2.5 centimeters). Within each minute of the statistical period, the average movement distance of all livestock is summarized and analyzed using a fast Fourier transform. The principal components of different frequency components are extracted, and a motion frequency distribution curve for the livestock group is constructed based on the principal frequency fluctuation range (0.2-1.5 Hz). In a specific experiment, for a group of 40 dairy cows, during the high-activity period from 8:00 AM to 10:00 AM each day, monitoring data showed a principal movement frequency of approximately 0.9 Hz, indicating a high-frequency movement of the group. The group movement frequency data is ultimately expressed as a movement frequency vector with timestamp as the main index. Each record contains the time point, the group average frequency value, the standard deviation, and the maximum fluctuation amplitude, which constitutes the livestock group movement frequency data for subsequent gas disturbance dispersion analysis.

[0079] Step S23: Acquire livestock farm structural design data; perform gas disturbance dispersion analysis on greenhouse gas fluctuation data based on livestock group movement frequency data and livestock farm structural design data to obtain greenhouse gas disturbance dispersion structure data;

[0080] In this embodiment of the present invention, the structural design data of the livestock farm is first retrieved, including the three-dimensional spatial dimensions of the farm (60 meters in length, 15 meters in width, and 5 meters in height), the distribution of door and window vents, ventilation equipment parameters (air volume of 3,000 cubic meters per hour, wind speed of 0.5 meters per second), and the thermal conductivity of the ceiling and floor materials (0.23 and 1.14 W / m·K, respectively). The initial boundary conditions for the gas disturbance path are determined by combining the channel distribution and the location of the feed placement area. The livestock group movement frequency data obtained in step S22 is mapped into a spatial coordinate system. Each livestock position is assigned a disturbance intensity factor based on its movement frequency. CFD (computational fluid dynamics) simulation technology is combined with the RNG k-ε turbulence model to establish a flow field disturbance model for air and greenhouse gases within the farm. The disturbance at the individual livestock position is considered a disturbance source node. The disturbance energy is calculated within a 1-meter radius around the cattle body as the disturbance radius. Within this radius, an initial velocity vector disturbance is introduced (the baseline disturbance velocity is 0.2 times the livestock movement velocity), and the disturbance is superimposed on the existing ventilation-induced flow field. During the simulation period (the simulation time is set to 120 seconds and the step size is 0.05 seconds), 、 The diffusion paths of greenhouse gases such as GHG under disturbance conditions are simulated and analyzed over multiple time periods. The gas concentration change surface under disturbance is extracted, and gradient boundaries are extracted for concentration gradients within different regions to generate gas disturbance dispersion structure data. This data is stored as a three-dimensional grid structure, with each grid node containing coordinates, a disturbance velocity vector, greenhouse gas concentration change values, and a volatility index. This data is used to accurately characterize the impact of air movement caused by livestock movement on the gas distribution structure, guiding subsequent ventilation and heating linkage control.

[0081] Step S24: performing spatial thermal energy vector loss deconstruction on the temperature and humidity fluctuation data and the greenhouse gas disturbance dispersion structure data according to the livestock farm structural design data to obtain spatial thermal energy vector loss deconstruction data;

[0082] In this embodiment of the present invention, in step S24, the structural design data for the livestock farm is first retrieved, including the internal spatial geometry of the barn (length 60 meters, width 15 meters, height 5 meters), the heat transfer coefficients of the walls and roof (0.45 W / m·K and 0.32 W / m·K, respectively), a floor thermal reflectivity of 0.12, the ventilation duct layout, an air velocity of 0.5 m / s at the air outlet, and the distribution coordinates of heat sources in each area (such as the waterer heat exchanger and feed heating equipment). Based on this data, a three-dimensional internal heat conduction model of the barn is constructed. The greenhouse gas perturbation dispersion structure data generated in step S23 is introduced as an interference term to participate in the heat diffusion path analysis. The greenhouse gas concentration gradient change rate of each grid cell in the perturbation structure is matched with the wind speed perturbation vector. In combination with the principles of thermal air convection and diffusion, the heat transfer direction and heat loss values within each area of the space are deconstructed. For temperature and humidity fluctuation data, a spatial temperature distribution map is created at each sampling time point. The current temperature value and the ambient humidity of the area are calibrated with a grid of 1 cubic meter. Based on the laws of thermal convection, the effect of humidity on heat exchange efficiency is calculated, and the influence of wind speed disturbances on the heat migration path is also considered. In practice, the temperature and humidity fluctuation data are aligned with the gas disturbance structure data through spatial grid nodes to form the heat input, output, and loss status within each unit spatial grid point. Subsequently, the heat flow path is analyzed for all grid points, and a heat flux density directional map is constructed using the finite volume method. The vector direction of heat flow from high-temperature to low-temperature areas in a specific region is derived based on boundary conditions. Finally, the spatial heat energy vector loss deconstruction data is output as a heat energy flow direction vector field in a three-dimensional coordinate system. Each vector represents the net direction and amount of heat energy loss per unit time in a specific spatial unit, expressed in watts per cubic meter, to guide subsequent body hypothermia analysis.

[0083] Step S25: performing a body hypothermia measurement integration between livestock with different body shapes on the livestock body shape difference data based on the spatial thermal energy vector loss deconstruction data to generate livestock body hypothermia measurement data.

[0084] In this embodiment of the present invention, in step S25, the spatial thermal energy vector loss deconstruction data obtained in step S24 is called and combined with the livestock posture difference data obtained in step S21. The heat loss value of the 1-cubic-meter thermal energy vector node at the livestock's location in space is extracted, using the livestock's individual spatial position as the reference point. Because the livestock's posture difference data already calibrates the displacement amplitudes of key joints and the dimensions of their posture movements, their heat dissipation efficiency can be influenced by the frequency and dimensions of their posture movements. Specifically, for each individual livestock, a posture activity index (e.g., the average sum of the displacements per second of the head, trunk, and limb joints) is extracted from its posture difference vector as a body surface area motion factor per unit time. This is then integrated with the direction of the individual's spatial thermal energy loss vector to estimate the individual's hypothermia intensity. In an actual experiment, the average displacement of key points on the head and shoulders of a 580 kg Simmental cattle was measured at 3.2 cm / s over a two-minute monitoring period. This corresponds to a posture activity factor of 3.2, and a unit volume heat loss of 12 watts / m³. Based on an average exposed body area of 2.6 square meters, the total heat loss per unit time was deduced to be 78.4 watts. This calculation process is performed on a herd-by-herd basis, with all individual hypothermia values indexed by their unique animal numbers, generating a body hypothermia measurement. This data is formatted as a one-dimensional structured array, with each element containing the animal number, the spatial heat loss vector, the posture factor, and the unit time heat loss value. These parameters serve as key parameters in the heating control process and contribute to the subsequent determination of heat compensation requirements.

[0085] Step S23 includes the following steps:

[0086] Step S231: Acquire livestock farm structural design data; perform gas circulation design path analysis on the livestock farm structural design data to obtain a gas circulation design path;

[0087] Step S232: performing group gait acceleration frequency analysis on the livestock group movement frequency data to obtain the group gait acceleration frequency;

[0088] Step S233: Calculating the continuous pressure difference of the peripheral airflow based on the group gait acceleration frequency to obtain the continuous pressure difference of the peripheral airflow;

[0089] Step S234: Deducing the change of azimuthal velocity difference of the gas flow section on the designed gas flow path according to the continuous pressure difference driven by the surrounding airflow, and obtaining the change data of azimuthal velocity difference of the gas flow section;

[0090] Step S235: performing mass / volume regional distribution difference analysis on the greenhouse gas fluctuation data to obtain the greenhouse gas mass / volume regional distribution difference;

[0091] Step S236: performing gas disturbance dispersion analysis on the greenhouse gas mass / volume regional distribution difference based on the flow section azimuthal velocity difference change data to obtain greenhouse gas disturbance dispersion structure data.

[0092] In this embodiment of the present invention, in step S231, construction drawings and building information modeling (BIM) data for the livestock farm are first retrieved. These include the overall structural dimensions of the barn (total length 65 meters, width 18 meters, eaves height 5.5 meters), internal functional area divisions, ventilation system component distribution, the specific locations and area distribution of air inlets and exhaust vents (including three west-side inlets, each with an area of 0.5 square meters, and four east-side exhaust vents, each with an area of 0.4 square meters), as well as the duct routing and cross-sectional dimensions (main duct diameter 60 centimeters, secondary duct diameter 30 centimeters). During the analysis process, a two-dimensional planar layout diagram is combined with a three-dimensional structural diagram to extract the structural boundary nodes and perform spatial mapping according to the coordinate system specified in the drawings. Based on the spatial finite meshing method, the entire barn is divided into equilateral cubic units (with a unit side length of 1 meter). The interconnectivity between each unit is calibrated based on the gas flow continuity conditions and the duct distribution path, thereby determining the main gas flow path from the air inlet to the exhaust. Obstacle areas such as walls and supports are further shielded to block out unit nodes that do not participate in airflow. By establishing a directed graph structure, all airflow units are path-numbered and topologically sorted, ultimately forming complete airflow design path data from the air inlet. The data format includes the airflow unit number, spatial coordinates, flow status, and upstream and downstream unit indexes, providing a spatial foundation for subsequent airflow disturbance modeling.

[0093] In step S232, the livestock group movement frequency data acquired in step S22 is retrieved. Each animal is considered a moving subject and its movement distance between consecutive frames in the monitored video is temporally differentiated to obtain an individual gait velocity time series. A single differentiation is performed on the velocity time series to extract the acceleration time series. Each individual acceleration series is then normalized using maximum normalization. Using a one-minute time window, the maximum acceleration and acceleration fluctuation frequency per unit time for each animal are calculated, and the main acceleration frequency components are extracted using a Fast Fourier Transform (FFT) transform. To eliminate data interference caused by occasional jumps or unnatural behavior, a median filter is used for sequence smoothing. The acceleration frequency of each animal sample is then linearly weighted averaged, with the weight being the inverse of the individual's time spent in the central area (i.e., higher activity levels result in higher weights). The group gait acceleration frequency for the current monitoring period is finally calculated. In actual testing, data from 40 beef cattle was collected for 15 minutes. The average acceleration frequency of individual gaits ranged from 0.15 to 0.45 Hz, resulting in a calculated group average acceleration frequency of 0.31 Hz. This value served as the input indicator for the disturbance energy intensity in the airflow disturbance simulation. The mass and momentum conservation equations from continuum mechanics were used to construct a peripheral airflow propulsion model. Using the gas flow path obtained in step S231 as the framework, a local disturbance source caused by livestock movement was introduced within each spatial unit. The disturbance source intensity was assigned based on the group gait acceleration frequency in step S232. Each animal was assumed to be a periodic vibration source, generating localized gas disturbances within its unit volume. The disturbance frequency was set to 0.31 Hz, corresponding to a disturbance velocity amplitude of 0.07 meters per second. In the continuous air medium, this disturbance generates periodic density fluctuations, which in turn lead to the spatial transmission of pressure fluctuations. The finite difference method was used to solve for the pressure variation at each unit spatial grid point, thereby deriving the pressure gradient field. Based on the pressure gradient value and direction, the acceleration direction and magnitude of the gas per unit volume under this pressure difference are calculated, and this value is then converted into a spatial vector field representing the airflow pressure difference. The specific calculation area is set as the central area of the barn (40 m × 10 m × 3 m), and the spatial granularity is 0.5-meter-long cubes. Ultimately, a continuous pressure difference dataset is generated within each grid, including coordinate location, disturbance frequency, instantaneous pressure change per unit volume, and airflow acceleration direction. This data serves as one of the input parameters for greenhouse gas dispersion modeling. This data reflects the direction and intensity of spatial airflow disturbance energy propagation caused by the dynamic behavior of livestock groups, providing a dynamic basis for subsequent barn environmental regulation.

[0094] Based on the continuous pressure differential data driven by the surrounding airflow obtained in step S233, a velocity gradient model is established within the adjacent gas flow units upstream and downstream of each spatial section (defined as a unit area section, measuring 1 square meter) in the designed gas flow path. Based on the pressure gradient and the continuity equation in aerodynamics, the local flow velocity at the center of the section is deduced. For example, the central passage of a barn has nine standard sections along the main gas flow path, with upstream and downstream spacing of 2 meters. The pressure difference between the front and rear of each section is estimated using a differential method. Combined with the air density of 0.001225 grams per cubic centimeter, the average flow velocity change driven by this pressure differential is calculated. If the pressure difference between section A and the front and rear is 15 Pa, then based on the velocity-pressure differential relationship, the resulting average flow velocity is calculated to be 1.4 meters per second, a significant difference from the 0.8 meters per second flow velocity at the upstream section. To further extract the directional change characteristics, the velocity vector of each section is projected according to the design direction of the airflow, and the component difference of the section airflow velocity in the direction of the ventilation design path is obtained. The increment or attenuation trend along the path advancement direction is recorded, thereby forming the section azimuth velocity difference change data. The data set is stored in the form of a matrix, and each item records the section number, coordinates, velocity change magnitude, velocity direction angle, and velocity difference with adjacent sections. 、 The greenhouse gas concentration data are processed by spatial regional difference. First, the entire livestock house is gridded according to the aforementioned spatial division units, and the monitoring data is mapped according to the spatial unit where each sensor point is located. On this basis, the gas mass value per unit volume of each area is calculated, and according to the gas density value (such as carbon dioxide density is 1.977 kilograms per cubic meter, methane density is 0.656 kilograms per cubic meter), the sensor sampling concentration is multiplied by the unit volume to obtain the actual gas mass, and then the total gas mass of all monitoring areas is calculated. Taking the 10m×10m central activity area as an example, there are 16 sensor nodes inside, and the measured The concentration value ranges from 410ppm to 580ppm, which is converted to the area The total mass is approximately 3.25 kg, corresponding to a volume of 300 cubic meters. The gas mass data in all areas are identified by area numbers, and then the difference between the gas masses of adjacent space units is calculated to extract the gradient change trend of gas mass in space. According to the spatial distribution trend diagram, it can be observed that the area on the north side of the livestock house The mass distribution is relatively more concentrated, while the mass gradient in the area near the south exhaust outlet is significantly reduced, thus forming a map of mass distribution differences between regions. All data are uniformly archived as a greenhouse gas mass / volume regional distribution difference dataset, with fields including unit number, center coordinates, gas type, mass value, volume size, and mass difference with adjacent units. The flow section azimuth velocity difference change data obtained in step S234 is used as the driving force basis for gas disturbances and is fused with the greenhouse gas mass / volume regional distribution difference data in step S235. The processing method adopts a spatial coupling algorithm: first, a directed channel structure of the gas diffusion direction is established with the spatial section as a node, and the velocity change vector of each section is mapped to the adjacent upstream and downstream spatial units, and the disturbance propagation path is determined by combining its corresponding velocity increment direction. Then, a disturbance transfer coefficient is introduced in each unit body. This coefficient is determined by the cross-section azimuth velocity difference and reflects the possibility of gas diffusion from the high-mass density area to the low-mass density area. Taking the northwest corner of the livestock house as an example, this is a structurally closed area, and the velocity difference of the ventilation section is negative, which means the air flow velocity is decreasing, and this area The mass distribution is 1.75 kg, and the adjacent unit is 1.05 kg, with a significant mass gradient. According to the disturbance intensity and flow velocity direction, the disturbance energy is distributed to the three adjacent units downstream, and the disturbance dispersion probability weights are assigned to 0.6, 0.3 and 0.1 respectively. Without introducing a probability calculation framework, discrete weights are used to describe the disturbance intensity transfer trend. Through the above method, a gas disturbance dispersion structure data map is established in the entire livestock house space. The data fields include unit space number, disturbance direction, disturbance amplitude, target transmission path, disturbance energy distribution structure, and the mass value mapping difference between the source unit and the target unit. The greenhouse gas disturbance dispersion structure data finally obtained is used for subsequent heating thermal energy optimization layout and local air quality regulation modeling input.

[0095] Step S24 includes the following steps:

[0096] Step S241: performing material thermal insulation performance degradation analysis on the livestock farm structural design data to obtain material thermal insulation performance degradation data;

[0097] Step S242: performing temperature and humidity spatial fluctuation distribution mapping on the temperature and humidity fluctuation data according to the livestock farm structural design data to obtain temperature and humidity spatial fluctuation distribution mapping data;

[0098] Step S243: Based on the material thermal insulation performance loss data, the spatial heat energy loss divergence per unit time is calculated for the temperature and humidity spatial fluctuation distribution mapping data and the greenhouse gas disturbance dispersion structure data to obtain spatial heat energy loss divergence data;

[0099] Step S244: performing spatial heat energy vector loss deconstruction based on the spatial heat energy loss divergence data to obtain spatial heat energy vector loss deconstruction data.

[0100] In this embodiment of the present invention, in step S241, a thermal insulation performance loss analysis is performed on the material information of the wall, roof, floor, and door and window components included in the livestock farm structural design data. First, based on the design drawings and engineering material statistics, the construction material type and parameter data are extracted, including material type (e.g., polyurethane composite panel, double-layer insulating glass, concrete structure), thickness (in millimeters), thermal conductivity (in watts per meter per Kelvin), and structural location. For each structural unit, a thermal insulation performance loss factor is quantified based on the environmental exposure it will experience during its service life. For a roof polyurethane sandwich panel, for example, with a design thickness of 75 mm and an initial thermal conductivity of 0.023 watts per meter per Kelvin, the annual increase in thermal conductivity is calculated using the empirical decay function method, given the local average annual temperature and humidity of 7.5 degrees Celsius and an average annual relative humidity of 82%. The current thermal conductivity is 0.031 watts per meter per Kelvin. Accordingly, the thermal insulation capacity is subtracted from the initial capacity, and the thermal insulation performance loss value for that structural unit is recorded. Similar analysis was performed for the 240 mm thick concrete structure of the south wall of the farmhouse, the double-glazed plastic-steel windows on the east side, and the composite insulation layer (including 100 mm foam board) on the floor, generating a table corresponding to structural partitions and insulation loss. All data was archived by spatial number into a material insulation loss dataset. Fields included structural unit number, material type, thickness, initial thermal conductivity, age, environmental exposure parameters, current thermal conductivity, and percentage of insulation loss, which were used for subsequent structured heat loss modeling. Based on the farm's structural design data and time series data collected by temperature and humidity sensor nodes deployed at different locations, spatial distribution mapping of temperature and humidity fluctuations was performed. First, the farm's interior was divided into regular spatial grid cells according to the structural diagram. Each cell was set to 2 m × 2 m × 2 m, corresponding to a discrete volume in three-dimensional space. Within each volume cell, data from the sensor sampling points that covered or were adjacent to it was aggregated at a sampling frequency of every 10 seconds. The data included temperature (in degrees Celsius) and relative humidity (in percentage). Taking a certain moment (for example, 4 a.m. every day) as a unified reference moment, the instantaneous readings of all sensor nodes in the entire field are extracted, and the nearest neighbor interpolation mapping is performed based on the three-dimensional coordinates of the sensors and the positions of the spatial units, and the temperature and humidity data are filled into all spatial units. For example, there are two sensors in the northern corner area of the farm, and the temperatures are 6.8 degrees Celsius and 7.1 degrees Celsius respectively. Then, the grid unit in this area is assigned a value of 6.95 degrees Celsius through the weighted average method. After completing the full-field mapping, all spatial units are statistically analyzed to identify the range with the largest temperature and humidity fluctuations. By repeating the above operation for the spatial mapping within 24 hours a day, 24 temperature and humidity distribution maps are generated for each day, and the temperature and humidity fluctuation amplitudes within the same spatial unit are calculated. The temperature fluctuation value is defined as the maximum temperature value minus the minimum temperature value, and the humidity fluctuation is defined as the maximum humidity value minus the minimum humidity value.Finally, the spatiotemporal distribution data of temperature and humidity in all units are integrated into a temperature and humidity spatial fluctuation distribution mapping dataset. The fields include spatial unit number, coordinate center, hourly temperature value, humidity value, temperature fluctuation amplitude, humidity fluctuation amplitude, etc., providing data support for subsequent heat energy vector loss analysis.

[0101] In step S243, the obtained material insulation performance loss data is first used as the basis for the heat conduction influencing factor. By correlating this data with the temperature and humidity spatial fluctuation distribution mapping data, the heat conduction direction and heat loss starting point are determined. Specifically, the farm's three-dimensional structure is divided into 1m x 1m x 1m volume cells according to the spatial grid cells. Each cell is associated with the structural material information and the temperature gradient change value in that area. Assuming the northwest corner area has a 240mm concrete wall structure with a thermal insulation performance loss rate of 23%, the daytime temperature in this area rises from 6.2°C in the morning to 13.7°C at noon, while the adjacent inner area is 14.3°C. The heat conduction direction is inward-outward. The heat loss per unit time is deduced by combining the thermal conduction loss factor, temperature gradient, and exposed area. The spatial unit heat loss divergence is defined as the amount of heat transferred to the external environment per unit time per grid cell, expressed in joules per cubic meter per second. At the same time, the airflow disturbance vectors from the greenhouse gas disturbance dispersion structure data are integrated, the gas disturbance frequency in the ventilation duct is compared with the pressure distribution trend diagram, and the heat exchange frequency within the disturbance zone is superimposed to complete the assessment of the auxiliary heat exchange loss caused by the gas disturbance. Based on the above analysis, the heat loss divergence is numerically deduced for each spatial unit with a time resolution of one minute, resulting in a complete spatial heat loss divergence dataset. This dataset includes fields such as the space number, boundary structure thermal conductivity, insulation loss rate, temperature gradient, heat loss rate, heat flow direction vector, and disturbance gain coefficient, which serve as input data for subsequent heat energy vector deconstruction. In step S244, based on the spatial heat energy loss divergence data obtained in step S243, the three-dimensional heat energy vector deconstruction operation is further performed. First, a heat energy vector field is established in each spatial volume unit based on its heat energy loss divergence value and corresponding conduction direction. Each heat energy vector is defined as oriented inwards and outwards, with its magnitude determined by the amount of heat loss per unit volume per unit time. For example, for the volume unit in the middle with a higher temperature, the heat loss divergence is set to 12.3 joules per cubic meter per second, and the direction of the heat energy vector points to the low-temperature block in the southeast. For the entire spatial field, a three-dimensional heat conduction structure map with multi-vector superposition is constructed. Subsequently, vector decomposition technology is used to decompose each heat energy vector into three axial components of X, Y, and Z. The distribution of heat energy transmission channels in the entire area is constructed based on the superposition of the components. For example, in the southeast corner of a large cattle shed, after superimposing the heat energy vectors, it was found that there was a trend of heat concentrated on the roof. Combined with the insulation loss rate of the roof structure of 31%, this channel was confirmed to be the main heat energy loss path. At the same time, through the vector coupling relationship between adjacent units, the concentrated heat energy discharge area and the weak heat return area were deduced, and the heat energy path was visualized and output.The resulting spatial heat energy vector loss deconstruction dataset contains fields such as heat energy vector direction, vector intensity, vector path number, vector superposition trend, region number, and structure-related material number, providing structural support for subsequent heat compensation and energy-saving control strategies of the heating system.

[0102] Step S243 includes the following steps:

[0103] The incremental coefficient of structural pore thermal conductivity is calculated based on the thermal insulation performance loss data of the material to obtain the incremental coefficient of structural pore thermal conductivity;

[0104] Based on the incremental thermal conductivity coefficient of structural pores, the interface heat transfer acceleration proportional identification is performed on the material insulation performance loss data to obtain the interface heat transfer acceleration proportional data;

[0105] The low temperature / low humidity relative change rate difference is calculated for the temperature and humidity spatial fluctuation distribution mapping data to generate the spatial distribution low temperature / low humidity relative change rate difference;

[0106] According to the relative change rate difference of low temperature / low humidity in spatial distribution, the spatial discrete heat loss density value increment of greenhouse gas disturbance dispersion structure data per unit time is integrated to obtain the spatial heat loss density increment value per unit time;

[0107] Based on the interface heat transfer acceleration proportional data and the spatial heat loss density increment value per unit time, the spatial heat energy loss divergence per unit time is calculated to obtain the spatial heat energy loss divergence data.

[0108] In this embodiment of the present invention, the incremental thermal conductivity of structural pores is calculated based on the material insulation performance degradation data. Based on the livestock farm structural design data, the building material types, service life, construction methods, and known physical aging patterns of each area are extracted. An empirical thermal coefficient growth function is then used to incrementally correct the original thermal conductivity over time. For example, the exterior wall, as specified in the structural design, uses expanded perlite-cement composite panels as the insulation layer. The initial thermal conductivity is 0.030 watts per meter per kelvin, and the design life is 10 years. Based on the aging data of this type of material from published building thermal performance test literature, a functional growth relationship between aging and porosity is constructed. As the porosity increases by 0.9% per year, the thermal conductivity increases proportionally by 0.0025 watts per meter per kelvin per one percent of porosity. By the seventh year of construction and use, the corresponding porosity increase is 6.3%, which leads to a deduced thermal conductivity increase of 0.01575 watts per meter per kelvin. This increment is added to the original value to obtain the structural pore thermal conductivity increment coefficient as a conversion value of the total growth rate, which is used to construct a set of time-varying parameters of material thermal conductivity, including wall number, original thermal conductivity value, service life, theoretical pore growth rate and thermal conductivity correction value. Based on the above structural pore thermal conductivity increment coefficient, the interface heat transfer accelerated proportional identification operation is performed on the material insulation performance degradation data. According to the design drawing annotations, the differences in thermal conductivity of each structural material at the wall-top, wall-ground and other joint nodes are analyzed, and the equivalent thermal resistance conversion method is used to deduce the heat flux concentration trend. Nodes with large thermal conductivity differences are identified as high-risk interfaces, and nodes with a unidirectional thermal conductivity difference greater than 0.010 watts per meter per Kelvin in the heat flow path are mathematically set as accelerated identification targets. Taking the northeast corner connection node as an example, the wall thermal conductivity is 0.045 watts per meter per Kelvin, and the ceiling is 0.085 watts per meter per Kelvin. This difference in thermal resistance creates a unidirectional acceleration channel. Based on the continuous boundary condition for heat transfer, the heat transfer rate is quantified and analyzed. By establishing a geometrically increasing mapping table, the growth rate of heat loss transfer from the beginning of the year to the end of the year is summarized as a geometric progression model. If the heat loss at the beginning of the year is q and at the end of the year is 1.5q, the annual growth geometric factor is 1.5. This generates a geometric data table for interface heat transfer acceleration, including node number, material thermal conductivity comparison, thermal conductivity difference, annual growth rate, and predicted geometric factor. The relative rate difference of low temperature / low humidity is calculated using the temperature and humidity spatial fluctuation distribution mapping data. Using the established temperature and humidity distribution grid data, the entire aquaculture space is divided into regular grid cells. Based on the spatial distribution of the lowest temperature and lowest humidity points within a day, the temperature and humidity changes per unit time within each grid cell are calculated. The spatial change difference matrix was established by calculating the gradient difference of adjacent time periods on 3-day continuous data and comparing the maximum rate change with the minimum rate change.For example, during the early morning to morning hours of a particular day, the area near the north exhaust duct experienced the highest temperature change rate, dropping by 2.1 degrees Celsius per hour, and the humidity decreased by 3.8%. In the central area, however, the temperature change rate was only 1.3 degrees Celsius per hour, and the humidity decreased by 1.1%. The differences in these values were recorded as a low-temperature rate difference of 0.8 degrees Celsius per hour and a low-humidity rate difference of 2.7%. By comparing the rate differences between different areas, we generated data on the relative rate difference of low temperature / low humidity, which served as the input parameter for the heat loss gradient. This data table structure included spatial grid number, daily change period, temperature and humidity change rate, difference gradient, and trend identification.

[0109] Based on the obtained spatially distributed relative change rate differences of low temperature and low humidity, the greenhouse gas perturbation dispersion structure data were numerically integrated with the spatial discrete heat loss density per unit time. The farm space was divided into three-dimensional structural units, each with a 1 m × 1 m × 1 m spatial volume as the basic grid unit, and each unit was labeled with a number. Using the relative change rate differences of low temperature and low humidity as the index variable, units with a temperature change rate of more than 1.5 degrees Celsius per hour and a humidity decrease rate of more than 2% per hour were defined as high-variance units. For each high-variance unit, the perturbation frequency, distribution density, and gas concentration gradient were extracted from the corresponding greenhouse gas perturbation dispersion structure data. Combined with the heat capacity constant of the perturbed gas, the heat perturbation contribution per unit time was estimated. The average frequency of greenhouse gas concentration perturbations observed in this unit was 12 per minute. The gases were primarily a mixture of carbon dioxide and ammonia, with an average specific heat capacity of approximately 0.9 kilojoules per kilogram per kelvin. The perturbations were primarily upward turbulent, and the volume of the perturbed area was approximately 0.8 cubic meters. Substituting the product of the perturbed volume and the local temperature gradient into the calculation process, combined with the heat capacity constant and the perturbation frequency, it is deduced that the heat loss density increment per unit volume per hour is 4.6 kilojoules per cubic meter. After processing the remaining spatial units in the same area in the same way, the heat loss density increments of each spatial unit are accumulated in the time dimension through the integration method to complete the numerical generation operation of the spatial heat loss density increment per unit time, forming a data matrix field including unit number, low temperature / low humidity change rate, perturbation structure characteristics, heat capacity parameters, and integrated cumulative heat value. Based on the interface heat transfer acceleration proportional data and the spatial heat loss density increment value per unit time, the spatial heat energy loss divergence per unit time is estimated. First, the acceleration interface position, proportional coefficient, and thermal conductivity difference index in the interface heat transfer acceleration proportional data are coordinate-mapped to the spatial structure, and all heat transfer sensitive joint areas are marked. In each joint region, a unidirectional heat transfer channel is established along the interface normal. Heat flux growth vectors for different time sections are generated based on the geometric acceleration factor. Combined with the incremental heat loss density values of adjacent cells, a recursive differential integration method is used to perform segmented heat transfer amplification along the heat transfer path. For joint region #P45, for example, the interface length is 3.2 meters, corresponding to the wall-ceiling junction. The thermal conductivity gradient is 0.042 watts per meter per Kelvin, and the annual heat flux acceleration geometric factor is 1.4. Calculations are performed on an hourly basis, with a base heat flux of 12 kilojoules per meter. Combined with the incremental heat loss density of adjacent cells at 4.2 kilojoules per cubic meter, a heat flux amplification path across the interface is formed. This acceleration data is recursively imported into five consecutive spatial cells along the interface normal, with different weighting factors assigned to each cell for incremental superposition. Ultimately, the direction and total divergence of heat energy loss emanating outward from the joint region are determined.The spatial heat energy loss divergence data is arranged in a coordinate grid to generate a data table. Each record item contains the interface number, acceleration factor, affected unit number, unit heat superposition value, spatial divergence direction vector and total heat energy divergence quantitative index, which serves as the basic input condition for subsequent heating path optimization.

[0110] Step S25 includes the following steps:

[0111] Step S251: performing an extreme value analysis of low temperature tolerance among livestock with different body shapes on the livestock body shape difference data to obtain the extreme values of low temperature tolerance among livestock with different body shapes;

[0112] Step S252: performing Fourier thermal spectrum transformation on the spatial thermal energy vector loss deconstruction data to obtain a frequency domain distribution spectrum of thermal energy loss;

[0113] Step S253: performing a time-varying nonlinear dynamic evolution of body surface heat energy loss among livestock of different body shapes on the livestock body shape difference data based on the heat energy loss frequency domain distribution spectrum and the extreme values of livestock low temperature tolerance among livestock of different body shapes, to obtain nonlinear time-varying data of body surface heat energy loss;

[0114] Step S254: performing convergence-constrained power series expansion on the nonlinear time-varying data of body surface heat energy loss to obtain a convergent power series of body surface heat energy loss;

[0115] Step S255: performing a body hypothermia measurement integration between livestock with different body shapes on the livestock body shape difference data according to the convergence power series of body surface heat energy loss to generate livestock body hypothermia measurement data.

[0116] In an embodiment of the present invention, when analyzing livestock body shape difference data, an interval mathematical clustering algorithm is first used to classify the body shape data. Body shape data includes, but is not limited to, parameters such as body surface area, subcutaneous fat thickness, back line length, and limb length. Specifically, data from 180 livestock at different growth stages were collected, with body surface areas ranging from 0.6 square meters to 1.9 square meters, subcutaneous fat thickness from 0.8 centimeters to 4.2 centimeters, and back line lengths from 45 centimeters to 90 centimeters. Interval centroid mean aggregation was used to define five typical body shape segments, each of which corresponded to a critical point of heat flux loss under a constant temperature environment. By comparing the heat flux changes under the time axis distribution with the actual sampled surface temperature difference data, the corresponding cold response time node for each body shape group (e.g., the time required for the surface temperature to drop to 34°C) was extracted. The statistical range method was then used to extract the extreme thermal threshold response for each body shape group, which was defined as the extreme cold tolerance value. When performing frequency domain processing on the spatial thermal energy vector loss deconstruction data, the data is first transformed into the three-dimensional coordinates (x, y, z) and the heat loss value per unit time t. Discrete calibration was performed and a four-dimensional heat tensor matrix was constructed. The heat energy vector corresponding to each location was decomposed into spatial axis components and a loss time series. A fast Fourier transform algorithm was then used to perform spectral conversion on the heat loss series at each location, extracting frequency response functions and calibrating the energy density distribution within different frequency bands. For example, after transforming the surface heat loss data, a dominant peak appeared in the frequency domain spectrum in the 0.1Hz-0.3Hz range, corresponding to the concentrated response of instantaneous heat loss caused by nighttime room temperature fluctuations. The spectrum was further constructed into a two-dimensional frequency domain spectrum with heat energy density as the vertical axis and spatial points as the horizontal axis, facilitating subsequent analysis of heat flow migration trends. By combining the heat loss frequency domain distribution spectrum with the extreme values of livestock cold tolerance, a nonlinear mapping relationship was first established between body parameters and heat loss data for the dynamic evolution of time-varying heat loss based on body posture data. By constructing a bivariate coupling matrix with body parameters (such as subcutaneous fat thickness) as the primary variables and heat energy spectrum density as the response variable, a spline function interpolation algorithm was used to extract the changing trends of heat loss rates within each time period. Taking a typical body sample with a subcutaneous fat thickness of 1.2 cm and a body surface area of 1.1 m2 as an example, the corresponding heat density at a frequency interval of 0.15 Hz is 2.4 kilojoules per square decimeter. Considering its extreme low-temperature tolerance range of 33.5°C, the heat loss curve per unit time is simulated. By fitting the derivative of its heat flow trend, a nonlinear time-varying curve of its surface heat energy loss is generated. This curve is then grouped into corresponding body postures, forming a body posture-based heat evolution time series data set, and the final nonlinear time-varying data of surface heat energy loss is output.

[0117] In the convergence-constrained power series expansion of nonlinear, time-varying data on surface heat loss, the time series curves are initially smoothed to remove spikes caused by short-term environmental fluctuations. Segmented spline interpolation is used for smoothing, and the change in heat loss per unit surface area for each animal body type is extracted at 15-minute intervals. For a sample with a surface area of 1.2 square meters and a subcutaneous fat thickness of 1.0 centimeter, the surface heat loss data points from t = 0 to t = 180 minutes are as follows: 0.11 kJ / dm² at the 1st minute, 0.45 kJ / dm² at the 15th minute, and 0.88 kJ / dm² at the 30th minute, with these values progressively increased to 180 minutes. This heat loss function, f(t), is then subjected to a Taylor series expansion, with the order of the expansion limited to 7 to account for the stability of the data environment. The expansion form uses a convergence constraint of a power function term coefficient decreasing rate of no less than 0.7. Seven power function terms are constructed by taking the derivatives of each order of this function at the initial time point. The expanded function is then verified to have a fitting error of less than 0.03% with the original nonlinear heat loss curve. This ultimately yields a convergent power series function for the surface heat loss of the livestock individual under the current temperature environment and its own body shape. This power series expansion step is repeated for all body shape samples, yielding a complete set of power series functions. When integrating the body heat loss carryover metric for livestock body shape difference data based on the convergent power series, the power series functions of the different body shape samples are first structurally aligned to ensure that the expanded form is uniformly 7th order or less. Time parameters for all samples are normalized to between 0 and 180 minutes. A body carryover factor is then defined, consisting of three parameters: body surface area, subcutaneous fat thickness, and dorsal length. It is calculated as the body surface area multiplied by the product of subcutaneous fat thickness and dorsal length. For example, sample A has a body surface area of 1.2 square meters, subcutaneous fat thickness of 1.0 centimeter, and a dorsal length of 80 centimeters. Its carrying factor is 1.2 × 1.0 × 80 = 96. This factor is then multiplied by the corresponding power series function, and the power function curve is integrated from 0 to 180 minutes to obtain the total surface heat energy loss per unit time for sample A. All samples are processed in the same manner, ultimately summarizing the body hypothermia data for each animal size exposed to an isothermal environment. The data is expressed in kilojoules, and the output structure is the body hypothermia at a specific temperature corresponding to each animal number.

[0118] Step S254 includes the following steps:

[0119] The instantaneous growth index of heat energy loss is calculated for the nonlinear time-varying data of body surface heat energy loss to obtain the instantaneous growth index of heat energy loss;

[0120] Based on the instantaneous growth index of heat energy loss, the relative entropy difference of local mutation of heat energy loss is analyzed to obtain the relative entropy difference of local mutation of loss;

[0121] Based on the relative entropy difference of the local mutation of loss, the convergence of the mutation point of the nonlinear time-varying data of the body surface heat energy loss is analyzed to obtain the convergence data of the loss mutation point;

[0122] According to the convergence data of loss mutation point, the nonlinear time-varying data of body surface heat energy loss is expanded with convergence constraint power series to obtain the convergence power series of body surface heat energy loss.

[0123] In the present embodiment, when calculating the instantaneous growth index of heat loss for nonlinear, time-varying data on body surface heat loss, the original time-varying series of heat loss is first segmented into equally spaced time points. The time interval is set to 5 minutes, and the time range is cut from minute 0 to minute 180, extracting a total of 37 time-point heat loss value sequences. For a cattle sample with a body surface area of 1.3 square meters and a subcutaneous fat thickness of 0.8 centimeters, the heat loss per unit area of the body surface at 0, 5, 10, and 15 minutes is 0.10 kJ / dm², 0.25 kJ / dm², 0.55 kJ / dm², and 0.95 kJ / dm², respectively. A logarithmic growth difference analysis is performed on the heat energy change within each adjacent time period. The unit time growth rate is defined as the natural logarithm of the ratio of the heat loss value at the subsequent time point to the heat loss value at the previous time point, thereby forming a set of growth indices in the form of a time series. Taking the 10- to 15-minute segment as an example, the growth index is the logarithm (0.95 divided by 0.55). Through time series analysis of this growth index, we extract the fluctuation trends of heat energy growth within different time periods. Furthermore, when the growth index significantly exceeds the overall average during specific periods, such as 30 to 45 minutes, these periods are identified as early warning points for sudden increases in heat energy loss. This process is repeated for all livestock samples, forming independent time series of growth indices for each sample, which are then used for local mutation identification. In the relative entropy difference analysis of local mutations in heat energy loss based on the instantaneous growth index of heat energy loss, the instantaneous growth index series of each sample is first tested for stationarity according to the time series structure. Using a sliding window of length 3, the local distribution probability of the growth index within each window is calculated across the entire time series. The statistical method discretizes the growth index values into equal-width intervals and normalizes the frequency of growth values falling within each interval. Then, in two adjacent sliding windows, the normalized probability distribution sequence was extracted respectively, and based on the Shannon entropy theory, the local entropy value difference in each window was calculated. Then, by further normalizing the entropy value difference between the previous and next windows, the relative entropy difference at that time point was obtained. Taking the sample cattle in the 45 to 60 minute segment as an example, the relative entropy difference reached its maximum at the 55th minute, which was 0.387, much higher than the previous average relative entropy difference of 0.114 for the entire sequence. This point was determined to be the mutation point of heat energy loss. All livestock samples were subjected to relative entropy difference analysis in the same way to identify their respective local mutation positions in the time series, forming a data set of relative entropy difference of local mutation loss. The output form is the time point with the maximum relative entropy difference corresponding to the livestock number and its entropy difference value.

[0124] In the process of analyzing the convergence of mutation points in nonlinear, time-varying data of body surface heat loss, a list of mutation time nodes is first established for each livestock sample type based on the relative entropy difference data of local loss mutations obtained in the previous step. For adult sows with sample number B17, for example, the heat loss growth index exhibits relative entropy difference mutation peaks at the 35th, 55th, 90th, and 110th minutes, respectively, with corresponding entropy difference values of 0.412, 0.387, 0.439, and 0.462. Convergence assessment is performed on the time series of these mutation nodes by first-order differencing the time interval sequence of the mutation points to generate a sequence of adjacent time intervals of 20 minutes, 35 minutes, and 20 minutes, respectively. Second-order differencing is then performed on this time interval sequence. If multiple consecutive values approach zero or the amplitude converges, the time position of the mutation point is considered to exhibit weak convergence in the time domain. On this basis, average trend analysis was performed on the differential data of mutation time intervals for samples within the same livestock body type group. If the second-order difference convergence trends of multiple samples were consistent, the livestock body type was determined to have a stable thermal energy mutation rhythm. Finally, a set of mutation point convergence data for each livestock type was output. The data structure consisted of a list of mutation time points corresponding to the livestock body type number and their first-order and second-order difference convergence trend values. In the process of performing a convergence-constrained power series expansion of the nonlinear time-varying data of surface thermal energy loss based on the loss mutation point convergence data, a local nonlinear time series function was constructed by first intercepting the thermal energy loss time-varying data segment ±20 minutes before and after the mutation point for each livestock sample within the time period of its mutation point. For example, for sow B17, which experienced a mutation at the 55th minute, the heat loss sequence from 35 to 75 minutes was intercepted, yielding nine data points: 0.38, 0.42, 0.49, 0.59, 0.68, 0.76, 0.81, 0.83, and 0.85 kilojoules per square decimeter. A power series fitting operation based on the Taylor series principle was performed on this local sequence, with the expansion point set at 55 minutes after the mutation, and a maximum expansion order of 5. The convergence trend values from the convergence data at the mutation point were used as weight constraints during the fitting process. The coefficients of each level were estimated using least squares methods. The coefficient set was solved through constrained optimization, ultimately constructing a power series representation, encompassing a 0th-order constant term to a 5th-order time exponential term, representing the nonlinear growth of heat loss over time for this individual animal during the mutation period. This operation was repeated for all animals in the same group, and the converged power series function coefficient set corresponding to the body type number was output for subsequent hypothermia measurement. The data types of the convergent power series are: livestock number, mutation point time, expansion point, coefficient of each order, fitting residual and convergence score.

[0125] Step S3 includes the following steps:

[0126] Step S31: normalizing the spatial thermal energy vector loss deconstruction data to obtain spatial thermal energy vector loss normalized data;

[0127] Step S32: performing logic learning processing on the livestock hypothermia measurement data to obtain livestock hypothermia learning data;

[0128] Step S33: performing spatial thermal energy replenishment rate matching according to the spatial thermal energy vector loss normalization data to obtain spatial thermal energy replenishment rate data;

[0129] Step S34: matching heating steam circulation requirements between livestock of different body shapes based on the livestock hypothermia learning data to obtain heating steam circulation requirements;

[0130] Step S35: Construct a heating regulation model based on the spatial heat energy replenishment rate data and the heating hot steam cycle demand to obtain a heating adaptive regulation model, and send the heating adaptive regulation model to the cloud platform to execute intelligent heating control of the livestock farm.

[0131] As an example of the present invention, refer to Figure 3 As shown, in this example, step S3 includes:

[0132] Step S31: normalizing the spatial thermal energy vector loss deconstruction data to obtain spatial thermal energy vector loss normalized data;

[0133] In this embodiment of the present invention, the normalization process for spatial heat energy vector loss deconstruction data requires establishing upper and lower limits for the maximum and minimum values of the spatial heat energy vector loss. Taking the heat energy vector loss deconstruction data for the southeastern area (numbered A3) of a livestock farm over a 24-hour period as an example, the measured heat energy vector values for this area at different time points ranged from 1.43 kJ / m2 to 7.85 kJ / m2. The global maximum value for the heat energy vector loss deconstruction data for this area was 8.12, and the minimum value was 1.23. Normalization utilizes a linear interval mapping method, mapping the original heat energy vector values from the interval [1.23, 8.12] to the standardized interval [0, 1]. This process includes point-by-point value acquisition, interval difference conversion, and mapping output. For example, if the heat energy vector loss at a certain moment is 6.22 kJ / m2, the normalized value is (6.22 - 1.23) divided by (8.12 - 1.23), resulting in a value of 0.765. This processing operation is carried out point by point on all spatial units and corresponding time dimensions, and finally forms a normalized data matrix of spatial thermal energy vector loss. The matrix dimensions correspond to spatial coordinates and time series, and the unit element value is between 0 and 1, indicating the relative thermal energy loss intensity level.

[0134] Step S32: performing logic learning processing on the livestock hypothermia measurement data to obtain livestock hypothermia learning data;

[0135] In this embodiment of the present invention, during the logical learning process of livestock hypothermia measurement data obtained in the aforementioned steps, the data is grouped by livestock body type classification number (e.g., adult boar P1, adult sow P2, growing pig P3, etc.). A logical classification is performed on each sample group based on the relationship between the hypothermia measurement value and the environmental factors of heat energy loss. For example, for adult sows, the hypothermia measurement data values are concentrated in the range of 0.67 to 0.83 between 10:00 AM and 2:00 PM daily. Combined with the fact that the normalized external thermal energy vector values during this period are concentrated in the range of 0.38 to 0.41, a logical relationship set L1 is constructed: when the normalized thermal energy vector value is between 0.38 and 0.41 and the livestock is in body type P2, the hypothermia measurement value is stable between 0.67 and 0.83. This logical classification utilizes discrete logic Boolean segmentation, with the classification criteria including the normalized thermal energy value range, body type number, and daytime. When multiple hypothermia-thermal energy matching logical linear intervals exist for samples of the same body type in different time periods, multiple logical condition subsets, such as L2 and L3, are generated. The final output of the logical learning data is a set of multiple condition matching expressions, consisting of upper and lower limits for hypothermia measurement, upper and lower limits for normalized thermal energy, body type number, and time period. Each condition subset includes sample support and logical consistency confidence, with values ranging from 0.00 to 1.00.

[0136] Step S33: performing spatial thermal energy replenishment rate matching according to the spatial thermal energy vector loss normalization data to obtain spatial thermal energy replenishment rate data;

[0137] In an embodiment of the present invention, when matching the spatial heat energy replenishment rate based on the normalized spatial heat energy vector loss data, the spatial heat source replenishment zoning of the livestock farm is first determined. Each zoning corresponds to a heat energy supply capacity parameter. For example, the maximum daily heat energy replenishment capacity of Southeast Zone A3 is 12.5 kilojoules per square meter. Based on this, the heat energy loss of each spatial unit in each time period is evaluated based on the normalized heat energy vector loss matrix data. Taking zone A3 at 11:00 as an example, the normalized heat energy loss value is 0.79, so its heat energy replenishment demand rate is the maximum supply rate multiplied by 0.79, or 9.875 kilojoules per square meter per hour. Similarly, all spatial units within the farm are processed sequentially by time period to form a spatial heat energy replenishment rate matrix. To avoid thermal disturbance rebound caused by local rate mutations, a third-order sliding average smoothing operation is performed on the rate matrix for each time period to reduce interference from short-term local outliers. The final output is the heat energy replenishment rate value corresponding to each spatial unit in each time period, in kilojoules / square meter / hour. The data structure is a three-dimensional matrix with the dimensions of the spatial coordinate XY axis and the time axis T. The matrix element values are strictly limited to the maximum heat replenishment rate range of each area.

[0138] Step S34: matching heating steam circulation requirements between livestock of different body shapes based on the livestock hypothermia learning data to obtain heating steam circulation requirements;

[0139] In an embodiment of the present invention, when matching heating steam cycle requirements between livestock of different body types based on livestock hypothermia learning data, it is first necessary to clearly define the logical intervals for heat supplementation corresponding to each livestock type. For example, P1 represents adult boars, P2 represents adult sows, and P3 represents growing and finishing pigs. The aforementioned logical learning data already contains the required heat supplementation intensity intervals and their confidence levels for each type under different normalized thermal energy environments. For example, for adult sows of type P2, if the normalized thermal energy environment range matched by the logical learning data for the current time period is 0.36 to 0.41, the corresponding heat supplementation intensity requirement range is 0.67 to 0.83. The interval median method is used here to quantify the required value, defining the required intensity as (0.67 + 0.83) divided by 2, which equals 0.75. Combined with the current spatial heat replenishment rate data, for example, if the regional heat replenishment rate is 11.2 kilojoules per square meter per hour and the unit energy output efficiency of hot steam is 2.5 kilojoules per cubic meter, the hot steam circulation demand is calculated as the heat demand divided by the output efficiency: 11.2 times 0.75 divided by 2.5, resulting in an hourly hot steam circulation demand of 3.36 cubic meters per square meter. This calculation is performed separately for each livestock type within the farm, generating a three-dimensional mapping table of type number, time period, and steam demand. The output data structure is a two-dimensional matrix of type number and spatial region index. The matrix value represents the hourly hot steam circulation demand in cubic meters per square meter per hour.

[0140] Step S35: Construct a heating regulation model based on the spatial heat energy replenishment rate data and the heating hot steam cycle demand to obtain a heating adaptive regulation model, and send the heating adaptive regulation model to the cloud platform to execute intelligent heating control of the livestock farm.

[0141] In an embodiment of the present invention, when constructing a heating regulation model based on spatial heat energy replenishment rate data and heating hot steam circulation demand, the two types of data must first be aligned in time and space coordinates. Taking the breeding area number A3 as an example, assume that the heat energy replenishment rate of this area is 9.875 kilojoules / square meter / hour, and the steam circulation demand corresponding to the body number P2 is 3.36 cubic meters / square meter / hour. According to the heating regulation logic, a piecewise linear matching method is used to take the heat energy rate and steam demand as input variables, construct a two-variable heat energy regulation association table, and construct a complete regulation sequence based on the 24-hour daily cycle data. The regulation model calculates the required opening time of the heating unit and the steam circulation power at each time point to ensure that the steam output meets the real-time heat energy demand intensity. The regulation model uses a discrete-time dynamic control method to handle fluctuations in the input variables, sets the sampling interval to once every 15 minutes, and updates the heating control instruction data. For example, at 12:15 PM, when the heat replenishment rate in zone A3 rises to 10.3 kilojoules per square meter per hour and the steam demand is 3.88 cubic meters per square meter per hour, the model outputs control parameters for steam valve opening of 72%, heating duration of 15 minutes, and steam pressure set at 0.5 MPa. The control parameters generated by the regulation model include fields such as spatial region index, timestamp, steam power setpoint, heating duration, and valve opening instruction. Ultimately, the complete set of heating adaptive regulation model control parameters is sent to the cloud platform control center via the MQTT standard interface, enabling cloud servers to complete command scheduling and execution, thereby achieving real-time closed-loop intelligent regulation of the heating system within the livestock farm.

[0142] The present invention also provides an intelligent heating control system for a livestock farm, which is used to execute the intelligent heating control method for a livestock farm as described above. The intelligent heating control system for a livestock farm comprises:

[0143] The data monitoring module is used to monitor the behavior and activity status of livestock in the livestock farm through electronic monitoring equipment to obtain livestock behavior and activity status monitoring videos; the temperature and humidity sensors and greenhouse gas sensors in the environmental sensing module are used to monitor the temperature and humidity fluctuations and greenhouse gas fluctuations in the livestock farm under extremely cold weather, and obtain temperature and humidity fluctuation data and greenhouse gas fluctuation data respectively;

[0144] The postural hypothermia measurement module is used to analyze postural differences in livestock behavior and activity status monitoring videos to generate livestock postural difference data; perform spatial thermal energy vector loss deconstruction on temperature and humidity fluctuation data and greenhouse gas fluctuation data to obtain spatial thermal energy vector loss deconstruction data; and perform postural hypothermia measurement integration between livestock of different postural shapes on livestock postural difference data based on the spatial thermal energy vector loss deconstruction data to generate livestock postural hypothermia measurement data;

[0145] The heating regulation model construction module is used to match the heating hot steam circulation demand between livestock of different body shapes based on livestock hypothermia learning data to obtain the heating hot steam circulation demand; the heating regulation model is constructed according to the heating hot steam circulation demand to obtain a heating adaptive regulation model, and the heating adaptive regulation model is sent to the cloud platform to implement intelligent heating control of livestock farms.

[0146] The foregoing description is intended only to provide specific embodiments of the present invention, which will enable those skilled in the art to understand and implement the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown herein, but is to be construed in the widest possible manner consistent with the principles and novel features disclosed herein.

Claims

1. An intelligent heating control method for livestock farms, characterized in that: The livestock farm is equipped with electronic monitoring equipment and an environmental sensing module, wherein the environmental sensing module includes a temperature and humidity sensor and a greenhouse gas sensor. The intelligent heating control method for the livestock farm includes the following steps: Step S1: Monitor the behavior and activity status of livestock in a livestock farm using electronic monitoring equipment to obtain a livestock behavior and activity status monitoring video; monitor the temperature and humidity fluctuations and greenhouse gas fluctuations in the livestock farm under extreme cold weather using the temperature and humidity sensors and greenhouse gas sensors in the environmental sensing module to obtain temperature and humidity fluctuation data and greenhouse gas fluctuation data respectively; Step S2: Performing posture difference analysis on livestock behavior and activity status monitoring videos to generate livestock posture difference data; performing spatial thermal energy vector loss deconstruction on temperature and humidity fluctuation data and greenhouse gas fluctuation data to obtain spatial thermal energy vector loss deconstruction data; performing posture hypothermia measurement integration between livestock of different postures on the livestock posture difference data based on the spatial thermal energy vector loss deconstruction data to generate livestock posture hypothermia measurement data; Step S3 includes: Step S31: normalizing the spatial thermal energy vector loss deconstruction data to obtain spatial thermal energy vector loss normalized data; Step S32: performing logic learning processing on the livestock hypothermia measurement data to obtain livestock hypothermia learning data; Step S33: performing spatial thermal energy replenishment rate matching according to the spatial thermal energy vector loss normalization data to obtain spatial thermal energy replenishment rate data; Step S34: matching heating steam circulation requirements between livestock of different body shapes based on the livestock hypothermia learning data to obtain heating steam circulation requirements; Step S35: Construct a heating regulation model based on the spatial heat energy replenishment rate data and the heating hot steam cycle demand to obtain a heating adaptive regulation model, and send the heating adaptive regulation model to the cloud platform to execute intelligent heating control of the livestock farm.

2. The intelligent heating control method for livestock farms according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: performing posture difference analysis on the livestock behavior and activity status monitoring video to generate livestock posture difference data; Step S22: performing group movement frequency analysis on the livestock behavior and activity status monitoring video to obtain livestock group movement frequency data; Step S23: Acquire livestock farm structural design data; perform gas disturbance dispersion analysis on greenhouse gas fluctuation data based on livestock group movement frequency data and livestock farm structural design data to obtain greenhouse gas disturbance dispersion structure data; Step S24: performing spatial thermal energy vector loss deconstruction on the temperature and humidity fluctuation data and the greenhouse gas disturbance dispersion structure data according to the livestock farm structural design data to obtain spatial thermal energy vector loss deconstruction data; Step S25: performing a body hypothermia measurement integration between livestock with different body shapes on the livestock body shape difference data based on the spatial thermal energy vector loss deconstruction data to generate livestock body hypothermia measurement data.

3. The intelligent heating control method for livestock farms according to claim 2, characterized in that: Step S23 includes the following steps: Step S231: Acquire livestock farm structural design data; perform gas circulation design path analysis on the livestock farm structural design data to obtain a gas circulation design path; Step S232: performing group gait acceleration frequency analysis on the livestock group movement frequency data to obtain the group gait acceleration frequency; Step S233: Calculating the continuous pressure difference of the peripheral airflow based on the group gait acceleration frequency to obtain the continuous pressure difference of the peripheral airflow; Step S234: Deducing the change of azimuthal velocity difference of the gas flow section on the designed gas flow path according to the continuous pressure difference driven by the surrounding airflow, and obtaining the change data of azimuthal velocity difference of the gas flow section; Step S235: performing mass / volume regional distribution difference analysis on the greenhouse gas fluctuation data to obtain the greenhouse gas mass / volume regional distribution difference; Step S236: performing gas disturbance dispersion analysis on the greenhouse gas mass / volume regional distribution difference based on the flow section azimuthal velocity difference change data to obtain greenhouse gas disturbance dispersion structure data.

4. The intelligent heating control method for livestock farms according to claim 3, characterized in that: Step S24 includes the following steps: Step S241: performing material thermal insulation performance degradation analysis on the livestock farm structural design data to obtain material thermal insulation performance degradation data; Step S242: performing temperature and humidity spatial fluctuation distribution mapping on the temperature and humidity fluctuation data according to the livestock farm structural design data to obtain temperature and humidity spatial fluctuation distribution mapping data; Step S243: Based on the material thermal insulation performance loss data, the spatial heat energy loss divergence per unit time is calculated for the temperature and humidity spatial fluctuation distribution mapping data and the greenhouse gas disturbance dispersion structure data to obtain spatial heat energy loss divergence data; Step S244: performing spatial heat energy vector loss deconstruction based on the spatial heat energy loss divergence data to obtain spatial heat energy vector loss deconstruction data.

5. The intelligent heating control method for livestock farms according to claim 4, characterized in that: Step S243 includes the following steps: The incremental coefficient of structural pore thermal conductivity is calculated based on the thermal insulation performance loss data of the material to obtain the incremental coefficient of structural pore thermal conductivity; Based on the incremental thermal conductivity coefficient of structural pores, the interface heat transfer acceleration proportional identification is performed on the material insulation performance loss data to obtain the interface heat transfer acceleration proportional data; The low temperature / low humidity relative change rate difference is calculated for the temperature and humidity spatial fluctuation distribution mapping data to generate the spatial distribution low temperature / low humidity relative change rate difference; According to the relative change rate difference of low temperature / low humidity in spatial distribution, the spatial discrete heat loss density value increment of greenhouse gas disturbance dispersion structure data per unit time is integrated to obtain the spatial heat loss density increment value per unit time; Based on the interface heat transfer acceleration proportional data and the spatial heat loss density increment value per unit time, the spatial heat energy loss divergence per unit time is calculated to obtain the spatial heat energy loss divergence data.

6. The intelligent heating control method for livestock farms according to claim 4, characterized in that: Step S25 includes the following steps: Step S251: performing an extreme value analysis of low temperature tolerance among livestock with different body shapes on the livestock body shape difference data to obtain the extreme values of low temperature tolerance among livestock with different body shapes; Step S252: performing Fourier thermal spectrum transformation on the spatial thermal energy vector loss deconstruction data to obtain a frequency domain distribution spectrum of thermal energy loss; Step S253: performing a time-varying nonlinear dynamic evolution of body surface heat energy loss among livestock of different body shapes on the livestock body shape difference data based on the heat energy loss frequency domain distribution spectrum and the extreme values of livestock low temperature tolerance among livestock of different body shapes, to obtain nonlinear time-varying data of body surface heat energy loss; Step S254: performing convergence-constrained power series expansion on the nonlinear time-varying data of body surface heat energy loss to obtain a convergent power series of body surface heat energy loss; Step S255: performing a body hypothermia measurement integration between livestock with different body shapes on the livestock body shape difference data according to the convergence power series of body surface heat energy loss to generate livestock body hypothermia measurement data.

7. The intelligent heating control method for livestock farms according to claim 6, characterized in that: Step S254 includes the following steps: The instantaneous growth index of heat energy loss is calculated for the nonlinear time-varying data of body surface heat energy loss to obtain the instantaneous growth index of heat energy loss; Based on the instantaneous growth index of heat energy loss, the relative entropy difference of local mutation of heat energy loss is analyzed to obtain the relative entropy difference of local mutation of loss; Based on the relative entropy difference of the local mutation of loss, the convergence of the mutation point of the nonlinear time-varying data of the body surface heat energy loss is analyzed to obtain the convergence data of the loss mutation point; According to the convergence data of loss mutation point, the nonlinear time-varying data of body surface heat energy loss is expanded with convergence constraint power series to obtain the convergence power series of body surface heat energy loss.

8. An intelligent heating control system for livestock farms, characterized in that: Used to execute the livestock farm intelligent heating control method according to claim 1, the livestock farm intelligent heating control system comprises: The data monitoring module is used to monitor the behavior and activity status of livestock in the livestock farm through electronic monitoring equipment to obtain livestock behavior and activity status monitoring videos; the temperature and humidity sensors and greenhouse gas sensors in the environmental sensing module are used to monitor the temperature and humidity fluctuations and greenhouse gas fluctuations in the livestock farm under extremely cold weather, and obtain temperature and humidity fluctuation data and greenhouse gas fluctuation data respectively; The postural hypothermia measurement module is used to analyze postural differences in livestock behavior and activity status monitoring videos to generate livestock postural difference data; perform spatial thermal energy vector loss deconstruction on temperature and humidity fluctuation data and greenhouse gas fluctuation data to obtain spatial thermal energy vector loss deconstruction data; and perform postural hypothermia measurement integration between livestock of different postural shapes on livestock postural difference data based on the spatial thermal energy vector loss deconstruction data to generate livestock postural hypothermia measurement data; Heating regulation model building blocks for: Normalizing the deconstructed data of the spatial thermal energy vector loss to obtain normalized data of the spatial thermal energy vector loss; Performing logic learning processing on livestock hypothermia measurement data to obtain livestock hypothermia learning data; Perform spatial thermal energy replenishment rate matching based on the spatial thermal energy vector loss normalization data to obtain spatial thermal energy replenishment rate data; Based on the livestock hypothermia learning data, the heating steam cycle demand between livestock of different body shapes is matched to obtain the heating steam cycle demand; A heating regulation model is constructed based on the spatial heat energy replenishment rate data and the heating hot steam cycle demand to obtain a heating adaptive regulation model, which is then sent to the cloud platform to execute intelligent heating control of livestock farms.

Images