Livestock and poultry epidemic disease transmission dynamics modeling and intelligent prevention and control decision-making method

Abstract

The invention belongs to the technical field of bioinformatics, and discloses a livestock epidemic disease transmission dynamics modeling and intelligent prevention and control decision-making method, which comprises a biological perception fusion module, utilizes non-contact respiratory metabolism detection and terahertz spectral imaging to identify respiratory metabolism characteristics of livestock individuals, and adopts a three-dimensional point cloud skeleton extraction method to extract the respiratory metabolism characteristics of the livestock individuals. Monitoring livestock and poultry group behaviors to obtain livestock and poultry monitoring data; the propagation dynamics modeling module is used for constructing an improved SEIR model containing a space-time dynamic propagation rate and a space diffusion coefficient based on livestock and poultry monitoring data, introducing a Transform network to carry out propagation trend prediction, and generating a space risk thermodynamic diagram; the physical field construction prevention and control module is used for simulating epidemic situation propagation in a complex environment by adopting a multi-band light sound field and environmental physical disturbance based on the space risk thermodynamic diagram, and testing whether a preset prevention and control strategy takes effect or not based on an autonomous evolutionary game mechanism; and dynamic and scientific livestock and poultry epidemic disease transmission management and prevention and control decisions are realized.

Description

Technical Field

[0001] This invention relates to the field of bioinformatics technology, and more specifically, to a method for modeling the dynamics of livestock and poultry disease transmission and making intelligent prevention and control decisions. Background Technology

[0002] Patent publication number CN111048214A discloses a method and device for early warning of the spread of diseases in introduced livestock and poultry. Specifically, the early warning method includes: acquiring quarantine data of introduced livestock and poultry; obtaining characteristic data of the disease to be predicted based on the quarantine data; determining the probability of the disease in the introduced livestock and poultry based on the characteristic data of the disease to be predicted using a disease probability model; wherein the disease probability model is trained using historical quarantine data of introduced livestock and poultry; obtaining a population risk coefficient for the disease to be predicted in the same introduced livestock and poultry species based on a preset probability and the probability of the disease to be predicted; determining the risk level of the entire batch of introduced livestock and poultry based on the characteristic data of the disease to be predicted and the population risk coefficient, and determining and pushing out an early warning plan based on the risk level. Through the early warning method and device of this invention, the spread of diseases in introduced livestock and poultry can be effectively predicted and warned.

[0003] Existing methods for modeling the transmission dynamics of livestock and poultry diseases and for intelligent prevention and control decision-making have the following shortcomings:

[0004] Current technologies primarily rely on contact or semi-contact detection, which not only stresses livestock and poultry but also makes real-time dynamic monitoring difficult. Traditional non-contact detection methods, such as infrared thermal imaging and single-band gas concentration detection, can only obtain rough indicators such as respiratory rate and body temperature, failing to deeply analyze changes in metabolic components in exhaled gases or distinguish the spectral response characteristics of characteristic molecules in secretions. Existing terahertz detection methods often employ static absorption coefficient assumptions, neglecting the dynamic fluctuations in water, protein, and carbohydrate components in secretions as livestock and poultry health status and environmental changes occur, resulting in significant errors in absorption feature identification.

[0005] Most existing spectroscopic detection methods are based on fixed absorption coefficients or thickness assumptions, failing to consider the impact of changes in animal health status and environmental conditions on the composition and state of secretions. This leads to biased results and insufficient reliability. The spectral response of livestock and poultry secretions is not constant; it is affected by various dynamic factors, including changes in secretion concentration and composition. Under different health conditions, the concentrations of water, protein, and sugars in oral and nasal secretions vary, causing changes in their absorption characteristics to terahertz waves. Changes in temperature and humidity alter the state of secretions, thus affecting their spectral absorption characteristics. In actual measurements, the thickness of secretions is difficult to keep constant.

[0006] In view of this, the present invention proposes a method for modeling the dynamics of livestock and poultry disease transmission and intelligent prevention and control decision-making to solve the above problems. Summary of the Invention

[0007] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, the present invention provides the following technical solution: a method for modeling the dynamics of livestock and poultry disease transmission and intelligent prevention and control decision-making, comprising:

[0008] S1. Non-contact respiratory metabolism detection and terahertz spectral imaging are used to identify the respiratory metabolism characteristics of individual livestock and poultry. A three-dimensional point cloud skeleton extraction method is used to monitor the behavior of livestock and poultry groups and obtain livestock and poultry monitoring data.

[0009] S2. Based on livestock and poultry monitoring data, an improved SEIR model is constructed that includes spatiotemporal dynamic propagation rate and spatial diffusion coefficient. A Transformer network is introduced to predict propagation trends and generate a spatial risk heat map.

[0010] S3. Based on the spatial risk heat map, multi-band photoacoustic field and environmental physical disturbance are used to simulate the spread of the epidemic in a complex environment, and based on the autonomous evolutionary game mechanism, the effectiveness of the preset prevention and control strategy is tested.

[0011] S4. Based on the predicted transmission trend and the effective preset prevention and control strategies, combined with the dynamic field test feedback, generate prevention and control paths, and provide interpretable prevention and control strategy decisions based on feature importance ranking.

[0012] S5. High-precision disinfection path planning is carried out using multi-line lidar and ultra-wideband positioning technology. The disinfection path is autonomously adjusted based on the spatial risk heat map, and electrolytic water spray is used for disinfection. The disinfection effect is fed back to the prevention and control decision terminal in real time.

[0013] Preferably, the method for identifying the respiratory metabolic characteristics of individual livestock and poultry includes:

[0014] By deploying infrared and laser gas sensor arrays in key areas of the farm, non-contact data is collected on the gases exhaled naturally by livestock and poultry during their daily activities and their respiratory frequency, thus obtaining respiratory metabolic detection data. During the data collection process, the metabolic intensity of the individual animal is automatically calibrated based on a preset respiratory cycle signal, and the metabolic fluctuations of the individual animal are assessed using the rate of change in gas concentration.

[0015] A broadband terahertz wave source and a high-speed reflector receiver array were deployed to scan the mouth and nose area of ​​individual livestock and poultry to obtain the absorption and reflection spectra of saliva and nasal secretions in the broadband terahertz wave band; the absorption and reflection spectra were analyzed by Fourier transform to convert the time information in the absorption and reflection spectra into spectral information.

[0016] When a terahertz beam irradiates the mouth and nose area of ​​livestock and poultry, it interacts with secretions. Different components have different absorption peaks and bandwidths at different frequencies. By analyzing the compositional characteristics of the oral and nasal secretions of individual livestock and poultry under broadband terahertz spectrum using Lambert-Beer's law, the respiratory and metabolic characteristics of individual livestock and poultry can be obtained.

[0017] Preferably, the method for generating the livestock and poultry monitoring data includes:

[0018] By deploying m depth cameras at different locations in the livestock and poultry farming environment, real-time three-dimensional spatial data of livestock and poultry groups is collected to generate high-density point cloud data. The high-density point cloud data contains the spatial distribution, body shape characteristics, and dynamic movement information of individual livestock and poultry. The collected high-density point cloud data is then filtered, denoised, and sparsified.

[0019] Spatial clustering algorithm is used to segment livestock and poultry individuals in the processed high-density point cloud data to distinguish the three-dimensional point cloud sets of different individuals; for each segmented livestock and poultry individual point cloud, skeleton extraction algorithm is used to process it to obtain the three-dimensional skeleton model of the livestock and poultry individual.

[0020] By analyzing the time series of different skeletal points of individual livestock and poultry, we can calculate joint angles, movement speeds, and gait period kinematic parameters to identify the basic behavioral states of individual livestock and poultry. Statistical analysis of individual behaviors in livestock and poultry groups yields individual behavior data. The individual behaviors extracted from the skeletal structure are then integrated with respiratory and metabolic monitoring data to form livestock and poultry monitoring data.

[0021] Preferably, the method for predicting propagation trends includes:

[0022] Based on livestock and poultry monitoring data, a three-dimensional skeleton model is used to extract the temporal behavioral state, respiratory and metabolic characteristics and environmental parameters of individual livestock and poultry, and the transmission rate is dynamically calculated. This transmission rate reflects the possibility of infection between livestock and poultry at different times and locations, and is adjusted in real time with changes in farm environment and group behavior.

[0023] Based on the spatial location and movement trajectory of individual livestock and poultry, the spatial diffusion coefficient is used to represent the speed of disease spread in the farm. The spatial diffusion coefficient is dynamically updated according to the density and movement pattern of livestock and poultry groups. The state variables of the improved SEIR model are constructed into spatiotemporal sequence data to reflect the dynamic process of disease spread.

[0024] A deep learning network based on the Transformer structure is used to jointly model the temporal behavioral state and environmental parameter features to learn the disease transmission pattern and potential spatiotemporal dependencies. By training and optimizing the Transformer network, the predicted values ​​of disease infection risk at various time points and spatial locations in the future are obtained.

[0025] Preferably, the method for generating the space risk heatmap includes:

[0026] The predicted risk values ​​of disease infection at each future time point are correlated with the spatial location of individual livestock and poultry to form a spatial risk distribution point set at each time point; the spatial risk distribution point set is interpolated, and a spatial interpolation algorithm is used to construct a continuous risk field within the farm.

[0027] The reconstructed risk field is visualized as a two-dimensional heat map of the farm space. The steps include dividing the farm space into grid cells according to a preset resolution; using the center point of each grid cell as an interpolation node to calculate its infection risk value; and assigning different colors according to the level of infection risk value to form an intuitive spatial risk heat map.

[0028] Preferably, the method for simulating the spread of an epidemic under complex conditions includes:

[0029] Based on the spatial risk heat map, multi-band photoacoustic field and environmental physical disturbance are further used to simulate the transmission path and speed of livestock and poultry diseases in complex environments. A multi-band photoacoustic generator at point L is set up in the farm to simulate the sound field and photoacoustic disturbance in complex environments, generate photoacoustic waves covering different frequencies and energy densities, and simulate the impact of different environmental factors on the carrying, deposition, diffusion and transmission path of virus particles in the air.

[0030] The multi-band photoacoustic generator adjusts the frequency, power and emission direction of the sound waves, superimposes acoustic disturbances to form an acoustic potential well and a local high-pressure zone, and simulates the obstruction and acceleration effects of different spatial locations in a farm on the movement of pathogen particles.

[0031] By combining the infection risk values ​​in the spatial risk heat map, acoustic perturbations and photoacoustic modulation are superimposed on the regions corresponding to different infection risk values ​​to simulate the dynamic spread of virus particles in a complex environment. Based on the actual layout and ventilation conditions of the farm, the local airflow field is adjusted by physical perturbation to restore the pathogen transmission path in a complex environment. Combined with the collection of local gas and aerosol concentrations by sensor array, the spread speed, spatial distribution and infection risk of the epidemic in a complex environment are dynamically simulated. The simulation results are combined with the spatial risk heat map to form a dynamically evolving spatiotemporal transmission trend map.

[0032] Preferably, the method for testing whether the preset prevention and control strategy is effective includes:

[0033] An autonomous evolutionary game framework is constructed, in which individual livestock and poultry are regarded as game players. Each game player selects different behavioral strategies to maximize its own survival benefits based on its own spatiotemporal location, health status, surrounding population density and environmental conditions.

[0034] In this autonomous evolutionary game framework, a pre-defined game payoff function is introduced, which includes health payoff, social payoff, energy consumption, and external constraints. A pre-defined prevention and control strategy is introduced, which serves as an external game intervention condition to dynamically affect the payoff function and the available strategy space of the game players.

[0035] Evolutionary game theory algorithms were used to simulate the behavioral evolution of livestock and poultry groups under different prevention and control strategies. Through iterative calculations at time steps, the changing trends of group behavior patterns and transmission paths were observed. The changes in group behavior and the spread of the epidemic were monitored during the game process.

[0036] If the game payoff after the introduction of the preset prevention and control strategy is greater than or equal to the preset game payoff threshold, and the epidemic spread speed is less than the preset epidemic spread speed threshold, then the preset prevention and control strategy is deemed effective; if the game payoff after the introduction of the preset prevention and control strategy is less than the preset game payoff threshold, and the epidemic spread speed is greater than or equal to the preset epidemic spread speed threshold, then the preset prevention and control game payoff threshold is deemed invalid.

[0037] Preferably, the method for generating the prevention and control path includes:

[0038] Based on the improved SEIR model and Transformer network prediction results, the disease transmission risk trend at various time points and spatial locations in the future is obtained, including the distribution of high-risk areas, potential transmission paths and transmission speed; the disease transmission risk trend is integrated with the spatiotemporal transmission trend map to dynamically correct the spatial risk heat map;

[0039] Based on the fused propagation trend data and the corrected spatial risk heat map, the boundary features, internal density distribution and propagation speed change trends of different risk areas are extracted to generate a prevention and control path planning model. Based on the spatiotemporal propagation trend map, a prevention and control path scheme including isolation zone layout, temporary blockade area, physical isolation barrier, guidance channel and optimized ventilation route is determined. The prevention and control path scheme is matched with the preset prevention and control scheme that has been verified and effective to form a dynamic prevention and control path.

[0040] Preferably, the method for providing interpretable prevention and control strategy decisions includes:

[0041] By analyzing the internal feature vectors of the Transformer network, a feature importance ranking algorithm is used to quantify the contribution of different input features to the prediction results of disease transmission and to rank the features by importance.

[0042] Based on the ranking of feature importance, a prevention and control strategy decision tree is generated to provide interpretable prevention and control strategy decisions for livestock and poultry farms, and to define the logical basis and influencing factors behind each prevention and control path and strategy decision.

[0043] Preferably, the method for autonomously adjusting the prevention and control path includes:

[0044] Multi-line lidar sensors and ultra-wideband positioning devices are deployed in livestock and poultry farms to construct a three-dimensional spatial environment model and mobile positioning network. The multi-line lidar scans the environment to collect real-time information on the three-dimensional geometric structure and obstacles of the farm, forming a high-resolution point cloud map. The ultra-wideband positioning device uses UWB base stations and mobile tags deployed in the farm to perform three-dimensional positioning of disinfection robots or human operators.

[0045] Based on spatial risk heat maps, we analyze the different color representations and spread trends of each region to identify high-risk areas and key prevention and control nodes; and we design disinfection path planning algorithms using the spatial distribution information of spatial risk heat maps.

[0046] The disinfection path planning algorithm aims to optimize the shortest path, cover the largest risk area, and minimize repeated disinfection. It comprehensively considers environmental obstacles, the layout of breeding facilities, and the animal activity area to calculate the initial disinfection path.

[0047] During the disinfection process, the current location and path execution status of the disinfection equipment are fed back in real time through ultra-wideband positioning devices, and high-resolution point cloud maps are continuously updated through multi-line lidar to capture information on environmental changes and temporary obstacles. The initial disinfection path is dynamically adjusted according to the latest spatial risk heat map, prioritizing coverage of newly generated or risk-increasing areas. Environmental constraints are introduced, and the energy efficiency of the initial disinfection path is optimized by combining the energy consumption and operation time limits of the disinfection equipment, and the disinfection path is adjusted autonomously.

[0048] The technical effects and advantages of the present invention, a method for modeling the transmission dynamics of livestock and poultry diseases and for intelligent prevention and control decision-making, are as follows:

[0049] This invention combines gas component detection with broadband terahertz spectral imaging using a multi-sensor array to achieve multi-dimensional and continuous monitoring of oral and nasal secretions and respiratory gases in livestock and poultry. This comprehensively reflects the health and metabolic status of individual animals and enhances early warning capabilities for disease transmission. By accurately identifying the absorption peaks of water in secretions and the broadband absorption bands of key molecules such as viral proteins and sugars, the invention significantly enhances the depth of respiratory metabolic feature recognition and spectral resolution, enabling sensitive capture of subtle changes in the early stages of pathogen infection. By dynamically adjusting the absorption coefficient based on real-time detection of changes in the water content of secretions, the invention effectively compensates for spectral feature shifts and errors caused by fluctuations in environmental temperature and humidity, as well as changes in animal physiological states. This ensures the accuracy and reliability of data input into the kinetic model and avoids the risk of misjudgment due to environmental interference. This dynamic adjustment mechanism improves the stability and applicability of terahertz spectral detection in complex farming environments.

[0050] Based on precise individual respiratory and metabolic characteristics, combined with multi-dimensional environmental parameters, a disease transmission dynamics model has been constructed that can more accurately describe the transmission rate and spatial diffusion path of pathogens, providing a solid basis for the scientific formulation of intelligent prevention and control strategies. The model can dynamically update the health status and transmission risk of each individual livestock and poultry, reflecting the real-time development trend of the epidemic within the farm. Through intelligent algorithms, prevention and control resources are optimized and dynamically adjusted, effectively reducing the risk of disease transmission and improving the overall biosecurity level and disease control efficiency of the farm. Simultaneously, this technical solution supports non-invasive health monitoring of large-scale livestock and poultry populations, reducing manual intervention and labor intensity, promoting the intelligent and digital transformation of livestock production, and providing strong support for the sustainable development of animal husbandry. Attached Figure Description

[0051] Figure 1 A schematic diagram of a method for modeling the transmission dynamics of livestock and poultry diseases and making intelligent prevention and control decisions.

[0052] Figure 2 A schematic diagram of the structure of a livestock and poultry disease transmission dynamics modeling and intelligent prevention and control decision-making system;

[0053] Figure 3 A method flowchart is provided for this invention. Detailed Implementation

[0054] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0055] Example 1

[0056] Please see Figure 1 and Figure 3 As shown, Example 1 further illustrates the livestock and poultry disease transmission dynamics modeling and intelligent prevention and control decision-making method proposed in this invention, including:

[0057] The spread of livestock and poultry diseases not only seriously threatens the economic benefits of the livestock industry but also relates to public health security. Accurate and real-time monitoring of the health status of livestock and poultry and the dynamics of disease transmission is crucial for effectively controlling outbreaks and spread. However, existing methods for modeling the dynamics of livestock and poultry disease transmission and for intelligent prevention and control decision-making still have many shortcomings.

[0058] Traditional livestock and poultry health monitoring relies heavily on contact or semi-contact detection technologies, such as body temperature measurement, blood sampling, and laboratory testing. While these methods can provide certain physiological and pathological indicators, they require manual operation, which can easily cause stress responses in individual livestock and poultry, affecting their normal behavior. Furthermore, they are difficult to implement on a large scale and continuously, thus limiting the ability to detect diseases early and track them in real time.

[0059] To mitigate the stress associated with contact-based detection, some research and applications have employed non-contact monitoring technologies, such as infrared thermal imaging and single-band gas concentration detection. While these methods enable long-distance measurements, the information they acquire is relatively limited, primarily focusing on coarse indicators like respiratory rate and surface body temperature. They lack in-depth analysis of the complex metabolic substances in exhaled gases and their dynamic changes, making it difficult to reflect the metabolic health and immune status of individual livestock and poultry, resulting in insufficient sensitivity and accuracy in disease early warning.

[0060] In recent years, terahertz spectroscopy has been introduced into the field of livestock and poultry health monitoring due to its ability to penetrate biological tissues and identify various biomolecules. However, existing terahertz detection methods are generally based on the static absorption coefficient assumption, which assumes that the spectral absorption characteristics of livestock and poultry oral and nasal secretions remain unchanged under different times and environmental conditions. This assumption ignores the dynamic fluctuations in components such as water, protein, and sugar in secretions as the health status of livestock and poultry and environmental changes occur, leading to significant errors in spectral feature identification and seriously affecting the reliability and practicality of the detection results.

[0061] Current spectral detection methods often employ models with fixed thickness and absorption coefficients, failing to adequately consider the impact of changes in animal health status, environmental temperature, humidity, and ventilation conditions on the composition and physical state of oral and nasal secretions. The concentration and thickness of secretions constantly change in actual farming environments. This dynamic characteristic makes the spectral absorption response complex and variable, making it difficult for traditional methods to accurately capture these dynamic changes. This limits the accuracy of health status assessment and disease transmission dynamics modeling based on spectral data.

[0062] Traditional disease transmission dynamics models are mostly classic SEIR models, which lack sufficient integration with the complex behavior of livestock and poultry groups and the variability of the breeding environment, and cannot dynamically reflect the spatiotemporal transmission characteristics of diseases. At the same time, existing intelligent prevention and control strategies mostly rely on experience or simple statistical methods, lacking deep learning and joint modeling of complex spatiotemporal data and environmental factors, making it difficult to accurately predict future disease transmission trends and provide scientific decision support.

[0063] Existing methods for modeling the dynamics of livestock and poultry disease transmission and for intelligent prevention and control decision-making have the following main technical shortcomings: First, the detection methods rely on contact or semi-contact technologies, which cannot achieve interference-free, continuous, and dynamic monitoring. Second, non-contact detection provides limited information and cannot deeply capture the dynamic spectral characteristics of metabolic components and secretions in exhaled gases. Third, spectral detection is based on static assumptions, ignoring the time-varying nature of secretion components and states, resulting in large identification errors and insufficient reliability. Fourth, disease transmission models lack integrated modeling of group behavior and dynamic environmental factors, and intelligent prevention and control decisions lack refined, personalized, and real-time response capabilities.

[0064] To effectively address the above problems, this invention proposes a method for modeling the dynamics of livestock and poultry disease transmission and for intelligent prevention and control decision-making, including:

[0065] S1. Non-contact respiratory metabolism detection and terahertz spectral imaging are used to identify the respiratory metabolism characteristics of individual livestock and poultry. A three-dimensional point cloud skeleton extraction method is used to monitor the behavior of livestock and poultry groups and obtain livestock and poultry monitoring data.

[0066] S2. Based on livestock and poultry monitoring data, an improved SEIR model is constructed that includes spatiotemporal dynamic propagation rate and spatial diffusion coefficient. A Transformer network is introduced to predict propagation trends and generate a spatial risk heat map.

[0067] S3. Based on the spatial risk heat map, multi-band photoacoustic field and environmental physical disturbance are used to simulate the spread of the epidemic in a complex environment, and based on the autonomous evolutionary game mechanism, the effectiveness of the preset prevention and control strategy is tested.

[0068] S4. Based on the predicted transmission trend and the effective preset prevention and control strategies, combined with the dynamic field test feedback, generate prevention and control paths, and provide interpretable prevention and control strategy decisions based on feature importance ranking.

[0069] S5. High-precision disinfection path planning is carried out using multi-line lidar and ultra-wideband positioning technology. The disinfection path is autonomously adjusted based on the spatial risk heat map, and electrolytic water spray is used for disinfection. The disinfection effect is fed back to the prevention and control decision terminal in real time.

[0070] Methods for identifying the respiratory and metabolic characteristics of individual livestock and poultry include:

[0071] Livestock and poultry emit gases (such as carbon dioxide, ammonia, methane, and volatile organic compounds) during respiration. These gases carry information about an individual's health and metabolic characteristics, reflecting physiological states (such as stress, infection, and metabolic abnormalities). Non-contact detection uses sensing devices (such as infrared gas sensors, laser gas analyzers, and mass spectrometers) to collect the concentration of specific gas components.

[0072] By deploying infrared and laser gas sensor arrays in key areas of the farm, non-contact data is collected on the gases exhaled naturally by livestock and poultry during their daily activities and their respiratory frequency, resulting in respiratory metabolic detection data. The respiratory metabolic detection data includes CO2, NH3, CH4, and VOCs (volatile organic compounds). During the data collection process, the metabolic intensity of individual animals is automatically calibrated based on preset respiratory cycle signals, and the metabolic fluctuations of individual livestock and poultry are assessed using the rate of change in gas concentration.

[0073] A broadband terahertz wave source and a high-speed reflector receiver array were deployed to scan the mouth and nose area of ​​individual livestock and poultry to obtain the absorption and reflection spectra of saliva and nasal secretions in the broadband terahertz wave band; the absorption and reflection spectra were analyzed by Fourier transform to convert the time information in the absorption and reflection spectra into spectral information.

[0074] When a terahertz beam irradiates the mouth and nose area of ​​livestock and poultry, it interacts with secretions. Different components have different absorption peaks and bandwidths at different frequencies. By analyzing the compositional characteristics of the oral and nasal secretions of individual livestock and poultry under broadband terahertz spectrum using Lambert-Beer's law, the respiratory and metabolic characteristics of individual livestock and poultry can be obtained.

[0075] Lambert-Beer's law is: I(v) = I0(v)·e -α(v)·d Where I(v) represents the intensity of the terahertz wave after passing through the nasal secretions of livestock and poultry at frequency v; I0(v) represents the terahertz wave intensity of the preset parameter; α(v) represents the absorption coefficient, which reflects the absorption capacity of the secretions for terahertz waves at frequency v; d represents the thickness of the secretions; and v represents the frequency of the terahertz electromagnetic wave.

[0076] However, the spectral response of livestock and poultry secretions is not constant; it is affected by a variety of dynamic factors, including changes in secretion concentration and composition. Under different health conditions, the concentrations of water, protein, and sugar in oral and nasal secretions will change, leading to changes in their absorption characteristics to terahertz waves. Changes in temperature and humidity will alter the state of secretions, thus affecting their spectral absorption characteristics. In actual measurements, the thickness of secretions is difficult to keep constant, which will cause deviations in the estimation of the absorption coefficient.

[0077] The absorption coefficient in Beer-Lambert law is dynamically adjusted using an absorption coefficient adjustment function, which is: α'(v)=α(v)·(1+k·ΔC) H2O (t)); where α'(v) represents the absorption coefficient after dynamic adjustment; ΔC H2O (t) represents the change in the moisture content of secretions at time t relative to the preset moisture content of secretions; k is the adjustment coefficient, which represents the sensitivity of the absorption coefficient to the rate of change in moisture content;

[0078] It should be noted that changes in water content can lead to changes in the material's ability to absorb terahertz waves, and water content is the main factor affecting absorption in the oral and nasal secretions of livestock and poultry; the effect of the absorption coefficient on water content can be approximated as a linear relationship at low concentrations or within a small fluctuation range. This assumption stems from the fact that in many spectroscopic experiments, physical quantities can be approximated as linear functions (the first term of the Taylor expansion) at small deviations.

[0079] This solves the following problems existing in the technology:

[0080] Current technologies primarily rely on contact or semi-contact detection (such as blood collection and sample analysis), which not only stresses livestock and poultry but also makes real-time dynamic monitoring difficult. Traditional non-contact detection methods, such as infrared thermography and single-band gas concentration detection, can only obtain rough indicators such as respiratory rate and body temperature, failing to deeply analyze changes in metabolic components in exhaled gases or distinguish the spectral response characteristics of characteristic molecules (such as water, proteins, and carbohydrates) in secretions. Existing terahertz detection methods mostly employ static absorption coefficient assumptions, ignoring the dynamic fluctuations in water, protein, and carbohydrate components in secretions with changes in livestock and poultry health status and environment, resulting in significant errors in absorption feature identification. Most existing spectroscopic detection methods are based on fixed absorption coefficient or thickness assumptions, failing to consider the influence of changes in animal health status and environmental conditions (temperature and humidity) on the composition and state of secretions, leading to biased detection results and insufficient reliability.

[0081] Compared to existing technologies, the beneficial effects are as follows: By combining gas component detection with broadband terahertz spectral imaging using a multi-sensor array, the system can continuously collect the absorption spectra and respiratory gas parameters of livestock and poultry oral and nasal secretions, comprehensively reflecting individual health and metabolic status, and improving the early warning capability for epidemic transmission. Accurate identification of moisture absorption peaks and absorption bands of key molecules such as viral proteins and sugars greatly enhances the depth and resolution of respiratory metabolic characteristics. By real-time detection of changes in secretion moisture content and adjustment of the absorption coefficient, the system effectively compensates for the impact of environmental temperature and humidity fluctuations and physiological state changes on spectral characteristics, ensuring the reliability of the input data for the kinetic model. Based on precise individual respiratory metabolic characteristics, the model can more accurately describe the pathogen transmission rate and diffusion path, providing a scientific basis for intelligent prevention and control strategies. Dynamically updating individual health status and transmission risk enables optimized allocation and real-time adjustment of prevention and control resources, reducing the risk of disease transmission and improving the safety level of livestock and poultry farming.

[0082] Methods for generating livestock and poultry monitoring data include:

[0083] By deploying m depth cameras (such as ToF cameras, structured light sensors, or laser scanners) at different locations in the livestock and poultry farming environment, real-time three-dimensional spatial data of livestock and poultry groups is collected to generate high-density point cloud data. The high-density point cloud data contains the spatial distribution, body shape characteristics, and dynamic movement information of individual livestock and poultry. The collected high-density point cloud data is then filtered, denoised, and sparsified.

[0084] Spatial clustering algorithms (such as density-based DBSCAN or Euclidean distance clustering) are used to segment livestock and poultry individuals in the processed high-density point cloud data to distinguish the three-dimensional point cloud sets of different individuals. For each segmented livestock and poultry individual point cloud, a skeleton extraction algorithm (such as a skeleton extraction method based on curvature analysis or a topology shrinkage algorithm) is used to process it to obtain the three-dimensional skeleton model of the livestock and poultry individual.

[0085] By analyzing the time series of different skeletal points of individual livestock and poultry, we can calculate joint angles, movement speeds, and gait period kinematic parameters to identify the basic behavioral states of individual livestock and poultry (such as walking, resting, feeding, fighting, etc.). We can also perform statistical analysis on the individual behaviors in the livestock and poultry population to obtain individual behavior data. Finally, we can integrate the individual behaviors extracted from the skeleton with respiratory and metabolic monitoring data to form livestock and poultry monitoring data.

[0086] Methods for predicting transmission trends include:

[0087] Based on livestock and poultry monitoring data, a three-dimensional skeleton model is used to extract the temporal behavioral state, respiratory and metabolic characteristics, and environmental parameter characteristics of individual livestock and poultry, and the transmission rate is dynamically calculated. This transmission rate reflects the possibility of infection between livestock and poultry at different times and locations, and is adjusted in real time with changes in the farm environment (such as temperature, humidity, and ventilation) and changes in group behavior.

[0088] Based on the spatial location and movement trajectory of individual livestock and poultry, the spatial diffusion coefficient is used to represent the speed of disease spread in the farm. The spatial diffusion coefficient is dynamically updated according to the density and movement pattern of livestock and poultry populations. The state variables (susceptible individuals, latent individuals, infected individuals, and recovered individuals) of the improved SEIR model are constructed into spatiotemporal sequence data to reflect the dynamic process of disease transmission.

[0089] A deep learning network based on the Transformer structure is used to jointly model the temporal behavioral state and environmental parameter features to learn the disease transmission pattern and potential spatiotemporal dependencies. By training and optimizing the Transformer network, the predicted values ​​of disease infection risk at various time points and spatial locations in the future are obtained.

[0090] Methods for generating space risk heat maps include:

[0091] The predicted risk values ​​of disease infection at each future time point are correlated with the spatial locations of individual livestock and poultry to form a spatial risk distribution point set at each time point. The spatial risk distribution point set is then interpolated using spatial interpolation algorithms (such as inverse distance weighting, Kriging interpolation, or Gaussian process regression) to construct a continuous risk field within the farm. This process calculates the weighted average risk value of each location point in space to obtain the predicted infection risk value at any location point within the farm.

[0092] The reconstructed risk field is visualized as a two-dimensional heat map of the farm space. The steps include dividing the farm space into grid cells according to a preset resolution; using the center point of each grid cell as an interpolation node to calculate its infection risk value; and assigning different colors according to the level of infection risk value to form an intuitive spatial risk heat map.

[0093] Methods for simulating the spread of an epidemic in complex environments include:

[0094] Based on the spatial risk heat map, multi-band photoacoustic field and environmental physical disturbance are further used to simulate the transmission path and speed of livestock and poultry diseases in complex environments. A multi-band photoacoustic generator at point L is set up in the farm to simulate the sound field and photoacoustic disturbance in complex environments, generate photoacoustic waves covering different frequencies and energy densities, and simulate the impact of different environmental factors (such as temperature and humidity gradients, air flow, pressure fluctuations, etc.) on the carrying, deposition, diffusion and transmission path of virus particles in the air.

[0095] The multi-band photoacoustic generator adjusts the sound wave frequency, power, and emission direction to superimpose acoustic disturbances to form an acoustic potential well and a local high-pressure zone, simulating the blocking and acceleration effects of different spatial locations in a farm on the movement of pathogenic particles (viruses, bacteria, aerosols).

[0096] By combining the infection risk values ​​in the spatial risk heat map, acoustic field perturbation and photoacoustic modulation are superimposed on the regions corresponding to different infection risk values ​​to simulate the propagation dynamics of virus particles in complex environments (such as high-density populations, strongly ventilated areas, and temperature and humidity gradient areas).

[0097] Based on the actual layout and ventilation conditions of the farm, the local airflow field is adjusted by physical disturbances (such as forced ventilation simulation, pressure fluctuation simulation, and temperature and humidity gradient simulation) to restore the pathogen transmission path under complex conditions. Combined with the collection of local gas and aerosol concentrations by sensor array, the speed of epidemic transmission, spatial distribution and infection risk under complex conditions are dynamically simulated. The simulation results are combined with the spatial risk heat map to form a dynamically evolving spatiotemporal transmission trend map.

[0098] Methods for testing whether a pre-set prevention and control strategy is effective include:

[0099] An autonomous evolutionary game framework is constructed, in which individual livestock and poultry are regarded as game players. Each game player selects different behavioral strategies (including avoidance, gathering, migration, etc.) to maximize its own survival benefits based on its own spatiotemporal location, health status, surrounding population density and environmental conditions.

[0100] In this autonomous evolutionary game framework, a pre-defined game payoff function is introduced, which includes health payoff (avoiding infection), social payoff (maintaining normal behavior patterns), energy consumption (movement, feeding, etc.), and external constraints (such as zookeeper intervention and isolation measures). Pre-defined prevention and control strategies (such as local lockdown, isolation zone division, ventilation adjustment, health screening, vaccination, etc.) are introduced and used as external game intervention conditions to dynamically affect the payoff function and the available strategy space of the game players.

[0101] Evolutionary game theory algorithms (such as evolutionary strategy updates based on replication dynamics and genetic algorithms) are used to simulate the behavioral evolution of livestock and poultry groups under different prevention and control strategies. Through iterative calculations at time steps, the changing trends of group behavior patterns and transmission paths are observed; changes in group behavior and the spread of the epidemic are monitored during the game process.

[0102] If the game payoff after the introduction of the preset prevention and control strategy is greater than or equal to the preset game payoff threshold, and the epidemic spread speed is less than the preset epidemic spread speed threshold, then the preset prevention and control strategy is deemed effective; if the game payoff after the introduction of the preset prevention and control strategy is less than the preset game payoff threshold, and the epidemic spread speed is greater than or equal to the preset epidemic spread speed threshold, then the preset prevention and control game payoff threshold is deemed invalid.

[0103] Methods for generating prevention and control pathways include:

[0104] Based on the improved SEIR model and Transformer network prediction results, the disease transmission risk trend at various time points and spatial locations in the future is obtained, including the distribution of high-risk areas, potential transmission paths and transmission speed; the disease transmission risk trend is integrated with the spatiotemporal transmission trend map to dynamically correct the spatial risk heat map;

[0105] Based on the fused transmission trend data and the corrected spatial risk heat map, the boundary features, internal density distribution and transmission speed change trends of different risk areas are extracted to generate a prevention and control path planning model. Based on the spatiotemporal transmission trend map, a prevention and control path scheme including isolation zone layout, temporary blockade area, physical isolation barrier, guidance channel and optimized ventilation route is determined. The prevention and control path scheme is matched with the verified and effective preset prevention and control schemes (including local isolation, environmental adjustment, vaccination and individual intervention) to form a dynamic prevention and control path.

[0106] Methods for providing explainable prevention and control strategy decisions include:

[0107] By analyzing the internal feature vectors of the Transformer network, feature importance ranking algorithms (such as SHAP value and attention weight distribution) are used to quantify the contribution of different input features (including individual livestock and poultry behavior, respiratory and metabolic features, environmental parameters, population density, temperature and humidity, wind direction, etc.) to the prediction results of disease transmission, and to rank the features by importance.

[0108] Based on the ranking of feature importance, a prevention and control strategy decision tree is generated to provide interpretable prevention and control strategy decisions for livestock and poultry farms, and to define the logical basis and influencing factors behind each prevention and control path and strategy decision.

[0109] Methods for independently adjusting prevention and control strategies include:

[0110] Multi-line lidar sensors and ultra-wideband positioning devices are deployed in livestock and poultry farms to construct a three-dimensional spatial environment model and mobile positioning network. The multi-line lidar scans the environment to collect real-time information on the three-dimensional geometric structure and obstacles of the farm, forming a high-resolution point cloud map. The ultra-wideband positioning device uses UWB base stations and mobile tags deployed in the farm to perform three-dimensional positioning of disinfection robots or human operators.

[0111] Based on spatial risk heat maps, we analyze the different color representations and spread trends of each region to identify high-risk areas and key prevention and control nodes; and we design disinfection path planning algorithms using the spatial distribution information of spatial risk heat maps.

[0112] The disinfection path planning algorithm aims to optimize the shortest path, cover the largest risk area, and minimize repeated disinfection. It comprehensively considers environmental obstacles, the layout of breeding facilities, and the animal activity area to calculate the initial disinfection path.

[0113] During the disinfection process, the current location and path execution status of the disinfection equipment are fed back in real time through ultra-wideband positioning devices, and high-resolution point cloud maps are continuously updated through multi-line lidar to capture information on environmental changes and temporary obstacles. The initial disinfection path is dynamically adjusted according to the latest spatial risk heat map, prioritizing coverage of newly generated or risk-increasing areas. Environmental constraints (such as ventilation channel restrictions and personnel safety zone divisions) are introduced, and combined with the energy consumption and operation time limits of the disinfection equipment, the energy efficiency of the initial disinfection path is optimized, and the disinfection path is adjusted autonomously.

[0114] A preset game payoff threshold is set by collecting different game payoffs and taking the average of multiple game payoffs as the preset game payoff threshold. Similarly, a preset epidemic spread speed threshold is set.

[0115] This embodiment combines gas component detection with broadband terahertz spectral imaging using a multi-sensor array to achieve multi-dimensional and continuous monitoring of oral and nasal secretions and respiratory gases in livestock and poultry. This comprehensively reflects the health and metabolic status of individual animals and enhances early warning capabilities for disease transmission. By accurately identifying the absorption peaks of water in secretions and the broadband absorption bands of key molecules such as viral proteins and sugars, the recognition depth and spectral resolution of respiratory metabolic characteristics are greatly enhanced, enabling sensitive capture of subtle changes in the early stages of pathogen infection. By dynamically adjusting the absorption coefficient based on real-time detection of changes in the water content of secretions, spectral feature shifts and errors caused by fluctuations in environmental temperature and humidity, as well as changes in animal physiological states, can be effectively compensated for. This ensures the accuracy and reliability of data input into the kinetic model and avoids the risk of misjudgment due to environmental interference. This dynamic adjustment mechanism improves the stability and applicability of terahertz spectral detection in complex farming environments.

[0116] Based on precise individual respiratory and metabolic characteristics, combined with multi-dimensional environmental parameters, a disease transmission dynamics model has been constructed that can more accurately describe the transmission rate and spatial diffusion path of pathogens, providing a solid basis for the scientific formulation of intelligent prevention and control strategies. The model can dynamically update the health status and transmission risk of each individual livestock and poultry, reflecting the real-time development trend of the epidemic within the farm. Through intelligent algorithms, prevention and control resources are optimized and dynamically adjusted, effectively reducing the risk of disease transmission and improving the overall biosecurity level and disease control efficiency of the farm. Simultaneously, this technical solution supports non-invasive health monitoring of large-scale livestock and poultry populations, reducing manual intervention and labor intensity, promoting the intelligent and digital transformation of livestock production, and providing strong support for the sustainable development of animal husbandry.

[0117] Example 2

[0118] Please see Figure 2 As shown, for parts not described in detail in this embodiment, please refer to the description in Embodiment 1. A livestock and poultry disease transmission dynamics modeling and intelligent prevention and control decision-making system is provided, including:

[0119] The biosensing fusion module uses non-contact respiratory metabolism detection and terahertz spectral imaging to identify the respiratory metabolism characteristics of individual livestock and poultry, and adopts a three-dimensional point cloud skeleton extraction method to monitor the behavior of livestock and poultry groups and obtain livestock and poultry monitoring data.

[0120] The propagation dynamics modeling module, based on livestock and poultry monitoring data, constructs an improved SEIR model that includes spatiotemporal dynamic propagation rate and spatial diffusion coefficient, and introduces Transformer network to predict propagation trend and generate spatial risk heat map.

[0121] The physical field constructs a prevention and control module. Based on the spatial risk heat map, it uses multi-band optical and acoustic fields and environmental physical disturbances to simulate the spread of the epidemic in a complex environment. Based on the autonomous evolutionary game mechanism, it tests whether the preset prevention and control strategies are effective.

[0122] The constraint perception and control module generates control paths based on predicted propagation trends and effective preset control strategies, combined with dynamic field test feedback, and provides interpretable control strategy decisions based on feature importance ranking.

[0123] The adaptive disinfection module uses multi-line lidar and ultra-wideband positioning technology to plan disinfection paths with high precision. It autonomously adjusts the disinfection path based on the spatial risk heat map, and uses electrolyzed water spray for disinfection. The disinfection effect is then fed back to the prevention and control decision terminal in real time.

[0124] Since the electronic device described in this embodiment is the electronic device used to implement the livestock and poultry disease transmission dynamics modeling and intelligent prevention and control decision-making method described in this application embodiment, those skilled in the art can understand the specific implementation method and various variations of the electronic device in this embodiment based on the livestock and poultry disease transmission dynamics modeling and intelligent prevention and control decision-making method described in this application embodiment. Therefore, how the electronic device implements the method in this application embodiment will not be described in detail here. As long as those skilled in the art implement the livestock and poultry disease transmission dynamics modeling and intelligent prevention and control decision-making method described in this application embodiment, the electronic device used is within the scope of protection of this application.

[0125] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0126] The above description is merely a preferred embodiment of the present invention, and the scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for users of ordinary technical skills, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A method for modeling the dynamics of livestock and poultry disease transmission and for intelligent prevention and control decision-making, characterized in that, include: S1. Non-contact respiratory metabolism detection and terahertz spectral imaging are used to identify the respiratory metabolism characteristics of individual livestock and poultry. A three-dimensional point cloud skeleton extraction method is used to monitor the behavior of livestock and poultry groups and obtain livestock and poultry monitoring data. S2. Based on livestock and poultry monitoring data, an improved SEIR model is constructed that includes spatiotemporal dynamic propagation rate and spatial diffusion coefficient. A Transformer network is introduced to predict propagation trends and generate a spatial risk heat map. S3. Based on the spatial risk heat map, multi-band photoacoustic field and environmental physical disturbance are used to simulate the spread of the epidemic in a complex environment, and based on the autonomous evolutionary game mechanism, the effectiveness of the preset prevention and control strategy is tested. S4. Based on the predicted transmission trend and the effective preset prevention and control strategies, combined with the dynamic field test feedback, generate prevention and control paths, and provide interpretable prevention and control strategy decisions based on feature importance ranking. S5. High-precision disinfection path planning is carried out using multi-line lidar and ultra-wideband positioning technology. The disinfection path is autonomously adjusted based on the spatial risk heat map, and electrolytic water spray is used for disinfection. The disinfection effect is fed back to the prevention and control decision terminal in real time.

2. The method for modeling the transmission dynamics of livestock and poultry diseases and making intelligent prevention and control decisions according to claim 1, characterized in that, The method for identifying the respiratory metabolic characteristics of individual livestock and poultry includes: By deploying infrared and laser gas sensor arrays in key areas of the farm, non-contact data is collected on the gases exhaled naturally by livestock and poultry during their daily activities and their respiratory frequency, thus obtaining respiratory metabolic detection data. During the data collection process, the metabolic intensity of the individual animal is automatically calibrated based on a preset respiratory cycle signal, and the metabolic fluctuations of the individual animal are assessed using the rate of change in gas concentration. A broadband terahertz wave source and a high-speed reflector receiver array were deployed to scan the mouth and nose area of ​​individual livestock and poultry to obtain the absorption and reflection spectra of saliva and nasal secretions in the broadband terahertz wave band; the absorption and reflection spectra were analyzed by Fourier transform to convert the time information in the absorption and reflection spectra into spectral information. When a terahertz beam irradiates the mouth and nose area of ​​livestock and poultry, it interacts with secretions. Different components have different absorption peaks and bandwidths at different frequencies. By analyzing the compositional characteristics of the oral and nasal secretions of individual livestock and poultry under broadband terahertz spectrum using Lambert-Beer's law, the respiratory and metabolic characteristics of individual livestock and poultry can be obtained.

3. The method for modeling the transmission dynamics of livestock and poultry diseases and making intelligent prevention and control decisions according to claim 2, characterized in that, The method for generating the livestock and poultry monitoring data includes: By deploying m depth cameras at different locations in the livestock and poultry farming environment, real-time three-dimensional spatial data of livestock and poultry groups is collected to generate high-density point cloud data. The high-density point cloud data contains the spatial distribution, body shape characteristics, and dynamic movement information of individual livestock and poultry. The collected high-density point cloud data is then filtered, denoised, and sparsified. Spatial clustering algorithm is used to segment livestock and poultry individuals in the processed high-density point cloud data to distinguish the three-dimensional point cloud sets of different individuals; for each segmented livestock and poultry individual point cloud, skeleton extraction algorithm is used to process it to obtain the three-dimensional skeleton model of the livestock and poultry individual. By analyzing the time series of different skeletal points of individual livestock and poultry, we can calculate joint angles, movement speeds, and gait period kinematic parameters to identify the basic behavioral states of individual livestock and poultry. Statistical analysis of individual behaviors in livestock and poultry groups yields individual behavior data. The individual behaviors extracted from the skeletal structure are then integrated with respiratory and metabolic monitoring data to form livestock and poultry monitoring data.

4. The method for modeling the transmission dynamics of livestock and poultry diseases and making intelligent prevention and control decisions according to claim 3, characterized in that, The method for predicting propagation trends includes: Based on livestock and poultry monitoring data, a three-dimensional skeleton model is used to extract the temporal behavioral state, respiratory and metabolic characteristics and environmental parameters of individual livestock and poultry, and the transmission rate is dynamically calculated. This transmission rate reflects the possibility of infection between livestock and poultry at different times and locations, and is adjusted in real time with changes in farm environment and group behavior. Based on the spatial location and movement trajectory of individual livestock and poultry, the spatial diffusion coefficient is used to represent the speed of disease spread in the farm. The spatial diffusion coefficient is dynamically updated according to the density and movement pattern of livestock and poultry groups. The state variables of the improved SEIR model are constructed into spatiotemporal sequence data to reflect the dynamic process of disease spread. A deep learning network based on the Transformer structure is used to jointly model the temporal behavioral state and environmental parameter features to learn the disease transmission pattern and potential spatiotemporal dependencies. By training and optimizing the Transformer network, the predicted values ​​of disease infection risk at various time points and spatial locations in the future are obtained.

5. The method for modeling the transmission dynamics of livestock and poultry diseases and making intelligent prevention and control decisions according to claim 4, characterized in that, The method for generating the space risk heatmap includes: The predicted risk values ​​of disease infection at each future time point are correlated with the spatial location of individual livestock and poultry to form a spatial risk distribution point set at each time point; the spatial risk distribution point set is interpolated, and a spatial interpolation algorithm is used to construct a continuous risk field within the farm. The reconstructed risk field is visualized as a two-dimensional heat map of the farm space. The steps include dividing the farm space into grid cells according to a preset resolution; using the center point of each grid cell as an interpolation node to calculate its infection risk value; and assigning different colors according to the level of infection risk value to form an intuitive spatial risk heat map.

6. The method for modeling the transmission dynamics of livestock and poultry diseases and making intelligent prevention and control decisions according to claim 5, characterized in that, The methods for simulating the spread of an epidemic under complex environments include: Based on the spatial risk heat map, multi-band photoacoustic field and environmental physical disturbance are further used to simulate the transmission path and speed of livestock and poultry diseases in complex environments. A multi-band photoacoustic generator at point L is set up in the farm to simulate the sound field and photoacoustic disturbance in complex environments, generate photoacoustic waves covering different frequencies and energy densities, and simulate the impact of different environmental factors on the carrying, deposition, diffusion and transmission path of virus particles in the air. The multi-band photoacoustic generator adjusts the frequency, power and emission direction of the sound waves, superimposes acoustic disturbances to form an acoustic potential well and a local high-pressure zone, and simulates the obstruction and acceleration effects of different spatial locations in a farm on the movement of pathogen particles. By combining the infection risk values ​​in the spatial risk heat map, acoustic field perturbation and photoacoustic modulation are superimposed on the regions corresponding to different infection risk values ​​to simulate the propagation dynamics of virus particles in a complex environment. Based on the actual layout and ventilation conditions of the farm, the local airflow field is adjusted by physical disturbance to recreate the pathogen transmission path in a complex environment. Combined with the collection of local gas and aerosol concentrations by sensor array, the spread speed, spatial distribution and infection risk of the epidemic in a complex environment are dynamically simulated. The simulation results are combined with the spatial risk heat map to form a dynamically evolving spatiotemporal transmission trend map.

7. The method for modeling the transmission dynamics of livestock and poultry diseases and making intelligent prevention and control decisions according to claim 6, characterized in that, The method for testing whether the preset prevention and control strategy is effective includes: An autonomous evolutionary game framework is constructed, in which individual livestock and poultry are regarded as game players. Each game player selects different behavioral strategies to maximize its own survival benefits based on its own spatiotemporal location, health status, surrounding population density and environmental conditions. In this autonomous evolutionary game framework, a pre-defined game payoff function is introduced, which includes health payoff, social payoff, energy consumption, and external constraints. A pre-defined prevention and control strategy is introduced, which serves as an external game intervention condition to dynamically affect the payoff function and the available strategy space of the game players. Evolutionary game theory algorithms were used to simulate the behavioral evolution of livestock and poultry groups under different prevention and control strategies. Through iterative calculations at time steps, the changing trends of group behavior patterns and transmission paths were observed. The changes in group behavior and the spread of the epidemic were monitored during the game process. If the game payoff after the introduction of the preset prevention and control strategy is greater than or equal to the preset game payoff threshold, and the epidemic spread speed is less than the preset epidemic spread speed threshold, then the preset prevention and control strategy is deemed effective; if the game payoff after the introduction of the preset prevention and control strategy is less than the preset game payoff threshold, and the epidemic spread speed is greater than or equal to the preset epidemic spread speed threshold, then the preset prevention and control game payoff threshold is deemed invalid.

8. The method for modeling the transmission dynamics of livestock and poultry diseases and making intelligent prevention and control decisions according to claim 7, characterized in that, The method for generating prevention and control paths includes: Based on the improved SEIR model and Transformer network prediction results, the disease transmission risk trend at various time points and spatial locations in the future is obtained, including the distribution of high-risk areas, potential transmission paths and transmission speed; the disease transmission risk trend is integrated with the spatiotemporal transmission trend map to dynamically correct the spatial risk heat map; Based on the fused propagation trend data and the corrected spatial risk heat map, the boundary features, internal density distribution and propagation speed change trends of different risk areas are extracted to generate a prevention and control path planning model. Based on the spatiotemporal propagation trend map, a prevention and control path scheme including isolation zone layout, temporary blockade area, physical isolation barrier, guidance channel and optimized ventilation route is determined. The prevention and control path scheme is matched with the preset prevention and control scheme that has been verified and effective to form a dynamic prevention and control path.

9. The method for modeling the transmission dynamics of livestock and poultry diseases and making intelligent prevention and control decisions according to claim 8, characterized in that, The methods for providing interpretable prevention and control strategy decisions include: By analyzing the internal feature vectors of the Transformer network, a feature importance ranking algorithm is used to quantify the contribution of different input features to the prediction results of disease transmission and to rank the features by importance. Based on the ranking of feature importance, a prevention and control strategy decision tree is generated to provide interpretable prevention and control strategy decisions for livestock and poultry farms, and to define the logical basis and influencing factors behind each prevention and control path and strategy decision.

10. The method for modeling the transmission dynamics of livestock and poultry diseases and making intelligent prevention and control decisions according to claim 9, characterized in that, The methods for autonomously adjusting the prevention and control path include: Multi-line lidar sensors and ultra-wideband positioning devices are deployed in livestock and poultry farms to construct a three-dimensional spatial environment model and mobile positioning network. The multi-line lidar scans the environment to collect real-time information on the three-dimensional geometric structure and obstacles of the farm, forming a high-resolution point cloud map. The ultra-wideband positioning device uses UWB base stations and mobile tags deployed in the farm to perform three-dimensional positioning of disinfection robots or human operators. Based on spatial risk heat maps, we analyze the different color representations and spread trends of each region to identify high-risk areas and key prevention and control nodes; and we design disinfection path planning algorithms using the spatial distribution information of spatial risk heat maps. The disinfection path planning algorithm aims to optimize the shortest path, cover the largest risk area, and minimize repeated disinfection. It comprehensively considers environmental obstacles, the layout of breeding facilities, and the animal activity area to calculate the initial disinfection path. During the disinfection process, the current location and path execution status of the disinfection equipment are fed back in real time through ultra-wideband positioning devices, and high-resolution point cloud maps are continuously updated through multi-line lidar to capture information on environmental changes and temporary obstacles. The initial disinfection path is dynamically adjusted according to the latest spatial risk heat map, prioritizing coverage of newly generated or risk-increasing areas. Environmental constraints are introduced, and the energy efficiency of the initial disinfection path is optimized by combining the energy consumption and operation time limits of the disinfection equipment, and the disinfection path is adjusted autonomously.

Images