Livestock and poultry farm biological safety key hidden danger evaluation method and system based on risk offset factor

Abstract

The invention relates to the technical field of livestock and poultry breeding management and biological safety prevention and control, and discloses a livestock and poultry farm biological safety key hidden danger evaluation method and system based on risk offset factors, and the method comprises the following steps: collecting environment data, animal health data and management measure data of a livestock and poultry farm in real time through Internet of Things equipment and a sensor; the method comprises the steps of collecting multi-dimensional data, preprocessing the collected multi-dimensional data, training a risk assessment model by using a deep learning algorithm based on the processed multi-dimensional data, optimizing a vaccine vaccination rate, a disinfection frequency and an isolation measure by using a particle swarm optimization algorithm or a genetic algorithm, and performing risk assessment based on an optimized risk offset factor and an obtained risk assessment result. According to the invention, the method combines the Internet of Things, deep learning and multi-dimensional data processing, achieves the efficient assessment of the biological safety risk of the livestock and poultry farm, carries out the analysis through the real-time collection of environment, health and management data, and employs a CNN or LSTM model, achieves the precise recognition of potential hazards, and provides a scientific basis for decision making.

Description

Technical Field

[0001] The present invention relates to the field of livestock and poultry breeding management and biosafety prevention and control technology, and specifically to a method and system for evaluating key biosafety hazards in livestock and poultry farms based on risk offset factors. Background Art

[0002] In the current livestock and poultry management system, biosecurity, as a core component of ensuring animal health and preventing the spread of disease, is receiving increasing attention. With the continued expansion of large-scale farming, the biosecurity risks faced by livestock and poultry facilities are becoming increasingly complex, involving not only environmental and animal health conditions but also daily management practices. Although some farming companies have introduced information systems for basic data monitoring, actual risk assessment and hazard identification still rely primarily on manual judgment and static criteria, significantly lacking the ability to dynamically perceive and intelligently intervene in unexpected risks.

[0003] Existing technologies typically treat various risk factors as independent variables, ignoring the potential interactions between them. This results in a lack of systematicity and predictive power in assessment results. Furthermore, risk response measures often rely on manual formulation and triggering, lacking automated control mechanisms, making it impossible to achieve a truly integrated closed-loop "front-end perception - mid-end assessment - end-point response." There is also a lack of quantitative guidance on the trade-off between management costs and prevention and control effectiveness, leading to inefficient resource allocation and unbalanced management and control. Summary of the Invention

[0004] In response to the shortcomings of the existing technology, the present invention provides a method and system for evaluating key biosafety hazards in livestock and poultry farms based on risk offset factors, which solves the problems of untimely identification, inaccurate assessment, and unintelligent response of biosafety hazards in livestock and poultry farms.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: a method for evaluating key biosafety hazards in livestock and poultry farms based on risk offset factors, comprising the following steps:

[0006] S1. Real-time collection of livestock and poultry farm environmental data, animal health data, and management measures data through IoT devices and sensors;

[0007] S2. Preprocessing the collected multi-dimensional data;

[0008] S3. Based on the processed multi-dimensional data, use deep learning algorithms to train the risk assessment model;

[0009] S4. Use particle swarm optimization or genetic algorithm to optimize vaccination rates, disinfection frequency, and isolation measures;

[0010] S5. Based on the optimized risk offset factors and the obtained risk assessment results, a multi-factor interaction matrix model is constructed to consider the interaction between different risk factors;

[0011] S6. Based on the comprehensive risk assessment results, use reinforcement learning algorithms to optimize emergency response strategies and automatically trigger emergency response measures.

[0012] Preferably, the data collection step includes collecting environmental data, animal health data, and management measures data of livestock farms through Internet of Things devices and sensors. The environmental data includes temperature, humidity, and air quality; the animal health data includes the spread of epidemics and animal health conditions; and the management measures data includes staff operating habits and vaccination records.

[0013] Preferably, the data preprocessing step includes denoising the collected multidimensional data, the denoising method is to use a Kalman filter or a mean filter, the missing values are filled by interpolation, and the data normalization method is maximum-minimum normalization or Z-score normalization.

[0014] Preferably, the multidimensional risk assessment step includes training a risk assessment model using a deep learning algorithm, wherein the deep learning algorithm includes a convolutional neural network or a long short-term memory network, which is used to analyze the nonlinear impact of each factor on biosafety risk.

[0015] Preferably, the machine learning algorithm is trained using historical data and real-time data, and dynamically adjusts the weight of each risk factor, and the weight adjustment is optimized using a back-propagation algorithm to achieve adaptive risk assessment.

[0016] Preferably, the risk offset factor optimization step includes using a particle swarm optimization algorithm or a genetic algorithm to optimize factors such as vaccination rate, disinfection frequency, and isolation measures, and the optimization goal is to minimize the risk assessment results and balance the costs of various management measures.

[0017] Preferably, the optimization objective function of the optimization algorithm is:

[0018]

[0019] Where r={r1,r2,…,r m} is the risk offset factor; R t is the risk assessment result at time t; C(r) is the cost of each management measure; α and β are the adjustment weight coefficients; T is the optimization period.

[0020] Preferably, the optimization algorithm uses a particle swarm optimization algorithm or a genetic algorithm to optimize the risk offset factor, and the specific process includes:

[0021] Particle swarm optimization algorithm, the update formulas of particle velocity and position are:

[0022] v i (t+1)=w·v i (t)+c1·r1·(p i -r i (t))+c2·r2·(g i -r i (t));

[0023] r i (t+1)=r i (t)+v i (t+1);

[0024] Among them, v i (t) is the velocity of the i-th particle at time t; w is the inertia weight; c1 and c2 are learning factors; r1 and r2 are random numbers; p i is the personal optimal position of the i-th particle; g i is the global optimal position; r i (t+1) is the new position of the i-th particle at time t+1.

[0025] The genetic algorithm includes crossover operation, mutation operation and individual selection. The crossover operation is used to generate offspring individuals. The fitness function f(r) is used as the selection basis. The optimization goal is to minimize biosafety risks and costs.

[0026] Preferably, the multi-factor interaction matrix modeling step includes constructing an n×n interaction matrix A, wherein the matrix elements A ij Represents factor x i and x j The interaction strength between them is determined by correlation analysis or regression analysis.

[0027] The present invention also provides a livestock and poultry farm biosafety key hazard assessment system based on risk offset factors, comprising:

[0028] Data collection module, used to collect real-time environmental data, animal health data, management measures data, etc. of livestock and poultry farms through IoT devices and sensors;

[0029] Data preprocessing module, used to perform denoising, missing value filling, normalization and standardization on the collected data;

[0030] The risk assessment module is used to construct a multi-dimensional risk assessment model based on processed data using machine learning algorithms to dynamically assess the biosafety risks of livestock and poultry farms;

[0031] Risk optimization module, which is used to adjust risk offset factors through optimization algorithms to minimize risk assessment results and balance the costs of management measures;

[0032] The multi-factor interaction matrix module is used to construct a multi-factor interaction matrix model based on risk assessment results and optimized risk offset factors to improve the accuracy of risk assessment;

[0033] The emergency response module is used to optimize emergency response strategies based on real-time risk assessment results through reinforcement learning algorithms and automatically trigger emergency response measures.

[0034] The present invention provides a method and system for evaluating key biosafety hazards in livestock and poultry farms based on risk offset factors. This method has the following beneficial effects:

[0035] 1. This invention combines IoT devices, deep learning algorithms, and multi-dimensional data processing to achieve efficient assessment of biosafety risks on livestock and poultry farms. By collecting real-time environmental data, animal health data, and management measures, and using convolutional neural networks (CNNs) or long short-term memory (LSTM) deep learning models for nonlinear analysis, it can more accurately identify potential biosafety hazards, providing decision makers with scientific and comprehensive risk assessment results.

[0036] 2. This invention uses particle swarm optimization (PSO) and genetic algorithms (GA) to optimize key factors such as vaccination rates, disinfection frequency, and quarantine measures to minimize risk assessment results and balance the costs of various management measures. This optimization process not only improves livestock and poultry farm management efficiency, but also reduces the risks and losses caused by management errors, further enhancing biosafety.

[0037] 3. By combining machine learning with deep learning technologies, this invention dynamically adjusts the weights of risk factors based on historical and real-time data. This adaptive adjustment mechanism makes risk assessment more flexible and timely, enabling timely adjustments to risk assessment strategies based on changes in the internal and external farm environments, ensuring that assessment results remain consistent with actual conditions.

[0038] 4. This invention incorporates a reinforcement learning algorithm to optimize emergency response strategies, automatically triggering emergency response measures based on real-time risk assessment results. By building a state-action response network, it enables more accurate and timely responses to emergencies, reduces human intervention and decision-making errors, and significantly improves the speed and effectiveness of emergency response, thereby ensuring biosafety on livestock and poultry farms.

[0039] 5. By establishing a multi-factor interaction matrix model, this invention systematically considers the interrelationships between risk factors, enabling a more comprehensive assessment of the interplay of these factors. This multi-dimensional modeling approach avoids the singularity and limitations of traditional assessment methods, providing more scientific and accurate decision-making support for livestock and poultry farm management, significantly enhancing the scientific nature and foresight of biosafety management. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 Schematic diagram of the method flow of the present invention;

[0041] Figure 2 Schematic diagram of the system structure of the present invention. DETAILED DESCRIPTION

[0042] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the present specification. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0043] Please see the attached Figure 1 The embodiment of the present invention provides a method for evaluating key biosafety hazards in livestock and poultry farms based on risk offset factors, comprising the following steps:

[0044] S1. Real-time collection of livestock and poultry farm environmental data, animal health data, and management measures data through IoT devices and sensors

[0045] In this example, data collection is achieved through Internet of Things (IoT) devices and sensors deployed in different areas of the livestock farm to monitor and record real-time data within the farm. Specifically, the data collection module includes submodules for environmental monitoring, animal health monitoring, and management measures monitoring. Each submodule collects data based on specific parameters, ensuring a comprehensive picture of the farm's biosafety risk factors.

[0046] Environmental data is primarily collected through devices such as temperature and humidity sensors and air quality monitors. These devices provide real-time monitoring of climatic conditions within livestock farms, such as temperature, humidity, and CO2 concentration. These environmental factors significantly impact livestock health and, in turn, biosafety risks. Therefore, real-time monitoring of this environmental data enables timely identification of factors contributing to disease transmission or environmental contamination.

[0047] The collection of animal health data relies on various sensor devices, such as temperature sensors and heart rate monitors. These devices can be used to monitor the health of each animal in real time, particularly changes in vital signs and possible abnormalities. Animal health monitoring can help identify early signs of disease, especially infectious diseases, allowing for timely isolation and treatment measures, thereby reducing the risk of disease spread.

[0048] The collection of management measures data involves monitoring staff behavior, such as disinfection frequency, vaccination records, and quarantine procedures. This data is typically collected using RFID technology, QR code scanning, and other methods to automatically record staff behavior and the implementation of each management measure. For example, sensors can automatically record disinfectant usage and vaccine administration dates and dosages. This data provides an effective basis for risk assessment and subsequent optimization of management measures.

[0049] All data collection devices communicate with the data processing platform in real time via wireless networks, ensuring that collected data can be quickly uploaded to the cloud or stored in a local database. During this process, all sensors and monitoring equipment must have efficient data transmission capabilities to ensure real-time and reliable data.

[0050] To ensure data accuracy and integrity, this embodiment employs a data verification and error correction mechanism. Specifically, each data point undergoes preliminary verification after collection to ensure there are no transmission errors or data loss. If any data anomalies are detected, the system automatically triggers an alarm and initiates corrections or re-collection. This mechanism ensures that the data ultimately used for biosafety risk assessment is accurate and complete.

[0051] During the data collection process, environmental, animal health, and management practice data are collected and time-stamped simultaneously, ensuring that all data can be accurately compared over time. This provides the basis for subsequent real-time risk assessment and historical data analysis.

[0052] In terms of data acquisition equipment selection, this embodiment prioritizes the use of highly sensitive, long-life sensors that offer long-term stable operation and strong anti-interference capabilities, ensuring stable and reliable data acquisition. Furthermore, the equipment's deployment locations were carefully selected to ensure coverage of key areas of the farm and minimize data blind spots.

[0053] This embodiment also preferably utilizes cloud computing and big data analysis technologies for data storage and processing. After data collection, it is uploaded to a cloud platform via a network connection, where it is further processed, analyzed, and stored. The cloud platform can flexibly scale its computing power based on the volume of data, ensuring efficient and stable processing. Furthermore, the cloud platform enables remote data access, facilitating remote monitoring of livestock and poultry farm biosafety.

[0054] Through the aforementioned data collection steps, the present invention can comprehensively and accurately collect various types of data within livestock and poultry farms, providing a solid data foundation for subsequent risk assessment, optimized decision-making, and intelligent emergency response. The real-time, accuracy, and completeness of the data collection process are key to subsequent steps (such as risk assessment, optimization, and emergency response). The efficient operation of the entire system depends on smooth and unimpeded data collection at all stages, ensuring that the system can respond promptly and take necessary safety measures.

[0055] Summarize:

[0056] In step S1 of this embodiment, by deploying advanced IoT devices and sensors, comprehensive monitoring of the farm's environment, animal health, and management practices is achieved. The efficient data acquisition module provides real-time, accurate, and comprehensive data support for subsequent risk assessment, optimization, and emergency response, forming a fundamental component of the entire farm biosafety management system. Through real-time data upload, storage, and processing, the system can continuously monitor the farm's biosafety status, providing timely safety warnings and enabling rapid response to decision-makers.

[0057] S2. Preprocessing the collected multi-dimensional data

[0058] In this embodiment, data preprocessing first includes a denoising process. Since the data collected by sensors and equipment contains noise caused by the external environment or equipment failure, it is necessary to remove this invalid information to ensure the accuracy of the data. To achieve this, a Kalman filter or a mean filter is used for denoising in this embodiment. The Kalman filter is a recursive filtering algorithm that can effectively extract valid signals from noisy signals and is particularly suitable for real-time data processing in dynamic systems. For static data, a mean filter can also be used to smooth the data and remove outliers caused by accidental factors. This denoising process can significantly improve data quality and provide reliable input for subsequent analysis and modeling.

[0059] In this embodiment, data preprocessing also includes missing value filling. In the actual data collection process, some data may be missing due to equipment failure, transmission error or external environmental interference. In order to avoid the impact of missing data on subsequent model analysis, this embodiment uses interpolation to fill missing values. Specifically, the linear interpolation method is preferably used to fill time series data. By using linear interpolation, the value of the missing data can be estimated based on the trend of the existing data points to ensure the continuity and consistency of the data. If there are many missing data, use a more complex interpolation method (such as spline interpolation, etc.) to fill in the missing data to restore the authenticity of the missing part to the greatest extent.

[0060] Next, data normalization and standardization are another key step in this embodiment. The data collection process involves a variety of different data sources with different scales and units. For example, the dimensions of temperature and humidity data differ from those of animal health data, and such differences can affect subsequent analysis and modeling. To eliminate this effect, this embodiment uses maximum-minimum normalization and Z-score standardization methods for unified data processing.

[0061] Maximum and minimum normalization: This method scales all data values to between 0 and 1 to make the dimensions of each feature consistent, thus avoiding the deviation caused by data scale differences. For any feature x, the normalization formula is:

[0062]

[0063] Where x′ is the normalized value; min(x) and max(x) are the minimum and maximum values of the feature, respectively.

[0064] Z-score standardization: This method converts the data into a form with zero mean and unit variance by subtracting the mean of each data point and dividing it by the standard deviation, so that data with different characteristics can be compared under the same standard. The Z-score standardization formula is:

[0065]

[0066] Where x is the original data value, μ is the mean of the data, and σ is the standard deviation of the data. This method is suitable for cases where the data is unevenly distributed and can eliminate differences in data scale and distribution.

[0067] In this embodiment, data preprocessing also takes into account the handling of outliers and outliers. During the processing, some data points may deviate significantly from the normal range. These data points are caused by sensor failure or human error. To address this issue, this embodiment uses box plot analysis to identify and handle outliers. By calculating the upper and lower quartiles of the data and determining outliers based on the interquartile range (IQR), these irregular data can be effectively eliminated, thereby ensuring data quality.

[0068] The ultimate goal of data preprocessing is to ensure the consistency, accuracy, and comparability of all input data. In this example, all data preprocessing operations are performed after data acquisition to provide high-quality data input for subsequent risk assessment, model training, and optimization steps. This preprocessed data will be used to construct risk assessment models, ultimately providing a scientific basis for biosafety risk management on livestock and poultry farms.

[0069] Through the above preprocessing process, this example effectively eliminates noise and anomalies in the data, fills in missing values, and converts data from different sources into a unified format, laying a solid foundation for subsequent analysis and modeling. This data processing method can ensure data quality during the livestock and poultry farm biosafety risk assessment process, thereby improving the accuracy and reliability of the overall biosafety management system.

[0070] Summarize:

[0071] In this embodiment, step S2 uses a series of data preprocessing methods (including denoising, missing value filling, data normalization and standardization, and outlier processing) to ensure that the raw data collected from the livestock and poultry farms can be effectively converted into high-quality input data, providing a solid foundation for subsequent risk assessment and optimization. By employing Kalman filtering, interpolation, and normalization techniques, this embodiment effectively addresses data inconsistencies, enabling the data to be used to construct an efficient and accurate biosafety risk assessment model.

[0072] S3. Based on the processed multi-dimensional data, use deep learning algorithms to train risk assessment models

[0073] In this embodiment, step S3 employs a machine learning algorithm to perform a multi-dimensional risk assessment. This primarily utilizes deep learning methods to process and analyze pre-processed data, ensuring a comprehensive assessment of the biosafety risk of the livestock and poultry farm. This process first requires the data acquired from step S2, including but not limited to environmental data, animal health data, and management measures data. Based on this data, a risk assessment model is established to identify potential biosafety hazards and dynamically assess the current biosafety risk level of the livestock and poultry farm.

[0074] During implementation, the model is primarily based on deep learning algorithms, preferably using convolutional neural networks (CNNs) or long short-term memory networks (LSTMs). These two algorithms are capable of effectively processing and analyzing data with highly nonlinear characteristics. Convolutional neural networks (CNNs) excel in image processing, but they also demonstrate powerful feature extraction capabilities when processing high-dimensional environmental and animal health data. Long short-term memory networks (LSTMs), on the other hand, are well-suited for processing time series data and are therefore suitable for analyzing the impact of changes in animal health and environmental data over time, thereby assessing biosecurity risks in livestock and poultry farms.

[0075] In this example, the deep learning algorithm takes as input the high-quality data obtained through denoising, missing value filling, and normalization in step S2, and outputs a comprehensive biosafety risk assessment. This result is a risk score for each time point, and the weights of risk factors can be adjusted to reflect the contribution of different factors to the overall risk.

[0076] Specifically, the deep learning model used in this embodiment has multiple hidden layers, each of which is used to extract different features from the data. The input layer receives preprocessed data, including various risk factors such as temperature and humidity, and animal health status. Through multiple convolutional layers or LSTM units, the model gradually extracts high-level features from the data and generates risk assessment results in the final output layer. The model is trained using historical data and automatically adjusts the weights of each risk factor, so that when new data is input, it can predict the current biosafety risk of livestock and poultry farms based on real-time data.

[0077] During the deep learning model training process, this embodiment uses a backpropagation algorithm to optimize weights and a gradient descent method to minimize the loss function. The loss function is calculated by comparing the error between the predicted risk assessment results and the actual results. Through training, the model is gradually optimized and can provide more accurate risk assessments based on new data.

[0078] The risk assessment results in this example not only reflect the current biosafety risk of a livestock farm but also dynamically adjust the weights of different risk factors based on changes, providing real-time risk assessment results. For example, if certain environmental factors or animal health conditions change, the model will be able to adjust the weights in a timely manner to reflect the increase or decrease in the current biosafety risk.

[0079] To further improve the accuracy of the model, this embodiment also introduces cross-validation techniques, using a validation set to evaluate the model's generalization ability. Cross-validation ensures that the model does not overfit, while improving its accuracy and robustness in practical applications. The validation process divides the dataset into multiple subsets, tests each subset in turn, and determines the model's performance by calculating the average error.

[0080] The multidimensional risk assessment model in this example comprehensively considers the impact of multiple factors, including the environment, animal health, and management practices, resulting in a more accurate biosafety risk assessment. By combining deep learning with time series analysis, the model effectively captures nonlinear relationships and time-varying characteristics in the data, providing highly accurate risk assessments and a scientific basis for subsequent risk management and decision-making.

[0081] Summarize:

[0082] Step S3 of this embodiment establishes a multidimensional risk assessment model through a deep learning algorithm, which can comprehensively assess livestock and poultry farm biosafety risks and provide accurate real-time risk predictions. By using deep learning methods such as convolutional neural networks (CNNs) or long short-term memory networks (LSTMs), this embodiment can process complex multidimensional data and adjust the weights of risk factors in real time, thereby improving the accuracy and effectiveness of risk assessments. The implementation of this step enables the present invention to provide more accurate and timely decision-making support for livestock and poultry farm biosafety management.

[0083] S4. Use particle swarm optimization or genetic algorithm to optimize vaccination rates, disinfection frequency, and isolation measures

[0084] In this embodiment, the core objective of step S4 is to adjust risk offset factors through an optimization algorithm to balance the costs of management measures while ensuring biosafety. Specifically, risk offset factors include factors such as vaccination rates, disinfection frequency, and quarantine measures, which directly impact the level of biosafety risk. By using an optimization algorithm, this embodiment can identify the optimal combination of risk offset factors based on existing risk assessment results, thereby minimizing biosafety risks on livestock and poultry farms.

[0085] In this embodiment, the form of the optimization objective function is:

[0086]

[0087] in, is the vector of risk offset factors, representing management measures such as vaccination rate, disinfection frequency, and isolation measures; R t is the biosafety risk assessment result at time t, obtained based on the risk assessment model used in step S3; is the cost function of each management measure, which represents the cost required to implement measures such as vaccination, disinfection and isolation; α and β are adjustment weight coefficients, which determine the importance of risk assessment results and cost function in the objective function; T is the optimization period, which is usually a longer period of time and is used to evaluate the long-term effects of management measures.

[0088] By minimizing this objective function, the optimization algorithm adjusts the values of the various risk-offsetting factors to effectively manage biosecurity risks on the farm while also minimizing the cost of associated management measures. Thus, the optimization process not only focuses on reducing risk but also achieves optimal biosecurity outcomes within an acceptable cost range.

[0089] In this embodiment, the optimization algorithm preferably uses a particle swarm optimization (PSO) algorithm or a genetic algorithm (GA). These two optimization algorithms have strong global search capabilities and can find the optimal solution in a complex objective function space. Specifically, the particle swarm optimization algorithm searches by simulating the movement of each individual in the group, and all individuals jointly search for the optimal solution through mutual influence. The genetic algorithm, by simulating the process of natural selection, uses operations such as crossover, mutation, and selection to continuously generate a new generation of solutions in the solution space, thereby gradually approaching the optimal solution.

[0090] In the application of the particle swarm optimization algorithm, assuming that each particle represents a possible risk offset factor combination, the particle speed and position update rules are as follows:

[0091] v i (t+1)=w·v i (t)+c1·r1·(p i -r i (t))+c2·r2·(g i -r i (t));

[0092] r i (t+1)=r i (t)+v i (t+1);

[0093] Among them, v i (t) is the velocity of the i-th particle at time t; r i (t) is the position of the i-th particle at time t, that is, the current risk offset factor combination; w is the inertia weight; c1 and c2 are learning factors; r1 and r2 are random numbers; p i is the individual optimal position of the particle; g i is the global optimal position; r i (t+1) is the new position of the i-th particle at time t+1. Through continuous iterative updates, the particle swarm can eventually find the optimal risk offset factor combination.

[0094] In genetic algorithms, the optimization process includes steps such as selection, crossover, and mutation. The selection process selects the best performing individuals for crossover by evaluating the fitness function, thereby generating a new generation of individuals. The fitness function in this case is the optimization objective function of the present invention. The crossover operation generates offspring individuals by combining the characteristics of parent individuals, while the mutation operation increases the diversity of the search space by randomly changing certain genes of individuals (i.e., certain values of the risk offset factor), thereby preventing the algorithm from falling into a local optimal solution.

[0095] In this example, using a particle swarm optimization algorithm or a genetic algorithm, we can fully consider the interactions between different risk factors and find the optimal combination of risk-offsetting factors. In livestock and poultry farm biosafety management, this optimization process allows for the scientific deployment of various prevention and control measures, thereby improving the overall biosafety level of the farm.

[0096] In addition, the results of the optimization algorithm will automatically adjust the value of the risk offset factor according to the calculation results of the objective function. For example, when the biosafety risk assessment result R t As increases, the optimization algorithm might adjust the vaccination rate r1 or increase the disinfection frequency r2 to reduce the risk and keep the management cost within an acceptable range.

[0097] Summarize:

[0098] Step S4 of this embodiment optimizes the risk offset factor for livestock and poultry farms using a particle swarm optimization algorithm or a genetic algorithm. This minimizes biosafety risks while also rationally controlling the costs of management measures. The optimization objective function combines risk assessment results with a balance between management costs to ensure optimal biosafety management for livestock and poultry farms. Through these optimization algorithms, the present invention enables flexible adjustment of various management measures to address the ever-changing risk environment, providing scientific and accurate decision-making support for livestock and poultry farm biosafety management.

[0099] S5. Based on the optimized risk offset factors and the obtained risk assessment results, a multi-factor interaction matrix model is constructed to consider the interaction between different risk factors.

[0100] In this embodiment, the goal of step S5 is to further improve the accuracy of the assessment of livestock farm biosafety risks through a multi-factor interaction matrix model. In step S3, a machine learning algorithm was used to conduct a preliminary assessment of the biosafety risks of livestock farms. However, these assessment results did not consider the interactions between individual risk factors. In reality, however, risk factors on livestock farms are often interrelated and mutually influential. Therefore, by introducing a multi-factor interaction matrix model, this embodiment can consider the interactions between different risk factors and improve the accuracy of the overall risk assessment results.

[0101] In this embodiment, the multi-factor interaction matrix model is implemented by quantitatively modeling the interaction relationship between risk factors. Specifically, the core of the model is an interaction matrix A, whose elements A ij Represents factor xi and x j The interaction strength between the two risk factors reflects how the two risk factors jointly affect the biosafety risk of livestock and poultry farms. The elements in the interaction matrix are obtained by analyzing historical data and real-time data using methods such as correlation analysis and regression analysis.

[0102] Specifically, assuming there are n risk factors in a livestock farm, the interaction matrix A is an n×n matrix, where each element A ij Represents factor x i and x j The interaction matrix is generated by performing regression or correlation analysis on a large amount of historical data to reveal the influence relationships between various factors. For example, temperature, humidity, and animal density may have a synergistic effect on disease transmission, while disinfection frequency and animal health may be negatively correlated. In this way, the model can accurately reflect the complex interactions between multiple factors.

[0103] In this embodiment, the comprehensive risk assessment result R after the interaction matrix is calculated interact It can be obtained by the following formula:

[0104]

[0105] Among them, R interact A is the comprehensive risk assessment result after comprehensive consideration of the interaction between factors; ij is the element in the interaction matrix, representing the factor x i and x j The interaction strength between i and x j are the values of the i-th and j-th factors, respectively. This formula multiplies the interaction strength between each two factors by the corresponding factor value and sums the impact of all factor pairs to obtain the final comprehensive risk assessment result.

[0106] The interaction matrix model in this example effectively captures the nonlinear relationships and complex interactions between multiple risk factors. For example, on a livestock farm, environmental factors (such as temperature and humidity) and animal health factors (such as animal immunity) may jointly influence the spread of disease, while management measures (such as vaccination rates) may also interact with other factors under different environmental conditions. The introduction of these interactive effects enables the model to more accurately assess the biosafety risks of livestock and poultry farms.

[0107] By introducing a multi-factor interaction matrix model, this embodiment can further consider the synergistic effects and mutual influences between factors based on the risk offset factors optimized in step S4, thereby improving the accuracy of the comprehensive risk assessment. This model can play an important role in livestock and poultry farm management, helping managers to more comprehensively understand potential risks and thus provide a more accurate basis for subsequent decision-making.

[0108] The interaction matrix in this embodiment is not limited to linear relationships; in some cases, the interactions can be nonlinear. To this end, this embodiment can further optimize the generation of the interaction matrix through nonlinear regression models or deep learning models, enhancing the model's adaptability to complex interactions. This flexibility enables the model to adapt to the management needs and environmental conditions of different livestock and poultry farms, thereby providing accurate risk assessments in various application scenarios.

[0109] Furthermore, this embodiment includes a mechanism for dynamically adjusting the interaction matrix. Because the farm's environment and management practices may change over time, the interactions within the interaction matrix may also change accordingly. Regularly recalculating and optimizing the interaction matrix ensures that the model promptly reflects changes in actual management and environmental conditions, thereby maintaining the accuracy and timeliness of biosafety risk assessment results.

[0110] Summarize:

[0111] Step S5 of this embodiment comprehensively considers the interactions between multiple risk factors by introducing a multi-factor interaction matrix model. This process effectively improves the accuracy of biosafety risk assessment, enabling the model to comprehensively consider the synergistic effects and nonlinear relationships of various factors, providing more accurate risk assessment results. By dynamically updating and optimizing the interaction matrix, the present invention can provide scientific decision-making support for livestock and poultry farm biosafety management, helping managers adjust strategies and measures in real time, thereby improving overall biosafety levels.

[0112] S6. Based on the comprehensive risk assessment results, use reinforcement learning algorithms to optimize emergency response strategies and automatically trigger emergency response measures.

[0113] In this embodiment, the core objective of step S6 is to automatically trigger and execute emergency response measures through an intelligent emergency response system when the biosafety risk on a livestock and poultry farm reaches a certain threshold. The primary task of this process is to monitor the current biosafety risk in real time based on the multi-factor interaction matrix obtained in step S5 and the risk offset factors optimized in step S4. When the risk level reaches a critical value, the corresponding emergency response strategy is automatically adjusted and initiated. Through this intelligent emergency response, this embodiment can respond to emergencies quickly and efficiently, minimizing biosafety risks.

[0114] In this embodiment, the optimization strategy for intelligent emergency response relies on a reinforcement learning algorithm. Reinforcement learning is an algorithm that optimizes decisions based on behavior and feedback. During this step, the system continuously interacts with the environment to learn the optimal emergency response strategy. Specifically, reinforcement learning uses feedback from real-time risk assessment results to enable the system to automatically adjust emergency response strategies based on current biosafety risks and make optimal decisions. By simulating a variety of possible emergency response scenarios, the system continuously optimizes and adjusts response strategies, ensuring that in actual operations, the most appropriate decisions are made based on the current risk situation.

[0115] In the reinforcement learning process, the goal of the system is to maximize the expected value of the biosafety assessment results. This goal can be expressed by the following formula:

[0116]

[0117] Among them, S t represents the emergency response strategy at time t; ε[R t |S] is based on the current biosafety risk assessment result R under strategy S. t In this way, the system can evaluate the effectiveness of different emergency response strategies in reducing risks and automatically select the most appropriate emergency response measures.

[0118] In the design of emergency response strategies, this embodiment generates emergency response decisions by simulating different scenarios. These simulation scenarios include but are not limited to:

[0119] Initiate animal quarantine measures;

[0120] Increase disinfection frequency;

[0121] Adjusting vaccination schedules;

[0122] Improve safety protection standards for staff, etc.

[0123] In different risk scenarios, the system dynamically adjusts its strategies by learning from historical data and real-time feedback. For example, in high-risk situations, the system might prioritize strengthening quarantine measures and increasing vaccination efforts, while in medium-risk situations, it might adjust disinfection frequency or optimize vaccine distribution.

[0124] The reinforcement learning algorithms used in this embodiment include Q-learning or Deep Q-Networks (DQNs). These algorithms evaluate the effectiveness of each policy by calculating a state-action value function (Q-value) and update the optimal policy through trial and error. In Q-learning, the system selects an action based on the current state and adjusts the weight of the action based on feedback. In Deep Q-Networks, a neural network model is used to approximate the Q-value, further improving the efficiency and accuracy of learning.

[0125] Furthermore, the system optimizes emergency response decisions by combining real-time data streams with historical data. When the actual situation on a livestock farm changes, the system can quickly extract new information from the data stream, recalculate the current risk assessment, and optimize the emergency response strategy based on the new risk data. This data-driven dynamic adjustment mechanism enables emergency response to not only respond quickly to emergencies but also continuously optimize itself as environmental and management conditions change.

[0126] The intelligent emergency response system in this embodiment also features automatic monitoring and early warning capabilities. It continuously tracks various biosafety data from livestock and poultry farms. If a biosafety risk approaches or exceeds a preset threshold, the system automatically activates an emergency response mechanism, ensuring the farm can implement appropriate prevention and control measures in the shortest possible time. Furthermore, the system sends early warning messages to management personnel, prompting them to take further control measures.

[0127] Summarize:

[0128] Step S6 of this embodiment uses a reinforcement learning algorithm to intelligently optimize emergency response strategies. Based on risk assessment and optimized management measures, the system automatically adjusts and executes emergency response strategies. By maximizing the expected value of biosafety assessment results, it ensures that livestock farms can take the most appropriate measures when faced with sudden biosafety risks. Through real-time monitoring, the integration of data streams, and historical feedback, the system can self-optimize emergency response strategies, providing efficient and flexible biosafety management support for livestock farms.

[0129] Please see the attached Figure 2 The present invention also provides a livestock and poultry farm biosafety key hazard evaluation system based on risk offset factors, comprising:

[0130] Data collection module, used to collect real-time environmental data, animal health data and management measures data of livestock and poultry farms through IoT devices and sensors;

[0131] Data preprocessing module, used to perform denoising, missing value filling, normalization and standardization on the collected data;

[0132] The risk assessment module is used to construct a multi-dimensional risk assessment model based on processed data using machine learning algorithms to dynamically assess the biosafety risks of livestock and poultry farms;

[0133] Risk optimization module, which is used to adjust risk offset factors through optimization algorithms to minimize risk assessment results and balance the costs of management measures;

[0134] The multi-factor interaction matrix module is used to construct a multi-factor interaction matrix model based on risk assessment results and optimized risk offset factors to improve the accuracy of risk assessment;

[0135] The emergency response module is used to optimize emergency response strategies based on real-time risk assessment results through reinforcement learning algorithms and automatically trigger emergency response measures.

[0136] The system of this embodiment can be used to execute the above method embodiments, and its principles and technical effects are similar, so they will not be repeated here.

[0137] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A method for evaluating key biosafety hazards in livestock and poultry farms based on risk offset factors, characterized by: The following steps are involved: S1. Real-time collection of livestock and poultry farm environmental data, animal health data, and management measures data through IoT devices and sensors; S2. Preprocessing the collected multi-dimensional data; S3. Based on the processed multi-dimensional data, use deep learning algorithms to train the risk assessment model; S4. Use particle swarm optimization or genetic algorithm to optimize vaccination rates, disinfection frequency, and isolation measures; S5. Based on the optimized risk offset factors and the obtained risk assessment results, a multi-factor interaction matrix model is constructed to consider the interaction between different risk factors; S6. Based on the comprehensive risk assessment results, use reinforcement learning algorithms to optimize emergency response strategies and automatically trigger emergency response measures.

2. The method for evaluating key biosafety hazards in livestock and poultry farms based on risk offset factors according to claim 1, characterized in that: The data collection step includes collecting environmental data, animal health data, and management measures data of livestock and poultry farms through Internet of Things devices and sensors. The environmental data includes temperature, humidity, and air quality. The animal health data includes the spread of epidemics and animal health conditions. The management measures data includes staff operating habits and vaccination records.

3. The method for evaluating key biosafety hazards in livestock and poultry farms based on risk offset factors according to claim 1, characterized in that: The data preprocessing step includes denoising the collected multidimensional data using a Kalman filter or a mean filter, filling missing values using an interpolation method, and data normalization using a maximum-minimum normalization or Z-score normalization method.

4. The method for evaluating key biosafety hazards in livestock and poultry farms based on risk offset factors according to claim 1, characterized in that: The multidimensional risk assessment step includes using a deep learning algorithm to train a risk assessment model, wherein the deep learning algorithm includes a convolutional neural network or a long short-term memory network, which is used to analyze the nonlinear impact of each factor on biosafety risks.

5. The method for evaluating key biosafety hazards in livestock and poultry farms based on risk offset factors according to claim 4, characterized in that: The machine learning algorithm is trained using historical data and real-time data, and dynamically adjusts the weights of each risk factor. The weight adjustment is optimized using a back-propagation algorithm to achieve adaptive risk assessment.

6. The method for evaluating key biosafety hazards in livestock and poultry farms based on risk offset factors according to claim 1, characterized in that: The risk offset factor optimization step includes using a particle swarm optimization algorithm or a genetic algorithm to optimize factors such as vaccination rates, disinfection frequency, and isolation measures. The optimization goal is to minimize the risk assessment results and balance the costs of various management measures.

7. The method for evaluating key biosafety hazards in livestock and poultry farms based on risk offset factors according to claim 6, characterized in that: The optimization objective function of the optimization algorithm is: Where r={r1,r2,…,r m } is the risk offset factor; R t is the risk assessment result at time t; C(r) is the cost of each management measure; α and β are the adjustment weight coefficients; T is the optimization period.

8. The method for evaluating key biosafety hazards in livestock and poultry farms based on risk offset factors according to claim 7, characterized in that: The optimization algorithm uses a particle swarm optimization algorithm or a genetic algorithm to optimize the risk offset factor. The specific process includes: Particle swarm optimization algorithm, the update formulas of particle velocity and position are: v i (t+1)=w·v i (t)+c1·r1·(p i -r i (t))+c2·r2·(g i -r i (t)); r i (t+1)=r i (t)+v i (t+10; Among them, v i (t) is the velocity of the i-th particle at time t; w is the inertia weight; c1 and c2 are learning factors; r1 and r2 are random numbers; p i is the personal optimal position of the i-th particle; g i is the global optimal position; r i (t+1) is the new position of the i-th particle at time t+1. The genetic algorithm includes crossover operation, mutation operation and individual selection. The crossover operation is used to generate offspring individuals. The fitness function f(r) is used as the selection basis. The optimization goal is to minimize biosafety risks and costs.

9. The method for evaluating key biosafety hazards in livestock and poultry farms based on risk offset factors according to claim 1, characterized in that: The multi-factor interaction matrix modeling step includes constructing an n×n interaction matrix A, where the matrix elements A ij Represents factor x i and x j The interaction strength between them is determined by correlation analysis or regression analysis.

10. A livestock and poultry farm biosafety key hazard assessment system based on risk offset factors, used to implement the method according to any one of claims 1 to 9, characterized in that: include: Data collection module, used to collect real-time environmental data, animal health data and management measures data of livestock and poultry farms through IoT devices and sensors; Data preprocessing module, used to perform denoising, missing value filling, normalization and standardization on the collected data; The risk assessment module is used to construct a multi-dimensional risk assessment model based on processed data using machine learning algorithms to dynamically assess the biosafety risks of livestock and poultry farms; Risk optimization module, which is used to adjust risk offset factors through optimization algorithms to minimize risk assessment results and balance the costs of management measures; The multi-factor interaction matrix module is used to construct a multi-factor interaction matrix model based on risk assessment results and optimized risk offset factors to improve the accuracy of risk assessment; The emergency response module is used to optimize emergency response strategies based on real-time risk assessment results through reinforcement learning algorithms and automatically trigger emergency response measures.

Images