Efficient breeding method for new beef cattle breed

Abstract

The application relates to the beef cattle breeding technical field and discloses an efficient beef cattle new variety breeding method, which comprises the following steps: S1, collecting phenotype data, genotype data and environment data of beef cattle, wherein the phenotype data comprises the body weight, body shape, exercise amount and health state of the beef cattle, the genotype data is collected through a gene chip or whole genome sequencing, and the environment data is monitored through a sensor to monitor environmental factors such as temperature and humidity and gas concentration in a cowshed; S2, pre-processing the collected phenotype data, genotype data and environment data, including data cleaning, missing value filling, standardization and feature extraction. The beef cattle breeding strategy is optimized by combining deep reinforcement learning with game theory, intelligent decision-making and dynamic adjustment are realized, the real-time feedback mechanism makes the breeding strategy more flexible, the breeding strategy can be continuously optimized according to the feedback data, and the precision, efficiency and adaptability of the beef cattle breeding are improved.

Description

Technical Field

[0001] This invention relates to the field of beef cattle breeding technology, specifically to an efficient method for breeding new beef cattle breeds. Background Technology

[0002] The breeding of new beef cattle breeds is an important issue in agriculture and animal husbandry. Traditional breeding methods mainly rely on artificial screening and gene analysis techniques to improve the production performance of beef cattle by evaluating their phenotypic and genetic characteristics. These methods have promoted the improvement of beef cattle breeds to a certain extent, especially in genetic optimization. However, with the diversification of breeding objectives and environmental conditions, the limitations of existing technologies in handling complex, multi-dimensional data and dynamically adjusting breeding strategies are gradually becoming apparent.

[0003] Existing technologies focus on optimizing single objectives, such as improving growth rate or meat quality. Many traditional methods fail to adequately consider the impact of environmental factors on beef cattle performance and lack comprehensive consideration of multi-objective optimization. Furthermore, traditional breeding methods rely on manual judgment and experience, making the process cumbersome and inefficient, and hindering real-time strategy adjustments to address unforeseen circumstances. Traditional methods also have limited capacity for processing multi-dimensional data, making it impossible to leverage modern big data technologies for in-depth analysis. These issues result in shortcomings in adaptability, accuracy, and real-time performance of existing breeding methods. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an efficient method for breeding new beef cattle breeds, solving the problems of lack of flexibility and dynamic adjustment mechanisms in the beef cattle breeding process in existing technologies.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an efficient method for breeding new beef cattle breeds, comprising the following steps: S1. Collect phenotypic, genotypic, and environmental data of beef cattle. Phenotypic data includes the weight, body size, activity level, and health status of beef cattle. Genotypic data is collected through gene chips or whole genome sequencing. Environmental data is collected by monitoring environmental factors such as temperature, humidity, and gas concentration in the cattle shed through sensors. S2. Preprocess the collected phenotypic data, genotypic data and environmental data, including data cleaning, missing value imputation, standardization and feature extraction, to obtain key features related to beef cattle performance; S3. Based on the preprocessed data, establish an optimization model. Through stochastic optimization, Markov decision process and nonlinear dynamic system model, dynamically adjust the breeding strategy of beef cattle and perform balance optimization among multiple objectives. The Markov decision process is used to define the state, action and reward function of beef cattle, evaluate the impact of each action on the performance of beef cattle, and dynamically adjust the breeding program according to the evaluation results. S4. Based on game theory models, analyze the selection and mating decisions among beef cattle, determine the optimal breeding strategy for each beef cattle, and optimize the breeding effect of the entire herd through Nash equilibrium. S5. The system will continuously provide feedback based on real-time data and use the feedback data to further optimize the linear regression model. With each round of breeding decisions and strategy adjustments, the model will be continuously updated. S6. Through a real-time data feedback system, dynamically monitor the performance of beef cattle and adjust breeding programs.

[0006] Preferably, the optimization model uses a stochastic optimization method to model the uncertainties of environmental and genetic factors and uses a Gaussian process to predict future breeding results.

[0007] Preferably, the Markov decision process defines the state, actions, and reward functions of beef cattle, evaluates the impact of each action on the performance of beef cattle, and dynamically adjusts the breeding strategy based on the evaluation.

[0008] Preferably, the nonlinear dynamic system model is based on nonlinear differential equations to simulate the growth process of beef cattle under different mating, feeding and environmental conditions, predict the performance changes of the beef cattle population, and optimize the breeding strategy.

[0009] Preferably, the game theory model is based on the Nash equilibrium principle, analyzes the decision-making strategy of each beef cattle in the breeding process, and optimizes the overall breeding strategy through game theory.

[0010] Preferably, the deep reinforcement learning algorithm is based on a deep Q-network model, which optimizes the breeding strategy by learning from historical data and real-time feedback, and adjusts the mating plan according to real-time feedback to optimize the performance of each generation of beef cattle.

[0011] Preferably, the real-time data feedback system can monitor the weight, body shape, health status and other performance data of beef cattle in real time, and feed the data back to the cloud system for use by subsequent decision-making models.

[0012] Preferably, the feature extraction includes using principal component analysis to extract key features related to beef cattle growth, meat quality, and disease resistance from high-dimensional phenotypic data, and using them for subsequent optimization decisions.

[0013] Preferably, the real-time data feedback system uses a deep learning model to predict real-time data and feeds the prediction results back to the breeding decision module to optimize the breeding process of each generation of beef cattle.

[0014] Preferably, the breeding strategy optimization method can adjust the selection weights of genotype and phenotype based on the performance results of each generation of cattle, thereby achieving long-term stable optimization and improvement of the cattle herd.

[0015] This invention provides an efficient method for breeding new beef cattle breeds. It has the following beneficial effects: 1. This invention adopts a technical solution that combines deep reinforcement learning with a multi-objective optimization model, achieving intelligent and efficient beef cattle breeding decision optimization. Compared with the traditional manual selection or simple genetic algorithm in the prior art, it solves the problem of lack of adaptability and flexibility when facing complex environments and multi-objective conflicts. Through real-time feedback and dynamic adjustment, the system can optimize each decision based on real-time data, ensuring that the performance of each generation of beef cattle continues to improve.

[0016] 2. This invention introduces a game theory model to optimize the breeding strategy of beef cattle herds, achieving the technical effect of collaborative optimization of group behavior. Compared with the existing technical solutions that only consider the selection of a single individual, this invention solves the problem of failing to effectively handle the interaction between individuals and maximize the overall benefits of the group. By combining Nash equilibrium and multi-agent game theory, it can ensure that the decision of each beef cattle is optimally coordinated within the group, thereby improving the overall breeding effect.

[0017] 3. This invention achieves dynamic adjustment during the breeding process through real-time data monitoring and feedback mechanisms, thus achieving the technical effect of optimizing decision-making and real-time response. Compared with conventional breeding methods in the prior art, traditional methods rely on periodic evaluation, which leads to a lack of response mechanisms when facing emergencies. However, this invention can adjust the mating and feeding strategies in real time according to the performance of beef cattle at each stage, ensuring that every link in the breeding process can be flexibly adjusted and that the herd performance is always at its best.

[0018] 4. This invention combines multiple data sources, including phenotypic data, genotypic data, and environmental data, and employs precise feature extraction and optimization models to achieve comprehensive and multi-dimensional analysis. Compared with existing technologies that rely solely on limited genetic data or single phenotypic evaluation, this invention solves the problem of neglecting the influence of environmental factors and genetic diversity during the breeding process. By comprehensively integrating multiple data sources, this invention can provide more accurate breeding decisions and improve the efficiency and stability of beef cattle breed improvement. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the method flow of the present invention. Detailed Implementation

[0020] The technical solutions in 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.

[0021] Please see the appendix Figure 1 This invention provides an efficient method for breeding new beef cattle breeds, comprising the following steps: S1. Collect phenotypic, genotypic, and environmental data of beef cattle. Phenotypic data includes the weight, body size, activity level, and health status of beef cattle. Genotypic data is collected through gene chips or whole genome sequencing. Environmental data is collected by monitoring environmental factors such as temperature, humidity, and gas concentration in the cattle shed through sensors. In the implementation of this invention, data acquisition is the fundamental module of the entire beef cattle breeding method. Data acquisition not only provides accurate input data for subsequent optimization models but also provides a reliable basis for every decision made throughout the breeding process. Accurate and comprehensive data acquisition can effectively improve the efficiency and precision of subsequent modules, thereby promoting the scientific and precise nature of the breeding process.

[0022] Generally, the data acquisition module includes three main aspects: phenotypic data acquisition, genotypic data acquisition, and environmental data acquisition. This data is collected through various devices and sensors, covering multiple dimensions such as the health status, growth rate, genetic background, and environmental conditions of beef cattle. This data will support subsequent modules such as data preprocessing, feature extraction, and optimization model construction.

[0023] Phenotypic data collection In this embodiment, phenotypic data is collected in real time using devices such as smart ear tags, weight sensors, and video monitoring systems. These devices enable dynamic monitoring of each beef cattle's weight, body shape, activity level, and health status. In particular, regularly capturing images of the cattle's body shape using high-definition cameras allows for the observation of morphological changes at different stages, including muscle distribution, fat deposition, and other vital signs. This data is crucial for subsequent analyses such as growth rate assessment, meat quality evaluation, and health status monitoring.

[0024] Specifically, the smart ear tags have built-in sensors that continuously track the cattle's movement and body temperature, and can wirelessly transmit the data to a data center in real time. Weight sensors are installed in the cattle sheds to periodically record the weight changes of each ox. The video monitoring system uses real-time video analysis to help determine changes in the cattle's body shape and activity levels. All these devices can collect data around the clock, ensuring the timeliness and accuracy of the data.

[0025] Genotype data collection In one possible implementation, genotypic data is collected using gene chips or whole-genome sequencing technology. Genomics techniques allow for the acquisition of the genome information of each beef cattle, aiding in the analysis of their genetic background. Genotypic data is crucial for precision breeding, as it reveals whether beef cattle carry gene markers associated with desirable traits. Analysis of genotypic data allows for the identification of individuals with good growth potential and disease resistance at an early stage.

[0026] Generally, genotypic data collection utilizes gene chip or whole-genome sequencing technologies. Gene chip technology employs probe design targeting specific gene regions to perform high-throughput screening of the cattle genome, obtaining genotypic information for each animal. Whole-genome sequencing, on the other hand, provides a more comprehensive and in-depth analysis of the beef cattle genome, identifying gene markers related to key traits such as meat quality, disease resistance, and fertility.

[0027] Environmental data collection In this embodiment, environmental data is collected through a sensor network. By deploying sensors such as temperature and humidity sensors, gas concentration sensors, and light intensity sensors within the cattle shed, environmental changes can be monitored in real time, and their impact on beef cattle growth can be recorded. Environmental factors, such as temperature, humidity, air quality, and light intensity, can significantly affect the growth rate, health status, and production performance of beef cattle. Therefore, the collection of environmental data plays a crucial role in assessing the adaptability of beef cattle and further optimizing breeding strategies.

[0028] Specifically, the sensor network will wirelessly transmit data to a central processing system in real time. The data processing system will analyze the environmental data in real time to ensure that beef cattle grow and reproduce under optimal environmental conditions. In some embodiments, the environmental data may include not only temperature, humidity, and gas concentration, but also noise levels and air circulation within the cattle shed.

[0029] Data acquisition equipment and data transmission During data acquisition, the selection of equipment and the reliability of data transmission are crucial. All collected data is transmitted in real time to the cloud data processing platform via wireless transmission technology. The efficient wireless network connection ensures the real-time nature and reliability of the data. During transmission, data is protected by encryption technology to ensure data security.

[0030] Generally, data transmission utilizes low-power wide-area network technologies such as Zigbee, Wi-Fi, and LoRa to ensure smooth data transmission even in complex environments. Furthermore, to ensure system scalability, standardized interface designs are used between devices and the platform, facilitating future system upgrades and expansions.

[0031] Equipment and tool selection during technology implementation Alternatively, the phenotypic data acquisition devices in this embodiment, including but not limited to smart ear tags, weight sensors, and video monitoring systems, are all commercially available products and can be well integrated with the data processing platform. The gene chips and whole-genome sequencing technologies used for genotypic data acquisition are also existing, mature technologies. The application of these devices effectively reduces human intervention in the breeding process and improves the accuracy and efficiency of data acquisition.

[0032] Data quality control and calibration To ensure the accuracy and completeness of the data, quality control is also performed during the data acquisition process. Data quality checks include, but are not limited to, the following: checking the normal operating status of the acquisition equipment to ensure that the data acquisition frequency and time periods meet predetermined standards; conducting preliminary analysis of the acquired data to ensure its rationality and consistency. All acquisition equipment is calibrated to ensure the reliability of its measurement results.

[0033] Connection and follow-up steps After the data acquisition module is completed, all collected phenotypic, genotypic, and environmental data will be input into the next data preprocessing and feature extraction module. In this module, the data will undergo cleaning, missing value imputation, and standardization, and then key features related to beef cattle performance will be extracted. These features will serve as input to support the construction of subsequent optimization models, ensuring the accuracy and effectiveness of optimization decisions.

[0034] S2. Preprocess the collected phenotypic data, genotypic data and environmental data, including data cleaning, missing value imputation, standardization and feature extraction, to obtain key features related to beef cattle performance; In the preceding steps, we completed the data acquisition module, collecting multi-dimensional data on beef cattle through smart ear tags, genotype data acquisition devices, and environmental sensor systems. The next task for this massive amount of raw data is preprocessing and feature extraction to ensure its maximum effectiveness in subsequent analysis and model optimization. Data preprocessing and feature extraction are not only for cleaning and standardizing the data, but also fundamental to subsequent optimization decisions and model accuracy.

[0035] In this embodiment, the main goal of data preprocessing is to ensure that the collected phenotypic, genotypic, and environmental data are accurate and complete, and can be transformed into effective features that can be used for model analysis. Data preprocessing steps include data cleaning, missing value imputation, and data standardization; the feature extraction step focuses on extracting features related to key traits such as beef cattle growth, meat quality, and health, providing input for subsequent optimization decisions and deep learning models.

[0036] Data cleaning and missing value imputation In this embodiment, data cleaning is the first step in data preprocessing. For the collected data, especially phenotypic and environmental data, we first use a quality control system to check its validity and consistency, and identify and remove outliers. Outliers are usually caused by factors such as equipment failure, data transmission problems, or human error. If not handled properly, they may seriously affect the accuracy of the model and the decision-making results.

[0037] Specifically, data cleaning begins by identifying outliers in the dataset using methods such as standard deviation detection or box plots. Generally, measurements outside the normal range are considered outliers and removed from the dataset. For example, if a cow's weight increases or decreases too rapidly in a short period, it might indicate sensor error or data loss and should be removed. After cleaning, the data undergoes quality verification to ensure its reliability.

[0038] After data cleaning, missing value imputation becomes the next step. Since data loss may occur during the data collection process, missing values ​​need to be appropriately imputed to avoid affecting subsequent analysis and modeling. As an option, this embodiment uses the K-Nearest Neighbors (KNN) imputation method to interpolate missing values. Specifically, the KNN algorithm calculates the distance between missing data points and other similar data points, selects the K nearest neighbors most similar to the missing point, and then performs interpolation based on the values ​​of these neighbors.

[0039] In addition, in some cases, if the missing values ​​are concentrated or the data volume is small, methods such as linear interpolation or mean imputation can be used. Each method has its applicable scenarios, and the imputation scheme should be chosen flexibly according to the specific situation.

[0040] Data standardization To ensure data consistency and comparability, especially in subsequent modeling and optimization processes, this embodiment standardizes all collected data. The goal of standardization is to transform data with different dimensions into data with the same scale, thereby avoiding unnecessary impacts from features with larger dimensions on subsequent algorithms.

[0041] Generally, data standardization includes two common methods: min-max standardization and Z-score standardization. As an option, this embodiment uses the Z-score standardization method. Specifically, the formula for calculating Z-score standardization is:

[0042] in, The original data, The mean of the data. The standard deviation of the data. These are the standardized values. This method standardizes the data by converting the mean of each feature to 0 and the standard deviation to 1, ensuring that each feature is analyzed and processed on the same scale.

[0043] In some embodiments, for certain phenotypic characteristics (such as body size data of beef cattle), we can also choose min-max normalization to scale the data to between 0 and 1. This normalization method is more suitable for situations where the data range is known and relatively concentrated.

[0044] Feature extraction After data standardization, we proceed to the feature extraction stage. In this step, this embodiment uses methods such as Principal Component Analysis (PCA) and Factor Analysis (FA) to extract the most representative features from the high-dimensional phenotypic data. Specifically, PCA is a commonly used method in dimensionality reduction techniques. It maps the original high-dimensional feature space to a new low-dimensional space through linear transformation while retaining as much information as possible. In PCA, the principal components to be retained are mainly based on the variance of the data.

[0045] Generally, the first step in the PCA process is to calculate the covariance matrix of the standardized data matrix. Then, eigenvalue decomposition is performed on the covariance matrix to obtain the principal components of the data. After obtaining the principal components, we can select the first few principal components based on the magnitude of their eigenvalues; these principal components will form a new feature space. This process maximizes the preservation of the data's variance, thereby reducing redundant information.

[0046] Specifically, suppose we have a phenotypic data matrix Its dimensions are ,in It is the number of samples in the data. It is the eigenvalue. First, we look at the data matrix... Standardization is performed. Then, the covariance matrix is ​​calculated. Eigenvalue decomposition is performed. The resulting eigenvectors are the new principal components. The principal components with the largest variances are selected for subsequent analysis and decision-making.

[0047] In some embodiments, genotype data may also be processed using similar dimensionality reduction techniques. However, since genotype data is usually relatively high in dimensionality, deep learning methods can be combined during feature extraction to extract biologically significant features from complex genomic data.

[0048] Connection and follow-up steps After the data preprocessing and feature extraction steps described above, all processed data will be used as input to enter the optimization model construction stage. In this stage, by using key features extracted based on PCA, combined with the previously acquired phenotypic, genotypic, and environmental data, an optimization model suitable for beef cattle breeding will be constructed. At this point, the results of feature extraction will directly affect the model's construction effectiveness and subsequent optimization decisions.

[0049] In this way, data preprocessing and feature extraction not only provide high-quality input data for subsequent optimization models and deep learning algorithms, but also ensure the consistency and accuracy of data processing, providing a solid foundation for subsequent decision-making.

[0050] S3. Based on the preprocessed data, establish an optimization model. Through stochastic optimization, Markov decision process and nonlinear dynamic system model, dynamically adjust the breeding strategy of beef cattle and perform balance optimization among multiple objectives. The Markov decision process is used to define the state, action and reward function of beef cattle, evaluate the impact of each action on the performance of beef cattle, and dynamically adjust the breeding program according to the evaluation results. In the preceding steps, the data acquisition and preprocessing modules have been successfully completed, generating high-quality data suitable for subsequent analysis. The feature extraction step further refined the data, providing clear and standardized input for the construction of the optimization model. Therefore, the next crucial step in this invention is the construction of the optimization model. The optimization model is the core of the entire breeding method, responsible for formulating the optimal beef cattle breeding program based on the collected and processed data. This step utilizes multiple mathematical optimization models to ensure the accuracy and effectiveness of the breeding decisions.

[0051] In this embodiment, the construction of the optimization model involves multiple technical modules, including stochastic optimization, Markov decision process (MDP), and nonlinear dynamic system modeling. Each optimization method plays a role in different decision-making scenarios, collectively supporting the breeding decisions for new beef cattle breeds.

[0052] Stochastic optimization and Markov decision processes (MDP) First, based on the phenotypic, genotypic, and environmental data collected in the preceding steps, a dynamic decision-making process is constructed. In this embodiment, a **Markov Decision Process (MDP)** is used to model the breeding process of beef cattle. MDPs are suitable for decision problems with temporal and uncertain characteristics, and can dynamically adjust breeding strategies by defining states, actions, and reward functions.

[0053] Specifically, in the Markov decision-making process, we first define "state" as the performance state of beef cattle during the breeding process, including relevant phenotypic characteristics such as weight, body shape, and health status. "Action" represents the decisions that may be made during the breeding process, such as mating, feeding methods, and environmental adjustments. "Reward" is the feedback to the breeding decisions, specifically manifested as the target traits such as growth rate and meat quality achieved by the beef cattle after a certain period. The reward function is quantified by a formula, with the parameters defined as follows:

[0054] in, It's a reward. This is the current state. It is an action. Function The MDP model describes the relationship between states and actions to measure the effectiveness and impact of breeding decisions. By continuously adjusting actions in different breeding scenarios, the MDP model can optimize breeding programs within each decision cycle, ensuring optimal performance of beef cattle at each stage.

[0055] In general, the MDP model is solved using the Bellman equation:

[0056] in For state The value function represents the expected reward that can be obtained from the current state; For possible actions; This is a discount factor, representing a discount on future rewards; Let be the state transition probability, representing the probability from the current state. Take action After transitioning to state The probability of.

[0057] In practical applications, the MDP model continuously updates its status and evaluates rewards based on real-time data, dynamically adjusting breeding decisions. Through repeated iterative optimization, optimal mating schemes and feeding strategies can be obtained, thereby achieving efficient breeding of the entire population.

[0058] Modeling of nonlinear dynamic systems In addition to using the MDP model for decision optimization, this embodiment also employs nonlinear dynamic system modeling for the growth process of beef cattle. The growth process of beef cattle is influenced by factors such as genetics and environment, and the relationships between these factors are often nonlinear. Therefore, based on principles of physics and biology, nonlinear differential equations are used to model the growth process of beef cattle, simulating its growth changes under different conditions.

[0059] Specifically, we use a nonlinear differential equation to describe the growth rate and weight changes of beef cattle under specific environmental conditions. Let the weight of the beef cattle be... ,in Given time, the growth rate can be expressed as a differential equation of the following form:

[0060] in, It is the growth rate constant. This is the maximum weight that beef cattle can reach (environmental carrying capacity). It is a decay constant related to the environment and feeding methods, representing the negative impact of health problems or other adverse factors on growth.

[0061] Using this nonlinear differential equation model, we can simulate the weight changes of beef cattle under different growth stages and environmental conditions, and further evaluate the performance of traits such as meat quality and reproductive capacity. Based on these simulation results, breeding strategies can be dynamically adjusted while ensuring the health and productivity of beef cattle.

[0062] Model fusion and optimization In practical applications, MDP (Multi-Level Planning) and nonlinear dynamic system models work together to optimize breeding strategies. In some embodiments, by combining the decision optimization of the MDP model with the growth prediction of the nonlinear dynamic system, the growth process of beef cattle can be accurately modeled and predicted. Within this model framework, the MDP model adjusts breeding decisions based on real-time feedback, while the nonlinear dynamic system predicts the performance of future generations based on the results of the current decisions. This fusion approach maximizes the accuracy of breeding strategies and reduces conflicts between different optimization objectives.

[0063] Connection and follow-up steps Through the aforementioned optimization model construction, the collected phenotypic, genotypic, and environmental data have been successfully transformed into an optimization model that aids in beef cattle breeding decisions. This optimized model will then guide breeding decisions in subsequent steps. Based on the optimization results of the aforementioned model, the system will be able to automatically adjust breeding strategies, including mating, feeding, and environmental management, to ensure continuous improvement in meat quality, body shape, growth rate, and disease resistance in each generation.

[0064] S4. Based on game theory models, analyze the selection and mating decisions among beef cattle, determine the optimal breeding strategy for each beef cattle, and optimize the breeding effect of the entire herd through Nash equilibrium. In the preceding steps, we have completed data collection, preprocessing, feature extraction, and the construction of the optimization model. The main task of these modules is to provide data support and a model framework for breeding decisions. The next step is to optimize multi-agent decisions for the beef cattle population using a game theory model. By applying game theory, this invention can analyze the interactions between individual beef cattle, optimize the mating decisions of each cattle, and improve the overall breeding performance of the herd.

[0065] The application of game theory in the breeding process, especially multi-agent game models, can handle the interactions between individual beef cattle and between individuals and the environment. The growth, health, and reproductive traits of different individual beef cattle may be interdependent, which requires game theory models to predict the decision-making behavior of individuals and their impact on the overall breeding outcome of the herd.

[0066] Introduction and selection of game theory models In this embodiment, the game theory model is constructed based on the Nash equilibrium principle. Generally, Nash equilibrium can be used to analyze how multiple participants (i.e., multiple individual beef cattle) can optimize the overall benefit of the group by choosing different strategies (such as breeding strategies). Each cattle chooses a "strategy" during the breeding process, i.e., a specific breeding or feeding plan. These decisions affect the growth, reproduction, and other characteristics of other cattle. Therefore, through the game theory model, we can find the optimal decision for each cattle and ensure that the overall benefit of the group is maximized.

[0067] Specifically, in multi-agent game models, each cow's choices depend not only on its own growth and health but also on the performance and selection strategies of other cows in the group. For example, if a cow is in good health, it may become a preferred breeding target, and other cows may adjust their breeding strategies accordingly. Through game theory analysis, we can predict the outcomes of these interactions and find the optimal breeding strategy within the group through Nash equilibrium.

[0068] Applications of Nash Equilibrium In this embodiment, the application of Nash equilibrium is mainly reflected in the mating decisions of individual beef cattle. In game theory, Nash equilibrium is a stable state in which each participant (i.e., each cattle) chooses its optimal strategy given the strategies of other participants. Specifically, in the context of beef cattle breeding, the strategy of the beef cattle is its chosen mating partner and feeding plan.

[0069] Generally, Nash equilibrium can be represented by a policy matrix, where each participant's choice corresponds to a "policy," and the outcome of each policy combination affects the overall group's performance. In this embodiment, the policy matrix can be represented tabularly, and each cow's policy combination will be evaluated to determine whether it achieves the optimal performance for the group. To find the Nash equilibrium, we need to solve the following equation:

[0070] in, Indicates the first The utility function of a cow. Indicates the first The selection strategy for cows Indicates except the first The strategy for cattle other than that one. This is another strategy choice. The formula means: the strategy chosen by each cow. It must satisfy the following condition: given other cattle strategies In this case, its utility function Cannot be used by other strategies Surpass.

[0071] By solving the Nash equilibrium, we can determine the optimal breeding strategy for the beef cattle herd and optimize the overall performance of the herd.

[0072] Cooperative game and non-cooperative game In some cases, cattle may not be completely independent, but rather exhibit cooperative or competitive relationships. Therefore, this embodiment also considers the differences and combinations of cooperative and non-cooperative game dynamics to further optimize the breeding strategy.

[0073] Specifically, in cooperative game theory, the interests of individual beef cattle can be shared. For example, in multi-generational breeding, if one cattle performs exceptionally well, other cattle can also benefit from this good performance. In this situation, individuals may choose to cooperate, that is, to improve the overall performance of the herd by optimizing breeding combinations.

[0074] In contrast, non-cooperative game theory considers the competitive relationships between individuals. For example, some beef cattle might choose to breed with weaker individuals based on differences in size or growth rate to enhance their genetic advantage in offspring. In non-cooperative game theory, each cattle's breeding choices maximize its own benefit, regardless of the overall performance of the herd.

[0075] Optimization and Practical Applications of Game Theory Models To enhance the model's practical applicability, this embodiment utilizes multiple simulations of a game theory model to explore the impact of different breeding combinations on the performance of beef cattle herds. Through repeated game iterations, the breeding strategy can be continuously adjusted to ensure that each generation of cattle achieves the expected performance. For example, through repeated game optimization, the most suitable breeding scheme for beef cattle under specific environmental conditions can be gradually found.

[0076] In practical applications, the results of the game theory model are used as input to a deep reinforcement learning (DRL) model to further optimize the decision-making process. By combining game theory models with reinforcement learning, the system can automatically adjust breeding strategies based on game outcomes, continuously optimizing the performance of each generation of beef cattle.

[0077] Connection and follow-up steps The application of game theory models provides an intelligent decision-making framework for beef cattle breeding, ensuring that the mating decisions for each cow achieve optimal balance within the herd. By combining game theory with optimization models, dynamic system modeling, and deep reinforcement learning, game theory models provide theoretical basis and decision support for the entire breeding process. The next step is to further optimize these decisions using deep reinforcement learning, achieving intelligent and dynamic adjustment of the beef cattle breeding process.

[0078] By introducing a game theory model, this invention can comprehensively consider the mutual influence between individual beef cattle and between individuals and the environment, making the breeding strategy more scientific and reasonable, and ultimately promoting the rapid breeding and optimization of new beef cattle breeds.

[0079] This invention optimizes the multi-agent decision-making process in beef cattle breeding using a game theory model. It leverages Nash equilibrium theory to address the competition and cooperation among individuals, thereby improving the accuracy and efficiency of the breeding process. By combining game theory with optimization models and deep reinforcement learning, game theory provides powerful decision support for beef cattle breeding, ensuring continuous improvement in the performance of each generation and ultimately achieving the goal of efficient new breed breeding.

[0080] S5. The system will continuously provide feedback based on real-time data and use the feedback data to further optimize the linear regression model. With each round of breeding decisions and strategy adjustments, the model will be continuously updated. In this embodiment, a linear regression model is used to learn the relationship between traits such as growth, health, and meat quality of beef cattle and selected breeding strategies using historical data. In this way, the system can predict future performance and adjust breeding and feeding strategies according to optimization goals. The advantages of linear regression lie in its simplicity and efficiency, making it suitable for rapid application on large-scale datasets. Although this model has certain limitations compared to deep Q-networks, its computational cost is lower, it is easier to implement, and it is sufficient to meet the needs of beef cattle breeding in practical applications.

[0081] Implementation of a linear regression model Linear regression models make predictions by establishing a linear relationship between input features and breeding objectives. In this model, we use phenotypic data, genotypic data, and environmental factors of beef cattle as input features, and breeding objectives (such as growth rate, meat quality score, and disease resistance) as output objectives. Specifically, the model will make predictions using the following formula:

[0082] in, Indicates the breeding objective (e.g., growth rate). This represents each input characteristic (such as weight, genotype, environmental factors, etc.). These are the regression coefficients. The goal of the model is to determine the optimal regression coefficients by minimizing the error (i.e., the difference between the actual and predicted values), thereby accurately predicting future breeding goals.

[0083] Model training and optimization During training, we use historical data to train the linear regression model. The least squares method is used to optimize the regression coefficients, minimizing the difference between the predicted results and the actual data. In this process, the model iteratively adjusts the regression coefficients until it converges to the optimal solution.

[0084] Specifically, the goal of the training process is to minimize the following loss function:

[0085] in, It is the first The true value of each sample It is the value predicted by the model. This refers to the sample size. By minimizing this loss function, the model can learn the optimal regression coefficients and make effective predictions on new data.

[0086] Optimization and Intelligent Decision Making By using a trained linear regression model, we can predict the performance of each beef cattle under different breeding and feeding conditions, and formulate optimization strategies based on these predictions. The model can predict the future performance of each beef cattle under different environmental conditions based on its current performance data (such as weight, health status, etc.), thereby adjusting breeding combinations and feeding strategies.

[0087] For example, if a cow's growth rate does not meet expectations, the model can adjust its breeding combination based on historical data, selecting to breed with cows of a higher growth rate genotype, or adjust feeding conditions based on environmental data, such as adjusting feed formulation or temperature and humidity. This real-time and flexible nature of optimization decision-making allows the breeding process to better adapt to actual conditions.

[0088] Through optimization of the linear regression model and intelligent decision-making, the system can provide rapid and accurate decision support for beef cattle breeding. Subsequent breeding processes will further optimize the model based on real-time data feedback to ensure that the growth, health, and other performance characteristics of each generation of beef cattle continue to improve.

[0089] By using a simple and efficient linear regression model, the system can reduce computational costs and simplify the implementation process while ensuring prediction accuracy, thus ensuring intelligent and refined management of the breeding process.

[0090] S6. Dynamically monitor the performance of beef cattle and adjust breeding programs through a real-time data feedback system; In the preceding steps, we have completed data collection, data preprocessing, feature extraction, optimization model construction, game theory model application, and deep reinforcement learning (DRL) training. Through these steps, we have been able to generate an optimized breeding strategy and adjust it using the deep reinforcement learning model. The next step is real-time feedback and dynamic adjustment, the goal of which is to ensure that the performance of each generation of beef cattle during the breeding process can be monitored in real time and that decisions are optimized through a feedback mechanism.

[0091] Real-time feedback and dynamic adjustment is a process that closely integrates breeding strategies with real-time data. Through real-time data feedback, the system can promptly understand the growth, health, and reproductive performance of beef cattle, and then adjust subsequent mating, feeding strategies, and environmental management based on this feedback. This step plays a "closed-loop feedback" role in the entire breeding process, making the breeding process more precise, efficient, and dynamic.

[0092] Real-time data monitoring and feedback In this embodiment, the real-time data feedback mechanism collects data through smart sensors, video surveillance equipment, and an environmental monitoring system. Smart ear tags, weight sensors, and a health monitoring system monitor indicators such as the cattle's weight, body shape, activity level, and health status in real time. The video surveillance equipment uses image recognition technology to periodically assess the cattle's body shape and meat quality.

[0093] Specifically, the weight data of beef cattle is uploaded to a cloud data platform in real time via wireless weight sensors. Body shape and meat quality are monitored through image acquisition and analysis using high-definition cameras and computer vision technology. Environmental data (such as temperature, humidity, and gas concentration) are monitored in real time by a sensor system installed in the cattle shed and the data is sent to the cloud, providing information on the impact of environmental factors on the performance of beef cattle.

[0094] Typically, all this data is transmitted to a central processing system for real-time analysis. The system combines information such as cattle health data, weight changes, and meat quality, and adjusts the current breeding strategy through feedback loops. The collection and analysis of real-time data not only helps understand the current performance of the herd but also provides a basis for dynamic adjustments to cope with uncertain environments and sudden changes.

[0095] Combining real-time feedback with DRL models In this embodiment, the real-time feedback mechanism is closely integrated with the deep reinforcement learning model. In the aforementioned steps, deep reinforcement learning optimizes breeding decisions by training on historical data and the current strategy. As real-time data is continuously input, the DQN model updates its strategy in each round of training, enabling the breeding program to be adjusted in real time based on the latest feedback data.

[0096] In the implementation, real-time feedback data is transmitted to the deep reinforcement learning system and used together with the current state data to update the Q-value function. Specifically, each performance of the beef cattle (such as weight, health status, etc.) serves as the state input, while feedback information (such as changes in growth rate, occurrence of diseases, etc.) serves as the reward input, thereby adjusting the breeding program or feeding management strategy.

[0097] Dynamically adjust breeding and feeding strategies In this embodiment, one of the main functions of real-time feedback is to dynamically adjust breeding strategies. Through real-time data, the system can promptly identify and adjust problems during the breeding process. For example, if a cow's weight gain is too slow, the system will automatically adjust its breeding combination or change its feeding strategy based on data analysis results. Through continuous feedback and adjustment, the system ensures that the performance of each generation of beef cattle is continuously optimized.

[0098] Specifically, suppose a cow has a slow growth rate, and the system learns from feedback data that the cow's body size, nutrient absorption, and other aspects are not performing well. In this case, the DRL model will adjust the breeding strategy based on the current state, selecting to breed with cows of a higher growth rate genotype, or adjusting feeding conditions based on environmental monitoring data, such as adjusting the temperature and humidity in the barn and the feed formula.

[0099] In addition, the system can optimize the overall health of the herd based on its collective performance. For example, when the overall health of the herd is poor, the system can automatically adjust its breeding strategy, prioritizing breeding of healthier cattle to improve the overall health of the herd. The system continuously adjusts its strategy based on feedback data to ensure that the health and growth of each generation of beef cattle meet the expected goals.

[0100] Combining feedback data with multi-objective optimization In this embodiment, the real-time feedback mechanism is not limited to optimizing a single objective, but also includes adjusting multiple objectives. In beef cattle breeding, there are typically multiple optimization objectives, such as growth rate, meat quality, disease resistance, and reproductive capacity. These objectives may conflict with each other or require a balance. Through a deep reinforcement learning model, the system can simultaneously consider these multiple objectives and dynamically adjust the weights of each objective based on feedback data to achieve the best overall effect.

[0101] Specifically, when updating the reward function, the reward for each objective is adjusted based on the current feedback data.

[0102] Real-time feedback and dynamic adjustments not only optimize individual breeding and husbandry strategies but also provide feedback data for subsequent steps, further guiding breeding decisions. After each round of feedback adjustments, the system continues to adjust the optimization model, optimizing strategies based on the latest data to form a closed-loop feedback loop.

[0103] Ultimately, through close integration with deep reinforcement learning models and optimization decision-making models, real-time feedback and dynamic adjustments ensure continuous optimization of each generation in the beef cattle breeding process, making the entire breeding process more intelligent, efficient, and adaptable.

[0104] In the preceding steps, we completed the various technical modules of the beef cattle breeding method, including data acquisition, preprocessing, optimization model construction, game theory model application, deep reinforcement learning optimization, and real-time feedback and dynamic adjustment. Each step provides precise guidance for beef cattle breeding, making the breeding process increasingly intelligent and optimized. Next, step 7, "Implementation and Evaluation," will put all the preceding strategies and decisions into practice, carry out the specific implementation of the breeding program, and continuously monitor and evaluate its effects.

[0105] In this embodiment, the core task of the implementation and evaluation phase is to apply the aforementioned optimized decision-making scheme to the actual breeding process of beef cattle populations and to evaluate the effectiveness of the breeding through continuous feedback. This process ensures the connection between theory and practice, helping us to verify the effectiveness and practicality of the optimized strategies designed in the previous steps in a real environment. Through implementation and evaluation, shortcomings in the scheme can be identified in a timely manner, and corresponding adjustments can be made to ensure that the expected breeding goals are ultimately achieved.

[0106] Application of the implementation plan In this embodiment, the implementation plan will first be applied to an experimental beef cattle herd. During this experiment, the breeding strategy is generated jointly by the aforementioned optimization model, game theory decision-making, and deep reinforcement learning model, serving as the basis for guiding the selection and mating of individual beef cattle. Specifically, preliminary screening is first conducted based on each cattle's genotype, phenotypic characteristics, and environmental adaptability, selecting cattle with excellent genetic background and phenotypic characteristics as breeding cattle. Then, based on the results of the deep reinforcement learning model and game theory analysis, an optimal mating strategy is formulated to ensure that each generation of offspring shows improved performance in growth, meat quality, and disease resistance.

[0107] Real-time monitoring and data collection during implementation During the implementation phase, real-time data acquisition and monitoring mechanisms remain crucial. Using the sensors and smart ear tags already installed in previous steps, the system will continuously monitor various indicators of beef cattle, including growth, health, and reproduction. Real-time data feedback will directly influence subsequent breeding decisions, enabling breeding strategies to dynamically adapt to changes in the beef cattle population.

[0108] Specifically, during implementation, sensors will continuously record phenotypic data for each cow, including weight, body shape, and health status, and upload this data in real time to a cloud-based data processing platform. The platform will analyze this data in real time, assess the performance of each cow, and generate reports based on the analysis results. These reports will be provided to breeders to guide future breeding decisions. Simultaneously, an environmental monitoring system will record data such as temperature, humidity, and air quality within the cattle sheds to ensure that the beef cattle's growing environment meets optimal conditions.

[0109] Evaluation scheme design and effectiveness evaluation In this embodiment, effectiveness evaluation is a crucial step in the implementation process. The system evaluates the effectiveness of the current breeding program through continuous monitoring and data analysis of the beef cattle herd. Evaluation indicators include, but are not limited to, the following aspects: Growth rate: The effect of the growth rate improvement was evaluated by comparing the weight changes of each cow at different time points.

[0110] Meat quality: The meat quality of beef cattle is evaluated through meat quality scores (such as lean meat percentage, fat distribution, etc.) to ensure that one of the breeding goals is achieved.

[0111] Health status: Assess the health status of beef cattle through health monitoring data (such as disease incidence, disease resistance, etc.).

[0112] Reproductive capacity: By analyzing indicators such as the reproductive success rate and mating effect of beef cattle, the impact of breeding strategies on reproductive capacity is evaluated.

[0113] Specifically, the evaluation process involves statistically analyzing the performance data of each generation of cattle to ensure the achievement of breeding objectives. By comparing the performance of the population under different breeding programs, the system can assess which strategies are most effective and adjust subsequent breeding programs accordingly.

[0114] Feedback and optimization of implementation results Feedback on evaluation results is crucial during implementation. By assessing the effectiveness of each stage, the system can adjust breeding strategies in real time to ensure the breeding process meets expected goals. Evaluation reports will provide decision support for breeders, ensuring that the growth and performance of each generation of beef cattle are optimized.

[0115] For example, if the evaluation report shows that the growth rate of the first generation of offspring has not met the target, the system will automatically adjust the breeding program, selecting higher-quality breeding cattle for mating, or adjusting conditions such as temperature and humidity in the cattle shed based on environmental data to improve the growth rate of the next generation. Such adjustments will ensure that the performance of each generation of beef cattle continues to improve, ultimately achieving rapid and effective breeding of new beef cattle breeds.

[0116] Connection and follow-up steps After the implementation and evaluation steps are completed, the next step will be long-term monitoring. The system will continue to collect performance data on beef cattle and regularly evaluate the effectiveness of breeding strategies. With the continuous accumulation of data and ongoing optimization based on feedback, the system will gradually improve the breeding program, providing more accurate decision support for the next round of breeding.

[0117] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A highly efficient method for breeding new beef cattle breeds, characterized in that, Includes the following steps: S1. Collect phenotypic, genotypic, and environmental data of beef cattle. Phenotypic data includes the weight, body size, activity level, and health status of beef cattle. Genotypic data is collected through gene chips or whole genome sequencing. Environmental data is collected by monitoring environmental factors such as temperature, humidity, and gas concentration in the cattle shed through sensors. S2. Preprocess the collected phenotypic data, genotypic data and environmental data, including data cleaning, missing value imputation, standardization and feature extraction, to obtain key features related to beef cattle performance; S3. Based on the preprocessed data, establish an optimization model. Through stochastic optimization, Markov decision process and nonlinear dynamic system model, dynamically adjust the breeding strategy of beef cattle and perform balance optimization among multiple objectives. The Markov decision process is used to define the state, action and reward function of beef cattle, evaluate the impact of each action on the performance of beef cattle, and dynamically adjust the breeding program according to the evaluation results. S4. Based on game theory models, analyze the selection and mating decisions among beef cattle, determine the optimal breeding strategy for each beef cattle, and optimize the breeding effect of the entire herd through Nash equilibrium. S5. The system will continuously provide feedback based on real-time data and use the feedback data to further optimize the linear regression model. With each round of breeding decisions and strategy adjustments, the model will be continuously updated. S6. Through a real-time data feedback system, dynamically monitor the performance of beef cattle and adjust breeding programs.

2. The efficient method for breeding new beef cattle breeds according to claim 1, characterized in that, The optimization model described uses a stochastic optimization method to model the uncertainties of environmental and genetic factors and uses a Gaussian process to predict future breeding results.

3. The efficient method for breeding new beef cattle breeds according to claim 1, characterized in that, The Markov decision process defines the state, actions, and reward functions of beef cattle, evaluates the impact of each action on the performance of beef cattle, and dynamically adjusts the breeding strategy based on the evaluation.

4. The efficient method for breeding new beef cattle breeds according to claim 1, characterized in that, The nonlinear dynamic system model described above is based on nonlinear differential equations to simulate the growth process of beef cattle under different mating, feeding and environmental conditions, predict the performance changes of the beef cattle population, and optimize the breeding strategy.

5. The efficient method for breeding new beef cattle breeds according to claim 1, characterized in that, The game theory model described is based on the Nash equilibrium principle, analyzes the decision-making strategy of each beef cattle in the breeding process, and optimizes the overall breeding strategy through game theory.

6. The efficient method for breeding new beef cattle breeds according to claim 1, characterized in that, The deep reinforcement learning algorithm described above is based on a deep Q-network model. It optimizes the breeding strategy by learning from historical data and real-time feedback, and adjusts the mating plan according to real-time feedback to optimize the performance of each generation of beef cattle.

7. The efficient method for breeding new beef cattle breeds according to claim 1, characterized in that, The real-time data feedback system can monitor the weight, body shape, and health status of beef cattle in real time and feed the data back to the cloud system for use by subsequent decision-making models.

8. The efficient method for breeding new beef cattle breeds according to claim 1, characterized in that, The feature extraction involves using principal component analysis to extract key features related to beef cattle growth, meat quality, and disease resistance from high-dimensional phenotypic data, and then using these features for subsequent optimization decisions.

9. The efficient method for breeding new beef cattle breeds according to claim 1, characterized in that, The real-time data feedback system uses a deep learning model to predict real-time data and feeds the prediction results back to the breeding decision module to optimize the breeding process of each generation of beef cattle.

10. The efficient method for breeding new beef cattle breeds according to claim 1, characterized in that, The aforementioned breeding strategy optimization method can adjust the selection weights of genotype and phenotype based on the performance results of each generation of cattle, thereby achieving long-term stable optimization and improvement of the cattle herd.

Images