Extreme weather ecological pasture carbon emission reduction intelligent optimization method

Abstract

The application provides an ecological pasture carbon emission reduction intelligent optimization method for extreme weather, and relates to the technical field of carbon emission reduction intelligent optimization. The method comprises the following steps: collecting initial monitoring data and preprocessing, constructing a multi-dimensional feature vector output state feature; using an LSTM time series prediction model to analyze the state of dairy cows and the activity of methanogenic bacteria in the rumen, using a DHI data analysis integrated gradient boosting tree model to analyze the amino acid balance of lactating dairy cows and precise nutrition technology, constructing a dairy cow manure production prediction model based on a Bayesian network, dynamically optimizing manure management strategies, and outputting manure management effect data; constructing a carbon emission reduction comprehensive evaluation model to evaluate the carbon emission reduction effect of the pasture. Through the research on the rumen microbial regulation technology of dairy cow carbon emission, the rumen flora structure is regulated, the rumen greenhouse gas emission is controlled, the protein level of the ruminant feed is reduced, and the methane emission is reduced.

Description

Technical Field

[0001] This invention relates to the field of intelligent optimization technology for carbon emission reduction, and in particular to an intelligent optimization method for carbon emission reduction in ecological pastures during extreme weather. Background Technology

[0002] Carbon emissions are a general term or abbreviation for greenhouse gas emissions. Against the backdrop of global climate change, carbon emission space will gradually become a key resource constraint for economic development, making low-carbon emissions in livestock farming imperative. Methane emissions from ruminant gut fermentation, as well as methane and nitrous oxide emissions from manure management systems, are major sources of greenhouse gas emissions from livestock farming, with dairy cows being the largest emitters of greenhouse gases among all ruminants.

[0003] Dairy cow nutrition management relies on experience and general standards, making it difficult to accurately grasp individual nutritional needs. Furthermore, the lack of dynamic optimization in formula feeding easily leads to nutritional imbalances. Manure management strategies are fixed, lacking the ability to dynamically adjust based on real-time conditions, and there are no accurate prediction and feedback mechanisms. Overall assessment indicators are one-sided, failing to comprehensively measure effectiveness. Adjustments after problems are identified rely on manual experience, lacking intelligent and continuous optimization capabilities. Carbon reduction technologies are applied in a fragmented manner, making it difficult to dynamically adapt them to achieve optimal emission reduction effects. Summary of the Invention

[0004] This invention provides an intelligent optimization method for carbon emission reduction in ecological pastures during extreme weather, which addresses the shortcomings of existing carbon emission reduction technologies, such as their fragmented application and difficulty in dynamically adapting to achieve optimal emission reduction results.

[0005] On the one hand, this invention provides a smart optimization method for carbon emission reduction in ecological pastures during extreme weather, comprising:

[0006] S1: Deploy monitoring nodes and a main control center on the ranch, set up monitoring devices for weather, dairy cow status, and manure treatment, and establish an encrypted communication link. After verifying the identity of the monitoring nodes, the main control center assigns initial monitoring tasks, synchronizes basic ranch information, and outputs initial monitoring data.

[0007] S2: Preprocess the initial monitoring data, construct a multi-dimensional feature vector through an attention mechanism, and dynamically allocate weights to output state features.

[0008] S3: Based on the state characteristics and historical additive adjustment effect dataset, use the LSTM time series prediction model to analyze the state of dairy cows and the activity of methanogens in the rumen, and output dynamically updated rumen microbiota state data.

[0009] S4: Using DHI data analysis with an integrated gradient boosting tree model, we obtained a low-carbon, low-protein diet for lactating dairy cows based on amino acid balance and precision nutrition technology.

[0010] S5: Construct a dairy cow manure production prediction model based on Bayesian networks. Input the low-carbon, low-protein dairy cow diet formulation and dynamically updated rumen microbiota status data into the prediction model, dynamically optimize manure management strategies, and output manure management effect data such as carbon and nitrogen emission reduction efficiency and resource conversion rate.

[0011] S6: Based on data on manure management effectiveness, rumen greenhouse gas emission control data, and low-carbon dairy cow low-protein diet formulation, construct a comprehensive carbon reduction assessment model to evaluate the effectiveness of farm carbon reduction.

[0012] According to the present invention, an intelligent optimization method for carbon emission reduction in ecological pastures during extreme weather conditions is provided. In step S1, feature extraction and classification are performed on the pasture weather data.

[0013] S11: Deploy monitoring nodes and a main control center. Set up meteorological monitoring stations, dairy cow status monitoring equipment, and manure treatment monitoring devices within the ranch, and establish encrypted communication links between each monitoring node and the control center.

[0014] S12: The main control center verifies the identity of each monitoring node. After successful verification, it assigns initial monitoring tasks, synchronizes basic ranch information, and outputs initial monitoring data.

[0015] According to the intelligent optimization method for carbon emission reduction in ecological pastures during extreme weather provided by the present invention, the specific steps for preprocessing the initial monitoring data in step S2 are as follows:

[0016] S21: Denoise the initial monitoring data by using a Kalman filter algorithm to smooth noise data from sensors such as weather and dairy cow status, and output the denoised data.

[0017] S22: Detect missing values ​​in the denoised data, use linear interpolation to fill in discrete missing data, and use cubic spline interpolation based on historical trends to fill in continuous missing data, and output the complete dataset.

[0018] S23: Standardize the complete dataset, map environmental data to the [0,1] interval, and use Z-score standardization for milk production and milk composition data to eliminate dimensional differences.

[0019] S24: Use an attention mechanism to construct a multi-dimensional feature vector, dynamically allocate weights according to the degree of impact of extreme weather, and output state features.

[0020] According to the intelligent optimization method for carbon emission reduction in ecological pastures during extreme weather provided by the present invention, the specific steps in step S24 of constructing a multi-dimensional feature vector using an attention mechanism are as follows:

[0021] S241: Based on the standardized complete dataset, construct the feature association matrix, calculate the mutual information values ​​between different features, and output the feature mutual information matrix.

[0022] S242: Based on the feature mutual information matrix, dynamic weights are assigned to motion state, environment perception, and communication quality features, and a weight vector is output.

[0023] S243: Perform a dot product operation between the standardized complete dataset and the weight vector to output the feature vector.

[0024] According to the intelligent optimization method for carbon emission reduction in ecological pastures during extreme weather provided by the present invention, the specific steps in step S3 for outputting dynamically updated rumen microbiota status data are as follows:

[0025] S31: Establish a basic library for correlation analysis based on status data and historical additive adjustment effect datasets.

[0026] S32: Based on the correlation analysis database and real-time data collected on the physiological indicators of dairy cows under extreme weather conditions and the activity of methanogens in the rumen, a dynamic monitoring sequence is formed.

[0027] S33: Based on the dynamic monitoring sequence, the real-time calculation module of the LSTM time series prediction model analyzes the changing trend of data with extreme weather and additive effects, and outputs dynamically updated rumen microbiota status data.

[0028] According to the intelligent optimization method for carbon emission reduction in ecological pastures under extreme weather conditions provided by the present invention, the specific steps in step S33, which involve analyzing the trend of data changes with extreme weather and additive effects using an LSTM time-series prediction model, are as follows:

[0029] S331: Standardize the dynamic monitoring sequence data and input it into the real-time calculation module of the LSTM time series prediction model.

[0030] S332: The LSTM-based memory cell structure captures the characteristics of data changes over time, distinguishing between the stage affected by extreme weather and the stage after adding different additives.

[0031] S333: By setting an appropriate time step, the LSTM model can focus on the changes in data over different time spans, and analyze whether the data in each stage is trending upward, downward, or stable.

[0032] S334: Optimize the fitting of data change trends, accurately extract the specific changes in data under the influence of extreme weather and additives, and obtain the data change trend with the influence of extreme weather and additives.

[0033] According to the intelligent optimization method for carbon emission reduction in ecological pastures during extreme weather provided by the present invention, the specific steps for obtaining a low-carbon, low-protein dairy cow diet formulation in step S4 are as follows:

[0034] S41: Extract categorical variables from the initial monitoring data. Categorical variables are discrete categories rather than continuous values.

[0035] S42: Encode categorical variables, standardize continuous variables, and construct a dataset containing DHI indicators and amino acid requirements.

[0036] S43: Train an integrated gradient boosting tree model, using historical DHI data as input, and train the model with the goal of meeting the balanced amino acid requirements and precise nutritional needs of dairy cows.

[0037] S44: Analyze the new DHI data using the trained model to output the types and requirements of essential amino acids for specific lactating cows under the current production state, the appropriate upper limit of protein level, and determine the maximum reduction of crude protein in low-protein diets based on low-carbon goals.

[0038] S45: Select raw materials with high protein utilization based on amino acid requirements, use mixed integer linear programming algorithm to calculate the optimal ratio, optimize the model based on feedback from small-scale monitoring data, repeatedly adjust the formula until the standard is met, and output a low-carbon, low-protein dairy cow diet formula.

[0039] According to the intelligent optimization method for carbon emission reduction in ecological pastures during extreme weather provided by the present invention, the specific steps in step S45 of calculating the optimal ratio using a mixed integer linear programming algorithm are as follows:

[0040] S451: Construct a multi-objective optimization function that minimizes dietary crude protein content, nitrogen excretion, and raw material costs.

[0041] S452: Establish a dynamic raw material database and use a mixed-integer linear programming algorithm to iteratively solve for the optimal raw material ratio through initialization, range narrowing, and sensitivity analysis.

[0042] S453: Evaluate the optimal raw material ratio and determine if it is greater than the threshold. If so, output the optimal raw material ratio; otherwise, backtrack and adjust.

[0043] According to the intelligent optimization method for carbon emission reduction in ecological pastures during extreme weather provided by the present invention, the specific steps in step S5 for outputting data on the effectiveness of manure management, including carbon and nitrogen emission reduction efficiency and resource conversion rate, are as follows:

[0044] S51: Collect historical data on manure production.

[0045] S52: Construct a Bayesian network model, determine the input nodes as diet formulation parameters and rumen microbiota status indicators, and the output nodes as manure production and characteristics. Train the model using historical data and quantify the correlation between various factors using conditional probability tables.

[0046] S53: Based on the model prediction results and combined with the current status of the sewage treatment system, set optimization targets for carbon and nitrogen emission reduction efficiency and resource conversion rate, use sewage treatment control parameters as decision variables, construct a multi-objective optimization function, and solve it through particle swarm optimization to obtain the optimal management strategy parameters.

[0047] S54: Real-time monitoring of biogas production, ammonia nitrogen residue, organic fertilizer conversion rate, and comparison with model predictions; determination of whether the difference exceeds the threshold; if so, updating the model with measured data, adjusting weights, and resolving to obtain data on the effectiveness of manure management.

[0048] According to the intelligent optimization method for carbon emission reduction in ecological pastures during extreme weather provided by the present invention, the specific steps for evaluating the effectiveness of pasture carbon emission reduction are as follows:

[0049] S61: Establish a carbon emission reduction assessment indicator system, including direct emission reduction indicators, indirect emission reduction indicators, and resource utilization indicators.

[0050] S62: Construct a comprehensive carbon emission reduction assessment model, train the model using historical emission reduction case data, and determine the mapping relationship between each variable and the assessment results.

[0051] S63: Analyze the mapping relationship between variables and evaluation results in real time using machine learning algorithms to assess the current carbon emission status of the ranch.

[0052] S64: Determine whether the carbon emissions of the ranch have reached the preset threshold; otherwise, adjust the emission reduction strategy.

[0053] This invention provides an intelligent optimization method for carbon emission reduction in ecological pastures during extreme weather. Through research on rumen microbial regulation technology for dairy cow carbon emissions, it regulates the rumen microbiota structure and controls rumen greenhouse gas emissions, achieving the following beneficial effects:

[0054] Improve protein-saving technologies. Low-protein diets help reduce the amount of protein-rich feed used in dairy farming and improve feed efficiency.

[0055] This invention addresses the issue of carbon and nitrogen greenhouse gas emissions from dairy farms. The yucca extract developed in this project can reduce methanogens and methane emissions, while low-protein diets increase nitrogen use efficiency and reduce nitrogen emissions. Regarding dairy manure waste, the most significant waste from large-scale dairy farms, the air nanobubble water technology developed in this project can significantly improve anaerobic fermentation efficiency, accelerate the degradation of organic matter and nitrogen, and effectively reduce pollutant emissions.

[0056] By saving on feed costs, the economic benefits of animal husbandry are directly improved. Through the use of the low-protein, amino acid-balanced diet developed in this project, the daily economic benefits per dairy cow have increased.

[0057] From a long-term perspective, this holistic optimization approach helps dairy farms improve their economic, ecological, and social benefits while coping with extreme weather and achieving carbon reduction targets. By optimizing aspects such as ensuring cow health, efficient resource utilization, and reduced environmental pollution, the invention enhances the competitiveness and adaptability of dairy farms, laying a solid foundation for their long-term development and aligning with the industry trend towards green, low-carbon, and sustainable development. Attached Figure Description

[0058] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0059] Figure 1 This is a flowchart illustrating an intelligent optimization method for carbon emission reduction in ecological pastures during extreme weather, provided by an embodiment of the present invention.

[0060] Figure 2 This is a roadmap of rumen regulation technology provided in the embodiments of the present invention;

[0061] Figure 3 This is a technical roadmap for low-protein diet formulation provided in the embodiments of the present invention;

[0062] Figure 4 This is a roadmap for carbon and nitrogen emission reduction technology in sewage management provided by an embodiment of the present invention. Detailed Implementation

[0063] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0064] Example 1: The following is combined with Figures 1-4 This invention describes an intelligent optimization method for carbon emission reduction in ecological pastures during extreme weather.

[0065] Figure 1This is a flowchart illustrating an intelligent optimization method for carbon emission reduction in ecological pastures during extreme weather, provided by an embodiment of the present invention.

[0066] like Figure 1 As shown in the embodiment of the present invention, an intelligent optimization method for carbon emission reduction in ecological pastures during extreme weather conditions is provided. The method includes:

[0067] S1: Deploy monitoring nodes and a main control center on the ranch, set up monitoring devices for weather, dairy cow status, and manure treatment, and establish an encrypted communication link. After verifying the identity of the monitoring nodes, the main control center assigns initial monitoring tasks, synchronizes basic ranch information, and outputs initial monitoring data.

[0068] S11: Deploy monitoring nodes and a main control center. Set up meteorological monitoring stations, dairy cow status monitoring equipment, and manure treatment monitoring devices within the ranch, and establish encrypted communication links between each monitoring node and the control center.

[0069] S12: The main control center verifies the identity of each monitoring node. After successful verification, it assigns initial monitoring tasks, synchronizes basic ranch information, and outputs initial monitoring data.

[0070] S2: Preprocess the initial monitoring data, construct a multi-dimensional feature vector through an attention mechanism, and dynamically allocate weights to output state features.

[0071] S21: Denoise the initial monitoring data by using a Kalman filter algorithm to smooth noise data from sensors such as weather and dairy cow status, and output the denoised data.

[0072] S22: Detect missing values ​​in the denoised data, use linear interpolation to fill in discrete missing data, and use cubic spline interpolation based on historical trends to fill in continuous missing data, and output the complete dataset.

[0073] S23: Standardize the complete dataset by mapping environmental data such as temperature and humidity to the [0,1] interval, and use Z-score standardization for milk production and milk composition data to eliminate dimensional differences.

[0074] S24: Employ an attention mechanism to construct a multi-dimensional feature vector, dynamically allocate weights based on the degree of impact of extreme weather, and output the fused collaborative state features.

[0075] S3: Based on the state characteristics and historical additive adjustment effect dataset, use the LSTM time series prediction model (Long Short-Term Memory Network Time Series Prediction Model) to analyze the state of dairy cows and the activity of methanogens in the rumen, and output dynamically updated rumen microbiota state data.

[0076] S31: Establish a basic database for correlation analysis based on state data and historical additive adjustment effect datasets. First, the system collects basic state data of the rumen microbiota in dairy cows, including the types, quantities, and proportions of microbiota at different physiological stages, specifically early lactation, mid-lactation, and dry periods, particularly the abundance and activity of key microbiota such as methanogens and cellulose-decomposing bacteria. Simultaneously, it compiles detailed parameters of various historically used rumen microbiota regulating additives, specifically yucca extract and probiotic preparations, covering additive type, dosage, duration of use, and corresponding microbiota adjustment effects, specifically the changes in methanogen counts, methane emission reductions, and cow physiological feedback. This data is structured and stored according to the correlation logic of "state parameters - additive parameters - effect indicators" to establish a basic database for correlation analysis. The environmental conditions corresponding to the data need to be labeled, such as normal weather, extreme high temperatures, and heavy rain. Outliers are removed through data cleaning to ensure the accuracy and consistency of the data in the database, providing a reliable data foundation for subsequent correlation analysis.

[0077] S32: A dynamic monitoring sequence is formed based on the association analysis database and real-time data collected on dairy cow physiological indicators and rumen methanogen activity under extreme weather conditions. The dynamic monitoring sequence is constructed based on existing data association patterns in the association analysis database and real-time collected information. Real-time data collection includes specific parameters of extreme weather, such as real-time temperature, duration, and humidity changes during high temperatures, and precipitation amount and intensity during heavy rain. Simultaneously, through smart wearable devices deployed on dairy cows and rumen sensors, real-time physiological indicators of dairy cows are acquired, such as rumination frequency, feed intake rate, rumen pH, rumen temperature, and rumen methanogen activity indicators. These real-time data are arranged sequentially according to timestamps, maintaining consistency with historical data formats for similar extreme weather conditions in the association analysis database, forming a multi-dimensional dynamic monitoring sequence that includes environmental parameters, physiological indicators, and microbial activity. This ensures that the data at each time point in the sequence corresponds completely and clearly reflects the real-time changes of each indicator under extreme weather and additive effects.

[0078] S33: Based on the dynamic monitoring sequence, the real-time calculation module of the LSTM time series prediction model analyzes the changing trend of data with extreme weather and additive effects, and outputs dynamically updated rumen microbiota status data.

[0079] The specific steps for analyzing the data trends under extreme weather and additive effects using an LSTM time-series prediction model are as follows:

[0080] S331: Standardize the dynamic monitoring sequence data and input it into the real-time calculation module of the LSTM time series prediction model.

[0081] S332: The LSTM-based memory cell structure captures the characteristics of data changes over time, distinguishing between the stage affected by extreme weather and the stage after adding different additives.

[0082] S333: By setting an appropriate time step, the LSTM model can focus on the changes in data over different time spans, and analyze whether the data in each stage is trending upward, downward, or stable.

[0083] S334: Optimize the fitting of data change trends, accurately extract the specific changes in data under the influence of extreme weather and additives, and obtain the data change trend with the influence of extreme weather and additives.

[0084] S4: Using DHI (Dairy Production Performance Index) data analysis with an integrated gradient boosting tree model, we obtained a low-carbon, low-protein diet for lactating dairy cows based on amino acid balance and precision nutrition technology.

[0085] S41: Dairy cow DHI data, including indicators such as milk yield, milk composition, somatic cell count, and feed intake, while also recording basic information such as the age, parity, and lactation stage of the dairy cow, as well as the protein, amino acid content and corresponding nitrogen excretion data in historical diet formulations.

[0086] S42: Preprocess the collected DHI data, remove outliers and missing values, encode categorical variables, standardize continuous variables and milk production, and construct a dataset that includes input features, DHI indicators, basic information of dairy cows, target variables, amino acid requirements, nitrogen utilization rate, and carbon emission-related indicators such as manure nitrogen emissions.

[0087] S43: Train an integrated gradient boosting tree model, using historical DHI data as input, and train the model with the goals of balanced amino acid requirements and precise nutritional requirements of dairy cows, as well as the appropriate amount of energy and protein. Adjust the model hyperparameters through cross-validation so that the model can accurately predict the amino acid requirements and nutritional metabolic efficiency of dairy cows under different DHI characteristics.

[0088] S44: Analyze the new DHI data using the trained model to output the types and requirements of essential amino acids for specific lactating cows under the current production state, the appropriate upper limit of protein level, and determine the maximum reduction of crude protein in low-protein diets in combination with low-carbon targets.

[0089] S45: Based on the model's predicted amino acid requirements, feed ingredients that can provide the corresponding amino acids are screened, prioritizing ingredients with high protein utilization. Using a formulation optimization algorithm, the optimal ratio of each ingredient is calculated while meeting amino acid balance and nutritional requirements, ensuring that the crude protein level in the diet is reduced without affecting the milk production performance and health of dairy cows. The initially designed low-protein diet formula is applied to a small-scale trial, monitoring the milk yield, changes in milk composition, and fecal nitrogen emissions of dairy cows. This actual data is fed back to the integrated gradient boosting tree model, and the model is retrained to optimize prediction accuracy. The diet formula is iteratively adjusted until the amino acid balance and nutritional supply predicted by the model are consistent with the actual low-carbon emission reduction effect. Finally, a low-carbon low-protein dairy cow diet formula that meets the requirements is determined, and the corresponding formula parameters and emission reduction effect data are output.

[0090] The specific steps for calculating the optimal ratio using a formula optimization algorithm are as follows:

[0091] S451: Construct a multi-objective optimization function with "lowest dietary crude protein level" as the core objective, while also incorporating "minimum nitrogen excretion" and "lowest raw material cost." Through weighted summation, the multi-objective function is transformed into a single-objective function, ensuring that the optimization direction aligns with low-carbon emission reduction goals. Based on the amino acid requirements output by the integrated gradient boosting tree model, minimum thresholds for essential amino acids such as lysine and methionine are defined, while the ranges for basic indicators such as metabolizable energy and neutral detergent fiber are limited. In terms of production performance, lower limits for milk yield, milk protein percentage, and milk fat percentage are set based on historical DHI data. Regarding raw material usage, the addition ranges for each raw material are specified to avoid impacting palatability or rumen function due to imbalances in raw material proportions.

[0092] S452: Establish a dynamic raw material database, inputting key parameters for 20-30 candidate raw materials, including crude protein content, amino acid composition and rumen availability, metabolizable energy value, unit price, and anti-nutritional factor content, and labeling the low-carbon attributes of the raw materials. Use a mixed-integer linear programming algorithm to solve the problem, setting the raw material ratio as the decision variable, and finding the minimum value of the objective function through algorithm iteration.

[0093] Initialize the feasible solution space, eliminating raw material combinations that clearly violate constraints. Use a branch-and-bound method to narrow the search range, focusing on optimizing raw materials that contribute significantly to amino acid content. Determine the tightness of each constraint through sensitivity analysis to ensure solution stability.

[0094] S453: First, use a nutrient composition calculator to check if the total amount and proportion of amino acids meet the standards. Then, use a rumen degradation model to simulate protein digestibility and nitrogen excretion. Finally, combine this with a milk yield prediction model to evaluate the impact of the formula on production performance. If any amino acid fails to meet the standard or the predicted milk yield is lower than the threshold, backtrack and adjust the raw material database or fine-tune the constraint weights, and rerun the algorithm. The crude protein level is reduced by 1.5-2 percentage points compared to the conventional formula, all essential amino acids reach the required values ​​of 105%-110%, nitrogen utilization is increased to over 68%, and small-scale trials have shown no significant decrease in milk production performance, thus ensuring that the formula achieves a balance between low carbon emissions and production performance.

[0095] S5: By combining low-carbon, low-protein dairy cow diets with extreme weather information, and using Bayesian networks to predict dairy cow manure production, the system dynamically optimizes manure management strategies such as the concentration of micro-nano bubble water and the frequency of acidifier spraying, and outputs manure management effectiveness data on carbon and nitrogen emission reduction efficiency and resource conversion rate.

[0096] S51: Collect historical manure production data, including average daily production, solid-liquid ratio, and carbon and nitrogen content.

[0097] S52: Construct a Bayesian network model, determine the input nodes as diet formulation parameters and rumen microbiota status indicators, and the output nodes as manure production and characteristics, specifically total weight, carbon-nitrogen ratio, and solid content. Train the model using historical data and quantify the correlation between various factors using conditional probability tables.

[0098] S54: Compare the measured values ​​with the model predictions, calculate the error, and if the error exceeds the threshold, update the conditional probability table of the Bayesian network with the measured data, retrain the model, adjust the weights of the optimization function, and solve again to obtain the new strategy. Repeat the above prediction-optimization-monitoring-feedback process until the carbon and nitrogen emission reduction efficiency and resource conversion rate stably meet the standards, and output the manure management effect data including specific carbon emission reduction, nitrogen emission reduction, resource conversion ratio, and corresponding control parameters.

[0099] S6: Based on data on manure management effectiveness and rumen greenhouse gas emission control, construct a comprehensive evaluation model to accurately assess the effectiveness of pasture carbon reduction and adjust and optimize strategies for each stage accordingly.

[0100] S61: Establish a carbon emission reduction assessment indicator system, including direct emission reduction indicators, indirect emission reduction indicators, and resource utilization indicators.

[0101] S62: Construct a comprehensive carbon emission reduction assessment model, train the model using historical emission reduction case data, and determine the mapping relationship between each variable and the assessment results.

[0102] S63: Analyze the dynamic correlation between data in each stage in real time through machine learning algorithms to assess the overall carbon emission status of the pasture and the actual effectiveness of various control measures.

[0103] S64: Based on the assessment results, if it is found that the carbon emission reduction effect does not meet the expected target, the system will automatically adjust the dosage of rumen microbiota additives for dairy cows, the specific ratio of low-carbon low-protein diet formulas for dairy cows, and key parameters such as the concentration of micro-nano bubble water added in the manure management strategy and the frequency of acidifier spraying. At the same time, it will provide feedback on adjustment suggestions to the main control center, which will then reallocate the subsequent monitoring tasks of the monitoring nodes accordingly, further optimizing the carbon emission reduction process of the entire farm under extreme weather conditions, forming a continuously iterative and intelligently optimized ecological farm carbon emission reduction operation mode, helping the farm achieve more efficient carbon emission reduction goals and sustainable development.

[0104] In summary, this embodiment provides an intelligent optimization method for carbon emission reduction in ecological pastures during extreme weather. Through research on rumen microbial regulation technology for dairy cow carbon emissions, it regulates the rumen microbiota structure and controls rumen greenhouse gas emissions, achieving the following beneficial effects:

[0105] Improve protein-saving technologies. Low-protein diets help reduce the amount of protein-rich feed used in dairy farming and improve feed efficiency.

[0106] This invention addresses the issue of carbon and nitrogen greenhouse gas emissions from dairy farms. The yucca extract developed in this project can reduce methanogens and methane emissions, while low-protein diets increase nitrogen use efficiency and reduce nitrogen emissions. Regarding dairy manure waste, the most significant waste from large-scale dairy farms, the air nanobubble water technology developed in this project can significantly improve anaerobic fermentation efficiency, accelerate the degradation of organic matter and nitrogen, and effectively reduce pollutant emissions.

[0107] Example 2: Effects of low-protein amino acid balanced diets on the production performance and fecal nitrogen content of Holstein dairy cows. The specific example is as follows:

[0108] The study selected 30 healthy, disease-free Chinese Holstein dairy cows of similar age, parity, lactation period, and milk yield, and randomly divided them into 3 groups of 10 cows each, and raised them under the same conditions.

[0109] Control group: fed the original dairy farm formula diet with a protein level of 17.08%.

[0110] Experimental Group I: Based on the control diet, the protein level was reduced by 0.80 percentage points, and RPLys 24.6 g / d + RPMet 10.4 g / d were added, resulting in a final protein level of 16.28%. Experimental Group II: Based on the control diet, the protein level was reduced by 1.66 percentage points, and RPLys 25.7 g / d + RPMet 9.4 g / d were added, resulting in a final protein level of 15.42%. The experiment lasted a total of 49 days, divided into a 7-day pre-feeding period and a 42-day trial period. Without affecting dairy cow production performance, the crude protein (CP) content in dairy cow diets was reduced. The nutritional value of the CP was assessed using the CNCPS system and nutrient composition analysis. CPM-Dairyv3.0 software was used to optimize the diet formulation, determining the dosage of RPLys and RPMet to balance the low-protein diet and explore its potential to reduce nitrogen excretion. This study aimed to mitigate environmental pollution. The effects of low-protein amino acid balanced diets on dry matter intake and dietary conversion rate in experimental groups I and II increased milk production by 5.10% and 9.07% respectively compared to the control group. Dry matter intake in experimental group I increased by 5.38% compared to the control group, and the yield of 4% standard milk was also slightly higher in group I.

[0111] The three different diets had no significant effect on the milk composition of dairy cows (P>0.05). The experimental group had slightly higher milk protein, lactose, and milk solids content than the control group. The experimental group had slightly lower milk somatic cell count than the control group, but overall, the milk somatic cell count was relatively low. These results indicate that diets formulated according to the ideal protein amino acid pattern for dairy cows can improve low-protein diets. In particular, by meeting the ideal protein or small intestinal amino acid requirements, the protein needs of dairy cows are fully met, thus not affecting their production performance. Furthermore, feeding a balanced amino acid diet has a positive impact on the production performance of dairy cows.

[0112] The three different diets had no significant effect on the apparent digestibility of DM, CP, NDF, and ADF in dairy cows (P>0.05), but the apparent digestibility of all parameters in experimental group II showed an increasing trend compared to the control group. The results of this study indicate that for amino acid-balanced diets, reducing the dietary CP content by 0.8–1.6 percentage points does not affect the apparent digestibility of various nutrients. Experimental group II had the largest reduction in crude protein, but its apparent digestibility of all parameters increased. This may be primarily because the experimental group used steam-flaked corn to replace protein feed, thus reducing the dietary CP content, resulting in higher non-fibrous carbohydrate (NFC) and starch content in the experimental group compared to the control group. Since the carbohydrate content and ratio in the diet are crucial for dietary protein degradation, rumen microbial growth, and rumen homeostasis, the experimental group's diet may be closer to an energy-nitrogen balance. Furthermore, the amino acid balance in the experimental group's diet improved nitrogen utilization. Therefore, under the dual influence of sufficient energy levels and amino acid balance, even if the dietary protein content of the experimental group is reduced, it does not affect its digestibility and utilization rate, and even has a positive effect.

[0113] The effects of low-protein, amino acid-balanced diets on nitrogen utilization and nitrogen emissions in dairy cows were investigated. In the three different dietary groups, group II showed a 9.51% decrease in nitrogen intake compared to the control group (P<0.05). Fecal nitrogen and urinary nitrogen in group II were also significantly lower than in the control group, decreasing by 10.21% (P<0.05) and 18.65% (P<0.05), respectively. Excreted nitrogen in both experimental groups was significantly lower than in the control group, decreasing by 7.67% (P<0.05) and 15.19% (P<0.05), respectively. While the protein content reduction in group I was relatively small, the amino acid balance did not negatively impact dairy cow performance; instead, it significantly reduced total nitrogen emissions. In group II, significant reductions in urinary nitrogen, fecal nitrogen, and total nitrogen excretion were observed, possibly due to a 1.66% reduction in dietary CP content and the efficient utilization of dietary nitrogen through amino acid and energy-nitrogen balance, which positively impacted dairy cow performance. The ratio of milk nitrogen to feed nitrogen in all three diets showed an increasing trend, with Group II showing a 17.58% increase compared to the control group (P<0.05). There were no significant differences in nitrogen utilization rate and nitrogen balance among the groups (P>0.05). Milk urea nitrogen (MUN) decreased significantly, with Groups I and II showing reductions of 8.17% (P<0.05) and 11.70% (P<0.05) compared to the control group, respectively. The milk nitrogen efficiency of Group II was significantly higher than that of the control group, and Group I was also slightly higher, possibly because the dietary protein content in the control group far exceeded the dairy cows' requirements, while the experimental groups made full use of feed nitrogen. The MUN content in the experimental groups was significantly lower than that in the control group, possibly due to higher protein utilization.

[0114] The economic benefits of Experiment I and Experiment II increased by 0.68% and 4.82% respectively compared to the control group. The daily dietary costs of Experiment I and Experiment II decreased by RMB 0.27 / (head·day) and RMB 0.84 / (head·day) respectively compared to the control group, while the economic benefits increased by RMB 0.45 / (head·day) and RMB 3.17 / (head·day) respectively. This indicates that a low-protein, amino acid-balanced diet can increase economic benefits.

[0115] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0116] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A smart optimization method for carbon emission reduction in ecological pastures during extreme weather, characterized in that, include: S1: Deploy monitoring nodes and a main control center in the ranch, set up monitoring devices for weather, dairy cow status, and manure treatment, and establish an encrypted communication link. After verifying the identity of the monitoring nodes, the main control center assigns initial monitoring tasks, synchronizes basic ranch information, and outputs initial monitoring data. S2: Preprocess the initial monitoring data, construct a multi-dimensional feature vector through an attention mechanism and dynamically allocate weights to output state features; S3: Based on the state characteristics and historical additive adjustment effect dataset, use the LSTM time series prediction model to analyze the state of dairy cows and the activity of methanogens in the rumen, and output dynamically updated rumen microbiota state data. The specific steps for outputting dynamically updated rumen microbiota status data are as follows: S31: Establish a basic library for correlation analysis based on status data and historical additive adjustment effect datasets; S32: Based on the correlation analysis database and real-time data collected on the physiological indicators of dairy cows under extreme weather conditions and the activity of methanogens in the rumen, a dynamic monitoring sequence is formed; S33: Based on the dynamic monitoring sequence, the real-time calculation module of the LSTM time series prediction model analyzes the changing trend of data with extreme weather and additive effects, and outputs dynamically updated rumen microbiota status data. S4: Using integrated gradient boosting tree model to analyze the production performance measurement data of dairy cows, we can obtain a low-carbon, low-protein diet for dairy cows based on amino acid balance and precision nutrition technology. The specific steps to obtain a low-carbon, low-protein dairy cow diet are as follows: S41: Extract the categorical variables from the initial monitoring data. Categorical variables are variables that are discrete categories rather than continuous values. S42: Encode categorical variables, standardize continuous variables, and construct a dataset containing indicators of dairy cow production performance and amino acid requirements; S43: Train the integrated gradient boosting tree model, using historical dairy cow production performance measurement data as input, and train the model with the goal of dairy cow amino acid balance requirements and precise nutritional requirements. S44: Analyze the production performance data of new dairy cows using the trained model, output the types and requirements of essential amino acids for specific lactating dairy cows under the current production state, the appropriate upper limit of protein level, and determine the maximum reduction of crude protein in low-protein diets based on low-carbon targets. S45: Select raw materials with high protein utilization based on amino acid requirements, use mixed integer linear programming algorithm to calculate the optimal ratio, optimize the model based on the feedback of small-scale monitoring data, repeatedly adjust the formula until the standard is met, and output a low-carbon dairy cow low-protein diet formula. S5: Construct a dairy cow manure production prediction model based on Bayesian network, input the low-carbon dairy cow low-protein diet formulation and dynamically updated rumen microbiota status data into the prediction model, dynamically optimize the manure management strategy, and output manure management effect data such as carbon and nitrogen emission reduction efficiency and resource conversion rate. The specific steps for outputting data on the effectiveness of manure management, including carbon and nitrogen emission reduction efficiency and resource conversion rate, are as follows: S51: Collect historical data on manure production; S52: Construct a Bayesian network model, determine the input nodes as diet formulation parameters and rumen microbiota status indicators, and the output nodes as manure production and characteristics. Train the model using historical data and quantify the correlation between various factors using conditional probability tables. S53: Based on the model prediction results and combined with the current status of the manure treatment system, set optimization targets for carbon and nitrogen emission reduction efficiency and resource conversion rate, use manure treatment control parameters as decision variables, construct a multi-objective optimization function, solve it through particle swarm optimization algorithm, and obtain the optimal management strategy parameters; S54: Real-time monitoring of biogas production, ammonia nitrogen residue, organic fertilizer conversion rate and model prediction values ​​are compared to determine whether the difference exceeds the threshold. If so, the model is updated with measured data, the weights are adjusted and the solution is recalculated to obtain data on the effectiveness of manure management. S6: Based on data on manure management effectiveness, rumen greenhouse gas emission control data, and low-carbon dairy cow low-protein diet formulation, construct a comprehensive carbon reduction assessment model to evaluate the effectiveness of farm carbon reduction.

2. The intelligent optimization method for carbon emission reduction in ecological pastures during extreme weather as described in claim 1, characterized in that, In step S1, feature extraction and classification are performed on the pasture weather data: S11: Deploy monitoring nodes and a main control center. Set up meteorological monitoring stations, dairy cow status monitoring equipment, and manure treatment monitoring devices in the ranch, and establish encrypted communication links between each monitoring node and the control center. S12: The main control center verifies the identity of each monitoring node. After successful verification, it assigns initial monitoring tasks, synchronizes basic ranch information, and outputs initial monitoring data.

3. The intelligent optimization method for carbon emission reduction in ecological pastures during extreme weather as described in claim 1, characterized in that, In step S2, the specific steps for preprocessing the initial monitoring data are as follows: S21: Denoise the initial monitoring data by using a Kalman filter algorithm to smooth the meteorological and dairy cow status data, and output the denoised data. S22: Detect missing values ​​in the denoised data, use linear interpolation to fill in discrete missing data, and use cubic spline interpolation based on historical trends to fill in continuous missing data, and output the complete dataset. S23: Standardize the complete dataset, map environmental data to the [0,1] interval, and use Z-score standardization for milk production and milk composition data to eliminate dimensional differences; S24: Use an attention mechanism to construct a multi-dimensional feature vector, dynamically allocate weights according to the degree of impact of extreme weather, and output state features.

4. The intelligent optimization method for carbon emission reduction in ecological pastures during extreme weather as described in claim 3, characterized in that, In step S24, the specific steps for constructing a multi-dimensional feature vector using the attention mechanism are as follows: S241: Based on the standardized complete dataset, construct the feature association matrix, calculate the mutual information values ​​between different features, and output the feature mutual information matrix; S242: Based on the feature mutual information matrix, assign dynamic weights to the motion state, environment perception, and communication quality features, and output a weight vector; S243: Perform a dot product operation between the standardized complete dataset and the weight vector to output the feature vector.

5. The intelligent optimization method for carbon emission reduction in ecological pastures during extreme weather as described in claim 1, characterized in that, In step S33, the specific steps for analyzing the data change trend with extreme weather and additive effects using the LSTM time series prediction model are as follows: S331: Standardize the dynamic monitoring sequence data and input it into the real-time calculation module of the LSTM time series prediction model; S332: The LSTM-based memory cell structure captures the characteristics of data changes over time, distinguishing between the stage affected by extreme weather and the stage after adding different additives; S333: By setting an appropriate time step, the LSTM model can focus on the changes in data within different time spans and analyze whether the data in each stage is rising, falling, or fluctuating steadily. S334: Optimize the fitting of data change trends, accurately extract the specific changes in data under the influence of extreme weather and additives, and obtain the data change trend with the influence of extreme weather and additives.

6. The intelligent optimization method for carbon emission reduction in ecological pastures during extreme weather as described in claim 1, characterized in that, In step S45, the specific steps for calculating the optimal allocation using the mixed-integer linear programming algorithm are as follows: S451: Construct a multi-objective optimization function that minimizes dietary crude protein content, nitrogen excretion, and raw material costs; S452: Establish a dynamic raw material database and use a mixed-integer linear programming algorithm to iteratively solve for the optimal raw material ratio through initialization, range narrowing, and sensitivity analysis; S453: Evaluate the optimal raw material ratio and determine whether it is greater than the threshold. If it is, output the optimal raw material ratio; otherwise, backtrack and adjust.

7. The intelligent optimization method for carbon emission reduction in ecological pastures during extreme weather as described in claim 1, characterized in that, The specific steps for assessing the effectiveness of pasture carbon reduction are as follows: S61: Establish a carbon emission reduction assessment indicator system, including direct emission reduction indicators, indirect emission reduction indicators, and resource utilization indicators; S62: Construct a comprehensive carbon emission reduction assessment model, train the model using historical emission reduction case data, and determine the mapping relationship between each variable and the assessment results; S63: Analyze the mapping relationship between variables and evaluation results in real time using machine learning algorithms to assess the current carbon emission status of the ranch; S64: Determine whether the carbon emission status of the ranch has reached the preset threshold; otherwise, adjust the emission reduction strategy.

Images