Automatic feed production regulation and control method and system

Abstract

The invention relates to the technical field of automatic feed distribution, in particular to an automatic feed production regulation and control method and system.The method comprises the following steps that healthy and growth rate data are collected and input into a data stream, the number of dynamic Bayesian network nodes and a connection mode are adjusted, real-time data are analyzed, and the learning rate and the transition probability are adjusted; and a prediction error matching feedback mechanism is established, key indexes are extracted, a distribution instruction is generated through a fuzzy logic controller, the feed ratio is adjusted to balance cost nutrition and environmental influence, and an optimal configuration result is obtained. According to the method, health and growth data are comprehensively utilized, feed feeding is more personalized and accurate, a feedback mechanism is analyzed and optimized in real time, it is ensured that the feed ratio responds to specific nutritional requirements of animals in real time, accordingly, resource waste is reduced, healthy growth of the animals is enhanced, feeding instructions are finely regulated and controlled through fuzzy logic, and the feeding efficiency is improved. The method adapts to different stages of animal growth, and a particle swarm optimization algorithm is adopted to balance the cost, nutrition and environmental influence.

Description

Technical Field

[0001] The present invention relates to the technical field of automatic feed distribution, and particularly to an automatic feed production regulation method and system. Background Art

[0002] The technical field of automatic feed distribution involves the development and application of various systems and devices to achieve the automatic supply and management of animal feed. These technologies generally include automatic control systems, sensors, computer networks, and software, aiming to optimize the distribution efficiency and accuracy of feed, while reducing the manpower requirements. These systems can automatically adjust the feed delivery according to specific time, quantity, or the needs of animals, ensuring that each animal or group receives appropriate feed according to its growth stage, health status, and nutritional requirements. In addition, the automatic feed distribution technology can also help monitor the storage and consumption of feed, thereby further improving the breeding efficiency and economic benefits.

[0003] Among them, the automatic feed production regulation method refers to controlling and optimizing the processes of feed preparation, storage, and distribution through automatic technologies. The main uses of this method are to improve the accuracy and efficiency of feed use, reduce waste, ensure that animals can obtain appropriate amounts of nutrients as needed, and thus promote their healthy growth. By implementing these automatic solutions, breeders can significantly improve productivity while reducing labor costs and the risk of operation errors.

[0004] The existing technologies mostly rely on fixed feed formulas, lack rapid response to individual differences, often lead to nutritional imbalance and resource waste, show insufficient flexibility in dealing with dynamic changes, and fail to adjust the feed ratio in a timely manner. For example, the failure to adjust the feed of sick and weak animals in a timely manner may delay their recovery and affect the breeding efficiency. In addition, the existing systems do not fully consider the balance between cost and environmental protection, lack comprehensive consideration of environmental impacts, and do not adopt effective environmental sustainability strategies, which is not conducive to resource optimization and environmental protection. Summary of the Invention

[0005] The purpose of the present invention is to solve the disadvantages existing in the prior art, and to propose an automatic feed production regulation method and system.

[0006] To achieve the above purpose, the present invention adopts the following technical solutions: An automatic feed production regulation method includes the following steps:

[0007] S1: Collect the health status data and growth rate data of piglets, input them into the data stream, receive the network structure adjustment instruction from the differential evolution, adjust the number of nodes and connection modes in the dynamic Bayesian network, and generate the network structure optimization result;

[0008] S2: Use the optimized result of the network structure to analyze real-time data, adjust the learning rate and transition probability of the dynamic Bayesian network, set network parameters through quantitative analysis, and perform genetic algorithm optimization of the network, including the selection of data points, crossover of network links, and mutation of random sites, to obtain the parameter optimization result;

[0009] S3: Based on the parameter optimization result, establish a matching feedback mechanism for prediction error and actual feed consumption. By defining the loss function of the prediction error, quantify the difference as a deviation index, and use the deviation to correct the fitness function to obtain the adaptive adjustment feedback;

[0010] S4: Extract key indicators from piglet health and growth data, define the membership function and fuzzy set of the key indicators, convert the input into a feed distribution instruction through a fuzzy logic controller, and use the adaptive adjustment feedback to cyclically optimize the input parameters to generate the fuzzy logic control optimization result;

[0011] S5: Apply the fuzzy logic control optimization result to adjust the feed ratio and feeding plan, balance costs, nutrition, and environmental impacts through a particle swarm optimization algorithm, and adjust the feeding time and amount in real time to obtain the multi-objective optimization configuration result.

[0012] The optimized result of the network structure is specifically the adjusted number of nodes and connection method. The parameter optimization result is specifically the learning rate, transition probability, and genetic algorithm parameters. The adaptive adjustment feedback includes the loss function, deviation index, and fitness function. The fuzzy logic control optimization result is specifically the membership function, fuzzy set, and feed distribution instruction. The multi-objective optimization configuration result includes the feed ratio, feeding time, and feeding amount.

[0013] As a further solution of the present invention, the steps for obtaining the optimized result of the network structure are specifically as follows:

[0014] S111: Collect piglet health status data and growth rate data, use the data to set the initial nodes and edge connections of the dynamic Bayesian network, and generate an initialized network structure;

[0015] S112: Use the differential evolution algorithm to adjust the number of nodes and connection method of the initialized network structure, map the characteristics of piglet health and growth data, and obtain the adjusted network structure;

[0016] S113: Based on the adjusted network structure, perform connection weight optimization, using the formula:

[0017]

[0018] Adjust the weight of each connection to obtain the optimized result of the network structure;

[0019] Among them, W ij represents the weight between node i and node j, and d ij represents the data difference degree between nodes. λ represents the sensitivity adjustment coefficient, which is used to balance the influence of connection weights. is a weight calculation formula based on the Gaussian function, which is used to calculate the weight according to the distance between nodes. The closer the distance, the greater the weight. is the sum of the weights between node i and all other nodes k.

[0020] As a further solution of the present invention, the steps for obtaining the parameter optimization result are specifically as follows:

[0021] S211: Based on the network structure optimization result, analyze the real-time data, adjust the learning rate and transition probability of the dynamic Bayesian network, determine the preliminary adjustment value according to the data deviation and the difference between the expected output, and generate an adjusted learning rate and transition probability scheme;

[0022] S212: Conduct a quantitative analysis on the adjusted learning rate and transition probability scheme, refine the learning rate and transition probability values, and generate a refined network parameter configuration;

[0023] S213: Utilize the refined network parameter configuration to perform genetic algorithm optimization, including selecting matching data points, performing crossover of network connections and mutation of random sites, and using the formula:

[0024]

[0025] Adjust the transition probability to generate the parameter optimization result;

[0026] Among them, P new represents the current transition probability, and P old represents the original transition probability. ΔE represents the energy change caused by parameter adjustment, and T is the control coefficient during the adjustment process, which adjusts the sensitivity of the mutation reaction.

[0027] As a further solution of the present invention, the steps for obtaining the adaptive adjustment feedback are specifically as follows:

[0028] S311: Based on the parameter optimization result, define a matching feedback mechanism between the prediction error and the actual feed consumption, establish a loss function of the prediction error, quantify the deviation between the prediction and the actual, and generate a preliminary deviation quantification index;

[0029] S312: Utilize the deviation quantification index to convert the prediction error into a deviation index, and generate a deviation value by comparing and calculating the difference between the actual data and the prediction data;

[0030] S313: According to the deviation value, correct the adaptive function, and use the formula:

[0031]

[0032] Adjust the adaptability function to obtain the adaptability adjustment feedback;

[0033] Among them, F new represents the current adaptability function, F old represents the original adaptability function, D is the deviation value calculated from the actual data and the predicted data, and β is an adjustment coefficient used to control the sensitivity of the deviation impact.

[0034] As a further solution of the present invention, the steps for obtaining the fuzzy logic control optimization result are specifically as follows:

[0035] S411: Extract key indicators from the piglet health and growth data, define the membership functions and fuzzy sets of these indicators, provide basic data and calculation models for subsequent fuzzy logic control, and generate the membership function and fuzzy set configuration;

[0036] S412: Apply the membership function and fuzzy set to the fuzzy logic controller, convert the input data into feed distribution instructions, optimize each feed distribution based on the current health and growth data, and generate a preliminary feed distribution plan;

[0037] S413: Apply the feed distribution instruction and the adaptability adjustment feedback, refine the input parameters, adjust the control parameters, and use the formula:

[0038]

[0039] Generate the fuzzy logic control optimization result;

[0040] Among them, P optimized is the optimized control parameter, P current is the current parameter setting, ΔP is the parameter increment based on the deviation, E target and E current are the target and current errors respectively, and γ is the sensitivity adjustment coefficient.

[0041] As a further solution of the present invention, the steps for obtaining the multi-objective optimization configuration result are specifically as follows:

[0042] S511: Apply the fuzzy logic control optimization result, adjust the feed ratio and feeding plan, set parameters for real-time adjustment of the feeding time and amount, and generate a preliminary adjustment plan;

[0043] S512: Refine the preliminary adjustment plan through the particle swarm optimization algorithm, balance the cost, nutrition and environmental impact, optimize the algorithm parameter settings, and generate an optimized feed ratio plan;

[0044] S513: In combination with the optimized feed ratio plan, adjust the feeding time and amount in real time, using the formula:

[0045]

[0046] Optimize the feeding plan to obtain the multi-objective optimization configuration result;

[0047] where T new represents the current feeding time, T old represents the original planned feeding time, ΔC is the cost change amount, which is used to adjust the feeding time to balance the cost efficiency, and C max and C min are the upper and lower limits of the cost, ensuring the economic feasibility of the feeding plan.

[0048] An automated feed production control system, which is used to execute the above-mentioned automated feed production control method. The system includes:

[0049] The data acquisition module collects the health status data and growth rate data of piglets, inputs the data stream, and adjusts the number of nodes and connection methods in the dynamic Bayesian network to obtain the network structure optimization index;

[0050] The network optimization module adjusts the learning rate and transition probability of the dynamic Bayesian network based on the network structure optimization index, and performs selection, crossover, and mutation operations to generate the parameter optimization result;

[0051] The parameter adjustment module establishes a matching mechanism between the prediction error and the actual feed consumption based on the parameter optimization result, and modifies the adaptation function by setting the loss function to generate the adjustment feedback result;

[0052] The prediction feedback module extracts key indicators from the health and growth data of piglets based on the adjustment feedback result, uses fuzzy logic processing to convert the key indicators into feed distribution instructions, and obtains the fuzzy logic control optimization result;

[0053] The fuzzy logic module circularly optimizes the feed ratio and feeding plan based on the fuzzy logic control optimization result, adjusts the input parameters, and generates the adaptive feed formulation plan;

[0054] The multi-objective optimization module adjusts the feeding time and amount based on the adaptive feed formulation plan by applying the particle swarm optimization algorithm to balance the cost, nutrition, and environmental impact, and obtains the feeding optimization configuration.

[0055] Compared with the prior art, the advantages and positive effects of the present invention are:

[0056] In the present invention, by comprehensively utilizing health and growth data, feed feeding is made more personalized and precise. The real-time analysis and optimization feedback mechanism ensures that the feed ratio responds in real time to the specific nutritional needs of animals, thereby reducing resource waste and enhancing the healthy growth of animals. Fuzzy logic is used to finely regulate feeding instructions to adapt to different stages of animal growth. The particle swarm optimization algorithm is used to balance costs, nutrition, and environmental impacts, effectively promoting the economic and environmental sustainable development of the aquaculture industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 It is a schematic diagram of the working process of the present invention;

[0058] Figure 2 It is a flowchart of the steps for obtaining the optimization result of the network structure of the present invention;

[0059] Figure 3 It is a flowchart of the steps for obtaining the optimization result of the parameters of the present invention;

[0060] Figure 4 It is a flowchart of the steps for obtaining the adaptive adjustment feedback of the present invention;

[0061] Figure 5 It is a flowchart of the steps for obtaining the optimization result of the fuzzy logic control of the present invention;

[0062] Figure 6 It is a flowchart of the steps for obtaining the optimization result of the multi-objective optimization configuration of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0063] In order to make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0064] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the accompanying drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. In addition, in the description of the present invention, the meaning of "a plurality" is two or more, unless otherwise specifically defined.

[0065] Embodiment 1

[0066] Please refer to Figure 1 , the present invention provides a technical solution: an automated feed production regulation method, including the following steps:

[0067] S1: Collect the data on the health status and growth rate of piglets, input the data stream, receive the network structure adjustment instruction from differential evolution, adjust the number of nodes and connection mode in the dynamic Bayesian network, and generate the network structure optimization result;

[0068] S2: Use the network structure optimization result to analyze the real-time data, adjust the learning rate and transition probability of the dynamic Bayesian network, set the network parameters by quantitative analysis, and perform the genetic algorithm optimization of the network, including the selection of data points, the crossover of network links, and the mutation of random sites, to obtain the parameter optimization result;

[0069] S3: Based on the parameter optimization result, establish a matching feedback mechanism for prediction error and actual feed consumption. By defining the loss function of the prediction error, quantify the difference into a deviation index, and use the deviation to correct the fitness function to obtain the adaptive adjustment feedback;

[0070] S4: Extract the key indicators from the health and growth data of piglets, define the membership function and fuzzy set of the key indicators, convert the input into a feed distribution instruction through a fuzzy logic controller, and use the adaptive adjustment feedback to cyclically optimize the input parameters to generate the fuzzy logic control optimization result;

[0071] S5: Apply the fuzzy logic control optimization result to adjust the feed ratio and feeding plan, balance the cost, nutrition, and environmental impact through the particle swarm optimization algorithm, and adjust the feeding time and amount in real time to obtain the multi-objective optimization configuration result.

[0072] The network structure optimization result specifically includes the adjusted number of nodes and connection mode. The parameter optimization result specifically includes the learning rate, transition probability, and genetic algorithm parameters. The adaptive adjustment feedback includes the loss function, deviation index, and fitness function. The fuzzy logic control optimization result specifically includes the membership function, fuzzy set, and feed distribution instruction. The multi-objective optimization configuration result includes the feed ratio, feeding time, and feeding amount.

[0073] Please refer to Figure 2 , and the specific steps for obtaining the network structure optimization result are as follows:

[0074] S111: Collect the data on the health status and growth rate of piglets, use the data to set the initial nodes and edge connections of the dynamic Bayesian network, and generate the initialized network structure;

[0075] When collecting data on piglet health status and growth rate, it is first necessary to configure sensing devices in different breeding environments in order to obtain multi-dimensional data on piglet weight, body temperature, and activity frequency. The data is transmitted to the central processing system, and data preprocessing algorithms are used to exclude noise interference and outliers to ensure the accuracy of subsequent data analysis. Through this process, the basic settings of the initial nodes and edge connections of the dynamic Bayesian network are carried out, providing data support for network structure optimization.

[0076] S112: Use the differential evolution algorithm to adjust the number of nodes and connection methods of the initialized network structure, map the characteristics of piglet health and growth data, and obtain the adjusted network structure;

[0077] Based on the initialized network structure, the differential evolution algorithm starts iterative calculation. First, evaluate the adaptability of the existing network structure, quantify the fitting degree of each network configuration to the data pattern through the matching degree function. Then, according to the matching degree results, select the network structure with a high matching degree as the parent generation, and generate the current network configuration through crossover and mutation operations. This process is repeated until the network structure that best suits the characteristics of piglet health status and growth data is captured, completing the key adjustment of the network and setting the basis for weight optimization.

[0078] S113: Based on the adjusted network structure, perform connection weight optimization, using the formula:

[0079]

[0080] Adjust the weight of each connection to obtain the optimized result of the network structure;

[0081] Among them, W ij represents the weight between node i and node j, d ij represents the data difference degree between nodes, and λ represents the sensitivity adjustment coefficient, which is used to balance the influence of connection weights. is a weight calculation formula based on the Gaussian function, which is used to calculate the weight according to the distance between nodes. The closer the distance, the greater the weight. is the sum of the weights between node i and all other nodes k.

[0082] Formula:

[0083]

[0084] The benefit of the formula is to adjust the network connection in the form of exponential weights, enhance the sensitivity of the model to key data features, and thus improve the accuracy of the overall network prediction.

[0085] Detailed explanation of the formula and the derivation process of formula calculation:

[0086] Let d ij= 5, λ = 2, and the calculation steps are as follows:

[0087] First, calculate

[0088]

[0089] Then, calculate the sum of other terms in the denominator and set it as So

[0090]

[0091] The result shows that the connection weight between node i and node j is 0.38, which reflects the influence degree of the data difference between the two nodes and guides the optimization of the network structure.

[0092] Please refer to Figure 3 , and the steps to obtain the parameter optimization results are specifically as follows:

[0093] S211: Based on the network structure optimization results, analyze the real-time data, adjust the learning rate and transition probability of the dynamic Bayesian network, determine the preliminary adjustment values according to the data deviation and the difference between the expected output, and generate a scheme for the adjusted learning rate and transition probability;

[0094] In the initial stage of network optimization, analyze the real-time data based on the network structure optimization results to adjust the learning rate and transition probability. This process includes the real-time collection and preprocessing of data. The processed data will be used to evaluate the effectiveness of the current network configuration, and the network parameters will be adjusted based on these evaluation results to better match the changing data characteristics. The adjustment strategy is based on a difference feedback loop, where the processing of each data point is associated with its impact on the network performance. The determined learning rate and transition probability will directly affect the adaptability and prediction accuracy of the network.

[0095] S212: Conduct a quantitative analysis on the adjusted learning rate and transition probability scheme, refine the values of the learning rate and transition probability, and generate a refined network parameter configuration;

[0096] Refine the settings of the learning rate and transition probability through quantitative analysis. The operations included in this step include calculating the deviation between the expected network output and the actual output theoretically, and adjusting the network parameters according to these deviations. This quantitative method relies on statistical algorithms to ensure that each adjustment is based on sufficient data support. The refined parameter configuration is tested through multiple iterations to verify its stability and accuracy under various operating conditions. The generated refined network parameter configuration aims to minimize the error and ensure the strong matching ability of the network when facing unknown data.

[0097] S213: Utilize the refined network parameter configuration to perform genetic algorithm optimization, including selecting matching data points, performing crossover of network connections, and mutation at random sites, using the formula:

[0098]

[0099] Adjust the transition probability to generate the parameter optimization result;

[0100] Among them, P new represents the current transition probability, P old represents the original transition probability, ΔE represents the energy change caused by parameter adjustment, and T is the control coefficient in the adjustment process, which adjusts the sensitivity of the mutation reaction.

[0101] Formula:

[0102]

[0103] The advantage of the formula is that by introducing the temperature parameter T and the energy difference ΔE to adjust the probability of network state transition, it increases the model's ability to match environmental changes, making the network optimization process closer to the physical annealing process, so as to obtain better results in global optimization.

[0104] Detailed explanation of the formula and the derivation process of formula calculation:

[0105] Set P old = 0.4, ΔE = 1, T = 300, then the formula becomes:

[0106]

[0107] Calculate the approximate value of e -1 / 300 :

[0108] e -1 / 300 ≈0.9967

[0109] Substitute into the formula:

[0110]

[0111] The result shows that the current transition probability P new is approximately 0.2, which means that the probability of network state transition is low under the current parameters and environment, indicating an enhanced stability after network parameter adjustment.

[0112] Please refer to Figure 4 , and the specific steps for obtaining the adaptive adjustment feedback are as follows:

[0113] S311: Based on the parameter optimization result, define the matching feedback mechanism between the prediction error and the actual feed consumption, establish the loss function of the prediction error, quantify the deviation between the prediction and the actual, and generate the preliminary deviation quantification index;

[0114] When defining the loss function for predicting errors, it is first necessary to collect actual feed consumption data and predicted data. These data are collected by feed consumption monitoring devices within different time periods, and each data point is the real-time recorded feed consumption. Through these data, the difference between the daily average feed consumption and the predicted value can be calculated. By this method, the prediction error can be accurately quantified, and the result of this quantification helps to adjust and optimize the prediction model to reduce the prediction error in future time periods.

[0115] S312: Use the deviation quantification index to convert the prediction error into a deviation index. By comparing and calculating the difference between the actual data and the predicted data, a deviation value is generated.

[0116] In the process of calculating the deviation value between the actual data and the predicted data, first compare the actual feed consumption data with the predicted data. The comparison of each data point is based on the difference between the actual consumption and the predicted result at the same time point. By statistically analyzing these differences, a total deviation value is obtained. This total deviation value is a key indicator to measure the accuracy of the prediction model. By this method, determine which factors cause the prediction error and adjust the model parameters accordingly.

[0117] S313: According to the deviation value, correct the fitness function, using the formula:

[0118]

[0119] Adjust the fitness function to obtain the fitness adjustment feedback.

[0120] Among them, F new represents the current fitness function, F old represents the original fitness function, D is the deviation value calculated from the actual data and the predicted data, and β is a regulation coefficient used to control the sensitivity of the deviation impact.

[0121] Formula:

[0122]

[0123] The advantage of the formula is that by introducing the deviation index and the regulation coefficient, the model is allowed to dynamically adjust the fitness function, so as to more flexibly reflect the relationship between the actual data and the prediction, and improve the matching ability and prediction accuracy of the model.

[0124] Detailed explanation of the formula and the derivation process of the formula calculation:

[0125] Set F old = 0.85 as the original fitness function value. Based on the comparison between the actual data and the predicted data, set the deviation index D = 10 and the regulation coefficient β = 0.5. In the formula, F newRepresents the adjusted adaptation function, calculated by inserting values:

[0126]

[0127] The results show that due to the large deviation, the adaptation function is significantly reduced, indicating that the model needs to be adjusted to reduce future prediction errors and enhance the model's adaptability to the actual situation.

[0128] Please refer to Figure 5 , and the specific steps for obtaining the optimization results of fuzzy logic control are as follows:

[0129] S411: Extract key indicators from the piglet health and growth data, define the membership functions and fuzzy sets of these indicators, provide basic data and calculation models for subsequent fuzzy logic control, and generate membership function and fuzzy set configurations;

[0130] Extract key indicators from the piglet health and growth data. The indicators reflect the overall health status and growth rate of the piglets. This process involves data screening and analysis to ensure that the selected indicators can effectively predict the feeding effect and provide real-time data for the fuzzy logic control system. Such analysis is not only based on previous data but also combines the current research on animal physiology.

[0131] The key indicators are extracted from the piglet health and growth data. By using data analysis techniques to determine the key components in the data, and by analyzing the daily weight gain data of the piglets and combining the feed intake, statistical methods are used to calculate the impact degree of each feed type on weight gain. These indicators serve as the basis for the membership function and fuzzy set, providing the necessary data support for the subsequent configuration of the fuzzy logic controller and ensuring that the logic of feed distribution is consistent with the actual growth data.

[0132] S412: Apply the membership function and fuzzy set to the fuzzy logic controller, convert the input data into feed distribution instructions, optimize each feed distribution based on the current health and growth data, and generate a preliminary feed distribution plan;

[0133] Based on the configuration results of the fuzzy set and membership function, define the control rules of fuzzy logic according to the differential performance of the growth indicators. These rules convert the growth data into corresponding feed distribution instructions by inputting them into the fuzzy logic controller. Through the algorithm of the logic controller, the input parameters of growth rate and health status are converted into the output of feed quantity, ensuring that the feed quantity allocated each time can maximize the support for the health and growth of the piglets.

[0134] S413: Apply the feed distribution instructions and adaptive adjustment feedback, refine the input parameters, adjust the control parameters, using the formula:

[0135]

[0136] Generate the optimized results of fuzzy logic control;

[0137] where P optimized is the optimized control parameter, P current is the current parameter setting, ΔP is the parameter increment based on the deviation, and E target and E current are the target and current errors respectively, and γ is the sensitivity adjustment coefficient.

[0138] Formula:

[0139]

[0140] The benefit of the formula is that it dynamically adjusts the feed distribution strategy to match the real-time changes in the health and growth data of piglets. By adjusting the coefficient γ, it can flexibly respond to the changing feed requirements of piglets, ensuring the timeliness and accuracy of feed distribution.

[0141] Detailed explanation of the formula and the derivation process of formula calculation:

[0142] Set the current parameter P current to 10 units, the parameter increment ΔP is planned to be 2 units, the target error E target is set to 1.5, the current error E current is 2.0, and the adjustment coefficient γ is set to 0.5. Then the formula calculation process is as follows:

[0143]

[0144] The result shows that the optimized control parameter is 6 units, which is less than the original parameter. This indicates that under the current error condition, it is necessary to reduce the feed distribution amount to match the current demand of piglets, and respond to the changes in growth data through real-time adjustment strategies.

[0145] Please refer to Figure 6 for the specific steps to obtain the multi-objective optimization configuration results:

[0146] S511: Apply the optimized results of fuzzy logic control to adjust the feed ratio and feeding plan, set parameters for real-time adjustment of feeding time and amount, and generate a preliminary adjustment plan;

[0147] Apply the optimized results of fuzzy logic control. By analyzing the current feed ratio data and animal growth model, adjust the feeding plan to ensure that each feed delivery meets the current nutritional requirements and cost-benefit analysis. Through this process, continuously update and optimize the feed ratio to make it more accurately meet the growth requirements and health indicators of piglets. While adjusting the ratio, monitor the changes in environmental factors including temperature and humidity, and feedback the data to the feeding system in real time to dynamically adjust the feeding time and amount to ensure the optimal use efficiency and growth effect of feed.

[0148] S512: Refine the preliminary adjustment plan through the particle swarm optimization algorithm, balance costs, nutrition, and environmental impacts, optimize the algorithm parameter settings, and generate an optimized feed formulation plan;

[0149] Balance the cost and nutrition requirements through the particle swarm optimization algorithm. For the price fluctuations of feed raw materials and the seasonal changes in nutritional components on the current market, conduct a cost-benefit analysis, combine real-time growth data and environmental monitoring data, and optimize the settings of feeding time and amount to achieve the optimal economic benefits and growth efficiency. By simulating different feed formulation plans and feeding schedules, evaluate their impacts on the growth rate and health status of piglets, so as to minimize the environmental burden while ensuring nutritional supply.

[0150] S513: Combine the optimized feed formulation plan, and adjust the feeding time and amount in real time, using the formula:

[0151]

[0152] Optimize the feeding plan to obtain the multi-objective optimization configuration result;

[0153] Among them, T new represents the current feeding time, T old represents the originally planned feeding time, ΔC is the cost change amount, used to adjust the feeding time to balance the cost efficiency, C max and C min are the upper and lower limits of the cost, ensuring the economic feasibility of the feeding plan.

[0154] Formula:

[0155]

[0156] The advantage of the formula is that by adjusting the feeding time, it directly reflects the real-time impact of cost changes on the feeding strategy, optimizes resource allocation, and improves cost-benefit.

[0157] Detailed explanation of the formula and the formula calculation derivation process:

[0158] Set the initial feeding time T old to 8 hours, the cost adjustment factor ΔC is 0.05 (5% cost increase), the cost maximum value C max is 1.0, the minimum value C min is 0.0, insert into the formula for calculation to get:

[0159]

[0160] The results show that the current feeding time needs to be increased from the original 8 hours to 8.4 hours to match the economic pressure brought about by the rising cost. The adjusted feeding time helps to maintain the balance between feeding cost and nutritional benefits, ensuring the healthy growth of animals and maximizing economic benefits.

[0161] An automated feed production regulation system, which is used to execute the above-mentioned automated feed production regulation method. The system includes:

[0162] The data acquisition module collects the health status data and growth rate data of piglets, inputs the data stream, adjusts the number of nodes and connection methods in the dynamic Bayesian network according to the data stream, and obtains the network structure optimization index.

[0163] The network optimization module adjusts the learning rate and transition probability of the dynamic Bayesian network based on the network structure optimization index, performs selection, crossover, and mutation operations, and generates the parameter optimization result.

[0164] The parameter adjustment module establishes a matching mechanism between the prediction error and the actual feed consumption based on the parameter optimization result, modifies the adaptation function by setting the loss function, and generates the adjustment feedback result.

[0165] The prediction feedback module extracts key indicators from the health and growth data of piglets based on the adjustment feedback result, uses fuzzy logic processing to convert the key indicators into feed distribution instructions, and obtains the fuzzy logic control optimization result.

[0166] The fuzzy logic module cyclically optimizes the feed ratio and feeding plan based on the fuzzy logic control optimization result, adjusts the input parameters, and generates an adaptive feed formulation plan.

[0167] The multi-objective optimization module adjusts the feeding time and amount based on the adaptive feed formulation plan by applying the particle swarm optimization algorithm, balances the cost, nutrition, and environmental impact, and obtains the feeding optimization configuration.

[0168] The above is only the preferred embodiment of the present invention, and does not limit the present invention in other forms. Any person skilled in the art may use the disclosed technical content to make changes or modifications into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change, and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical content of the present invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. An automated feed production control method, characterized in that: The following steps are involved: Collect data on piglet health and growth rate, input data streams, receive network structure adjustment instructions from differential evolution, adjust the number of nodes and connection methods in the dynamic Bayesian network, and generate network structure optimization results; Using the network structure optimization results, analyzing real-time data, adjusting the learning rate and transition probability of the dynamic Bayesian network, using quantitative analysis to set network parameters, performing genetic algorithm optimization of the network, including data point selection, network link crossover and random site mutation, and obtaining parameter optimization results; Based on the parameter optimization results, a matching feedback mechanism between the prediction error and the actual feed consumption is established, the difference is quantified as a deviation index by defining a loss function of the prediction error, and the deviation is used to correct the fitness function to obtain adaptive adjustment feedback; Extract key indicators from piglet health and growth data, define membership functions and fuzzy sets of key indicators, convert inputs into feed distribution instructions through a fuzzy logic controller, use the adaptive adjustment feedback to cyclically optimize input parameters, and generate fuzzy logic control optimization results; The fuzzy logic control optimization results are applied to adjust the feed ratio and feeding plan, balance the cost, nutrition and environmental impact through the particle swarm optimization algorithm, adjust the feeding time and amount in real time, and obtain the multi-objective optimization configuration results.

2. The automated feed production control method according to claim 1, characterized in that: The network structure optimization result specifically includes the number of nodes and connection mode after adjustment; the parameter optimization result specifically includes the learning rate, transition probability, and genetic algorithm parameters; the adaptive adjustment feedback includes loss function, deviation index, and adaptability function; the fuzzy logic control optimization result specifically includes membership function, fuzzy set, and feed distribution instruction; the multi-objective optimization configuration result includes feed ratio, feeding time, and feeding amount.

3. The automated feed production control method according to claim 2, characterized in that: The steps for obtaining the network structure optimization result are specifically as follows: Collect data on the health status and growth rate of piglets, use the data to set the initial nodes and edge connections of the dynamic Bayesian network, and generate an initialized network structure; Using a differential evolution algorithm to adjust the number of nodes and connection mode of the initialized network structure, mapping the health and growth data characteristics of the piglets, and obtaining an adjusted network structure; Based on the adjusted network structure, the connection weights are optimized using the formula: Adjust the weight of each connection to obtain the network structure optimization result; Among them, W ij represents the weight of node i and node j, d ij represents the data difference between nodes, and λ represents the sensitivity adjustment coefficient, which is used to balance the influence of connection weights. It is a weight calculation formula based on Gaussian function, which is used to calculate the weight according to the distance between nodes. The closer the distance, the greater the weight. is the sum of the weights between node i and all other nodes k.

4. The automated feed production control method according to claim 3, characterized in that: The steps for obtaining the parameter optimization results are specifically as follows: Based on the network structure optimization results, real-time data is analyzed, the learning rate and transition probability of the dynamic Bayesian network are adjusted, preliminary adjustment values ​​are determined according to data deviations and expected output differences, and an adjusted learning rate and transition probability scheme is generated; Quantitatively analyzing the adjusted learning rate and transition probability scheme, refining the learning rate and transition probability values, and generating a refined network parameter configuration; Using the refined network parameter configuration, a genetic algorithm optimization is performed, including selecting matching data points, performing crossover of network connections and mutation of random sites, using the formula: Adjust the transition probability and generate parameter optimization results; Among them, P new represents the current transition probability, P old represents the original transfer probability, ΔE represents the energy change caused by parameter adjustment, and T is the control coefficient in the adjustment process, which adjusts the sensitivity of the variation response.

5. The automated feed production control method according to claim 4, characterized in that: The steps of obtaining the adaptive adjustment feedback are specifically as follows: Based on the parameter optimization results, a matching feedback mechanism between the prediction error and the actual feed consumption is defined, a loss function of the prediction error is established, the deviation between the prediction and the actual is quantified, and a preliminary deviation quantification index is generated; The prediction error is converted into a deviation index by using the deviation quantification index, and a deviation value is generated by comparing and calculating the difference between the actual data and the predicted data; According to the deviation value, the adaptability function is modified using the formula: Adjust the adaptability function to obtain adaptive adjustment feedback; Among them, F new Represents the current fitness function, F old represents the original adaptive function, D is the deviation value calculated from the actual data and the predicted data, and β is an adjustment coefficient used to control the sensitivity of the deviation impact.

6. The automated feed production control method according to claim 5, characterized in that: The steps for obtaining the fuzzy logic control optimization result are specifically as follows: Extract key indicators from piglet health and growth data, define the membership functions and fuzzy sets of these indicators, provide basic data and calculation models for subsequent fuzzy logic control, and generate membership functions and fuzzy set configurations; Applying the membership functions and fuzzy sets to a fuzzy logic controller, converting input data into feed allocation instructions, optimizing each feed allocation based on current health and growth data, and generating a preliminary feed allocation plan; Apply the feed distribution instruction and the adaptive adjustment feedback, refine the input parameters, adjust the control parameters, and use the formula: Generate fuzzy logic control optimization results; Among them, P optimized is the optimized control parameter, P current is the current parameter setting, ΔP is the parameter increment based on the deviation, and E target 、E current are the target and current errors respectively, and γ is the sensitivity adjustment coefficient.

7. The automated feed production control method according to claim 6, characterized in that: The steps for obtaining the multi-objective optimization configuration result are specifically as follows: Apply the fuzzy logic control optimization results to adjust the feed ratio and feeding plan, set parameters for real-time adjustment of feeding time and amount, and generate a preliminary adjustment plan; Refining the preliminary adjustment plan through a particle swarm optimization algorithm, balancing cost, nutrition and environmental impact, optimizing algorithm parameter settings, and generating an optimized feed ratio plan; Combined with the above-mentioned optimized feed ratio scheme, the feeding time and amount are adjusted in real time, using the formula: Optimize the feeding plan and obtain multi-objective optimization configuration results; Among them, T new Represents the current feeding time, T old represents the original planned feeding time, ΔC is the cost change, which is used to adjust the feeding time to balance cost efficiency, and C max , C min It is the upper and lower limits of cost, ensuring the economic feasibility of the feeding plan.

8. An automated feed production control system, characterized in that: According to any one of claims 1 to 7, the automated feed production control method comprises: The data collection module collects the health status data and growth rate data of the piglets, inputs the data stream, adjusts the number of nodes and connection mode in the dynamic Bayesian network according to the data stream, and obtains the network structure optimization index; The network optimization module adjusts the learning rate and transition probability of the dynamic Bayesian network based on the network structure optimization index, performs selection, crossover and mutation operations, and generates parameter optimization results; The parameter adjustment module establishes a matching mechanism between the prediction error and the actual feed consumption based on the parameter optimization result, and generates an adjustment feedback result by correcting the adaptability function by setting a loss function; The prediction feedback module extracts key indicators from the health and growth data of the piglets based on the adjustment feedback results, uses fuzzy logic processing to convert the key indicators into feed distribution instructions, and obtains fuzzy logic control optimization results; The fuzzy logic module cyclically optimizes the feed ratio and feeding plan based on the fuzzy logic control optimization result, adjusts the input parameters, and generates an adaptive feed preparation plan; The multi-objective optimization module uses a particle swarm optimization algorithm to adjust the feeding time and amount based on the adaptive feed formulation scheme, balance the cost, nutrition and environmental impact, and obtain the optimal feeding configuration.

Images