Livestock feed intelligent quantitative proportioning management system based on big data
Patent Information
- Application Number
- CN202610739118.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-05-27
- Publication Date
- 2026-08-18
AI Technical Summary
饲料配方生成多以人工调配为主，未基于精细化特征构建多目标营养需求模型，配方筛选未在饲料原料空间中开展自动化优化计算
[0016]与现有技术相比，本发明的优点和积极效果在于：
Smart Images
Figure CN122597100A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of big data management technology for livestock farming, and in particular to an intelligent quantitative proportioning management system for livestock feed based on big data. Background Technology
[0002] Current livestock feed formulation management relies heavily on manual experience, with livestock information collection primarily based on scattered records. There is a lack of multi-dimensional information integration considering livestock breed, growth stage, body condition, and historical feed intake. This results in inconsistent data dimensions, missing information, and the absence of standardized livestock records. Feed nutrient requirement calculations often use fixed-stage standards without matching them to standard growth cycle maps. Body condition characteristics are simply recorded without physiological parameterization, and historical feed intake is only statistically analyzed without periodic breakdown. Feed formulation generation is largely manual, lacking the development of multi-objective nutrient requirement models based on refined characteristics. Formulation selection also lacks automated optimization calculations within the feed ingredient space.
[0003] Current methods for collecting and processing livestock information cannot guarantee the completeness and consistency of data; information bias and omissions directly affect the accuracy of livestock condition assessment. Fixed-standard nutrient calculation methods cannot match the individual growth status and feeding habits of livestock, resulting in discrepancies between calculated nutrient requirements and actual conditions. Manual feed formulation methods struggle to balance the dual constraints of nutrient requirements and raw material composition, failing to achieve precise and intelligent feed ratios.
[0004] To address the issues of disorganized and incomplete information dimensions in livestock farming, it is necessary to align multi-dimensional information, fill in missing values, and form structured farming records. To address the problem of inaccurate judgment of growth stages, body condition, and feeding patterns, it is necessary to complete the matching of standard growth cycle maps, the parameterization of body condition and physiological parameters, the periodic decomposition of feed intake time series, and to achieve automated ratio optimization of feed ingredients within the space by relying on multi-objective nutritional requirement models and formula optimization engines. Summary of the Invention
[0005] The purpose of this invention is to address the shortcomings of existing technologies by proposing a smart quantitative feed formulation management system for livestock based on big data.
[0006] To achieve the above objectives, the present invention adopts the following technical solution: a big data-based intelligent quantitative proportioning management system for livestock feed, comprising: The livestock breeding information processing module collects the original breeding information of the target livestock and performs multi-dimensional information alignment and missing value filling to generate a structured breeding archive containing livestock breed identification, growth stage label, body condition characteristic sequence and historical feed intake time series. The feature analysis module performs standard growth cycle map matching on the growth stage labels to obtain growth status assessment results, performs physiological parameterization processing on the body condition feature sequence to obtain standardized body condition feature vectors, and performs periodic decomposition on the historical feed intake time series to obtain feed intake pattern features. The nutrition requirement modeling module obtains nutrient composition data of feed ingredients from a feed ingredient database, which contains nutrient composition data of various feed ingredients. Based on the growth status assessment results, the standardized body condition feature vector, the feeding pattern characteristics, and the nutrient composition data, a multi-objective nutrition requirement model is constructed, and the target nutrition requirement spectrum is calculated through the multi-objective nutrition requirement model. The intelligent formula generation module takes the target nutrient requirement spectrum as the optimization target and the nutrient component data as the constraint basis. It performs optimization calculations in the feed raw material space through the formula optimization engine to generate a set of candidate formulas containing specific raw material ratios.
[0007] As a further aspect of the present invention, the growth stage labels are matched with standard growth cycle maps to obtain growth status assessment results, including: Obtain a pre-established standard growth cycle map corresponding to the livestock breed identifier. The standard growth cycle map defines the standard growth stage division and corresponding standard physiological parameter range of the livestock breed corresponding to the livestock breed identifier at different days or months. The growth stage labels are mapped to the standard growth cycle map to determine the current standard growth stage of the target livestock. Compare the actual body condition data in the body condition feature sequence with the standard physiological parameter range corresponding to the standard growth stage, and calculate the deviation between the actual growth level and the standard growth level. By combining the deviation with the standard growth stage, a growth status assessment result is generated that characterizes the position and progress of the target livestock's actual growth status relative to the standard status.
[0008] As a further aspect of the present invention, based on the growth status assessment results, the standardized body condition feature vector, the feeding pattern characteristics, and the nutrient component data, a multi-objective nutritional requirement model is constructed, and a target nutritional requirement spectrum is calculated using the multi-objective nutritional requirement model, including: In the multi-objective nutrient requirement model, basic nutrient requirement targets are determined based on the growth status assessment results. The basic nutrient requirement targets include at least maintenance requirements, growth requirements, and reproductive requirements. Based on the standardized body condition feature vector, a body condition adjustment factor is introduced onto the basic nutritional requirement target to make a weighted adjustment to the basic nutritional requirement target; Based on the feeding pattern characteristics, the expected total feed intake in the future stage is predicted. The weighted adjusted basic nutrient requirements target is converted into concentration based on the expected total feed intake to obtain the concentration target of various nutrients required per unit feed. The concentration targets for all nutrients are summarized, and a final balance is achieved by combining synergistic and antagonistic relationships among nutrients, forming a comprehensive spectrum of the target nutrient requirements.
[0009] As a further aspect of the present invention, using the target nutrient requirement spectrum as the optimization objective and the nutrient composition data as the constraint basis, an optimization calculation is performed in the feed ingredient space through a formula optimization engine to generate a candidate formula set containing specific ingredient ratios, including: A solution space is defined in the formula optimization engine, where each point represents a formula composed of different feed ingredients in proportion. A fitness function is constructed in the solution space, which is used to measure the degree to which any formulation meets the target nutritional requirement spectrum; Constraints are set in the solution space, including: total nutrient content range constraints calculated based on the nutrient composition data, upper and lower limits of single raw material usage, and incompatibilities between raw materials. The optimal value of the fitness function is searched within the region that satisfies the constraints using an optimization algorithm, and the multiple high-performance formulation points evaluated during the search process are used as the candidate formulation set.
[0010] As a further aspect of the present invention, the system further includes: The intelligent decision-making module performs cost assessment and process feasibility verification on each candidate formula in the candidate formula set, selects the optimal feed formula that meets the cost and process constraints, and converts the optimal feed formula into an executable proportioning instruction for the feed preparation equipment. The executable proportioning instruction includes the amount of each feed ingredient to be fed. The execution control module sends the executable proportioning instruction to the feed preparation equipment and receives the feeding confirmation information returned by the feed preparation equipment; For each candidate formulation in the candidate formulation set, a cost assessment and process feasibility verification are performed, specifically including: Obtain current market feed ingredient price information, calculate the unit weight raw material cost of each candidate formula based on the proportion of various feed ingredients in each candidate formula, and complete the cost assessment. Obtain the process parameters of the feed preparation equipment, including requirements for particle size, mixing uniformity, and pelleting conditions; Verify whether the physical properties of each candidate formulation meet the process parameters of the feed preparation equipment. The physical properties include the particle size distribution, mixing properties, and binding properties of the raw materials. Record the results of each candidate formulation's cost assessment and process feasibility verification, including whether it passed and the specific reasons for failure.
[0011] As a further aspect of the present invention, the step of screening for the optimal feed formulation that satisfies cost and process constraints includes: A preset cost threshold and a list of process constraints, wherein the list of process constraints contains all process parameter conditions that must be met; From the candidate formula set, candidate formulas whose unit weight raw material cost exceeds the cost threshold are removed to obtain a cost-qualified formula subset. From the cost-qualified formula subset, candidate formulas that fail to fully meet all conditions in the process constraint list are removed, resulting in a process-qualified formula subset. From the set of qualified feed formulations, the candidate formulation with the highest fitness function value is selected and determined as the optimal feed formulation.
[0012] As a further aspect of the present invention, the optimal feed formulation is converted into an executable proportioning instruction for the feed preparation equipment, including: The optimal feed formula is analyzed to extract the type identifier and corresponding mass ratio of each feed ingredient; Based on the total mass requirement of the target batch of feed to be produced, the mass ratio is converted into the specific feed mass required for each feed ingredient. Based on the instruction format identified by the feed preparation equipment, the type identifier of each feed ingredient and its corresponding specific feeding quantity are encoded into a series of sequentially executed control commands, forming the executable proportioning instruction.
[0013] As a further aspect of the present invention, the executable proportioning instruction further includes a raw material addition sequence, wherein the method for determining the raw material addition sequence includes: The priority of adding ingredients is determined based on their physical form, addition ratio, and interaction characteristics during the mixing process. For trace components that require premixing, a premixing instruction is generated to mix them first, and the premixing instruction is added as an independent instruction segment to the executable mixing instruction; The instructions for adding major bulk raw materials are sorted according to the preset process logic to form the main feeding instruction sequence; The premixing instruction, the main feeding instruction sequence, and the final mixing and discharging instruction are assembled in chronological or logical order to form the executable proportioning instruction containing a complete time sequence.
[0014] As a further aspect of the present invention, after sending the executable proportioning instruction to the feed preparation equipment, the system further includes: The monitoring module receives real-time operating data returned by the feed preparation equipment during instruction execution. The real-time operating data includes at least the type and weight of the raw materials added and the operating status of the mixing equipment. The real-time operating data is compared with the expected execution steps in the executable ratio instruction to determine whether there is any execution deviation. If there is an execution deviation, a deviation correction instruction or an execution interruption instruction is generated according to the type and degree of the deviation, and the deviation correction instruction or execution interruption instruction is sent to the feed preparation equipment.
[0015] As a further aspect of the present invention, the system further includes: The formula management module collects all information related to the preparation task after a complete feed preparation task is completed. The information includes the optimal feed formula used, the executable ratio instructions, the real-time operating data, and the sampling and testing data of the final feed product. All the information from this preparation task is linked and stored with the growth monitoring data of the target livestock over a subsequent period to form a complete record of formula application. The application records of the formulas are summarized into a historical database for subsequent iterative optimization of the algorithm parameters in the multi-objective nutritional requirement model and the formula optimization engine.
[0016] Compared with the prior art, the advantages and positive effects of the present invention are as follows: The original breeding information of the target livestock is aligned with multiple dimensions and missing values are filled in to generate a structured breeding archive containing livestock breed identification, growth stage labels, body condition characteristic sequences and historical feed intake time series. Deviations in breeding information in different dimensions are corrected, missing information is supplemented, and various types of breeding data form a regular and unified structured form. The completeness and regularity of breeding information are improved, and the form of data storage and retrieval is consistent.
[0017] The growth stage labels are matched with standard growth cycle maps to obtain growth status assessment results. Physiological parameterization is performed on the body condition feature sequence to obtain standardized body condition feature vectors. The historical feed intake time series is periodically decomposed to obtain feed intake pattern features. A multi-objective nutritional requirement model is constructed by combining feed ingredient nutrient composition data and the target nutritional requirement spectrum is calculated. With the target nutritional requirement spectrum as the optimization objective and nutrient composition data as the constraint basis, optimization calculations are carried out in the feed ingredient space through a formula optimization engine to generate a candidate formula set. The growth stage determination conforms to the standard cycle basis, the body condition features are converted into a unified parameter form, the periodic characteristics of the feed intake pattern are fully extracted, the input of the nutritional requirement model conforms to the actual state of livestock, the formula calculation can cover the entire raw material space, and the generation of candidate formulas meets the constraints of nutritional requirements and raw material composition. Attached Figure Description
[0018] Figure 1 This is a timeline diagram of the intelligent quantitative proportioning management system for livestock feed based on big data as described in this invention. Figure 2 A flowchart of the method for generating growth status assessment results; Figure 3 A flowchart illustrating the method for generating a candidate recipe set for a recipe optimization engine. Detailed Implementation
[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the 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 merely illustrative and not intended to limit the invention.
[0020] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.
[0021] See Figure 1This invention provides a big data-based intelligent quantitative feed formulation management system for livestock, specifically implemented as follows: The system includes a livestock information processing module, a feature analysis module, a nutritional requirement modeling module, and an intelligent formula generation module. The livestock information processing module collects the original livestock breeding information and performs multi-dimensional information alignment and missing value imputation to generate a structured breeding profile containing livestock breed identification, growth stage labels, body condition feature sequences, and historical feed intake time series. The feature analysis module performs standard growth cycle mapping matching on the growth stage labels to obtain growth status assessment results, performs physiological parameterization processing on the body condition feature sequences to obtain standardized body condition feature vectors, and performs periodic decomposition on the historical feed intake time series to obtain feed intake pattern characteristics. The nutritional requirement modeling module obtains nutrient composition data of feed ingredients from a feed ingredient database containing nutrient composition data of various feed ingredients. Based on the growth status assessment results, standardized body condition feature vectors, feed intake pattern characteristics, and nutrient composition data, a multi-objective nutritional requirement model is constructed, and the target nutritional requirement spectrum is calculated through the multi-objective nutritional requirement model. The intelligent formula generation module takes the target nutrient requirement spectrum as the optimization objective and nutrient composition data as the constraint basis. It uses the formula optimization engine to perform optimization calculations in the feed raw material space to generate a set of candidate formulas containing specific raw material ratios.
[0022] In one embodiment of the present invention, growth stage labels are matched with standard growth cycle maps to obtain growth status assessment results, see reference. Figure 2 This process includes acquiring a pre-established standard growth cycle map corresponding to livestock breed identifiers. The standard growth cycle map defines the standard growth stages and corresponding standard physiological parameter ranges for livestock of the breed at different ages (days or months). Growth stage labels are mapped onto the standard growth cycle map to determine the current standard growth stage of the target livestock. The actual body condition data in the body condition feature sequence is compared with the standard physiological parameter ranges corresponding to the standard growth stages. The deviation between the actual growth level and the standard growth level is calculated. Combining the deviation with the standard growth stage, a growth status assessment result representing the position and progress of the target livestock's actual growth state relative to the standard state is generated.
[0023] In practical implementation, the process of matching the growth stage labels described in this embodiment with standard growth cycle maps to obtain growth status assessment results can be illustrated with an example scenario, such as using fattening pigs identified as "Duroc pigs" as the operation object. In practical implementation, the system pre-establishes a standard growth cycle map corresponding to the "Duroc pig" breed identification. This standard growth cycle map defines the standard growth stage divisions and corresponding standard physiological parameter ranges for Duroc pigs at different ages. The standard growth stage divisions include lactation, nursery, growth, and fattening periods, and the standard physiological parameter ranges include weight range, backfat thickness range, and daily weight gain range. In practical implementation, the structured breeding file generated by the breeding information processing module contains the target livestock's growth stage label as "growth period," and the body condition feature sequence includes actual weight, actual backfat thickness, and actual daily weight gain. In practice, the feature analysis module maps the growth stage label "growth period" to the Duroc pig standard growth cycle map, determines that the target livestock is currently in the standard growth stage as the growth period, and reads the standard physiological parameter range corresponding to the standard growth stage, such as the standard weight range of 70 kg to 90 kg, the standard backfat thickness range of 10 mm to 15 mm, and the standard daily weight gain range of 800 g to 1000 g.
[0024] In some embodiments, when comparing actual body condition data in a body condition feature sequence with the standard physiological parameter range corresponding to a standard growth stage, it is necessary to calculate the deviation between the actual growth level and the standard growth level. In a specific implementation, for each physiological parameter, the relative deviation between the actual body condition data and the value within the standard range of that parameter is calculated, and weights are assigned according to the importance of the parameter. Finally, the overall deviation is obtained by summing these weights. It can be understood that the calculation of the deviation can be expressed using the formula:
[0025] in: Indicates the overall deviation. Indicates the number of physiological parameters, Indicates the first Actual physical condition data, Indicates the first The median of the standard physiological parameter range, Indicates the first The preset weights of the parameters and satisfy the following conditions: In practice, if the target livestock's actual weight is 85 kg, its actual backfat thickness is 12 mm, and its actual daily weight gain is 750 g, while the standard median values are 80 kg, 12.5 mm, and 900 g respectively, with preset weights of 0.5, 0.3, and 0.2, the overall deviation is calculated using a formula. In practice, the calculated overall deviation reflects the degree of deviation between the target livestock's actual body condition and the standard growth level.
[0026] In some embodiments, when generating a growth status assessment result by combining deviation and standard growth stage, the growth status assessment result can be a structured data object. In specific implementations, this data object includes a standard growth stage identifier, a calculated comprehensive deviation value, and individual deviation vectors for various physiological parameters. Optionally, the growth status assessment result can also include a classification label, such as "normal growth," "lagging growth," or "advanced growth," which is determined by comparing the comprehensive deviation with a preset threshold. Optionally, if the comprehensive deviation is below a first threshold, the classification label is "normal growth"; if the comprehensive deviation is between the first and second thresholds, the classification label is "lagging growth"; if the comprehensive deviation is above the second threshold, the classification label is "advanced growth." It can be understood that the growth status assessment result characterizes the position and progress of the target livestock's actual growth status relative to the standard state, providing direct input for subsequent nutritional requirement modeling. In specific implementations, through standard growth cycle map matching and deviation calculation, the system can quantitatively assess the growth status of individual livestock, avoiding the limitations of relying solely on growth stage labels for rough judgment.
[0027] In one embodiment of the present invention, a multi-objective nutritional requirement model is constructed based on growth status assessment results, standardized body condition feature vectors, feeding pattern characteristics, and nutrient composition data. The target nutritional requirement spectrum is calculated using this model. This includes determining basic nutritional requirement targets based on growth status assessment results within the model. These basic nutritional requirement targets include at least maintenance requirements, growth requirements, and reproductive requirements. Body condition regulation factors are introduced into the basic nutritional requirement targets based on the standardized body condition feature vectors to weight and adjust them. Based on feeding pattern characteristics, the expected total feed intake for future stages is predicted. The weighted basic nutritional requirement targets are then converted to concentrations according to the expected total feed intake to obtain the concentration targets of various nutrients required per unit of feed. All nutrient concentration targets are summarized and finally balanced based on synergistic and antagonistic relationships to form a comprehensive target nutritional requirement spectrum. Using the target nutritional requirement spectrum as the optimization objective and nutrient composition data as the constraint, a formulation optimization engine performs optimization calculations in the feed ingredient space to generate a candidate formulation set containing specific ingredient ratios. (See reference...) Figure 3The process involves defining a solution space within the formulation optimization engine. Each point in the solution space represents a formulation composed of different feed ingredient ratios. A fitness function is constructed within the solution space to measure the degree to which any formulation meets the target nutrient requirement spectrum. Constraints are set in the solution space, including constraints on the range of total nutrient content calculated based on nutrient composition data, upper and lower limits on the amount of a single ingredient, and incompatibilities between ingredients. An optimization algorithm is used to search for the optimal value of the fitness function within the region that satisfies the constraints. Multiple high-performance formulation points evaluated during the search process are used as a candidate formulation set.
[0028] In practical implementation, the process of constructing the multi-objective nutritional requirement model and generating the candidate formula set described in this embodiment can be illustrated with an example scenario. The inputs of this scenario include the growth status assessment results of the target livestock, standardized body condition feature vectors, feeding pattern characteristics, and nutrient composition data of various raw materials such as corn, soybean meal, wheat bran, and premix obtained from a feed ingredient database. In practical implementation, the multi-objective nutritional requirement model first determines the basic nutritional requirement targets based on the growth status assessment results. For example, for a fattening livestock in a "normal growth" state, the basic nutritional requirement targets include maintenance requirements and growth requirements, which are quantitatively expressed as the absolute daily requirements for various nutrients such as digestible energy, crude protein, lysine, calcium, and phosphorus. In practice, based on the standardized body condition feature vector, a body condition adjustment factor is introduced on the basic nutritional requirement target. The body condition adjustment factor is a set of weighted coefficients calculated based on the deviation of each parameter in the standardized body condition feature vector from the ideal value. This set of weighted coefficients is used to adjust the nutrient requirements in the basic nutritional requirement target. For example, when the standardized body condition feature vector indicates that body fat deposition is too fast, the body condition adjustment factor will reduce the nutritional requirement for fat synthesis.
[0029] In some embodiments, the expected total feed intake for future stages is predicted based on feeding pattern characteristics, which include historical feed intake trends and periodic information. In a specific implementation, by analyzing the time series of historical feed intake, the total feed intake of the target livestock in the next feeding cycle is predicted to be 250 kg. The weighted adjusted basic nutrient requirements target, i.e., the sum of daily requirements for various nutrients, is converted into concentration targets for various nutrients per unit of feed based on this expected total feed intake, such as the digestible energy, crude protein grams, or megajoules per kilogram of feed. In a specific implementation, the concentration targets of all nutrients are summarized, and a final balance is achieved by considering synergistic and antagonistic relationships, such as adjusting the calcium-phosphorus ratio and balancing the amino acid pattern, to form a comprehensive target nutrient requirement spectrum. This spectrum specifies in detail the concentration or content requirements of each nutrient indicator in the final feed product in tabular form.
[0030] Optionally, using the target nutrient requirement spectrum as the optimization objective and nutrient composition data as the constraint basis, an optimization calculation is performed in the feed ingredient space through a formulation optimization engine. The formulation optimization engine defines each point in the solution space as a formulation composed of different proportions of feed ingredients such as corn and soybean meal. In specific implementation, a fitness function is constructed in the solution space. The fitness function measures the degree to which any formulation meets the target nutrient requirement spectrum; the smaller the function value, the smaller the deviation between the formulation's nutrient concentration and the target nutrient requirement spectrum. One specific calculation method for the fitness function can be understood as follows:
[0031] in: This represents the fitness function value. This indicates the number of different types of nutrients considered. This represents the predicted concentration of nutrient J in the formula, calculated based on the formula and nutritional data. This indicates the first in the target nutrient requirement spectrum. Target concentrations of various nutrients This represents the preset importance coefficient of the Jth nutrient.
[0032] In some embodiments, constraints are set in the solution space, including constraints on the range of total nutrient content calculated based on nutrient composition data, upper and lower limits on the amount of a single ingredient, and incompatibilities between ingredients. Specifically, the total nutrient content range constraint ensures that the total amount of nutrients provided by the formula meets the animal's needs without exceeding the limits; the upper and lower limits on the amount of a single ingredient ensure that the amount added to each ingredient is within the feasible and safe range; and the incompatibilities between ingredients prevent the joint use of certain ingredients due to anti-nutritional factors or other issues. In specific embodiments, an optimization algorithm is used to search for the optimal value of the fitness function within the region that satisfies the constraints. The optimization algorithm can be a genetic algorithm or a particle swarm optimization algorithm. During the search process, the algorithm evaluates tens of thousands of different formulation points. In specific embodiments, the multiple high-performance formulation points evaluated during the search process—that is, those formulations with low fitness function values that satisfy all constraints—are output as a candidate formulation set containing specific proportions of ingredients such as corn, soybean meal, wheat bran, and premix. Optionally, the candidate formulation set typically includes ten to twenty alternative formulations that are similar in nutritional satisfaction but differ in ingredient structure or cost, for subsequent decision-making. It is understandable that the formula optimization engine uses systematic mathematical optimization to automatically and efficiently locate feasible formula regions that meet complex nutritional goals in a vast space of raw material combinations. This process avoids the tediousness and limitations of traditional manual calculations.
[0033] In one embodiment of the present invention, the system further includes an intelligent decision-making module and an execution control module. The intelligent decision-making module performs cost evaluation and process feasibility verification on each candidate formula in the candidate formula set, selects the optimal feed formula that meets the cost and process constraints, and converts the optimal feed formula into an executable proportioning instruction for the feed preparation equipment. The executable proportioning instruction includes the feeding amount of each feed ingredient. The execution control module sends the executable proportioning instruction to the feed preparation equipment and receives feeding confirmation information returned by the feed preparation equipment. The cost evaluation and process feasibility verification for each candidate formula in the candidate formula set specifically includes obtaining current market feed ingredient price information, calculating the unit weight raw material cost of the candidate formula based on the proportion of various feed ingredients in each candidate formula, completing the cost evaluation, and obtaining the process parameters of the feed preparation equipment, including particle size requirements, mixing uniformity requirements, and pelleting condition requirements. The system verifies whether the physical properties of each candidate formula meet the process parameters of the feed preparation equipment. Physical properties include the particle size distribution, mixing characteristics, and binding characteristics of the raw materials. The system records the results of each candidate formula passing the cost evaluation and process feasibility verification, including whether it passed and the specific reasons for failure. The optimal feed formulation that meets both cost and process constraints is selected through a screening process. This includes a preset cost threshold and a list of process constraints, which contains all necessary process parameters. From the candidate formulation set, formulations with raw material costs exceeding the cost threshold are removed, resulting in a cost-compliant subset. From this cost-compliant subset, formulations that do not fully meet all conditions in the process constraint list are removed, resulting in a process-compliant subset. Finally, the candidate formulation with the highest fitness function value is selected from this process-compliant subset and determined as the optimal feed formulation.
[0034] In practical implementation, the operation of the intelligent decision-making module and execution control module described in this embodiment can be illustrated with an example scenario. The intelligent decision-making module receives a candidate formula set containing eight candidate formulas from the intelligent formula generation module. Each candidate formula lists the specific percentage proportions of raw materials such as corn, soybean meal, fish meal, and limestone powder. In practical implementation, the intelligent decision-making module performs cost assessment and process feasibility verification for each candidate formula in the candidate formula set. During cost assessment, the intelligent decision-making module obtains current market feed raw material price information, which is stored in the system in the form of a price list and updated periodically. In practical implementation, based on the proportions of various feed raw materials in each candidate formula and combined with the raw material prices, the unit weight raw material cost of the candidate formula is calculated. The calculation is completed using the following formula:
[0035] in: This indicates the cost of raw materials per unit weight. This indicates the types and quantities of ingredients used in the formula. Indicates the first The current market price of the raw materials, Indicates the first The proportions of each raw material in the formula are determined. After cost assessment, the intelligent decision-making module verifies the feasibility of the process and obtains the process parameters of the feed preparation equipment. These parameters include requirements for particle size distribution, mixing uniformity, and pelleting conditions. For example, the particle size requirement is that all particles must pass through a 2.5 mm sieve, and the mixing uniformity requirement is that the coefficient of variation is no greater than 7%. In practice, the physical properties of each candidate formula are verified to meet the process parameters of the feed preparation equipment. These physical properties include the particle size distribution, mixing characteristics, and binding properties of the raw materials. The system obtains the inherent physical properties of each raw material in the formula by querying the raw material physical property database, and predicts the physical properties of the final mixture through weighted calculation. The predicted results are then compared with the process parameters.
[0036] In some embodiments, the results of cost assessment and process feasibility verification for each candidate formulation are recorded. The results include whether the formulation passed or failed, and the specific reasons for failure, such as "cost exceeds budget" or "estimated mixing uniformity does not meet standards." Table 1, which shows a simplified table of candidate formulation cost and process verification results, is provided. Table 1: Candidate Formulation Cost and Process Validation Results
[0037] In practice, the optimal feed formulation that meets both cost and process constraints is selected. A cost threshold and a list of process constraints are preset. The cost threshold is set, for example, at 3.30 yuan per kilogram. The list of process constraints includes all required process parameters, such as "particle size after grinding ≤ 2.5mm", "mixing uniformity CV ≤ 7%", and "pellet formation rate ≥ 95%". In practice, from the candidate formulation set, formulations with raw material costs exceeding the cost threshold are removed, resulting in a cost-compliant subset (e.g., formulation B is removed). From the cost-compliant subset, candidate formulations that do not fully meet all conditions in the process constraint list are removed, resulting in a process-compliant subset (e.g., formulation C is removed). From the process-compliant subset, the candidate formulation with the highest fitness function value (derived from the formulation optimization engine's evaluation) is selected as the optimal feed formulation. For example, from the remaining formulations A and D, formulation A with the better fitness function value is selected. Optionally, the intelligent decision-making module converts the optimal feed formula into executable proportioning instructions for the feed preparation equipment. These instructions include the feed ingredient quantities. The execution control module sends these instructions to the feed preparation equipment and receives confirmation of feed ingredient input from the equipment. In some embodiments, the specific operation of process feasibility verification involves complex material characteristic simulations, such as simulating mixing uniformity using pre-stored raw material flowability parameters. Optionally, the cost assessment may consider the minimum order quantity or inventory level of raw materials, adding a penalty coefficient to formulas corresponding to raw materials that cannot be obtained in sufficient quantities immediately. It can be understood that through dual verification of cost and process, the system can select a final solution that combines economic viability and production feasibility from multiple nutritionally feasible candidate formulas. The execution control module ensures that the decision-making solution is accurately delivered to the execution terminal.
[0038] In one embodiment of the present invention, the optimal feed formulation is converted into an executable proportioning instruction for the feed preparation equipment. This includes parsing the optimal feed formulation, extracting the type identifier and corresponding mass ratio of each feed ingredient, and converting the mass ratio into the specific feed ingredient mass required for each ingredient based on the total mass requirement of the target batch of feed to be produced. According to the instruction format recognized by the feed preparation equipment, the type identifier and corresponding specific feed mass of each feed ingredient are encoded into a series of sequentially executed control commands, forming an executable proportioning instruction. The executable proportioning instruction also includes the order of ingredient addition. The method for determining the order of ingredient addition includes determining the addition priority of various feed ingredients based on their physical form, addition ratio, and interaction characteristics during the mixing process. For trace components that need to be premixed, a premixing instruction is generated to mix them first, and the premixing instruction is added as an independent instruction segment to the executable proportioning instruction. The instructions for adding major raw materials are sorted according to the preset process logic to form the main feeding instruction sequence. The premixing instruction, the main feeding instruction sequence, and the final mixing and discharging instruction are assembled according to time or logical order to form an executable proportioning instruction containing a complete time sequence.
[0039] In practical implementation, the process of converting the optimal feed formula described in this embodiment into executable proportioning instructions for feed preparation equipment can be illustrated with an example scenario. Assume the determined optimal feed formula includes corn, soybean meal, limestone powder, salt, vitamin premix, and trace element premix, and the planned production batch has a total mass requirement of 1000 kg of compound feed. In practical implementation, the optimal feed formula is parsed to extract the type identifier and corresponding mass ratio of each feed ingredient. The type identifier is a unique code identifying the ingredient within the system, and the mass ratio is expressed as a percentage. In practical implementation, based on the total mass requirement of the target batch of feed to be produced, the mass ratio is converted into the specific feed mass required for each feed ingredient. The conversion is performed using a formula. Completed, among which This indicates the specific feed quantity required for a particular feed ingredient. This indicates the total mass requirement for the target batch of feed. This indicates the mass ratio of the raw material in the formula. See Table 2 for a detailed formula analysis and mass conversion table: Table 2: Optimal Feed Formulation Analysis and Feed Mass Conversion Table
[0040] In practical implementation, based on the instruction format identified by the feed preparation equipment, the type identifier of each feed ingredient and its corresponding specific feeding quality are encoded into a series of sequentially executed control commands. The instruction format identified by the feed preparation equipment may be a string or binary instruction based on a specific protocol. In practical implementation, the resulting executable proportioning instructions are a series of commands arranged in sequence, such as "BEGIN_BATCH,1000", "DISPENSE(MAIZE,650.0)", "DISPENSE(SBM,285.0)", etc., where "DISPENSE" is the feeding instruction, and the parameters in parentheses are the ingredient type identifier and the specific feeding quality.
[0041] In some embodiments, the executable formulation instruction further includes the order of raw material addition. The order of raw material addition is determined by determining the priority of addition among the raw materials based on their physical form, addition ratio, and interaction characteristics during the mixing process. In a specific implementation, for trace components that need to be premixed, such as vitamin premixes and trace element premixes, a premixing instruction is generated to mix them first. The premixing instruction can be "PRE_MIX(VIT_PREMIX,25.0,TRACE_PREMIX,25.0)" and this premixing instruction is added as an independent instruction segment to the executable formulation instruction. In specific implementation, the addition instructions for major bulk raw materials are ordered according to a preset process logic. The preset process logic may be to add raw materials with a larger proportion first, followed by raw materials with a smaller proportion to facilitate dispersion. Therefore, the main feeding instruction sequence is "DISPENSE(MAIZE,650.0)", "DISPENSE(SBM,285.0)", "DISPENSE(LIMESTONE,12.0)", and "DISPENSE(SALT,3.0)". Optionally, when assembling the complete instruction, the premixing instruction, the main feeding instruction sequence, and the final mixing and discharging instruction are assembled in a time or logical order. In some embodiments, the timing of the complete executable proportioning instruction is as follows: first, the premixing instruction is executed to generate the premix, then the bulk raw material addition instructions are executed sequentially, then the instruction "ADD_PREMIX" is executed to add the premix to the mixer, and finally "MIX(180)" is executed for 180 seconds of mixing and "DISCHARGE" is executed for discharging. It is understandable that determining the order of raw material addition is to optimize mixing efficiency and prevent trace components from remaining or unevenly distributed in the mixer. Optionally, for certain raw materials with special functions, such as oils that need to be added later, their addition instructions will be scheduled after the main dry materials are mixed. In practice, executable proportioning instructions containing a complete time sequence ensure that the feed preparation equipment can add raw materials in the correct order and quantity, thereby ensuring the consistency of the final feed product with the optimal feed formulation.
[0042] In one embodiment of the present invention, after the executable proportioning instruction is sent to the feed preparation equipment, the system further includes an execution monitoring module. The execution monitoring module receives real-time operating data returned by the feed preparation equipment during instruction execution. The real-time operating data includes at least the type and weight of the raw materials added and the operating status of the mixing equipment. The real-time operating data is compared with the expected execution steps in the executable proportioning instruction to determine if there is an execution deviation. If an execution deviation exists, a deviation correction instruction or an execution interruption instruction is generated based on the type and degree of the deviation, and this instruction is sent to the feed preparation equipment. The system also includes a formula management module. After a complete feed preparation task is completed, the formula management module collects all information related to the task, including the optimal feed formula used, the executable proportioning instruction, real-time operating data, and sampling and testing data of the final feed product. All information from the preparation task is associated with and stored with the growth monitoring data of the target livestock over a subsequent period, forming a complete formula application record. This formula application record is summarized in a historical database for subsequent iterative optimization of the algorithm parameters in the multi-objective nutritional requirement model and the formula optimization engine.
[0043] In this specific implementation, the execution monitoring and formula management process described in this embodiment is illustrated with an example scenario. After the intelligent decision-making module generates an executable proportioning instruction and sends it to the feed preparation equipment by the execution control module, the system enters the execution and monitoring phase. In this implementation, the execution monitoring module receives real-time operating data returned by the feed preparation equipment during instruction execution. This real-time operating data is acquired through sensors and controllers on the equipment and continuously uploaded according to the agreed communication protocol. The real-time operating data includes at least the type and weight of the raw materials added, and the operating status of the mixing equipment, such as current, speed, and mixing time. In this implementation, the execution monitoring module internally maintains a list of expected execution steps synchronized with the sent executable proportioning instructions. The list details the type, weight, and operation sequence of the expected raw materials added at each step. The execution monitoring module compares the real-time operating data with the expected execution steps in the executable proportioning instructions. For example, it compares the received "corn added, weight 650.5 kg" with the expected step "corn added 650.0 kg".
[0044] In some embodiments, the existence of execution deviation is determined based on preset tolerance rules, which set acceptable deviation ranges for different types of operations. In specific implementations, for the weight of raw materials fed, the tolerance range might be ±2% of the target value; for timing, the tolerance range might be ±10 seconds of the expected time point. The execution monitoring module determines the deviation by comparing the difference between real-time data and the expected value to see if it exceeds the corresponding tolerance range. In specific implementations, if an execution deviation exists, a corresponding instruction is generated based on the type and severity of the deviation. For example, when the weight of a raw material fed exceeds the upper limit but does not reach the danger value, a deviation correction instruction "ADJUST(SBM,-5.0)" containing a correction amount is generated and sent to the feed preparation equipment, requiring a compensatory reduction in the amount of soybean meal fed in the next raw material feeding. In specific implementations, if the deviation is severe, such as an abnormally excessive mixer operating current, an execution interruption instruction "EMERGENCY_STOP" is immediately generated and sent to stop the current production process.
[0045] Optionally, after a complete feed preparation task is completed, from the receipt of the instruction to the discharge of the finished feed, the formulation management module is triggered to collect all information related to this preparation task. In specific implementations, all information includes the optimal feed formulation used, executable mixing instructions, real-time operating data, and sampling and testing data of the final feed product. The sampling and testing data can be obtained from rapid near-infrared analysis of the finished feed to obtain actual nutrient composition values. In specific implementations, the formulation management module associates all information from this preparation task with the growth monitoring data of the target livestock over a subsequent period. The growth monitoring data includes records of the livestock's daily weight gain, feed intake, and health status during the feeding period of this batch of feed. This association is achieved by creating a unique batch identifier for this task and binding all relevant data records to this identifier, thus forming a complete formulation application record. In some embodiments, the association operation can be implemented through foreign key association in a relational database or through a time series alignment algorithm. It can be understood that aggregating the formulation application records into a historical database constitutes a continuously expanding dataset, which is used for subsequent iterative optimization of the algorithm parameters in the multi-objective nutritional requirement model and the formulation optimization engine. In practice, the iterative optimization process can be initiated periodically. Utilizing a large database of formulation application records, machine learning algorithms can be used to recalibrate the parameters of body condition regulation factors in the multi-objective nutritional requirement model, or adjust the weight coefficients of the fitness function in the formulation optimization engine. Optionally, comparing real-time operating condition data from formulation application records with finished product testing data can be used to correct nutrient composition data or raw material utilization coefficients in the feed ingredient database, making model predictions more closely reflect actual production results. Through this closed-loop data collection and model iteration mechanism, the system's recommendation and decision-making capabilities can gradually evolve with the accumulation of data.
[0046] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.
Claims
1. A smart quantitative proportioning management system for livestock feed based on big data, characterized in that, The system includes: The livestock breeding information processing module collects the original breeding information of the target livestock and performs multi-dimensional information alignment and missing value filling to generate a structured breeding archive containing livestock breed identification, growth stage label, body condition characteristic sequence and historical feed intake time series. The feature analysis module performs standard growth cycle map matching on the growth stage labels to obtain growth status assessment results, performs physiological parameterization processing on the body condition feature sequence to obtain standardized body condition feature vectors, and performs periodic decomposition on the historical feed intake time series to obtain feed intake pattern features. The nutrition requirement modeling module obtains nutrient composition data of feed ingredients from a feed ingredient database, which contains nutrient composition data of various feed ingredients. Based on the growth status assessment results, the standardized body condition feature vector, the feeding pattern characteristics, and the nutrient composition data, a multi-objective nutrition requirement model is constructed, and the target nutrition requirement spectrum is calculated through the multi-objective nutrition requirement model. The intelligent formula generation module takes the target nutrient requirement spectrum as the optimization target and the nutrient component data as the constraint basis. It performs optimization calculations in the feed raw material space through the formula optimization engine to generate a set of candidate formulas containing specific raw material ratios.
2. The intelligent quantitative proportioning management system for livestock feed based on big data as described in claim 1, characterized in that, The growth stage labels are matched with standard growth cycle maps to obtain growth status assessment results, including: Obtain a pre-established standard growth cycle map corresponding to the livestock breed identifier. The standard growth cycle map defines the standard growth stage division and corresponding standard physiological parameter range of the livestock breed corresponding to the livestock breed identifier at different days or months. The growth stage labels are mapped to the standard growth cycle map to determine the current standard growth stage of the target livestock. Compare the actual body condition data in the body condition feature sequence with the standard physiological parameter range corresponding to the standard growth stage, and calculate the deviation between the actual growth level and the standard growth level. By combining the deviation with the standard growth stage, a growth status assessment result is generated that characterizes the position and progress of the target livestock's actual growth status relative to the standard status.
3. The intelligent quantitative proportioning management system for livestock feed based on big data as described in claim 1, characterized in that, Based on the growth status assessment results, the standardized body condition feature vector, the feeding pattern characteristics, and the nutrient composition data, a multi-objective nutritional requirement model is constructed. The target nutritional requirement spectrum is calculated using this model, including: In the multi-objective nutrient requirement model, basic nutrient requirement targets are determined based on the growth status assessment results. The basic nutrient requirement targets include at least maintenance requirements, growth requirements, and reproductive requirements. Based on the standardized body condition feature vector, a body condition adjustment factor is introduced onto the basic nutritional requirement target to make a weighted adjustment to the basic nutritional requirement target; Based on the feeding pattern characteristics, the expected total feed intake in the future stage is predicted. The weighted adjusted basic nutrient requirements target is converted into concentration based on the expected total feed intake to obtain the concentration target of various nutrients required per unit feed. The concentration targets for all nutrients are summarized, and a final balance is achieved by combining synergistic and antagonistic relationships among nutrients, forming a comprehensive spectrum of the target nutrient requirements.
4. The intelligent quantitative proportioning management system for livestock feed based on big data as described in claim 1, characterized in that, Using the target nutrient requirement spectrum as the optimization objective and the nutrient composition data as the constraint basis, an optimization calculation is performed in the feed ingredient space through a formula optimization engine to generate a candidate formula set containing specific ingredient ratios, including: A solution space is defined in the formula optimization engine, where each point represents a formula composed of different feed ingredients in proportion. A fitness function is constructed in the solution space, which is used to measure the degree to which any formulation meets the target nutritional requirement spectrum; Constraints are set in the solution space, including: total nutrient content range constraints calculated based on the nutrient composition data, upper and lower limits of single raw material usage, and incompatibilities between raw materials. The optimal value of the fitness function is searched within the region that satisfies the constraints using an optimization algorithm, and the multiple high-performance formulation points evaluated during the search process are used as the candidate formulation set.
5. The intelligent quantitative proportioning management system for livestock feed based on big data according to claim 4, characterized in that, The system also includes: The intelligent decision-making module performs cost assessment and process feasibility verification on each candidate formula in the candidate formula set, selects the optimal feed formula that meets the cost and process constraints, and converts the optimal feed formula into an executable proportioning instruction for the feed preparation equipment. The executable proportioning instruction includes the amount of each feed ingredient to be fed. The execution control module sends the executable proportioning instruction to the feed preparation equipment and receives the feeding confirmation information returned by the feed preparation equipment; For each candidate formulation in the candidate formulation set, a cost assessment and process feasibility verification are performed, specifically including: Obtain current market feed ingredient price information, calculate the unit weight raw material cost of each candidate formula based on the proportion of various feed ingredients in each candidate formula, and complete the cost assessment. Obtain the process parameters of the feed preparation equipment, including requirements for particle size, mixing uniformity, and pelleting conditions; Verify whether the physical properties of each candidate formulation meet the process parameters of the feed preparation equipment. The physical properties include the particle size distribution, mixing properties, and binding properties of the raw materials. Record the results of each candidate formulation's cost assessment and process feasibility verification, including whether it passed and the specific reasons for failure.
6. The intelligent quantitative proportioning management system for livestock feed based on big data as described in claim 5, characterized in that, The process of selecting the optimal feed formulation that meets both cost and process constraints includes: A preset cost threshold and a list of process constraints, wherein the list of process constraints contains all process parameter conditions that must be met; From the candidate formula set, candidate formulas whose unit weight raw material cost exceeds the cost threshold are removed to obtain a cost-qualified formula subset. From the cost-qualified formula subset, candidate formulas that fail to fully meet all conditions in the process constraint list are removed, resulting in a process-qualified formula subset. From the set of qualified feed formulations, the candidate formulation with the highest fitness function value is selected and determined as the optimal feed formulation.
7. The intelligent quantitative proportioning management system for livestock feed based on big data according to claim 5, characterized in that, Converting the optimal feed formulation into executable proportioning instructions for the feed preparation equipment includes: The optimal feed formula is analyzed to extract the type identifier and corresponding mass ratio of each feed ingredient; Based on the total mass requirement of the target batch of feed to be produced, the mass ratio is converted into the specific feed mass required for each feed ingredient. Based on the instruction format identified by the feed preparation equipment, the type identifier of each feed ingredient and its corresponding specific feeding quantity are encoded into a series of sequentially executed control commands, forming the executable proportioning instruction.
8. The intelligent quantitative proportioning management system for livestock feed based on big data according to claim 7, characterized in that, The executable proportioning instruction also includes the order of raw material addition, and the method for determining the order of raw material addition includes: The priority of adding ingredients is determined based on their physical form, addition ratio, and interaction characteristics during the mixing process. For trace components that require premixing, a premixing instruction is generated to mix them first, and the premixing instruction is added as an independent instruction segment to the executable mixing instruction; The instructions for adding major bulk raw materials are sorted according to the preset process logic to form the main feeding instruction sequence; The premixing instruction, the main feeding instruction sequence, and the final mixing and discharging instruction are assembled in chronological or logical order to form the executable proportioning instruction containing a complete time sequence.
9. A smart quantitative proportioning management system for livestock feed based on big data as described in claim 5, characterized in that, After sending the executable proportioning command to the feed preparation equipment, the system further includes: The monitoring module receives real-time operating data returned by the feed preparation equipment during instruction execution. The real-time operating data includes at least the type and weight of the raw materials added and the operating status of the mixing equipment. The real-time operating data is compared with the expected execution steps in the executable ratio instruction to determine whether there is any execution deviation. If there is an execution deviation, a deviation correction instruction or an execution interruption instruction is generated according to the type and degree of the deviation, and the deviation correction instruction or execution interruption instruction is sent to the feed preparation equipment.
10. A smart quantitative proportioning management system for livestock feed based on big data as described in claim 9, characterized in that, The system also includes: The formula management module collects all information related to the preparation task after a complete feed preparation task is completed. The information includes the optimal feed formula used, the executable ratio instructions, the real-time operating data, and the sampling and testing data of the final feed product. All the information from this preparation task is linked and stored with the growth monitoring data of the target livestock over a subsequent period to form a complete record of formula application. The application records of the formulas are summarized into a historical database for subsequent iterative optimization of the algorithm parameters in the multi-objective nutritional requirement model and the formula optimization engine.