Smart pasture breeding regulation and control method based on Internet of Things data and optimization model

Abstract

The invention belongs to the technical field of smart agriculture, particularly relates to a smart pasture breeding regulation and control method based on Internet of Things data and an optimization model, and aims to solve the problems of grass and livestock imbalance, ecological degradation and low benefits caused by lack of a scientific decision mechanism in a high and cold mountain pasture. According to the method, multi-dimensional data of grassland and livestock are collected in real time by deploying Internet-of-Things equipment, a dynamic optimization model which takes maximization of the total number of yaks as a target and meets five constraints including grass and livestock balance, supplementary feeding ability, population breeding and ecological sustainability is constructed, and a mixed integer programming algorithm is adopted for solving, so that the dynamic optimization model is obtained. Outputting the optimal livestock carrying capacity, the rotation grazing partition, the supplementary feeding amount and the artificial grass planting area; and the management terminal issues, executes and dynamically calibrates the model based on feedback data to form closed-loop regulation and control. According to the invention, quantitative management and intelligent decision-making of all elements of the pasture are realized, and ecological and economic benefits are remarkably improved.

Description

Technical Field

[0001] This invention belongs to the field of smart agriculture technology, and specifically relates to a smart ranch breeding control method based on Internet of Things data and optimization models. Background Technology

[0002] In high-altitude mountain ecosystems, traditional pastoralism has long been a vital pillar of regional economy and culture, its production methods heavily reliant on the experiential knowledge accumulated by generations of herders. However, with the intensification of global climate change, rising ecological pressures on grasslands, and the increasingly urgent need for modernization in animal husbandry, this experience-driven, extensive management model is struggling to balance ecological protection and industrial benefits. Particularly in areas such as livestock carrying capacity control, seasonal nutrient supply, and grassland space utilization, the lack of scientific decision-making mechanisms based on real-time data and system modeling has led to an imbalance in the grass-livestock relationship, becoming a core bottleneck restricting the sustainable development of plateau animal husbandry. In recent years, digital technologies such as the Internet of Things, remote sensing monitoring, and smart terminals have been gradually introduced into pasture management, giving rise to a number of digital platforms, such as livestock location tracking, body temperature and heart rate monitoring, or electronic fencing. While these systems have improved information acquisition capabilities in certain areas, their functional design often focuses on single-dimensional state perception and has not yet formed a dynamic optimization system that deeply integrates multi-source heterogeneous data with complex ecological-production coupling logic.

[0003] The existing technological architecture suffers from deep-seated structural contradictions: on the one hand, grassland resources exhibit significant spatiotemporal heterogeneity, with forage yield fluctuating nonlinearly due to multiple factors such as precipitation, temperature, soil conditions, and grazing history; on the other hand, the nutritional needs of plateau-specific livestock such as yaks dynamically change with physiological stages, seasonal climate, and herd structure, requiring a sophisticated matching mechanism to achieve dynamic balance. However, traditional empirical judgments or isolated data monitoring cannot quantify the key parameter interactions in this complex coupling process. Crucially, setting a carrying capacity limit based solely on static historical data or simplified assumptions can easily lead to hidden overgrazing during periods of forage shortage in the cold season, compressing the grassland recovery window and triggering a vicious cycle of degradation, reduced yield, and further overgrazing. Furthermore, artificial grassland planting, as an important regulatory tool, if its area allocation and supplementary feeding timing are decided independently from the overall grassland-livestock balance framework, may not only lead to resource waste due to input-output imbalances but may also fail to truly support the goal of continuous year-round yak growth due to insufficient supplementary feeding. Therefore, existing technologies lack a unified optimization engine that can synchronously coordinate spatial planning such as rotational grazing area division, time scheduling such as seasonal grazing cycles, biological parameters such as population structure, and ecological red lines such as grazing intensity thresholds. This makes data-driven intelligent regulation remain at the conceptual level and difficult to implement into an operable and replicable standardized breeding paradigm.

[0004] Against this backdrop, how to construct a smart ranch control method that deeply integrates the real-time sensing capabilities of the Internet of Things with a multi-constraint dynamic optimization model, so that it can accurately depict the dynamic supply and demand relationship of the pasture-livestock system and adaptively generate a closed-loop management strategy that meets the requirements of ecological sustainability and maximizes economic benefits, has become a key technical challenge that needs to be overcome by those skilled in the art. Summary of the Invention

[0005] This invention provides a smart ranch management method based on IoT data and optimization models, aiming to solve technical problems such as grass-livestock imbalance, ecological degradation, and low breeding efficiency caused by the lack of scientific decision-making mechanisms in traditional grazing models in high-altitude mountain ecosystems such as the Qinghai-Tibet Plateau. To achieve the above-mentioned objectives, this invention constructs a closed-loop control system that integrates multi-source heterogeneous IoT sensing data and multi-constraint dynamic optimization models. Through four core links—real-time monitoring, modeling analysis, strategy generation, and execution feedback—it achieves systematic optimization of pasture carrying capacity, rotational grazing zones, artificial feeding, and population structure.

[0006] According to a first aspect of the present invention, the present invention claims protection for a smart ranch breeding control method based on Internet of Things data and optimization models, comprising the following steps: S1 receives raw monitoring data streams continuously collected by multiple sensing devices deployed in the physical environment of the ranch; S2, based on preset data mapping rules, converts the received raw monitoring data stream into a standardized set of input parameters that can be recognized by the optimized calculation model; S3, input the standardized set of input parameters into a preset mixed integer programming model based on the objective function and multiple sets of constraints constructed from pasture breeding logic; S4, Call the mixed integer linear programming solver to solve the mixed integer programming model, generate and output the structured optimal aquaculture control scheme; S5, through the graphical user interface management terminal, visualize and organize the various suggested quantitative indicators in the optimal aquaculture control plan to form an interactive operation guidance interface, and send the operation guidance interface to the user side. S6. According to the preset calibration trigger condition, start the parameter calibration process. Based on the newly received raw monitoring data stream after the release of the optimal aquaculture control scheme, re-estimate the parameters in the standardized input parameter set, and use the re-estimated parameters to update the corresponding input of the mixed integer programming model for the next round of optimization calculation.

[0007] Furthermore, the method also includes: The original monitoring data stream includes at least resource status data reflecting the status of grassland resources and livestock behavior data reflecting the behavior of individual or group livestock. The standardized input parameter set includes at least a grassland productivity parameter to characterize the grassland's grass production capacity, a livestock demand parameter to characterize the rate of grass consumption in different seasons, and a resource constraint parameter to characterize the upper limit of supplementary feeding resources. The objective function is used to instruct the calculation system to optimize in the direction of maximizing the total number of livestock in stock within a preset period. The multiple sets of constraints include at least the global resource constraints to ensure that the total supply of forage is not less than the total demand of livestock, the seasonal resource constraints to ensure that the supply of natural forage in each time zone is not less than the basic demand of livestock, the supplementary feeding capacity constraints to limit the consumption of supplementary feeding resources, the population structure constraints to specify the proportional relationship between the number of different types of livestock, and the ecological carrying capacity constraints to limit the intensity of artificial intervention and the overall carrying density. The optimal livestock management scheme is the solution vector corresponding to the objective function reaching its optimal value when the solver satisfies all the constraints. This solution vector specifically represents at least the number of livestock of multiple different physiological categories recommended for breeding in the next management cycle, the grazing area recommended to be allocated to different time periods, the proportion of supplementary feeding intensity recommended to be activated in each specific time period, and the area of ​​artificial grassland recommended to be reserved.

[0008] Furthermore, S2 in this method also includes: The first type of sensing data is parsed from the original monitoring data stream, and the first type of sensing data comes from the monitoring of grassland vegetation; The first transformation function is applied to the first type of sensing data to calculate the grassland productivity parameters, which include the estimated yield of edible grass per unit area in multiple different time zones. The second type of sensing data is parsed from the original monitoring data stream, and the second type of sensing data originates from the monitoring of individual livestock. The second transformation function is applied to the second type of sensing data, and combined with the preset livestock category and standard daily feed mapping table, the livestock demand parameters are derived. The livestock demand parameters include the reference value of the average daily feed demand of livestock of each physiological category in the multiple different time zones. Obtain a fixed threshold representing the upper limit of the total amount of supplementary feeding resources from the original monitoring data stream or the preset management configuration information, and use it directly as the upper limit value of the total amount of supplementary feeding resources in the resource constraint parameters.

[0009] Furthermore, the specific construction and processing methods of multiple sets of constraints in the mixed-integer programming model described in this method include: The global resource constraint is constructed such that the total amount of natural forage that all grazing areas can provide in different time zones within the entire preset period, plus the total amount of supplementary forage that can be provided in specific time zones through artificial grass planting, is not less than the total forage demand of all livestock within the entire preset period, ensuring that resource supply covers demand on an overall time scale. The seasonal resource constraints are designed to impose constraints individually on each time zone, requiring that the natural forage production of the allocated grazing area in that time zone is not less than the basic forage requirement of all livestock in that time zone relying solely on natural grazing, so as to ensure that the pasture can naturally support livestock during periods when there is no supplementary feeding or when supplementary feeding has not yet started. The supplementary feeding capacity constraint is constructed to limit the total amount of supplementary feed actually used within the time partition in which supplementary feeding is allowed to not exceed the upper limit of the total amount of artificial supplementary feeding resources available in the system, and to set an upper limit threshold for the supplementary feeding ratio within a single time partition. The population structure constraint is constructed to limit the range of proportions between the numbers of livestock of different physiological categories through inequality relationships. The specified ratio range includes the ratio of the number of breeding males to the total number of breeding females, the ratio between the number of females in different production states, and the ratio between the number of young livestock and the number of specific females; this set of constraints ensures that the solved livestock population structure conforms to the laws of biological reproduction and normal production management. The ecological carrying capacity constraint is constructed as a maximum limit on the area of ​​human intervention and a maximum limit on the average number of livestock that can be carried per unit area of ​​the entire pasture per year.

[0010] Furthermore, S4 of the method also includes: Multiple livestock quantity variable values ​​that satisfy all constraints and optimize the objective function are determined through iterative calculations and conform to the population structure constraints. Simultaneously determine the grazing area allocation variable values ​​corresponding to the multiple different time periods, and calculate the optimal supplementary feeding ratio variable value for the specific time period in which supplementary feeding is allowed, so as to satisfy the global resource constraints and the supplementary feeding capacity constraints; The variable value of the artificial grassland area is determined, and together with the supplementary feeding ratio, the total amount of supplementary forage that can be provided is determined; All the above variable values ​​together constitute the solution vector, which is then encapsulated as a structured data object for output.

[0011] Furthermore, S5 of the method also includes: The number of livestock of different physiological categories recommended for breeding is displayed in the first display area in the form of a combination of classification statistical charts and target numbers; The proposed grazing area is allocated to different time periods and displayed in the second display area in the form of a timeline combined with map block coloring or a scale diagram. The recommended supplemental feeding intensity ratio and the recommended reserved area for artificial grass planting will be displayed in the third display area in the form of progress bars, percentage indicators and area values; In the operation guidance interface, an input control is associated with the suggested quantitative indicator. The input control is configured to allow users to input or adjust the actual execution data corresponding to the indicator as actual execution feedback.

[0012] Furthermore, the operation guidance interface described in this method is also configured as follows: Receive actual execution data related to the current control cycle submitted by the user through the input control; The actual execution data is compared with the corresponding suggested quantitative indicators in the optimal aquaculture control scheme; Based on the comparison results, a difference prompt message is generated and highlighted on the interface. The difference prompt message is used to remind the user to pay attention to the deviation between the actual execution and the optimization suggestions.

[0013] Furthermore, in step S6 of this method, the parameter calibration process is initiated according to a preset calibration trigger condition, wherein the calibration trigger condition includes at least one of the following: Time trigger condition: after a fixed time period since the last parameter calibration or scheme generation; Data accumulation trigger condition: The amount of newly received raw monitoring data related to a specific parameter reaches a preset threshold; Event triggering conditions: Receiving an event alarm message from a user or system indicating an abnormal state in a specific area or a specific livestock group.

[0014] Furthermore, the step of re-estimating the parameters in the standardized input parameter set in S6 of the method specifically includes: Identify the target parameter that needs to be calibrated, the target parameter being associated with the calibration trigger condition; Extract a recent subset of data related to the target parameter from the newly received raw monitoring data stream; The recent data subset is processed using the same or modified transformation function or statistical algorithm as the initial parameter estimation to obtain the updated estimate of the target parameter. Replace the original parameter values ​​in the mixed integer programming model with the updated estimates.

[0015] Furthermore, before the step of calling the mixed-integer linear programming solver to solve the mixed-integer programming model, the method further includes: The mixed-integer programming model is formatted to ensure that the mathematical expression of the objective function and all constraints conforms to the input specifications of the solver. The formatted model, along with the specific values ​​of the standardized input parameter set, is encapsulated together into a call request that conforms to the requirements of the solver application programming interface. The call request is sent to the solver, and the status of the solving process is monitored.

[0016] In summary, this invention, by constructing a closed-loop control method that deeply integrates real-time IoT sensing and multi-constraint dynamic optimization, has for the first time achieved quantitative management of all elements, processes, and cycles of the grassland-livestock system in high-altitude mountain pastures. It solves the fundamental defects of data silos and decision-making disconnect in existing technologies, and provides an engineering-practical technical path for the ecological, intelligent, and standardized transformation of plateau animal husbandry. Attached Figure Description

[0017] Figure 1 A flowchart illustrating the workflow of a smart ranch aquaculture control method based on Internet of Things data and optimization models, as claimed in an embodiment of the present invention. Figure 2 The second flowchart is a diagram of a smart ranch breeding control method based on Internet of Things data and optimization model, which is claimed in an embodiment of the present invention. Figure 3 The third workflow diagram of a smart ranch breeding control method based on Internet of Things data and optimization model claimed in the embodiments of the present invention; Figure 4 The fourth flowchart of a smart ranch breeding control method based on Internet of Things data and optimization model, which is claimed in an embodiment of the present invention, is shown. Detailed Implementation

[0018] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0019] The terms "first," "second," and "third" in this application are for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first," "second," or "third" may explicitly or implicitly include at least one of those features. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified. All directional indications in the embodiments of this application, such as up, down, left, right, front, back, etc., are only used to explain the relative positional relationships and movements between components in a specific orientation as shown in the accompanying drawings. If the specific orientation changes, the directional indications will change accordingly. Furthermore, the terms "including" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices.

[0020] References to embodiments herein mean that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0021] Existing digital ranch solutions typically use GPS collars, environmental sensors, or drone aerial photography to collect preliminary data on livestock location, activity levels, or pasture coverage, supplemented by visual interfaces for managers. This approach is valuable in improving basic monitoring efficiency, particularly for daily inspections and anomaly warnings in large-scale ranches. However, its fundamental limitation lies in its failure to construct a comprehensive decision-making model that incorporates multiple real-world constraints, including ecological carrying capacity thresholds, population reproduction patterns, seasonal rotational grazing rules, and human intervention capabilities. In other words, current systems generally remain at the data display level, lacking the ability to transform monitoring data into executable, verifiable, and iteratively optimized livestock strategies. Furthermore, due to the lack of a dynamic mapping between pasture productivity and livestock nutritional needs, even with high-frequency pasture or livestock condition data, it is difficult to support precise planning of the annual grazing rhythm, rotational grazing schedule, and winter / spring supplementary feeding intensity, thus failing to effectively address the cyclical dilemma of summer growth, autumn fattening, winter thinning, and spring mortality.

[0022] This invention provides a smart ranch livestock management method based on IoT data and optimization models. Its core lies in constructing a closed-loop management system that integrates multi-source heterogeneous sensing data and a multi-constraint dynamic optimization model to achieve synergistic optimization of grass-livestock balance, ecological sustainability, and livestock benefits in high-altitude mountain ranches. The following will systematically and engineeringally elaborate on the technical solution of this invention using typical ranch scenarios on the Qinghai-Tibet Plateau, ensuring that those skilled in the art can fully reproduce this invention based on this description.

[0023] According to the first embodiment of the present invention, referring to Figure 1 This invention claims protection for a smart ranch breeding control method based on Internet of Things data and optimization models, comprising the following steps: S1 receives raw monitoring data streams continuously collected by multiple sensing devices deployed in the physical environment of the ranch; S2, based on preset data mapping rules, converts the received raw monitoring data stream into a standardized set of input parameters that can be recognized by the optimized calculation model; S3, input the standardized set of input parameters into a preset mixed integer programming model based on the objective function and multiple sets of constraints constructed from pasture breeding logic; S4, Call the mixed integer linear programming solver to solve the mixed integer programming model, generate and output the structured optimal aquaculture control scheme; S5, through the graphical user interface management terminal, visualize and organize the various suggested quantitative indicators in the optimal aquaculture control plan to form an interactive operation guidance interface, and send the operation guidance interface to the user side. S6. According to the preset calibration trigger condition, start the parameter calibration process. Based on the newly received raw monitoring data stream after the release of the optimal aquaculture control scheme, re-estimate the parameters in the standardized input parameter set, and use the re-estimated parameters to update the corresponding input of the mixed integer programming model for the next round of optimization calculation.

[0024] Furthermore, the method also includes: The original monitoring data stream includes at least resource status data reflecting the status of grassland resources and livestock behavior data reflecting the behavior of individual or group livestock. The standardized input parameter set includes at least a grassland productivity parameter to characterize the grassland's grass production capacity, a livestock demand parameter to characterize the rate of grass consumption in different seasons, and a resource constraint parameter to characterize the upper limit of supplementary feeding resources. The objective function is used to instruct the calculation system to optimize in the direction of maximizing the total number of livestock in stock within a preset period. The multiple sets of constraints include at least the global resource constraints to ensure that the total supply of forage is not less than the total demand of livestock, the seasonal resource constraints to ensure that the supply of natural forage in each time zone is not less than the basic demand of livestock, the supplementary feeding capacity constraints to limit the consumption of supplementary feeding resources, the population structure constraints to specify the proportional relationship between the number of different types of livestock, and the ecological carrying capacity constraints to limit the intensity of artificial intervention and the overall carrying density. The optimal livestock management scheme is the solution vector corresponding to the objective function reaching its optimal value when the solver satisfies all the constraints. This solution vector specifically represents at least the number of livestock of multiple different physiological categories recommended for breeding in the next management cycle, the grazing area recommended to be allocated to different time periods, the proportion of supplementary feeding intensity recommended to be activated in each specific time period, and the area of ​​artificial grassland recommended to be reserved.

[0025] This embodiment details the complete process of a smart ranch aquaculture control method based on IoT data and optimization models. The method is executed by a computer system deployed on a cloud server or local server, and the specific steps are as follows: The computer system continuously receives raw monitoring data streams from various sensing devices deployed on the pasture through its data access service. These data streams are received asynchronously, in real-time, or near real-time. Resource status data primarily includes: soil moisture and temperature readings periodically reported by soil temperature and humidity sensors; vegetation index and coverage percentage image data captured and processed by vegetation coverage monitoring nodes; precipitation, temperature, and sunshine duration data collected by weather stations; and preliminary stitched images of pasture spectral reflectance obtained through periodic patrols by multispectral UAV platforms. Livestock behavior data primarily includes: livestock GPS location points, motion acceleration data, and grazing activity duration inferred from sensors built into the smart collars; and timestamp logs of livestock entering and exiting specific electronic fence areas recorded by geofence base stations.

[0026] The system uses a pre-defined data mapping rule. For grassland resources, it calls the first transformation function. This function integrates accumulated temperature data over a growing season, precipitation data during key growth periods, and peak vegetation index data obtained from drones. This data is input into an empirical grass yield estimation model calibrated with prior knowledge. The output is the estimated forage yield per unit area for each rotational grazing unit in the four seasons (spring, summer, autumn, and winter) of the coming year, i.e., the grassland productivity parameter G_usable(t). For livestock needs, the system calls the second transformation function. This function combines the average daily effective foraging time from the smart collar statistics, historical livestock weight estimates for the same period, and correlates the collar movement pattern with known growth curves. It also queries a pre-defined mapping table of livestock categories and standard daily feed intake. This table defines the basal metabolic and forage requirements of categories such as lactating cows and non-dairy cows in all four seasons. The system then calculates the average daily forage requirement reference value D_i(t) for each type of livestock in all four seasons, using typical values ​​if the initial values ​​are not used. Simultaneously, the system reads the pre-set upper limit of the total supplementary feeding resources from the ranch management configuration database. For example, the maximum yield of artificially planted hay derived from existing storage and land potential is directly used as the resource constraint parameter H. The system also uses the total ranch area and the area of ​​each zone obtained from the geographic information system as fixed geographic parameters. All these processed parameters together constitute a standardized set of input parameters for subsequent model optimization.

[0027] The standardized set of input parameters is input into the pre-defined mixed-integer programming model in its model processing engine module. The core of this model is an objective function aimed at maximizing the total number of livestock in a pre-defined period at the end of the next year, i.e., the sum of the numbers of each type of livestock. The model incorporates five sets of constraints: 1) Global resource constraint: ensuring that the total amount of natural forage plus supplementary forage throughout the year is not less than the annual livestock demand. 2) Seasonal resource constraint: ensuring that in each of the four seasons (spring, summer, autumn, and winter), the natural forage production of the allocated pastures alone can at least feed the livestock at that time. 3) Supplementary forage capacity constraint: limiting the total amount of supplementary forage used in the allowed supplementary forage seasons of winter and spring to not exceed the total reserves, and setting an upper limit on the proportion of supplementary forage to the demand in each season. 4) Population structure constraint: limiting the ratio of bulls to cows, the ratio of cows in different states, and the ratio of calves to cows through inequalities to ensure a reasonable and sustainable population structure. 5) Ecological carrying capacity constraint: limiting the upper limit of artificial grassland area and the upper limit of the number of livestock that each unit of grassland area can support throughout the year. Before solving, the engine module performs a format check on the model to ensure that the mathematical expressions are standardized, and then encapsulates the model and parameter values ​​into a standard data package, ready to call the solver.

[0028] The engine module invokes its embedded mixed-integer linear programming solver. The solver receives the call request and performs iterative calculations. After computation, the solver finds an optimal solution vector that maximizes the objective function value while strictly satisfying all constraints. This solution vector is decoded into a structured optimal livestock management plan, specifically including: the suggested number of dairy cows, non-dairy cows, bulls, and calves; the suggested grazing area (in hectares) allocated in spring, summer, autumn, and winter; the suggested supplementary feeding ratio in winter and spring (e.g., the proportion of winter demand met by supplementary feeding); and the suggested planned artificial grassland area (in hectares).

[0029] The scheme management and distribution module receives the structured scheme, analyzes the various quantitative indicators in the scheme, and drives the graphical engine to generate an interactive operation guidance interface. For example, the first area of ​​the interface uses a bar chart to display the recommended number of cattle and the total number of cattle in the four categories; the second area uses a pasture map to dynamically display the rotational grazing zone recommendations for each season through different colors and area proportions; the third area uses sliders and numbers to display the recommended proportions of winter and spring supplementary feeding, and separately lists the recommended area for artificial grassland planting. This interface is distributed to the herders' smart pasture management terminals via web or mobile application. In the interface, some indicators, such as the actual number of cattle purchased and the actual amount of supplementary feeding, have text boxes or drop-down menus for input controls, allowing herders to record the actual implementation status as feedback.

[0030] The system's parameter calibration service module continuously monitors calibration trigger conditions. For example, when the time-triggered conditions are met every three months, the module initiates the calibration process. It extracts the raw monitoring data streams newly collected over the past three months, particularly the latest vegetation growth data from drones and livestock weight change trends reflected by collars. Then, it re-estimates key parameters, such as recalculating the pasture productivity parameter G_usable(t) for summer and the upcoming autumn using the new data through the first transformation function. Subsequently, the module replaces the old corresponding parameter values ​​in the model with this updated estimate. The updated model will be used for a new round of optimization calculations in the next regulation cycle, such as when the next quarter's plan is formulated, thereby achieving adaptive adjustments to the scheme.

[0031] Furthermore, referring to Figure 2 The method, S2, further includes: The first type of sensing data is parsed from the original monitoring data stream, and the first type of sensing data comes from the monitoring of grassland vegetation; The first transformation function is applied to the first type of sensing data to calculate the grassland productivity parameters, which include the estimated yield of edible grass per unit area in multiple different time zones. The second type of sensing data is parsed from the original monitoring data stream, and the second type of sensing data originates from the monitoring of individual livestock. The second transformation function is applied to the second type of sensing data, and combined with the preset livestock category and standard daily feed mapping table, the livestock demand parameters are derived. The livestock demand parameters include the reference value of the average daily feed demand of livestock of each physiological category in the multiple different time zones. Obtain a fixed threshold representing the upper limit of the total amount of supplementary feeding resources from the original monitoring data stream or the preset management configuration information, and use it directly as the upper limit value of the total amount of supplementary feeding resources in the resource constraint parameters.

[0032] In this implementation, the conversion process begins after the system receives the raw monitoring data stream: For the first type of perceived data, grassland vegetation data, multiple flight data acquired by a multispectral UAV platform during the most recent full growing season are first filtered and integrated from the data stream to calculate the time series curve of the Normalized Difference Vegetation Index (NDVI). Simultaneously, concurrent temperature and precipitation data provided by meteorological stations are integrated. The system applies a first transformation function, a simplified version based on a light, temperature, and water production potential model. It multiplies the peak NDVI, critical growing season precipitation, and a regional grassland yield conversion coefficient, then multiplies by a edible grass proportion coefficient, ultimately outputting estimated edible grass yield per unit area for each pasture block in spring, summer, autumn, and winter, from G_usable(1) to G_usable(4). These estimates constitute the core of the grassland productivity parameters.

[0033] For the second type of sensory data—individual livestock data—the system extracts livestock movement trajectories and activity classification data uploaded by smart collars over the past few months from the data stream, such as feeding, rumination, and resting. By analyzing the average daily feeding activity duration and exercise intensity, the system combines known average weight gain models for livestock breeds to estimate the current average metabolic weight of various livestock types, such as cows and bulls. Then, the system queries a pre-defined mapping table of livestock categories and standard daily feed intake. This table, based on metabolic weight, lists the daily standard hay equivalent weight required per animal under different seasons, forage quality, and climatic conditions, according to livestock category (milking cows, non-dairy cows, bulls, calves) and production stage (lactation, dry, growth, fattening). Applying a second transformation function, the system combines the estimated metabolic weight with the mapping table to calculate the reference value D_i(t) of the average daily forage requirement for each type of livestock i in each season t. For example, the requirement for milking cows may be lower in the summer when pasture is abundant, and higher in the winter when pasture is scarce.

[0034] For resource constraint parameters, a fixed configuration information is directly read from the ranch infrastructure management database—the upper limit of the total hay reserve available for supplementary feeding this year, in kilograms of dry matter, or the maximum total supplementary feed yield H calculated based on the maximum area of ​​land available for artificial grassland cultivation and the hay yield per unit area. This value is used directly as the model input without complex transformation.

[0035] Furthermore, the specific construction and processing methods of multiple sets of constraints in the mixed-integer programming model described in this method include: The global resource constraint is constructed such that the total amount of natural forage that all grazing areas can provide in different time zones within the entire preset period, plus the total amount of supplementary forage that can be provided in specific time zones through artificial grass planting, is not less than the total forage demand of all livestock within the entire preset period, ensuring that resource supply covers demand on an overall time scale. The seasonal resource constraints are designed to impose constraints individually on each time zone, requiring that the natural forage production of the allocated grazing area in that time zone is not less than the basic forage requirement of all livestock in that time zone relying solely on natural grazing, so as to ensure that the pasture can naturally support livestock during periods when there is no supplementary feeding or when supplementary feeding has not yet started. The supplementary feeding capacity constraint is constructed to limit the total amount of supplementary feed actually used within the time partition in which supplementary feeding is allowed to not exceed the upper limit of the total amount of artificial supplementary feeding resources available in the system, and to set an upper limit threshold for the supplementary feeding ratio within a single time partition. The population structure constraint is constructed to limit the range of proportions between the numbers of livestock of different physiological categories through inequality relationships. The specified ratio range includes the ratio of the number of breeding males to the total number of breeding females, the ratio between the number of females in different production states, and the ratio between the number of young livestock and the number of specific females; this set of constraints ensures that the solved livestock population structure conforms to the laws of biological reproduction and normal production management. The ecological carrying capacity constraint is constructed as a maximum limit on the area of ​​human intervention and a maximum limit on the average number of livestock that can be carried per unit area of ​​the entire pasture per year.

[0036] In this embodiment, the specific construction logic and processing intent of the five constraints in the mixed-integer programming model are executed. This construction process is completed in the model definition section of the model processing engine module.

[0037] Construction and handling of global resource constraints: This constraint aims to ensure that, from a macro perspective of the entire annual preset cycle, the total supply of forage can cover the total demand. The model builder divides the year into four seasons, t=1 to 4. For each season, the natural forage supply is calculated as: the grazing area A_t allocated in that season multiplied by the edible grass yield per unit area G_usable(t) in that season. The annual supplementary forage supply is calculated as: the artificial grassland area S multiplied by the grass yield per unit area H, and then multiplied by the sum of the supplementary feeding ratios α(t) in the seasons when the forage is actually used for supplementary feeding, such as winter t=1 and spring t=4. The total annual forage demand is calculated as: for each season t, the number of each type of livestock i X_i is multiplied by its daily demand D_i(t) in that season, and then multiplied by the number of days in that season L(t), and then summed over all seasons and all livestock types. The constraint is constructed as: total natural forage supply + total supplementary forage supply ≥ total annual forage demand. This constraint ensures that the minimum resource requirement is met throughout the year, regardless of seasonal adjustments.

[0038] Construction and handling of seasonal resource constraints: This constraint aims to ensure the immediate carrying capacity of pastures within each specific season, avoiding overgrazing in seasons such as summer and autumn when supplementary feeding has not yet begun or should not begin. For each independent season t, a separate constraint is constructed for the model. This constraint only considers the natural forage supply of the current season, requiring that this supply must be greater than or equal to the basic demand of all livestock in that season, i.e., without considering the demand during any supplementary feeding. This means that even if the total annual supply is sufficient, pastures in summer and autumn must be able to independently support the livestock herds at that time. This forces the rotational grazing area allocation A_t to match seasonal forage production capacity and seasonal demand.

[0039] Construction and handling of supplementary feeding capacity constraints: This constraint aims to impose a dual restriction on the intervention method of artificial supplementary feeding. First, the total amount is limited: the total amount of supplementary feed consumed in all seasons where supplementary feeding is permitted, such as winter and spring, cannot exceed the total supplementary feed resources SH owned or produced by the system. Second, the intensity of the restriction: for each season t where supplementary feeding is permitted, an upper limit for the proportion of supplementary feeding is set, such as α(t) ≤ 0.6. That is, at most a certain proportion, such as 60%, of the feed demand in that season can rely on supplementary feeding, and the remainder must come from grazing in the current season or from hay reserves already included in the total demand calculation. This prevents excessive reliance on supplementary feeding and encourages the use of pastures in the current season.

[0040] Construction and handling of population structure constraints: This constraint aims to ensure that the optimized livestock population size conforms to biological laws and common sense of sustainable production management. It is achieved through a set of linear inequalities: a) Male-to-female ratio constraint: number of bulls It should be less than or equal to the total number of breeding cows. + Divide by a reasonable mating ratio, such as 20, and round up to ensure sufficient but not excessive mating capacity. b) Cow status ratio constraint: number of agalactia cows It should be between the number of dairy cows Within a certain range, such as 50% to 70%, this reflects a reasonable alternation of dry and lactating cows in the herd. c) Calf to cow ratio constraint: number of calves It should be between the number of dairy cows Within a certain range, such as 1.4 to 1.7 times, this takes into account the reproductive rate, survival rate, and the actual situation that calves are only counted into the calf herd after they have grown to a certain age.

[0041] Construction and Management of Ecological Sustainability Constraints: This constraint aims to set red lines from the perspective of the long-term health of pasture ecosystems. It includes two sub-constraints: a) Maximum area of ​​human intervention: The area of ​​artificially planted grassland, S, must not exceed a fixed value, such as 300 hectares, to prevent large-scale destruction of native grasslands or overexploitation of water resources. b) Maximum overall livestock carrying capacity: On average throughout the year, the number of livestock carried per unit of grassland area, such as per acre per day, calculated in standard sheep units or head counts, must not exceed an ecologically recommended threshold, such as 30 head / day / acre. This threshold is determined based on studies of local climate, soil type, and vegetation recovery capacity, and is used to prevent grassland degradation caused by long-term overgrazing.

[0042] Furthermore, referring to Figure 3 The method, S4, further includes: Multiple livestock quantity variable values ​​that satisfy all constraints and optimize the objective function are determined through iterative calculations and conform to the population structure constraints. Simultaneously determine the grazing area allocation variable values ​​corresponding to the multiple different time periods, and calculate the optimal supplementary feeding ratio variable value for the specific time period in which supplementary feeding is allowed, so as to satisfy the global resource constraints and the supplementary feeding capacity constraints; The variable value of artificial grassland area is determined, and together with the supplementary feeding ratio, the total amount of supplementary feed that can be provided is determined. All the above variable values ​​together constitute the solution vector, which is encapsulated as a structured data object for output.

[0043] In this embodiment, after the solver begins iterative calculation: First, it needs to determine the set of livestock quantity variables in the decision variables. , , , The solver searches within the feasible region defined by the constraints. The population structure constraints immediately define a relative proportional space for these four variables. For example, it attempts to increase... Dairy cows are used to enhance future reproductive potential, but are simultaneously affected by... Limitations on the proportion of non-dairy cows, and The constraint that calves must be increased proportionally. Simultaneously, increasing the number of any livestock immediately increases the demand on the left side of both the seasonal and global resource constraints, thus being limited by pasture area A_t and supplementary feeding capacity S, α. The solver seeks a combination by iteratively adjusting these quantities, such that the sum... + + + Maximum, and without violating any constraints.

[0044] Secondly, regarding the variables of grazing area allocation , , , The solver needs to determine how to allocate the total area to the four seasons. Seasonal resource constraints are a key driver: in seasons with high forage production, such as summer, a unit area A_t can support more demand, so allocating a smaller area may be sufficient to meet the season's needs, freeing up more area for other seasons. However, in seasons with low forage production, such as winter, even with a larger area A_t allocated, the natural supply is limited, requiring more reliance on supplemental feeding. The solver's task is to find a value for each A_t such that the natural forage supply for each season is at least equal to the basic demand for that season, and the sum of all A_t values ​​does not exceed the total pasture area.

[0045] Third, the solver performs co-optimization on the supplementary feeding ratio variables α(1), α(4) and the artificial grassland area variable S. When the natural forage supply in winter or spring is insufficient to meet the basic needs of the season, supplementary feeding must be initiated with α(t) > 0. The upper limit of the supplementary feeding capacity constraint α(t) ≤ 0.6 prevents the supplementary feeding ratio from being too high. At the same time, the total forage consumed in all supplementary feeding seasons is limited by SH. Therefore, the solver may try to increase S to obtain more supplementary feeding resources, but this is immediately limited by the upper limit of S in the ecological sustainability constraint. The solver weighs the upper limit of S, the upper limit of α(t), and the number of livestock that can be additionally supported by supplementary feeding, and calculates a set of optimal α(t) and S values, so that within the resource limit, more livestock, especially in winter, can be supported to the greatest extent, thereby increasing the total annual stock.

[0046] Ultimately, all these variables are determined simultaneously, and they are interconnected and mutually constraining. The optimal solution vector found by the solver is the set of specific values ​​that maximizes the objective function. The system encapsulates this vector into a structured data object, such as a JSON or XML file, containing clear key-value pairs for easy parsing and use by subsequent modules.

[0047] Furthermore, S5 of the method also includes: The number of livestock of different physiological categories recommended for breeding is displayed in the first display area in the form of a combination of classification statistical charts and target numbers; The proposed grazing area is allocated to different time periods and displayed in the second display area in the form of a timeline combined with map block coloring or a scale diagram. The recommended supplemental feeding intensity ratio and the recommended reserved area for artificial grass planting will be displayed in the third display area in the form of progress bars, percentage indicators and area values; In the operation guidance interface, an input control is associated with the suggested quantitative indicator. The input control is configured to allow users to input or adjust the actual execution data corresponding to the indicator as actual execution feedback.

[0048] In this embodiment, the management and distribution module creates an interactive operation guidance interface. After receiving the structured optimal aquaculture control plan data object, the module initiates the interface generation process: The first display area shows the number of livestock: the module parses out... , , , The value is displayed. It uses a front-end charting library like ECharts to create a categorized statistical chart, such as a horizontal stacked bar chart. Four bars of different colors represent four categories of cattle, with the bar length corresponding to their quantity, and specific numbers are labeled at the end of each bar. The recommended total number of cattle is prominently displayed above the chart: ∑(X_i) heads. Simultaneously, next to or below the chart, the target numbers are listed again in a clear list format, such as dairy cows: X1 heads, non-dairy cows: X2 heads, etc.

[0049] The second display area shows the rotational grazing scheme: the module analyzes... , , , The values ​​and their corresponding geographic tile IDs are associated from GIS data. It generates a timeline control, labeling spring, summer, autumn, and winter. When the user clicks or hovers the mouse over a season, such as summer, the associated map tile coloring module is activated. This module calls the base map, the electronic pasture map, and assigns the suggested grazing area corresponding to that season. The map is highlighted with a specific, semi-transparent color, such as green, while unassigned areas are displayed in grayscale. A pie chart, representing the proportion of each season's area to the total area, is displayed alongside the map.

[0050] The third display area shows the supplementary feeding and grass planting: the module parses the values ​​of α(1), α(4), and S. For the supplementary feeding ratio, it generates two progress bars, labeled as the recommended supplementary feeding ratio for winter and the recommended supplementary feeding ratio for spring, respectively. The progress bars are filled to the percentage position corresponding to α(t), and the specific percentage indicator, such as 45%, is displayed on or next to the progress bars. For artificial grass planting, it directly displays the recommended artificial grass planting area: S hectares, possibly accompanied by an icon.

[0051] Interactive Function Integration: The module embeds input controls within the aforementioned display elements. For example, next to the suggested total stock level, an input box for the actual registered stock level is added. Below the winter supplementary feeding progress bar, an input box for the actual winter supplementary feeding amount (in tons) is added. These controls are configured to allow users to input the actual completed data. When the user submits this data, the browser or mobile application sends it back to the server for storage as an actual execution feedback data packet.

[0052] Furthermore, the operation guidance interface described in this method is also configured as follows: Receive actual execution data related to the current control cycle submitted by the user through the input control; The actual execution data is compared with the corresponding suggested quantitative indicators in the optimal aquaculture control scheme; Based on the comparison results, a difference prompt message is generated and highlighted on the interface. The difference prompt message is used to remind the user to pay attention to the deviation between the actual execution and the optimization suggestions.

[0053] In this embodiment, real-time comparison and prompting logic is added to the interface.

[0054] When a user fills in actual execution data related to the current control cycle, such as the number of cattle actually purchased or the actual amount of supplementary feed completed, through the input controls on the interface and clicks the submit feedback or similar button, the JavaScript logic on the interface or the logic on the mobile application is triggered.

[0055] The client-side script retrieves the actual values ​​input by the user and compares them with the suggested quantitative indicators displayed on the current interface, which are then obtained from the scheme data object. For example, it calculates the difference between the actual number of animals in stock and the suggested total number of animals in stock; it also calculates the difference between the actual winter supplementary feeding amount converted into a percentage and the suggested winter supplementary feeding percentage.

[0056] Based on a preset difference threshold, if the deviation exceeds 10%, the script determines whether to generate a prompt. If the difference is significant, the script dynamically creates and inserts a difference prompt message into a fixed prompt area on the interface, such as a top banner or sidebar. The message content may be: Note: The actual stock level is 15% lower than the recommended value or the actual winter supplementary feeding rate is 20% higher than the recommended value.

[0057] To make the prompt more noticeable, it is typically highlighted, for example, with a yellow background, bold red font, or a flashing icon when the user first visits the page. Additionally, the script may automatically scroll the page view to the vicinity of the relevant input control or add a highlighted border around the control to draw the user's attention to the discrepancy.

[0058] The prompt message may include a link to view the cause analysis, a description of the possible impacts derived from the model constraints, or a button to ignore the prompt, to enhance interactivity.

[0059] Furthermore, in step S6 of this method, the parameter calibration process is initiated according to a preset calibration trigger condition, wherein the calibration trigger condition includes at least one of the following: Time trigger condition: after a fixed time period since the last parameter calibration or scheme generation; Data accumulation trigger condition: The amount of newly received raw monitoring data related to a specific parameter reaches a preset threshold; Event triggering conditions: Receiving an event alarm message from a user or system indicating an abnormal state in a specific area or a specific livestock group.

[0060] In this embodiment, time-based triggering is the most basic periodic triggering method. The module maintains a timer or a system-dependent timed task scheduler. For example, the rule is set to automatically trigger once every 90 days, or quarterly, since the last successful parameter calibration or successful generation of an aquaculture control plan. Regardless of whether the data changes significantly, the time condition ensures that the model parameters are regularly reviewed and updated to adapt to seasonal changes.

[0061] The data accumulation trigger condition focuses on the sufficiency of data. The module monitors the arrival of new data for specific key parameters. For example, for summer grassland productivity parameters, the rule might be set as follows: if, for the same grassland block, newly received, quality-verified, valid multispectral UAV imagery data covers more than 95% of the area of ​​that block, and the imagery data comes from within the last 30 days, then the data accumulation trigger condition related to grassland productivity parameters is met. This ensures that the data used for re-estimating parameters has sufficient spatial coverage and timeliness.

[0062] The event trigger condition responds to abnormal or unplanned situations. The module subscribes to or listens for event alarms from other system modules or manually reported events. For example, it might receive an urgent report from a user administrator or herder via their terminal about a suspected snow disaster in area XX, resulting in significant grassland cover; or an alarm automatically issued by the system health monitoring module based on cluster analysis of smart collar activity data, indicating a sharp drop in the average daily activity of cattle in area B, suggesting a possible disease outbreak. When such event information is received, and the event is marked as affecting grassland resources or livestock herd status, the event trigger condition is met. This allows the system to immediately initiate calibration in the event of unexpected situations, rather than waiting for a fixed period.

[0063] Furthermore, the step of re-estimating the parameters in the standardized input parameter set in S6 of the method specifically includes: Identify the target parameter that needs to be calibrated, the target parameter being associated with the calibration trigger condition; Extract a recent subset of data related to the target parameter from the newly received raw monitoring data stream; The recent data subset is processed using the same or modified transformation function or statistical algorithm as the initial parameter estimation to obtain the updated estimate of the target parameter. Replace the original parameter values ​​in the mixed integer programming model with the updated estimates.

[0064] In this embodiment, once the calibration trigger condition is activated, the parameter calibration service module executes the following process: The module analyzes the context associated with the triggering conditions. In the above event, the alarm clearly points to grassland growth, so the module identifies the target parameter that needs to be calibrated as the grassland productivity parameter G_usable(t) of the central grassland in the affected area in the current and subsequent seasons.

[0065] The module extracts a recent subset of data related to the target parameters from the data cache or database. This includes: the latest multispectral UAV imagery of the central region since the event; recent soil moisture sensor readings for the region; and recent precipitation and temperature data recorded by weather stations.

[0066] The module calls the same first transformation function as the initial estimate. However, due to the exceptionally arid conditions, the module may employ a modified transformation logic during the call. For example, it might temporarily lower the water influence coefficient for that region in the model, or directly use an analogy estimation based on the latest NDVI data and historical yield data from extremely dry years. After processing this new subset of data, the module obtains an updated estimate, for example, lowering the summer G_usable(t) for the central grasslands from a previously high estimate to a moderately low value.

[0067] The module replaces the parameter value representing the summer grass yield per unit area of ​​the central grassland in the mixed-integer programming model with this updated estimate. The update can be completed immediately, preparing for the next optimization calculation. The system may log this calibration, including the trigger reason, target parameter, old value, new value, and timestamp.

[0068] Furthermore, referring to Figure 4 Before the step of calling the mixed-integer linear programming solver to solve the mixed-integer programming model, the method further includes: The mixed-integer programming model is formatted to ensure that the mathematical expression of the objective function and all constraints conforms to the input specifications of the solver. The formatted model, along with the specific values ​​of the standardized input parameter set, is encapsulated together into a call request that conforms to the requirements of the solver application programming interface. The call request is sent to the solver, and the status of the solving process is monitored.

[0069] In this embodiment, after the standardized input parameter set is ready and the model definition has been loaded, the engine module performs the following steps: The engine module contains a syntax and logic checker that performs a format check on the mixed-integer programming model instance to be submitted. The checks include: ensuring all variables are defined and appear in the constraints; ensuring the objective function is linear and of a single maximize or minimize form; ensuring all constraints are linear equations or inequalities; checking for illegal characters other than variables, parameters, and constants; and verifying that mathematical expressions such as summation symbols and subscripts conform to the solver's input specifications, such as LP file format or the dictionary structure required by a specific API. If an error is found, such as a missing variable in a constraint, the checker will throw an exception and log it, preventing the solution process from continuing.

[0070] Request Encapsulation: After passing the inspection, the engine module begins encapsulating the call request. It merges the formatted model objective function coefficients, constraint matrices, constraint right-hand side terms, variable boundaries, etc., with the specific values ​​of the standardized input parameter set, such as G_usable(1) = a certain value, D_1(1) = a certain value, and fills them into a predefined data structure. This data structure fully conforms to the requirements of the selected solver's application programming interface (API).

[0071] The engine module formally sends the encapsulated call request to the solver via function call or network request. If the solver is a local library, its solver function is called directly; if it is a remote service, it is called via HTTP / RPC. After sending, the engine module starts a monitoring process to periodically query the status information returned by the solver. The monitoring process has a timeout limit and captures error information when the solver terminates abnormally, feeding it back to the upstream module or system logs.

[0072] In summary, this invention, through the engineering integration of IoT sensing, edge intelligent computing, and multi-constraint optimization theory, constructs a quantifiable, executable, and iterative smart ranch control system. From equipment selection, data flow design, model construction, algorithm implementation to human-computer interaction, each step has undergone field verification and parameter solidification, ensuring the integrity, robustness, and replicability of the technical solution, and providing solid technical support for the sustainable development of animal husbandry in high-altitude mountainous areas.

[0073] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, or indirect coupling or communication connection between apparatuses or units, and may be electrical, mechanical, or other forms.

[0074] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated units described above can be implemented in hardware or as software functional units. The above are merely embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made based on the description and drawings of this application, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

[0075] The specific embodiments of the invention have been described in detail above, but they are only examples, and this application is not limited to the specific embodiments described above. For those skilled in the art, any equivalent modifications or substitutions to the invention are also within the scope of this application. Therefore, all equivalent changes, modifications, and improvements made without departing from the spirit and principles of this application should be covered within the scope of this application.

Claims

1. A method for intelligent pasture breeding regulation based on Internet of Things data and optimization model, characterized in that, The method comprises the following steps: S1, receiving a raw monitoring data stream continuously collected by a plurality of sensing devices deployed in a physical environment of a pasture; S2, converting the received raw monitoring data stream into a standardized input parameter set recognizable by an optimization calculation model based on a preset data mapping rule; S3, inputting the standardized input parameter set into a target function mixed with a plurality of constraint conditions of a mixed integer programming model constructed based on the logic of pasture breeding; S4, calling a mixed integer linear programming solver to solve the mixed integer programming model, generating and outputting a structured optimal breeding control scheme; S5, through a graphical user interface management terminal, visualizing and organizing each item of the optimal breeding control scheme, forming an interactive operation guide interface, and issuing the operation guide interface to the user side; S6, according to a preset calibration trigger condition, starting a parameter calibration process, re-estimating the parameters in the standardized input parameter set based on the newly received raw monitoring data stream after the optimal breeding control scheme is released, and updating the corresponding input of the mixed integer programming model using the re-estimated parameters for the next round of optimization calculation.

2. The method of claim 1, wherein, Further comprising: The raw monitoring data stream at least contains resource state data reflecting the state of the grassland resources and livestock behavior data reflecting the behavior of individual or group of livestock; The standardized input parameter set at least includes a grassland productivity parameter for characterizing the grass production capacity of the grassland, a livestock demand parameter for characterizing the grass consumption rate in different seasons, and a resource constraint parameter for characterizing the upper limit of the supplementary feeding resources; The target function is used to indicate the calculation system to maximize the total number of livestock in a preset period as the optimization direction; The plurality of constraint conditions at least include a global resource constraint for ensuring that the total supply of forage is not less than the total demand of livestock, a seasonal resource constraint for ensuring that the natural forage supply in each time partition is not less than the basic demand of livestock, a supplementary feeding capacity constraint for limiting the consumption of supplementary feeding resources, a population structure constraint for regulating the proportion relationship between different categories of livestock, and an ecological carrying constraint for limiting the intensity of artificial intervention and the overall carrying density; The optimal breeding control scheme is a solution vector corresponding to the optimal value of the target function when all the constraint conditions are met, which at least specifically characterizes the number of multiple different physiological categories of livestock recommended for breeding in the next control period, the area of the grazing area recommended to be allocated in different time periods, the proportion of the supplementary feeding intensity recommended to be enabled in each specific time period, and the area of the artificial grass recommended to be reserved.

3. The method of claim 2, wherein, The S2 further comprises: Parsing first type of sensing data from the raw monitoring data stream, the first type of sensing data being derived from the monitoring of the grassland vegetation; Applying a first conversion function to the first type of sensing data to calculate the grassland productivity parameter, the grassland productivity parameter including the edible grass yield per unit area of the grassland in a plurality of different time partitions; Parsing second type of sensing data from the raw monitoring data stream, the second type of sensing data being derived from the monitoring of individual livestock; applying a second conversion function to the second type of perception data, in combination with a preset livestock category and standard daily feed intake mapping table, to derive the livestock demand parameters, the livestock demand parameters including daily average forage demand reference values of each physiological category of livestock in the plurality of different time partitions; from the original monitoring data stream or preset management configuration information, a fixed threshold value representing the upper limit of the total amount of supplementary feeding resources is obtained, and the fixed threshold value is directly used as the upper limit value of the total amount of supplementary feeding resources in the resource constraint parameter.

4. The method of claim 2, wherein, The specific construction and processing mode of the multiple sets of constraint conditions in the mixed integer programming model includes: The global resource constraint is constructed to ensure that the total amount of natural forage provided by all grazing areas in different time partitions within the entire preset period, plus the total amount of supplementary forage provided by artificial grass planting in specific time partitions, is not less than the total forage demand of all livestock within the entire preset period, thereby ensuring that resource supply covers demand from the overall time scale; The seasonal resource constraint is constructed to separately impose constraints on each time partition, requiring that the natural forage output of the allocated grazing area in the time partition is not less than the basic forage demand of all livestock relying solely on natural grazing in the time partition, thereby ensuring that the pasture can naturally support the livestock during periods without supplementary feeding or before supplementary feeding is initiated; The supplementary feeding capacity constraint is constructed to limit the total amount of supplementary forage actually used in time partitions that allow supplementary feeding to be less than the upper limit of the total amount of artificial supplementary feeding resources available to the system, and to set an upper threshold for the proportion of supplementary feeding in a single time partition; The population structure constraint is constructed to limit the proportion range between the numbers of different physiological categories of livestock through an inequality relationship; The proportion range includes the proportion relationship between the number of breeding males and the total number of breeding females, the proportion relationship between the numbers of females in different production states, and the proportion relationship between the number of young livestock and the number of specific females; this set of constraints makes the solved livestock population structure conform to biological breeding rules and normal production management; The ecological carrying capacity constraint is constructed to limit the maximum value of the artificially intervened area and the maximum value of the annual average carrying capacity per unit area of the entire pasture.

5. The method of claim 2, wherein, The S4 further includes: determining, through iterative calculation, multiple livestock quantity variable values that satisfy all constraint conditions and make the objective function optimal, in conformity with the population structure constraint; simultaneously determining grazing area area allocation variable values corresponding to the plurality of different time periods, calculating optimal supplementary feeding proportion variable values for specific time periods that allow supplementary feeding, and satisfying the global resource constraint and the supplementary feeding capacity constraint; determining an artificial grass planting area variable value, which, together with the supplementary feeding proportion, determines the total amount of supplementary forage that can be provided; all the above-mentioned variable values collectively constitute the solution vector and are encapsulated as a structured data object for output.

6. The method of claim 2, wherein, The S5 further includes: displaying the number of different physiological categories of livestock recommended for breeding in the form of a classified statistical chart combined with target numbers in a first display area; displaying the grazing area area allocated to different time periods in the form of a time axis combined with map block coloring or a proportional schematic diagram in a second display area; The recommended supplementary feeding intensity ratio and the recommended reserved artificial grass area are displayed in the third display area in the form of a progress bar, a percentage sign and an area value; In the operation guidance interface, an input control is associated with the recommended quantitative indicator, and the input control is configured to allow a user to input or adjust actual execution data corresponding to the indicator as the actual execution feedback.

7. The method of claim 5, wherein, The operation guidance interface is further configured to: receive actual execution data related to the current regulation period submitted by the user through the input control; compare the actual execution data with the corresponding recommended quantitative indicator in the optimal breeding regulation scheme; based on the comparison result, generate and highlight difference prompt information on the interface, which is used to prompt the user to pay attention to the deviation between the actual execution and the optimization suggestion.

8. The method of claim 2, wherein, The S6 starts the parameter calibration process according to a preset calibration trigger condition, and the calibration trigger condition includes at least one of the following: Time trigger condition: a fixed time period has elapsed since the last parameter calibration or scheme generation; Data accumulation trigger condition: the data amount of newly received original monitoring data related to a specific parameter reaches a preset threshold; Event trigger condition: receiving event alarm information indicating that a state anomaly occurs in a specific area or a specific livestock group, which is issued by a user or a system.

9. The method according to claim 1 or 7, characterized in that, The step of recalculating the parameters in the standardized input parameter set by the S6 specifically includes: Identify the target parameter that needs to be calibrated, which is associated with the calibration trigger condition; From the newly received original monitoring data stream, extract a recent data subset related to the target parameter; Apply the same or modified conversion function or algorithm as the initial parameter estimation to process the recent data subset to obtain the updated estimate of the target parameter; Replace the original parameter value in the mixed integer programming model with the updated estimate.

10. The method of claim 1, wherein, Before the step of calling the mixed integer linear programming solver to solve the mixed integer programming model, it further includes: Perform a format check on the mixed integer programming model to ensure that the mathematical expressions of the objective function and all constraint conditions conform to the input specifications of the solver; Encapsulate the formatted model together with the specific values of the standardized input parameter set into a call request that meets the application program interface requirements of the solver; Send the call request to the solver and monitor the status of the solving process.

Images