Grass and livestock integrated intelligent management method and system with cooperation of grass original ecology protection and production

Abstract

The invention discloses a grass and livestock integrated intelligent management system and method with cooperation of grass original ecology protection and production. The system integrates a space-air-ground integrated monitoring network, a hybrid communication network, a cloud platform and an intelligent analysis model, and realizes comprehensive acquisition, transmission, analysis and decision of grassland environment, pasture growth and livestock behavior data; a scientific grazing scheme and graded early warning are generated through dynamic grassland bearing capacity evaluation, a livestock group optimization grazing strategy and a grass and livestock balance early warning model, and decision making, feedback collection and continuous model optimization are performed by means of an intelligent mobile terminal and an intelligent electronic fence system; according to the invention, the problems of disjunction of monitoring and execution, imbalance of grass and livestock and the like in traditional grassland management are solved, and intelligent cooperation and dynamic balance of ecological protection and animal husbandry production are realized.

Description

Technical Field

[0001] This invention relates to the fields of smart agriculture and ecological information technology, and in particular to an integrated intelligent management method and system for grassland ecological protection and production coordination. Background Technology

[0002] Existing grassland monitoring technologies, such as satellite remote sensing and drone patrols, mostly remain at the macro-level, and the data obtained lacks effective connection with specific grazing management activities. This results in grassland monitoring data not being directly transformed into actionable grazing management decisions. Current grassland management systems often focus only on one aspect of grassland monitoring or livestock management, lacking a collaborative management mechanism that organically combines the two. Although technologies such as the Internet of Things and big data have been applied in the agricultural field, they have poor adaptability to the special environment of grassland pastoral areas, especially lacking technical solutions that can achieve intelligent linkage between grassland status and livestock behavior. In addition, traditional physical fencing methods are fixed and lack flexibility, unable to be dynamically adjusted according to grassland conditions, while emerging virtual fencing technologies are mostly limited to simple livestock behavior control and have failed to establish an effective feedback mechanism with grassland ecological conditions.

[0003] Therefore, there is an urgent need for an intelligent management method and system that can deeply integrate grassland ecological monitoring with livestock production management to achieve coordinated development of grassland and livestock, in order to solve the above-mentioned technical problems and promote the ecological environmental protection and modernization of animal husbandry in grassland areas. Summary of the Invention

[0004] The purpose of this invention is to provide an integrated intelligent management method for grassland ecological protection and production coordination, comprising the following steps:

[0005] Data collection involves gathering grassland environmental data, pasture growth data, and livestock behavior data through an integrated air-space-ground monitoring network.

[0006] Data transmission involves transmitting the collected data to the cloud platform via a hybrid communication network.

[0007] Data analysis and decision-making: Data is processed in the cloud platform through intelligent analysis models to generate grassland-livestock balance management decision information. The intelligent analysis models include dynamic grassland carrying capacity assessment models, livestock herd optimization grazing strategy models, and grassland-livestock balance early warning models.

[0008] The dynamic grassland carrying capacity assessment model is used to dynamically calculate the theoretically suitable carrying capacity based on real-time grassland conditions.

[0009] The livestock herd optimization grazing strategy model is used to generate zone rotational grazing schemes and livestock transfer routes; path planning is performed based on the A* algorithm; multi-objective optimization is considered, including terrain slope, distance to water source, and pasture quality; and zone rotational grazing schemes and optimal transfer routes are generated.

[0010] The grassland-livestock balance early warning model is based on the grassland environment data and pasture growth data. It uses time series analysis and machine learning prediction methods to determine the grassland-livestock balance status and generate multi-level early warning information.

[0011] Decision execution and feedback involves disseminating decision information to user terminals and field control equipment, collecting execution feedback data, and optimizing the intelligent analysis model.

[0012] Furthermore, the integrated air-space-ground monitoring network includes data from satellite remote sensing terminals, unmanned aerial vehicle (UAV) equipment, IoT sensor networks, and smart livestock collars.

[0013] Furthermore, the hybrid communication network includes: low-power wide-area network and mobile communication network; the cloud platform is built on cloud computing architecture, including distributed cloud server cluster and big data storage center, the cloud server cluster is configured with GPU computing resources and deploys containerized services; the big data storage center adopts HDFS distributed file system, stores sensor data through time series database and manages geographic information data through spatial database.

[0014] As a preferred embodiment, the dynamic grassland carrying capacity assessment model calculates the corrected theoretical yield of pasture based on real-time monitoring data and historical experience data, and then obtains the appropriate carrying capacity through a hybrid modeling method that combines mechanistic formulas and random forest algorithms.

[0015] Specifically: the physical baseline yield is calculated using the normalized vegetation index, soil moisture, and precipitation according to the mechanistic formula; then, the deviation between historical measurements and the baseline is learned using random forest, and the deviation compensation value is dynamically output by introducing environmental complexity characteristics, and the empirical coefficient of the mechanistic formula is calibrated in real time to obtain the corrected theoretical yield of forage grass.

[0016] Preferably, the corrected theoretical yield of forage grass is expressed as follows: ,

[0017] in, This is the corrected theoretical yield of forage grass. The initial coefficients are calculated based on empirical formulas of mechanism. It is a bias compensation value predicted by the random forest algorithm, used for dynamic correction. NDVI is the real-time normalized vegetation index. It is the soil volumetric moisture content.

[0018] Preferably, the data analysis and decision-making are integrated and coupled through the data flow and decision-making logic of the dynamic grassland carrying capacity assessment model, the livestock herd optimization grazing strategy model, and the grassland-livestock balance early warning model: wherein, the livestock carrying capacity data output by the dynamic grassland carrying capacity assessment model provides scale constraints for the livestock herd optimization grazing strategy model, the grazing paths and zoning schemes generated by the livestock herd optimization grazing strategy model provide spatial assessment for the grassland-livestock balance early warning model, and the early warning information issued by the grassland-livestock balance early warning model is used to trigger parameter recalibration and strategy readjustment of the dynamic grassland carrying capacity assessment model and the livestock herd optimization grazing strategy model;

[0019] Preferably, the on-site control equipment includes an intelligent electronic fence system, comprising a virtual fence management module, an intelligent collar guidance device, and a feedback learning module. The virtual fence management module provides a graphical boundary definition tool, supports dynamic adjustment of the fence range, and batch management of fence rules. The intelligent collar guidance device has a built-in sound warning device and an electric shock stimulation device, employing multimodal gradient stimulation: a continuous buzzing sound at 50 meters from the boundary, an intermittent rapid buzzing sound at 20 meters, and a 12V, 0.5-second safety electric pulse at 5 meters. The feedback learning module is used to collect guidance effect data and optimize the guidance strategy.

[0020] This invention provides an integrated intelligent management system for grassland ecological protection and production coordination, comprising:

[0021] The perception layer is an integrated air-space-ground monitoring network used to collect grassland environmental data, pasture growth data, and livestock behavior data.

[0022] The network layer is a hybrid communication network used to transmit data collected by the perception layer to the platform layer.

[0023] The platform layer is a cloud platform with a cloud computing architecture, which deploys intelligent analysis models to process data and generate grassland-livestock balance management decision information.

[0024] The application layer includes smart mobile terminals with dedicated apps installed and smart electronic fence systems, which are used to receive and execute decision information and provide feedback on the execution data.

[0025] As a preferred option, smart mobile terminals are used to display grazing navigation, supplementary feeding suggestions, and early warning information to herders.

[0026] As a preferred approach, among the intelligent analysis models, the random forest algorithm of the dynamic grassland carrying capacity assessment model can dynamically adjust the weights of NDVI and soil moisture input factors to adapt to different grassland types and phenological periods; the livestock herd optimization grazing strategy model optimizes path parameters for sheep herds, cattle herds and mixed herds, with a path grid width of ≥3m for sheep herds and ≥5m for cattle herds; the grassland-livestock balance early warning model establishes an early warning feedback mechanism and optimizes the early warning threshold and prediction model parameters every quarter.

[0027] The beneficial effects of this invention are as follows:

[0028] By constructing an integrated air-space-ground monitoring network, a dynamic grassland carrying capacity assessment model, a livestock herd optimization grazing strategy model, a grass-livestock balance early warning model, and an intelligent electronic fence system, combined with a fully closed-loop automated process of hierarchical early warning push from intelligent mobile terminals, comprehensive and high-fidelity data collection of grassland environment, pasture growth, and livestock behavior has been achieved. This improves the prediction of suitable carrying capacity and grassland utilization efficiency, reduces grassland degradation rate, effectively reduces the daily management time cost for herders, and ultimately achieves a win-win situation for grassland ecological protection and livestock production. Attached Figure Description

[0029] To better understand and implement this application, the technical solution is described in detail below with reference to the accompanying drawings.

[0030] Figure 1 A flowchart illustrating an integrated intelligent management method for grassland ecological protection and production coordination provided in Embodiment 1 of this application;

[0031] Figure 2 A schematic diagram of the early warning process for an integrated intelligent management method for grassland ecological protection and production coordination provided in Embodiment 1 of this application;

[0032] Figure 3 This is a schematic diagram of the structure of an integrated intelligent management system for grassland ecological protection and production coordination provided in Embodiment 2 of this application. Detailed Implementation

[0033] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, exemplary embodiments will be described in detail below, examples of which are illustrated in the accompanying drawings. In the following description relating to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of methods and systems consistent with some aspects of this application as detailed in the appended claims.

[0034] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The singular forms “a,” “the,” and “the” used in this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0035] The following detailed description of the specific implementation methods, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided in detail.

[0036] Example 1

[0037] Please refer to Figure 1-2 This embodiment specifically describes an integrated intelligent management method for grassland ecological protection and production coordination, comprising the following steps:

[0038] S1: Data Acquisition: Collect grassland environmental data, pasture growth data, and livestock behavior data through an integrated air-space-ground monitoring network, specifically including:

[0039] Satellite remote sensing terminals are used for large-scale monitoring of grassland vegetation conditions; they are configured to use multispectral remote sensing satellites; spatial resolution reaches 10-30 meters; equipped with a hyperspectral imager, the spectral range covers 400-2500 nm; the revisit cycle is 3-5 days, enabling continuous monitoring of large-scale grassland vegetation conditions; acquiring parameters such as vegetation index (NDVI), leaf area index (LAI), and vegetation cover;

[0040] The drone equipment, equipped with a multispectral camera, is used for detailed inspection of key areas; it is equipped with a hexacopter drone platform with a flight time of ≥45 minutes; it carries a multispectral camera and a visible light camera, with a ground resolution of 5cm; it has autonomous flight path planning capabilities to achieve detailed inspection of key areas; it can identify pasture types, measure grass height, and detect degraded patches;

[0041] An IoT sensor network is deployed at key nodes in the grassland to collect soil and meteorological data. It adopts the LoRaWAN communication protocol and has a transmission distance of up to 10km. Soil temperature and humidity sensors are deployed to measure depths of 0-50cm. Weather stations are configured to monitor parameters such as temperature, humidity, wind speed, and precipitation. The node deployment density is 1-2 per 100 hectares to achieve full coverage of key areas.

[0042] The intelligent livestock collar integrates a GPS / BeiDou positioning module, an accelerometer, and a solar charging unit for tracking livestock location and behavior. It features integrated BeiDou / GPS dual-mode positioning with an accuracy of ≤2 meters; a built-in three-axis accelerometer with a sampling frequency of 10Hz; a solar panel with a conversion efficiency of ≥20%; and a waterproof and dustproof design with an IP67 protection rating. S2: Data Transmission: The collected data is transmitted to the cloud platform via a hybrid communication network, which specifically includes:

[0043] Low-power wide-area network: Deploy LoRa gateways with a coverage radius of 5-10km; data transmission rate of 0.3-50kbps; supports star network topology with a maximum number of connections ≥1000 nodes;

[0044] Mobile communication network: Employs 4G / 5G communication modules; supports TCP / IP protocol stack; features automatic network switching function;

[0045] Furthermore, the cloud platform is built on a cloud computing architecture, specifically including:

[0046] Cloud server clusters: adopt a distributed architecture and support elastic scaling; configure GPU computing resources for deep learning model training; deploy containerized services to implement a microservice architecture;

[0047] Big Data Storage Center: Employs a distributed file system (HDFS); uses a time-series database to store sensor data; and establishes a spatial database to manage geographic information data.

[0048] S3: Data Analysis and Decision Making: Data is processed through intelligent analysis models in the cloud platform to generate grassland-livestock balance management decision information; the intelligent analysis models include: dynamic grassland carrying capacity assessment model, livestock herd optimization grazing strategy model, and grassland-livestock balance early warning model;

[0049] The dynamic grassland carrying capacity assessment model is used to dynamically calculate the theoretically suitable carrying capacity based on real-time grassland conditions. Based on real-time monitoring data and historical experience data, the corrected theoretical forage yield is calculated using a hybrid modeling method that combines mechanistic formulas with random forest algorithms, and then the suitable carrying capacity is obtained.

[0050] Specifically, the physical baseline yield is first calculated using the normalized vegetation index, soil moisture, and precipitation according to the mechanistic formula. Then, random forest is used to learn the deviation between historical measurements and the baseline, and the deviation compensation value is dynamically output by introducing environmental complexity characteristics. The empirical coefficients of the mechanistic formula are calibrated in real time, thereby obtaining a corrected theoretical forage yield that combines ecological interpretability and phenological adaptability. The corrected theoretical forage yield is finally used to estimate the appropriate carrying capacity, taking into account both physical interpretability and complex ecological adaptability.

[0051] The corrected theoretical yield of forage grass is expressed as follows: ,in, This is the corrected theoretical yield of forage grass. The initial coefficients are calculated based on empirical formulas of mechanism. It is a bias compensation value predicted by the random forest algorithm, used for dynamic correction. NDVI is the real-time normalized vegetation index. P is the soil volumetric moisture content, and P is the cumulative precipitation over the past 7 days;

[0052] Based on the corrected theoretical forage yield Y, combined with the daily forage growth rate h(t) fitted from historical data and the standard sheep's daily feed intake C (2.5 kg), the suitable carrying capacity St for the next t days (7 ≤ t ≤ 30) is dynamically calculated:

[0053]

[0054] Where, A: Grazing area (hectares); R: Forage retention rate (≥20%, ensuring regeneration, i.e., R=0.2); h(t): Forage growth coefficient for the next t days (based on historical data fitting, warm season h(t)=1+0.05t, cold season h(t)=1-0.02t); C: Average daily feed intake per standard sheep unit (2.5kg dry matter / day); t: Number of predicted days (7≤t≤30).

[0055] Differentiated calculation logic is adapted for different grassland types: grass height correction item is added for typical grassland areas (based on UAV measured data), soil moisture weight is increased to 30% for desert grassland areas, and groundwater level influence factor is added for wetland grassland areas.

[0056] The dynamic grassland carrying capacity assessment model solves the problem of mismatch between traditional static carrying capacity and real-time changes in grassland, ensuring a dynamic balance between grazing intensity and grassland regeneration capacity.

[0057] Livestock grazing optimization strategy model: used to generate zoned rotational grazing schemes and livestock transfer routes; path planning based on A* algorithm; considering multi-objective optimization such as terrain slope, water source distance, and pasture quality; generating zoned rotational grazing schemes and optimal transfer routes;

[0058] Specifically, the livestock herd optimization grazing strategy model is based on satellite remote sensing NDVI data, detailed drone surveys of grass height (≥5cm for high-quality areas, 3-5cm for suitable areas, and <3cm for rest areas), and soil moisture data. The K-means clustering algorithm is used to divide the pasture area into 3-5 rotational grazing units, with the area of ​​each unit determined according to the herd size (15-20 hectares for every 50 standard sheep units). The division must consider the following: vegetation cover difference between units ≤15%, straight-line distance from water sources ≤3km, and the existence of natural or artificial pathways between units (width ≥5m, suitable for herd passage).

[0059] The core optimization objectives and their weights were determined as follows: uniformity of pasture grazing (weight 40%, ensuring forage residue height ≥ 2cm within the unit), energy consumption for livestock movement (weight 30%, total path length ≤ 5km / day), terrain adaptability (weight 20%, avoiding areas with slopes > 25°, swamps, and bare rock areas), and water accessibility (weight 10%, maximum distance from the herd to the water source during grazing ≤ 2km). Constraints were set: grazing duration for a single rotational grazing unit ≤ 7 days (warm season), ≤ 15 days (cool season), and interval between two grazing periods ≥ 21 days (forage regeneration cycle); single rotational grazing unit... Maximum grazing time ( The number of days (in days) is determined by the initial grass height, herd grazing intensity, and pasture regeneration threshold:

[0060] in, Initial grass layer height of unit i (cm, measured by UAV); Minimum retention height of pasture (≥2cm, i.e., Hmin=2); Unit i: Average daily growth height of pasture (cm / day, warm season) =0.5−1.0, cold season =0.1−0.3); S: herd size (standard sheep unit); α: average daily feed intake height coefficient per standard sheep unit (0.002 cm) -1 ·head -1 ·sky -1 This means that the grass height decreases by 0.002 cm per sheep per day due to grazing.

[0061] A gridded map (5m×5m grid size) is constructed for the livestock herd optimization grazing strategy model. Terrain slope, pasture quality, and path width are converted into cost functions (grid cost is set to infinity for slopes > 25°, 1 for high-quality pasture, and 2 for suitable areas). Starting from the current herd location and ending at the center of the target rotational grazing unit, the A* algorithm searches for the path with the minimum cost, while simultaneously optimizing the number of path inflection points (≤3) to reduce herd detours. Path parameters are optimized for different livestock species: sheep path grid width ≥ 3m, cattle ≥ 5m, and for mixed herds, the main path is designed according to cattle standards, while simultaneously planning branch grazing paths for sheep.

[0062] The total cost of the optimal path is a weighted sum of terrain, pasture quality, and distance:

[0063] Wherein, weight: ; Slope cost (rasterization, slope > 25°) =∞, otherwise =0.1 × slope). The cost of pasture quality (premium areas) =1, =2, =5); Distance cost (unit grid distance × 0.5, grid size 5m × 5m, i.e., cost 0.5 per 5m);

[0064] The livestock herd optimization grazing strategy model is based on the A* algorithm to build an intelligent grazing decision system, which achieves the triple goals of full utilization of pasture, balanced grazing of livestock, and restoration of grassland ecology, and solves the problems of uneven grassland utilization, excessive livestock energy consumption, and expansion of degraded patches in traditional grazing.

[0065] Grassland-livestock balance early warning model: Based on grassland environment and pasture growth data collected by an integrated air-space-ground monitoring network, the model uses time series analysis and machine learning methods to predict the grassland-livestock balance status 3-7 days in advance and generate risk warning information at levels 1-4, realizing the intelligent management transformation from "post-event response" to "pre-event intervention".

[0066] The grassland-livestock balance early warning model operates through a closed-loop logic of data fusion, indicator prediction, index calculation, level matching, and feedback optimization. The specific process is as follows:

[0067] Multi-source data fusion and indicator preprocessing:

[0068] The model performs data fusion every 2 hours, and the core monitoring indicators and frequencies are as follows:

[0069] NDVI (daily, satellite remote sensing); soil moisture (hourly, IoT sensor); pasture grazing rate (daily, data fusion of drone and smart collar data); livestock density (daily, smart collar statistics); cumulative precipitation over the past 10 days, pasture regeneration rate, and historical warning frequency for the same period (used to optimize accuracy).

[0070] The original indicators are standardized and mapped uniformly to the [0,1] interval:

[0071]

[0072]

[0073] (Feed intake rate is a negative indicator);

[0074] Hybrid prediction model:

[0075] A hybrid model combining ARIMA and LSTM is used to predict key indicators for the next 3 to 7 days (k=3~7): ARIMA model: captures linear trends and seasonal patterns; LSTM neural network: learns non-linear dependencies and long-term memory effects; weighted fusion: the prediction results of the two algorithms are weighted to generate the final predicted value. (NDVI forecast for day k) (Predicted soil moisture value for the kth day) (Predicted feed intake rate on day k).

[0076] Comprehensive Index Calculation and Early Warning Level Classification:

[0077] Substitute the forecast indicators into the composite index formula to calculate the grass-livestock balance index for the k-th day in the future. :

[0078]

[0079] according to The values ​​are used to classify warning levels. : The system takes the highest warning level in the next 3-7 days as the final warning output.

[0080] Early warning triggering and model linkage

[0081] After the warning information is generated, the push method is determined according to the warning level: Level 1-2 warnings are pushed via text notification through the APP; Level 3-4 warnings, in addition to being pushed via APP, SMS, etc., will also trigger parameter recalibration and strategy readjustment of other intelligent models.

[0082] when When the value is ≥3, the pasture retention rate is dynamically adjusted. : High-level warnings (≥3) will be simultaneously sent to county-level grassland management departments to initiate joint intervention.

[0083] Conditions for lifting the warning

[0084] To ensure the rigor and ecological safety of early warning cancellation, the system is equipped with a tiered, multi-condition cancellation mechanism; an early warning of the corresponding level can only be cancelled when the following real-time monitoring conditions and predicted trend conditions are met simultaneously:

[0085] Level 1 warning lifted: The NDVI value that needs to be monitored in real time reaches 0.4 or above, the soil moisture reaches 20% or above, and the forage consumption rate drops to below 25%, and this state is maintained stably for 3 consecutive days; at the same time, the average grass-livestock balance comprehensive index predicted by the model for the next 3 days is not lower than 0.7 (safe threshold).

[0086] Level 2 warning lifted: The NDVI value that needs to be monitored in real time reaches 0.3 or above, the soil moisture reaches 15% or above, the forage consumption rate drops to below 30%, and this state is maintained stably for 5 consecutive days; at the same time, the average grass-livestock balance index predicted by the model for the next 3 days is not lower than 0.5.

[0087] Level 3 warning lifted: The NDVI value that needs to be monitored in real time reaches 0.25 or above, the soil moisture reaches 12% or above, and the forage consumption rate drops to below 35%, and this state is maintained stably for 7 consecutive days; at the same time, the average grass-livestock balance index predicted by the model for the next 3 days is not lower than 0.3, and it is verified that the relevant rotational grazing plan does not need to be adjusted urgently.

[0088] Level 4 warning lifted: The NDVI value that needs to be monitored in real time reaches 0.2 or above, the soil moisture reaches 10% or above, the forage consumption rate drops to below 40%, and this state is maintained stably for 10 consecutive days; at the same time, the average grass-livestock balance comprehensive index predicted by the model for the next 3 days is not lower than 0.3 and shows a continuous upward trend, supplemented by field verification of vegetation restoration in degraded areas.

[0089] The lifting of all warning levels requires that the three core indicators of NDVI, soil moisture, and foraging rate be met simultaneously, and the required duration of stability increases with the level of warning to ensure that the ecosystem can fully recover.

[0090] Feedback learning and model optimization

[0091] The model establishes a quarterly feedback mechanism to collect data on the effects of herders implementing adjustment measures (such as changes in NDVI 10 days after reducing livestock carrying capacity). This data is then used to optimize the early warning threshold and prediction model parameters, forming a self-improving closed loop.

[0092] The data flow and decision-making logic of the dynamic grassland carrying capacity assessment model, the livestock herd optimization grazing strategy model, and the grassland-livestock balance early warning model are integrated and coupled: the livestock carrying capacity data output by the dynamic grassland carrying capacity assessment model provides scale constraints for the livestock herd optimization grazing strategy model, the grazing paths and zoning schemes generated by the livestock herd optimization grazing strategy model provide spatial assessment for the grassland-livestock balance early warning model, and the early warning information issued by the grassland-livestock balance early warning model is used to trigger parameter recalibration and strategy readjustment of the dynamic grassland carrying capacity assessment model and the livestock herd optimization grazing strategy model.

[0093] Furthermore, the data flow and decision-making logic of the dynamic grassland carrying capacity assessment model, the livestock herd optimization grazing strategy model, and the grassland-livestock balance early warning model are integrated and coupled, which is achieved through the following specific steps:

[0094] Step 1: Generation of Grazing Strategies Driven by Carrying Capacity Data

[0095] Data Release: After the dynamic grassland carrying capacity assessment model calculates the theoretical suitable carrying capacity St for the next t days, the data is encapsulated in JSON format and released as a data service through a RESTful API interface.

[0096] Strategy Invocation and Constraints: Before initiating the rotational grazing scheme calculation, the livestock herd optimization grazing strategy model first calls the above API to obtain the latest St data; this carrying capacity is used as a rigid scale constraint and is substituted into the rotational grazing unit area calculation model (unit area = St × area required per unit of livestock) to ensure that the total area of ​​the divided rotational grazing units strictly matches the current carrying capacity of the grassland.

[0097] Step Two: Grazing Plan Supporting Early Warning Spatial Assessment

[0098] Solution storage: The rotational grazing schemes and optimal transfer paths (including rasterized cost maps) generated by the livestock herd optimization grazing strategy model are written into the spatial database of the cloud platform.

[0099] Spatial Assessment: When performing comprehensive index calculation and early warning judgment, the grassland-livestock balance early warning model obtains the boundaries of rotational grazing units planned for grazing in the next 3-7 days in real time through a spatial query interface. The model focuses the early warning analysis on these specific spatial ranges, realizing spatial assessment from the overall average to the precise grazing area, significantly improving the pertinence and accuracy of the early warning.

[0100] Step 3: Early warning information triggers model parameter recalibration and strategy readjustment

[0101] Event Trigger: When the grassland-livestock balance early warning model issues a Level 3 or Level 4 early warning, the system will publish the event to the message queue;

[0102] Model recalibration: The dynamic grassland carrying capacity assessment model subscribes to this message topic. Once a message is received, the parameter recalibration process is initiated. For example, the pasture retention rate R is dynamically increased from the baseline value of 0.2 to 0.25 (level 3) or 0.3 (level 4) according to the warning level. This directly reduces the suitable carrying capacity in subsequent calculations, leaving more room for grassland restoration.

[0103] At the same time, it can trigger its internal random forest algorithm to start an incremental learning round using the latest data, quickly update the bias compensation value prediction model, and adapt to the rapid changes in the environment.

[0104] Strategy Reorganization: The livestock herd optimization grazing strategy model also subscribes to the early warning message; upon receiving the message, it immediately initiates the strategy reorganization process; the model first discards the original grazing paths and zoning schemes involving the early warning area; then, starting from the current livestock position, it re-executes multi-objective path planning based on the A* algorithm, and sets an extremely high passage cost for the early warning area in the cost function, thereby generating new grazing paths and zoning schemes that bypass the early warning area, achieving proactive risk avoidance.

[0105] Through the above three steps, a closed-loop intelligence is formed, which uses data feedforward to drive decision-making and early warning feedback to trigger optimization. This enables the system to have full-link adaptive capabilities in perception, decision-making, execution, and optimization, effectively solving the core problem of the disconnect between monitoring and execution in traditional management.

[0106] S4 Decision Execution and Feedback Steps: Decision information is sent to user terminals and on-site control equipment, execution feedback data is collected, and the intelligent analysis model is optimized; including: an intelligent mobile terminal with a dedicated APP installed, used to display grazing navigation, supplementary feeding suggestions, and early warning information to herders; and an intelligent electronic fence system, including a virtual fence management module and a livestock guidance device, which can activate multimodal gradient stimulation guidance based on livestock location information.

[0107] Specifically, the intelligent electronic fence system includes:

[0108] The virtual fence management module provides a graphical boundary definition tool; supports dynamic adjustment of fence range; enables batch management of fence rules; and is used for graphically defining and dynamically adjusting fence boundaries on a digital map. The intelligent collar guidance device has a built-in sound warning device and electric shock stimulation device, which activates corresponding guidance stimuli based on the relative position of the livestock and the virtual boundary. It employs multimodal gradient stimulation guidance: 50 meters from the boundary: emits a continuous buzzing sound; 20 meters from the boundary: emits an intermittent rapid buzzing sound; 5 meters from the boundary: emits a safe electric pulse (12V, duration 0.5 seconds). A feedback learning module is used to collect guidance effect data and optimize guidance strategies.

[0109] A dedicated app pushes grazing navigation, supplementary feeding suggestions, and early warning information to herders, while the intelligent electronic fence system uses graphical boundary management, multimodal gradient stimulus guidance, and feedback learning mode to accurately guide the grazing range of livestock.

[0110] This embodiment discloses an integrated intelligent management method for grassland and livestock, focusing on grassland ecological protection and production synergy. It constructs a closed-loop intelligent system encompassing perception, transmission, analysis, execution, and feedback. Its core logic involves comprehensively collecting grassland environment, pasture growth, and livestock behavior data through an integrated air-space-ground monitoring network. This data is then transmitted to a cloud platform with a cloud computing architecture via a hybrid communication network. Based on three intelligent analysis models, precise management decisions are generated. Finally, these decisions are implemented through intelligent mobile terminals and intelligent electronic fence systems. The models are continuously optimized based on feedback data to achieve a dynamic balance between ecology and production.

[0111] Example 2

[0112] Please see Figure 3 This embodiment provides an integrated intelligent management system for grassland ecological protection and production coordination, applied to an integrated intelligent management method for grassland ecological protection and production coordination, specifically including:

[0113] The perception layer, an integrated air-space-ground monitoring network, is used to collect grassland environmental data, pasture growth data, and livestock behavior data, specifically including:

[0114] Satellite remote sensing terminal: It adopts a multispectral remote sensing satellite with a spatial resolution of 10–30 meters, equipped with a hyperspectral imager with a spectral range covering 400–2500 nm and a revisit period of 3–5 days, and is used to acquire vegetation parameters such as NDVI, LAI, and vegetation cover.

[0115] Unmanned aerial vehicle (UAV) equipment: Equipped with a hexacopter UAV platform, with a flight time of ≥45 minutes, equipped with a multispectral camera and a visible light camera, with a ground resolution of 5cm, and has autonomous flight path planning function, used to identify pasture types, measure grass height, and detect degraded patches;

[0116] IoT sensor network: Adopting the LoRaWAN communication protocol, the transmission distance reaches 10km. Soil temperature and humidity sensors (measuring depth 0-50cm) and weather stations (monitoring temperature, humidity, wind speed, and precipitation) are deployed, with a node deployment density of 1-2 per 100 hectares;

[0117] Intelligent livestock collar: integrates a Beidou / GPS dual-mode positioning module (positioning accuracy ≤2 meters), a three-axis accelerometer (sampling frequency 10Hz) and a solar charging unit (conversion efficiency ≥20%), with a protection level of IP67, used to track the location and behavior of livestock.

[0118] The network layer, a hybrid communication network, is used to reliably transmit data collected by the perception layer to the platform layer, and specifically includes:

[0119] Low-power wide-area network: Deploy LoRa gateways with a coverage radius of 5–10km, data transmission rate of 0.3–50kbps, support for star network topology, and a maximum number of connections ≥1000 nodes;

[0120] Mobile communication network: It adopts 4G / 5G communication modules, supports TCP / IP protocol stack, and has automatic network switching function to ensure the real-time performance and stability of data transmission.

[0121] The platform layer, a cloud platform with a cloud computing architecture, deploys intelligent analysis models to process data and generate grassland-livestock balance management decision information. The dynamic grassland carrying capacity assessment model, the livestock herd optimization grazing strategy model, and the grassland-livestock balance early warning model within the intelligent analysis models are coupled through data interfaces and service calls, forming a closed-loop collaborative decision-making chain, specifically including:

[0122] Cloud server clusters: adopt a distributed architecture, support elastic scaling, configure GPU computing resources, deploy containerized services, and implement a microservice architecture;

[0123] Big Data Storage Center: Employs the HDFS distributed file system, uses a time-series database to store sensor data, and establishes a spatial database to manage geographic information data;

[0124] Intelligent analysis models, including:

[0125] Dynamic grassland carrying capacity assessment model: This model employs a hybrid modeling approach combining mechanistic formulas and random forest algorithms to dynamically calculate the theoretically suitable carrying capacity based on real-time grassland conditions. Input parameters include NDVI, soil moisture, precipitation, and historical carrying capacity. Output dynamic carrying capacity recommendations for the next 7–30 days, with a prediction accuracy of ≥85%.

[0126] Livestock herd optimization grazing strategy model: Based on the A* algorithm, multi-objective path planning is used to generate zonal rotational grazing schemes and livestock transfer paths with the lowest cost. Considering factors such as terrain slope, water source distance, and pasture quality, zonal rotational grazing schemes and optimal transfer paths are generated, which are suitable for sheep flocks (path width ≥ 3m), cattle herds (≥ 5m), and mixed livestock herds.

[0127] Grassland-livestock balance early warning model: It predicts the core indicators for the next 3-7 days using a hybrid prediction model (ARIMA and LSTM) and classifies the early warning levels into 1-4 based on the comprehensive prediction index. The model is coupled with the dynamic grassland carrying capacity assessment model and the livestock herd optimization grazing strategy model, which can trigger parameter recalibration and strategy reorganization. It also has tiered early warning cancellation conditions and a quarterly feedback optimization mechanism.

[0128] The application layer receives and executes decision information from the platform layer, and also provides feedback on the execution data. Specifically, it includes:

[0129] Smart mobile terminal: Install a dedicated APP to display grazing navigation, supplementary feeding suggestions and graded early warning information to herders (levels 1-2 are text prompts, level 3 adds SMS and voice reminders, and level 4 is pushed to the county-level management department).

[0130] The intelligent electronic fence system includes:

[0131] Virtual Fence Management Module: Provides a graphical boundary definition tool, allowing users to dynamically define, adjust, and manage electronic fence boundaries and rules, and supports dynamic adjustment of fence range and batch management of rules;

[0132] Intelligent collar guidance device: Built-in sound warning and electric shock stimulation device. Based on the distance between the livestock and the virtual boundary, it activates multimodal gradient stimulation to guide the behavior of the herd. Multimodal gradient stimulation guidance is used: a continuous buzzing sound is emitted at 50 meters from the boundary, an intermittent rapid buzzing sound is emitted at 20 meters, and a 12V, 0.5-second safe electric pulse is emitted at 5 meters.

[0133] Feedback Learning Module: Collects guidance effect data and continuously optimizes guidance strategies based on this data to improve the system's adaptability.

[0134] The system described in this embodiment comprehensively collects data through the perception layer, reliably transmits it to the platform layer for intelligent analysis and decision-making through the network layer, and finally executes and provides feedback through the application layer, forming a closed-loop intelligent management process of perception, transmission, analysis, execution, and feedback, thereby realizing the synergistic optimization of grassland ecological protection and animal husbandry production.

Claims

1. A smart management method for integrated grassland ecological protection and production coordination, characterized in that, Includes the following steps: Data collection involves gathering grassland environmental data, pasture growth data, and livestock behavior data through an integrated air-space-ground monitoring network. Data transmission involves transmitting the collected data to a cloud platform with a cloud computing architecture via a hybrid communication network. Data analysis and decision-making: Data is processed in the cloud platform through intelligent analysis models to generate grassland-livestock balance management decision information. The intelligent analysis models include dynamic grassland carrying capacity assessment models, livestock herd optimization grazing strategy models, and grassland-livestock balance early warning models. The dynamic grassland carrying capacity assessment model is used to dynamically calculate the theoretically suitable carrying capacity based on real-time grassland conditions. The livestock herd optimization grazing strategy model is used to generate zone rotational grazing schemes and livestock transfer paths; path planning is performed based on the A* algorithm to generate zone rotational grazing schemes and optimal transfer paths; The grassland-livestock balance early warning model is based on the grassland environment data and pasture growth data. It uses time series analysis and machine learning prediction methods to determine the grassland-livestock balance status and generate multi-level early warning information. Decision execution and feedback involves disseminating decision information to user terminals and field control equipment, collecting execution feedback data, and optimizing the intelligent analysis model.

2. The integrated intelligent management method for grassland ecological protection and production coordination according to claim 1, characterized in that, The integrated air-space-ground monitoring network includes data from satellite remote sensing terminals, unmanned aerial vehicle (UAV) equipment, IoT sensor networks, and smart livestock collars.

3. The integrated intelligent management method for grassland ecological protection and production coordination as described in claim 1, characterized in that, The hybrid communication network includes a low-power wide-area network and a mobile communication network; the cloud platform is built on a cloud computing architecture, including a distributed cloud server cluster and a big data storage center, the cloud server cluster is configured with GPU computing resources and deploys containerized services; the big data storage center adopts the HDFS distributed file system, stores sensor data through a time-series database, and manages geographic information data through a spatial database.

4. The integrated intelligent management method for grassland ecological protection and production coordination according to claim 1, characterized in that, The dynamic grassland carrying capacity assessment model, based on real-time monitoring data and historical experience data, calculates the corrected theoretical yield of pasture through a hybrid modeling method that combines mechanistic formulas with random forest algorithms, and then obtains the appropriate carrying capacity. Specifically: the physical baseline yield is calculated using the normalized vegetation index, soil moisture, and precipitation according to the mechanistic formula; then, the deviation between historical measurements and the baseline is learned using random forest, and the deviation compensation value is dynamically output by introducing environmental complexity characteristics, and the empirical coefficient of the mechanistic formula is calibrated in real time to obtain the corrected theoretical yield of forage grass.

5. The integrated intelligent management method for grassland ecological protection and production coordination as described in claim 4, characterized in that, The corrected theoretical yield of forage grass is expressed as follows: , in, This is the corrected theoretical yield of forage grass. The initial coefficients are calculated based on empirical formulas of mechanism. It is a bias compensation value predicted by the random forest algorithm, used for dynamic correction. NDVI is the real-time normalized vegetation index. It is the soil volumetric moisture content.

6. The integrated intelligent management method for grassland ecological protection and production coordination according to claim 1, characterized in that, The data analysis and decision-making are integrated and coupled through the data flow and decision-making logic of the dynamic grassland carrying capacity assessment model, the livestock herd optimization grazing strategy model, and the grassland-livestock balance early warning model. Specifically, the livestock carrying capacity data output by the dynamic grassland carrying capacity assessment model provides scale constraints for the livestock herd optimization grazing strategy model, the grazing paths and zoning schemes generated by the livestock herd optimization grazing strategy model provide spatial assessment for the grassland-livestock balance early warning model, and the early warning information issued by the grassland-livestock balance early warning model is used to trigger parameter recalibration and strategy readjustment of the dynamic grassland carrying capacity assessment model and the livestock herd optimization grazing strategy model.

7. The integrated intelligent management method for grassland ecological protection and production coordination according to claim 1, characterized in that, The on-site control equipment includes an intelligent electronic fence system, which comprises a virtual fence management module, an intelligent collar guidance device, and a feedback learning module. The virtual fence management module provides a graphical boundary definition tool, supports dynamic adjustment of the fence range, and batch management of fence rules. The intelligent collar guidance device has a built-in sound warning device and an electric shock stimulation device, employing multimodal gradient stimulation: a continuous buzzing sound at 50 meters from the boundary, an intermittent rapid buzzing sound at 20 meters, and a 12V, 0.5-second safety electric pulse at 5 meters. The feedback learning module is used to collect guidance effect data and optimize the guidance strategy.

8. A grassland-livestock integrated intelligent management system for implementing the grassland ecological protection and production synergy intelligent management method as described in any one of claims 1-7, characterized in that, include: The perception layer is an integrated air-space-ground monitoring network used to collect grassland environmental data, pasture growth data, and livestock behavior data. The network layer is a hybrid communication network used to transmit data collected by the perception layer to the platform layer. The platform layer is a cloud platform with a cloud computing architecture, which deploys intelligent analysis models to process data and generate grassland-livestock balance management decision information. The application layer includes smart mobile terminals with dedicated apps installed and smart electronic fence systems, which are used to receive and execute decision information and provide feedback on the execution data.

9. The integrated intelligent management system for grass and livestock as described in claim 8, characterized in that, The smart mobile terminal is used to display grazing navigation, supplementary feeding suggestions, and early warning information to herders.

Images