Cattle and sheep grassland digital management method and system

Abstract

The invention belongs to the technical field of cattle and sheep positioning, and discloses a cattle and sheep grassland digital management method and system, and the method comprises the steps: obtaining a pre-established grassland geographic information rasterization model, combining with positioning data which is uploaded by a sparse cattle and sheep positioning device and is provided with a positioning label and a time label, and carrying out the digital management of a cattle and sheep grassland according to the time of uploading the positioning data; obtaining initial position information of cattle and sheep at the moment; according to the initial position information of the cattle and the sheep, the position of the cattle and the sheep in the rasterized model at the next moment is predicted by fusing the movement speed and the movement direction of the cattle and the sheep through a Kalman filtering algorithm, and the predicted position of the cattle and the sheep is obtained; according to the cattle and sheep correction position and the cattle and sheep initial position, adopting a nearest neighbor algorithm to match the cattle and sheep correction position with a historical activity track, and if the deviation of the matched track is greater than a preset threshold value, triggering an abnormal behavior alarm; and determining the distribution state of the cattle and sheep groups in the rasterized model by using a density clustering algorithm according to the corrected positions of all cattle and sheep, and obtaining the center position and range of the cattle and sheep group distribution.

Description

Technical Field

[0001] The present invention relates to the technical field of cattle and sheep positioning, and specifically relates to a method and system for digital management of cattle and sheep pastures. Background Art

[0002] In the digital management of cattle and sheep pastures, the core technical problem lies in how to accurately, continuously and at low cost track the movement trajectories of individual cattle and sheep in a vast and ever-changing pasture environment, while ensuring the stability of the system during long-term operation and the reliability of data.

[0003] Specifically, the pasture environment is complex, with many interfering factors such as terrain undulations, vegetation occlusion, and weather changes. Traditional positioning technologies based on the Global Positioning System work well in open areas, but in areas with weak signals or severe occlusion, their positioning accuracy and stability will drop significantly. At the same time, equipping each cattle and sheep with high-precision positioning devices is too costly, and the battery life of the devices is also a challenge. How to infer the precise positions of each cattle and sheep through algorithms using limited positioning information, combined with the movement patterns of the cattle and sheep group, under the premise of low power consumption and low cost, and handle the problem of data loss caused by signal loss or delay is a key technical problem.

[0004] In addition, the activities of cattle and sheep are highly random and unpredictable. How to dynamically adjust the grazing strategy based on the real-time tracked position information of cattle and sheep, avoid overgrazing of local areas of the pasture, and ensure the activity range and safety of cattle and sheep is also a complex problem that needs to be solved. Summary of the Invention

[0005] To solve the above technical problems, the present invention provides a method and system for digital management of cattle and sheep pastures. The core lies in achieving refined management of individual and group cattle and sheep through sparse positioning data, effectively solving the problems of sparse positioning data and extensive management in traditional animal husbandry, and realizing the rational utilization of pastures and the healthy breeding of cattle and sheep.

[0006] A method for digital management of cattle and sheep pastures, the method comprising:

[0007] Obtain a pre-established rasterized model of pasture geographic information, combine the positioning data with positioning tags and time tags uploaded by sparse cattle and sheep positioning devices, and obtain the initial position information of cattle and sheep at this moment according to the time when the positioning data is uploaded;

[0008] According to the initial position information of cattle and sheep, fuse the movement speed and movement direction of cattle and sheep, predict the position of cattle and sheep in the rasterized model at the next moment, and obtain the predicted position of cattle and sheep;

[0009] Obtain the positioning signal strength uploaded by the positioning device, and according to the signal propagation model, judge the positioning signal strength of the predicted positions of cattle and sheep. If the positioning signal strength is lower than the preset threshold, then adopt the positioning algorithm based on the time difference of arrival, use the time difference of signal reception by multiple positioning devices to calculate the positions of cattle and sheep, and obtain the corrected positions of cattle and sheep;

[0010] According to the corrected positions of cattle and sheep and the initial positions of cattle and sheep, adopt the nearest neighbor algorithm to match the corrected positions of cattle and sheep with the historical activity trajectories. If the deviation of the matched trajectories is greater than the preset threshold, then trigger an abnormal behavior alarm;

[0011] According to the corrected positions of all cattle and sheep, use the density clustering algorithm to determine the distribution state of the cattle and sheep group in the rasterized model, and obtain the central position and range of the cattle and sheep group distribution;

[0012] According to the central position and range of the cattle and sheep group distribution, combined with the pre-established grassland carrying capacity model, judge the grassland pressure in the local area. If the grassland pressure in the local area exceeds the preset threshold, then generate a new activity range for cattle and sheep according to the historical activity trajectories of cattle and sheep.

[0013] Preferably, obtain the pre-established rasterized model of grassland geographic information, combine the positioning data with positioning tags and time tags uploaded by sparse cattle and sheep positioning devices, and according to the time when the positioning data is uploaded, obtain the initial position information of cattle and sheep at this time, including:

[0014] Obtain the rasterized model of grassland geographic information, judge the time tag of the data according to the data uploaded by the cattle and sheep positioning device, and determine the corresponding time of the data time tag;

[0015] According to the corresponding time of the data time tag, extract the grassland geographic information at this time from the rasterized model of grassland geographic information, judge the distribution of positioning tags in the grassland geographic information at this time, and obtain the positioning tag distribution information at this time;

[0016] Adopt the positioning tag distribution information at this time, and according to the positioning tag in the data uploaded by the cattle and sheep positioning device, judge the cattle and sheep corresponding to the positioning tag, and obtain the information of the cattle and sheep corresponding to the positioning tag;

[0017] According to the information of the cattle and sheep corresponding to the positioning tag, combined with the grassland geographic information at this time, if the cattle and sheep information matches the positioning tag distribution information in the grassland geographic information, then adopt the positioning tag distribution information in the grassland geographic information to obtain the initial position of cattle and sheep;

[0018] According to the positioning tag distribution information in the grassland geographic information at this time, if the cattle and sheep information does not match the positioning tag distribution information in the grassland geographic information, then adopt the positioning tag in the data uploaded by the cattle and sheep positioning device to obtain the initial position of cattle and sheep.

[0019] Preferably, based on the initial position information of cattle and sheep, by integrating the movement speed and direction of cattle and sheep, predict the positions of cattle and sheep in the rasterized model at the next moment. Obtaining the predicted positions of cattle and sheep includes:

[0020] Step 1: Collect the position of cattle, the position of sheep, the speed of cattle, the speed of sheep, the direction of cattle, and the direction of sheep at the initial moment of cattle and sheep. Divide the area where cattle and sheep are located into a rasterized model, and record the current timestamp;

[0021] Step 2: Based on the position of cattle and the position of sheep at the initial moment of cattle and sheep, determine the initial grid coordinates of cattle and sheep in the rasterized model. According to the speed of cattle, the speed of sheep, the direction of cattle, and the direction of sheep, combined with the Kalman filter algorithm, obtain the motion state vector and state transition matrix of cattle and sheep. The state vector includes the grid coordinates and speed of cattle and sheep, and the state transition matrix describes the transition probability between grids of cattle and sheep;

[0022] Step 3: Through the Kalman filter algorithm, integrate the motion state vector and state transition matrix of cattle and sheep to obtain a predicted value, which includes the prediction of the grid coordinates and speed of cattle and sheep at the next moment;

[0023] Step 4: When the system obtains the observed values of cattle and sheep, the observed values include the grid coordinates and speed actually measured by cattle and sheep at the next moment. Calculate the difference between the predicted value and the observed value to obtain a residual vector, and update the state estimate and covariance matrix in the Kalman filter algorithm by combining the residual vector and the covariance matrix;

[0024] Step 5: Based on the updated state estimate and covariance matrix, judge the predicted positions of cattle and sheep at the next moment to obtain more accurate grid coordinates and speed of cattle and sheep at the next moment;

[0025] Step 6: If the timestamp changes, use the updated grid coordinates and speed of cattle and sheep as the new initial state, and repeat Steps 3 to 5 for iterative prediction;

[0026] Step 7: Generate the continuous movement trajectories of cattle and sheep in the rasterized model based on the grid coordinates of cattle and sheep obtained in each iteration, and determine the activity range of cattle and sheep.

[0027] Preferably, obtain the positioning signal strength uploaded by the positioning device. According to the signal propagation model, judge the positioning signal strength of the predicted positions of cattle and sheep. If the positioning signal strength is lower than the preset threshold, use the positioning algorithm based on the time difference of arrival. Utilize the time difference of signal reception by multiple positioning devices to calculate the positions of cattle and sheep, and obtain the corrected positions of cattle and sheep, including:

[0028] Obtain the periodic positioning signal strength data uploaded by multiple positioning devices, and determine the functional relationship between the signal strength and the distance according to the attenuation model of signal propagation in space;

[0029] According to the historical position data of cattle and sheep and the functional relationship between signal strength and distance, a signal propagation model is used to estimate the positioning signal strength value of the predicted position of cattle and sheep. If the positioning signal strength value is lower than the preset threshold, the time difference of arrival-based positioning algorithm is activated;

[0030] The receiving time of the positioning signal sent by cattle and sheep reaching multiple positioning devices is obtained through the positioning device, and according to the receiving time of multiple positioning devices, the time difference of the cattle and sheep positioning signal reaching different positioning devices is calculated;

[0031] According to the time difference of the cattle and sheep positioning signal reaching different positioning devices, a hyperbolic positioning model is used to calculate multiple hyperbolic equations;

[0032] According to multiple hyperbolic equation systems, the least squares method is used to iteratively solve the preliminary position of cattle and sheep to obtain the preliminary position coordinates of cattle and sheep;

[0033] According to the preliminary position coordinates of cattle and sheep and combined with the position information of the positioning device, the Kalman filtering algorithm is used to perform data fusion on the preliminary position coordinates of cattle and sheep to obtain the fused position coordinates of cattle and sheep;

[0034] According to the fused position coordinates of cattle and sheep and combined with the space vector finite element method, it is judged whether the fused position coordinates of cattle and sheep generate space vector distortion. If distortion occurs, the fused position coordinates of cattle and sheep are corrected to obtain the corrected position of cattle and sheep.

[0035] Preferably, according to the corrected position of cattle and sheep and the initial position of cattle and sheep, the nearest neighbor algorithm is used to match the corrected position of cattle and sheep with the historical activity trajectory. If the deviation of the matched trajectory is greater than the preset threshold, an abnormal behavior alarm is triggered, including:

[0036] The initial position coordinates and time stamp of cattle and sheep are obtained, and according to the corrected position coordinates and time stamp of cattle and sheep, the initial position and corrected position of each cattle and sheep are determined;

[0037] For each cattle and sheep, based on the initial position and corrected position, the displacement vector of cattle and sheep is calculated to obtain the displacement direction and distance of cattle and sheep;

[0038] The historical activity trajectory data of cattle and sheep is obtained. This data includes historical position coordinates and time stamps. The nearest neighbor algorithm is used to calculate the Euclidean distance between the corrected position and the position coordinates in the historical trajectory, and the historical trajectory point corresponding to the minimum distance is determined;

[0039] If the minimum distance is greater than the preset distance threshold, it is judged that there is a trajectory jump for this cattle and sheep, and the time stamp of this trajectory jump is recorded;

[0040] For the cattle and sheep with trajectory jumps, the time difference between the time stamp and the current time is calculated to obtain the time interval when the trajectory jump occurs;

[0041] If the time interval is less than the preset time threshold, it is determined that the cattle and sheep have multiple trajectory jumps in a short period of time, triggering the abnormal behavior determination process;

[0042] According to the abnormal behavior determination process, obtain the video surveillance data corresponding to the cattle and sheep, detect the cattle and sheep targets in the video through the target detection algorithm, use the optical flow method to track the movement state of the cattle and sheep, determine whether there are abnormal behaviors of the cattle and sheep, and obtain the final abnormal behavior alarm result.

[0043] Preferably, according to the corrected positions of all cattle and sheep, using the density clustering algorithm, determine the distribution state of the cattle and sheep group in the rasterized model, and obtain the central position and range of the cattle and sheep group distribution, including:

[0044] Obtain the spatial coordinates of the corrected positions of all cattle and sheep, and allocate the positions of the cattle and sheep to the corresponding grid cells according to the spatial coordinates;

[0045] According to the number of cattle and sheep in each grid cell, use the DBSCAN density clustering algorithm to obtain the clustering center of the cattle and sheep group distribution;

[0046] If the distance between the spatial coordinates of the clustering centers is less than the preset threshold, merge the multiple clustering centers to determine the final central position of the cattle and sheep group;

[0047] According to each clustering center and the number of cattle and sheep it contains, judge whether each clustering is a cattle group or a sheep group, and obtain the number of cattle groups and the number of sheep groups;

[0048] According to the number of cattle and sheep in each grid cell and the area of the grid cell, calculate the cattle and sheep group density in each grid cell;

[0049] According to the number of cattle groups, the number of sheep groups, and the group density, obtain the group scale, and obtain the distribution state of the cattle and sheep group in the raster model;

[0050] According to the distribution state of the cattle and sheep group in the raster model, determine the spatial range of the cattle and sheep group distribution.

[0051] Preferably, according to the central position and range of the cattle and sheep group distribution, combined with the pre-established grassland carrying capacity model, judge the grassland pressure in the local area. If the grassland pressure in the local area exceeds the preset threshold, generate a new activity range for the cattle and sheep according to the historical activity trajectories of the cattle and sheep, including:

[0052] Obtain the position information of the cattle and sheep, periodically collect the central position data of the cattle and sheep group through the Beidou navigation system, and synchronously collect the data of the distribution range of the cattle and sheep group within a predetermined time period to obtain the cattle and sheep group density distribution map;

[0053] Obtain rainfall data and temperature data through meteorological satellite remote sensing data of specific areas, combine the grassland area and grassland type data collected by UAV aerial photography, and calculate the theoretical livestock carrying capacity of the area according to the pre-established grassland carrying capacity model;

[0054] According to the cattle and sheep position information, the number of cattle and sheep, and the grassland area data, calculate the number of cattle and sheep per unit area of the grassland, and determine whether the number of cattle and sheep per unit area exceeds the preset threshold. If it exceeds, it is determined that the grassland pressure in this area is too high;

[0055] If the grassland pressure does not exceed the preset threshold, then according to the cattle and sheep position information, combined with the electronic fence technology, maintain the current activity range of the cattle and sheep to obtain a new cattle and sheep population density distribution map;

[0056] If the grassland pressure exceeds the preset threshold, then retrieve the historical trajectory data of cattle and sheep in this area, and simulate the foraging behavior of cattle and sheep according to the pre-established cattle and sheep behavior model to obtain the predicted data of the activity range of cattle and sheep in the next stage;

[0057] Obtain the predicted data of the activity range of cattle and sheep in the next stage, and judge whether the grassland pressure in the new range exceeds the preset threshold through the grassland carrying capacity model. If it does not exceed, then determine this range as the new activity range of cattle and sheep;

[0058] If the grassland pressure in the new range still exceeds the preset threshold, then according to the historical migration law of cattle and sheep, combined with the number of cattle and sheep and the grassland carrying capacity, use the linear regression algorithm to predict multiple new activity ranges after the cattle and sheep group are dispersed to obtain a new cattle and sheep population density distribution map, use the particle swarm optimization algorithm for range optimization, use the long short-term memory network algorithm to predict the long-term grassland pressure change trend, and determine the final activity range of cattle and sheep according to the grassland change trend.

[0059] The present invention also discloses a digital management system for cattle and sheep grasslands. The system is used to implement any one of the above methods. The system includes: an initial position acquisition module, a position prediction module, a positioning correction module, an abnormal behavior alarm module, a group distribution determination module, and a grassland pressure assessment module;

[0060] The initial position acquisition module is used to obtain the pre-established rasterized model of grassland geographic information, combine the positioning data with positioning tags and time tags uploaded by the sparse cattle and sheep positioning devices, and obtain the initial position information of cattle and sheep at this moment according to the moment when the positioning data is uploaded;

[0061] The position prediction module is used to predict the position of cattle and sheep in the rasterized model at the next moment according to the initial position information of cattle and sheep, and fuse the movement speed and movement direction of cattle and sheep to obtain the predicted position of cattle and sheep;

[0062] The positioning correction module is used to obtain the positioning signal strength uploaded by the positioning device, judge the positioning signal strength of the predicted positions of cattle and sheep according to the signal propagation model. If the positioning signal strength is lower than the preset threshold, the positioning algorithm based on time difference of arrival is adopted, and the time difference of signal reception by multiple positioning devices is used to calculate the positions of cattle and sheep, so as to obtain the corrected positions of cattle and sheep.

[0063] The abnormal behavior alarm module is used to match the corrected positions of cattle and sheep with the historical activity trajectories by using the nearest neighbor algorithm according to the corrected positions of cattle and sheep and the initial positions of cattle and sheep. If the deviation of the matched trajectory is greater than the preset threshold, an abnormal behavior alarm is triggered.

[0064] The group distribution determination module is used to determine the distribution state of the cattle and sheep group in the rasterized model by using the density clustering algorithm according to the corrected positions of all cattle and sheep, so as to obtain the central position and range of the cattle and sheep group distribution.

[0065] The grassland pressure assessment module is used to judge the grassland pressure of the local area according to the central position and range of the cattle and sheep group distribution and in combination with the pre-established grassland carrying capacity model. If the grassland pressure of the local area exceeds the preset threshold, a new activity range for cattle and sheep is generated according to the historical activity trajectories of cattle and sheep.

[0066] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0067] The present invention discloses a digital management method and system for cattle and sheep on grasslands, which ingeniously integrates multiple links such as sparse positioning data of cattle and sheep, grassland geographic information, motion prediction, signal strength analysis, abnormal behavior detection, group distribution analysis, and grassland pressure assessment to form a complete closed-loop management system. This method first uses the pre-established rasterized model of grassland geographic information, combines the positioning data with positioning tags and time tags uploaded by sparse cattle and sheep positioning devices, predicts the positions of cattle and sheep through the Kalman filtering algorithm, and judges whether it is necessary to use the positioning algorithm based on time difference of arrival for correction according to the signal strength. Further, the corrected positions are matched with the historical trajectories by the nearest neighbor algorithm to detect abnormal behaviors. In addition, the density clustering algorithm is used to analyze the distribution of the cattle and sheep group, the local grassland pressure is evaluated in combination with the grassland carrying capacity model, and a new activity range for cattle and sheep is generated when necessary. Finally, the cattle and sheep are guided back to the new activity range through the electronic fence technology. The core of the present invention is to achieve refined management of individual and group cattle and sheep through sparse positioning data, effectively solve the problems of sparse positioning data and extensive management in traditional animal husbandry, and realize the reasonable utilization of grasslands and the healthy breeding of cattle and sheep. Description of the Drawings

[0068] To more clearly illustrate the technical solution of the present invention, the accompanying drawings required for use in the embodiments will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can be obtained based on these drawings.

[0069] Figure 1 Schematic diagram of the process of a digital management method for cattle and sheep pastures according to an embodiment of the present invention;

[0070] Figure 2 Schematic diagram of the model structure according to an embodiment of the present invention. Detailed implementation manners

[0071] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0072] It should be noted that unless otherwise defined, the technical terms or scientific terms used in the embodiments of the present disclosure should have the ordinary meaning understood by those of ordinary skill in the art in the field to which the present disclosure belongs. The "first", "second" and similar terms used in the embodiments of the present disclosure do not indicate any order, quantity or importance, but are only used to distinguish different components. The terms such as "including" or "comprising" mean that the elements or objects appearing before this word cover the elements or objects listed after this word and their equivalents, without excluding other elements or objects. The terms such as "connected" or "coupled" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The terms such as "upper", "lower", "left" and "right" are only used to represent relative positional relationships. When the absolute position of the object being described changes, the relative positional relationship may also change accordingly.

[0073] To make the above objects, features and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below in conjunction with the accompanying drawings and specific implementation manners.

[0074] Embodiment 1

[0075] As Figure 1 shown, the embodiment of the present invention provides a digital management method for cattle and sheep pastures, and the method includes:

[0076] Obtain the pre-established rasterized model of grassland geographical information, combine the positioning data with positioning tags and time tags uploaded by sparse cattle and sheep positioning devices, and obtain the initial position information of cattle and sheep at that moment according to the time when the positioning data is uploaded;

[0077] According to the initial position information of cattle and sheep, fuse the movement speed and movement direction of cattle and sheep, predict the position of cattle and sheep in the rasterized model at the next moment, and obtain the predicted position of cattle and sheep;

[0078] Obtain the positioning signal strength uploaded by the positioning device, judge the positioning signal strength of the predicted position of cattle and sheep according to the signal propagation model. If the positioning signal strength is lower than the preset threshold, then use the positioning algorithm based on time difference of arrival, utilize the time difference of signal reception by multiple positioning devices, calculate the position of cattle and sheep, and obtain the corrected position of cattle and sheep;

[0079] According to the corrected position of cattle and sheep and the initial position of cattle and sheep, adopt the nearest neighbor algorithm to match the corrected position of cattle and sheep with the historical activity trajectory. If the deviation of the matched trajectory is greater than the preset threshold, then trigger an abnormal behavior alarm;

[0080] According to the corrected positions of all cattle and sheep, use the density clustering algorithm to determine the distribution state of the cattle and sheep group in the rasterized model, and obtain the central position and range of the cattle and sheep group distribution;

[0081] According to the central position and range of the cattle and sheep group distribution, combine the pre-established grassland carrying capacity model to judge the grassland pressure in the local area. If the grassland pressure in the local area exceeds the preset threshold, then generate a new activity range for cattle and sheep according to the historical activity trajectory of cattle and sheep.

[0082] In this embodiment, obtaining the pre-established rasterized model of grassland geographical information, combining the positioning data with positioning tags and time tags uploaded by sparse cattle and sheep positioning devices, and obtaining the initial position information of cattle and sheep at that moment according to the time when the positioning data is uploaded includes:

[0083] Obtain the rasterized model of the grassland geographical information. According to the data uploaded by the livestock positioning device, judge the time tag of the data, and determine the corresponding moment of the time tag of the data. According to the corresponding moment of the time tag of the data, extract the grassland geographical information at that moment from the rasterized model of the grassland geographical information, judge the distribution of the positioning tags in the grassland geographical information at that moment, and obtain the positioning tag distribution information at that moment. Use the positioning tag distribution information at that moment, and according to the positioning tags in the data uploaded by the livestock positioning device, judge the livestock corresponding to the positioning tag, and obtain the livestock information corresponding to the positioning tag. According to the livestock information corresponding to the positioning tag, combined with the grassland geographical information at that moment, if the livestock information matches the positioning tag distribution information in the grassland geographical information, then use the positioning tag distribution information in the grassland geographical information to obtain the initial positions of the livestock. According to the positioning tag distribution information in the grassland geographical information at that moment, if the livestock information does not match the positioning tag distribution information in the grassland geographical information, then use the positioning tags in the data uploaded by the livestock positioning device to obtain the initial positions of the livestock. According to the positioning tags in the data uploaded by the livestock positioning device, combined with the initial positions of the livestock, judge the grid unit where the initial positions of the livestock are located, and obtain the grid unit information where the initial positions of the livestock are located. According to the grid unit information where the initial positions of the livestock are located, combined with the grassland geographical information at that moment, fuse the initial positions of the livestock and the grassland geographical information at that moment to obtain the fused information at that moment.

[0084] Specifically, obtain the rasterized model of the grassland geographical information. This is like a huge electronic map that divides the vast grassland into small squares, and each small square records the grassland information of the area, such as the type, height, density, etc. of the grass. Assume that this model divides the grassland into grid units of 100 meters × 100 meters, and each unit has a unique number, such as "A1", "B2", etc. The type of grass, such as "Leymus chinensis", "Stipa capillata", etc.; the height of the grass, such as "10 cm", "15 cm", etc.; and the density of the grass, such as "sparse", "dense", etc. are detailedly recorded in each grid unit.

[0085] In this embodiment, according to the initial position information of the livestock, fuse the movement speed and movement direction of the livestock, and predict the position of the livestock in the rasterized model at the next moment. The steps to obtain the predicted position of the livestock include:

[0086] Step 1: Collect information on the initial positions, speeds, and directions of cattle and sheep. Divide the area where the cattle and sheep are located into a rasterized model and record the current timestamp. Step 2: Determine the initial raster coordinates of the cattle and sheep in the rasterized model based on their initial positions. Combine the speeds and directions of the cattle and sheep with the Kalman filter algorithm to obtain the motion state vectors and state transition matrices of the cattle and sheep. The state vectors include the raster coordinates and speeds of the cattle and sheep, and the state transition matrices describe the transition probabilities of the cattle and sheep between rasters. Step 3: Use the Kalman filter algorithm to fuse the motion state vectors and state transition matrices of the cattle and sheep to obtain predicted values, which include predictions of the raster coordinates and speeds of the cattle and sheep at the next moment. Step 4: When the system obtains the observed values of the cattle and sheep, which include the actually measured raster coordinates and speeds of the cattle and sheep at the next moment, calculate the difference between the predicted values and the observed values to obtain the residual vector, and update the state estimate and covariance matrix in the Kalman filter algorithm in combination with the residual vector and covariance matrix. Step 5: Based on the updated state estimate and covariance matrix, judge the predicted positions of the cattle and sheep at the next moment to obtain more accurate raster coordinates and speeds of the cattle and sheep at the next moment. Step 6: If the timestamp changes, use the updated raster coordinates and speeds of the cattle and sheep as the new initial state, and repeat Steps 3 to 5 for iterative prediction. Step 7: Generate the continuous motion trajectories of the cattle and sheep in the rasterized model based on the raster coordinates of the cattle and sheep obtained in each iteration to determine the activity ranges of the cattle and sheep.

[0087] In this embodiment, the positioning signal strength uploaded by the positioning device is obtained. According to the signal propagation model, the positioning signal strength at the predicted positions of the cattle and sheep is judged. If the positioning signal strength is lower than the preset threshold, the positioning algorithm based on the time difference of arrival is adopted, and the time difference of the signals received by multiple positioning devices is used to calculate the positions of the cattle and sheep, and the corrected positions of the cattle and sheep are obtained, including:

[0088] Obtain the periodic positioning signal strength data uploaded by multiple positioning devices. According to the attenuation model of signal propagation in space, determine the functional relationship between the signal strength and the distance f(x)=y, where x is the distance and y is the signal strength. Based on the historical position data of the cattle and sheep and f(x)=y, use the signal propagation model to estimate the positioning signal strength value at the predicted positions of the cattle and sheep. If the positioning signal strength value is lower than the preset threshold, activate the positioning algorithm based on the time difference of arrival. Obtain the reception times of the positioning signals sent by the cattle and sheep reaching multiple positioning devices through the positioning devices. According to the reception times of the multiple positioning devices, calculate the time differences of the cattle and sheep positioning signals reaching different positioning devices. According to the time differences of the cattle and sheep positioning signals reaching different positioning devices, use the hyperbolic positioning model to calculate multiple hyperbolic equations. In the multiple equations, t i is the time when the i-th positioning device receives the signal, t jis the time when the j-th positioning device receives the signal, c is the signal propagation speed, (x i， y i ) is the coordinate of the i-th positioning device, (x j， y j ) is the coordinate of the j-th positioning device, (x, y) is the coordinate of the cattle and sheep, then there is c(t i -t j ) = √[(x - x i ) 2 +(y - y i ) 2 -√[(x - x j ) 2 +(y - y j ) 2 . According to multiple hyperbola equations, the least squares method is used for iterative solution to obtain the preliminary position of the cattle and sheep, and the preliminary position coordinates (x, y) of the cattle and sheep are obtained. According to the preliminary position coordinates (x, y) of the cattle and sheep, combined with the position information of the positioning device, the Kalman filtering algorithm is used to perform data fusion on the preliminary position coordinates (x, y) of the cattle and sheep to obtain the fused position coordinates of the cattle and sheep. According to the fused position coordinates of the cattle and sheep, combined with the space vector finite element method, it is judged whether the fused position coordinates of the cattle and sheep generate space vector distortion. If distortion occurs, the fused position coordinates of the cattle and sheep are corrected to obtain the corrected position of the cattle and sheep.

[0089] In this embodiment, according to the corrected position and the initial position of the cattle and sheep, the nearest neighbor algorithm is used to match the corrected position of the cattle and sheep with the historical activity trajectory. If the deviation of the matched trajectory is greater than the preset threshold, an abnormal behavior alarm is triggered, including:

[0090] Obtain the initial position coordinates and time stamps of the cattle and sheep. According to the corrected position coordinates and time stamps of the cattle and sheep, determine the initial position and corrected position of each cattle and sheep. For each cattle and sheep, based on the initial position and the corrected position, calculate the displacement vector of the cattle and sheep to obtain the displacement direction and distance of the cattle and sheep. Obtain the historical activity trajectory data of the cattle and sheep, which includes historical position coordinates and time stamps. Using the nearest neighbor algorithm, calculate the Euclidean distance between the corrected position and the position coordinates in the historical trajectory, and determine the historical trajectory point corresponding to the minimum distance. If the minimum distance is greater than the preset distance threshold, it is judged that there is a trajectory jump for this cattle and sheep, and record the time stamp of this trajectory jump. For the cattle and sheep with trajectory jumps, calculate the time difference between the time stamp and the current time to obtain the time interval when the trajectory jump occurs. If the time interval is less than the preset time threshold, it is judged that this cattle and sheep has multiple trajectory jumps in a short time, and the abnormal behavior determination process is triggered. According to the abnormal behavior determination process, obtain the video surveillance data corresponding to this cattle and sheep, detect the cattle and sheep targets in the video through the target detection algorithm, use the optical flow method to track the movement state of the cattle and sheep, and judge whether the cattle and sheep have abnormal behaviors to obtain the final abnormal behavior alarm result.

[0091] In this embodiment, according to the corrected positions of all cattle and sheep, using the density clustering algorithm, the distribution state of the cattle and sheep population in the rasterized model is determined, and the central position and range of the cattle and sheep population distribution are obtained, including:

[0092] Obtain the spatial coordinates of the corrected positions of all cattle and sheep, and allocate the positions of cattle and sheep to the corresponding grid cells according to the spatial coordinates. According to the number of cattle and sheep in each grid cell, use the DBSCAN density clustering algorithm to obtain the clustering center of the cattle and sheep population distribution. If the distance between the spatial coordinates of the clustering centers is less than the preset threshold, then merge multiple clustering centers to determine the final central position of the cattle and sheep population. According to each clustering center and the number of cattle and sheep it contains, determine whether each clustering is a cattle herd or a sheep herd, and obtain the number of cattle herds and the number of sheep herds. According to the number of cattle and sheep in each grid cell and the area of the grid cell, calculate the density of the cattle and sheep population in each grid cell. According to the number of cattle herds, the number of sheep herds, and the population density, obtain the population size, and get the distribution state of the cattle and sheep population in the grid model. According to the distribution state of the cattle and sheep population in the grid model, determine the spatial range of the cattle and sheep population distribution.

[0093] In this embodiment, according to the central position and range of the cattle and sheep population distribution, combined with the pre-established grassland carrying capacity model, judge the grassland pressure in the local area. If the grassland pressure in the local area exceeds the preset threshold, then generate a new activity range for the cattle and sheep according to the historical activity trajectories of the cattle and sheep, including:

[0094] Obtain the position information of cattle and sheep. Periodically collect the central position data of the cattle and sheep group through the Beidou navigation system, and synchronously collect the distribution range data of the cattle and sheep group within a certain period of time to obtain the cattle and sheep group density distribution map. Obtain rainfall data and temperature data through meteorological satellite remote sensing data of a specific area, combine the grassland area and grassland type data collected by UAV aerial photography, and calculate the theoretical livestock carrying capacity of the area according to the pre-established grassland carrying capacity model. According to the cattle and sheep position information, the number of cattle and sheep, and the grassland area data, calculate the number of cattle and sheep per unit area of the grassland, and determine whether the number of cattle and sheep per unit area of the grassland exceeds the preset threshold. If it exceeds, it is determined that the grassland pressure in this area is too high. If the grassland pressure does not exceed the preset threshold, according to the cattle and sheep position information, combined with the electronic fence technology, maintain the current activity range of the cattle and sheep to obtain a new cattle and sheep group density distribution map. If the grassland pressure exceeds the preset threshold, retrieve the historical trajectory data of the cattle and sheep in this area, and simulate the foraging behavior of the cattle and sheep according to the pre-established cattle and sheep behavior model to obtain the predicted data of the next stage of the activity range of the cattle and sheep. Obtain the predicted data of the next stage of the activity range of the cattle and sheep, and judge whether the grassland pressure in the new range exceeds the preset threshold through the grassland carrying capacity model in step 2. If it does not exceed, it is determined that this range is the new activity range of the cattle and sheep. If the grassland pressure in the new range still exceeds the preset threshold, according to the historical migration law of the cattle and sheep, combined with the number of cattle and sheep and the grassland carrying capacity, use the linear regression algorithm to predict multiple new activity ranges after the cattle and sheep group is dispersed to obtain a new cattle and sheep group density distribution map, use the particle swarm optimization algorithm for range optimization, use the long short-term memory network algorithm to predict the long-term grassland pressure change trend, and determine the final activity range of the cattle and sheep according to the grassland change trend.

[0095] Specifically, as Figure 2 shown, both the grassland carrying capacity model and the cattle and sheep behavior model include: a CNN layer, an LSTM-ATT-LSTM layer, and an output layer;

[0096] Among them, the CNN layer is used to extract the feature X input ∈R N×T×F of the training set based on the one-dimensional convolutional layer, and perform downsampling on the extracted features using the max-pooling method to obtain key features; the calculation formula is as follows:

[0097] X conv [i, j, k] = ReLU(X input [i, j:j + k, :] * W conv [k, :, :] + b conv ),

[0098] Perform a max-pooling operation on the output data, retain the maximum value within each window to reduce the dimension of the data, and at the same time retain the key feature information. The output length information:

[0099] X pool[i, j] = max(X conv [i, 2j, :], X conv [i, 2j + 1, :]),

[0100] where X input is the data sequence of the training set of the input model, N is the number of samples, T is the sequence length, and F is the number of features of each sequence; W conv is the convolutional kernel, with a shape of k × F × F conv , F conv is the number of convolutional kernels, b conv is the bias term; ReLU(z) = max(0, z) is the ReLU activation function, X conv [i, j, k] is the output of the convolutional layer, and X pool [i, j] is the key feature output after pooling; combined with the training and optimization of this model, the output shape is X conv ∈R 32×8×150 , X pool ∈R 32×4×150 .

[0101] The LSTM-ATT-LSTM layer includes a first LSTM layer, an attention mechanism layer, and a second LSTM layer, which are used to process the key features to obtain feature vectors; among them, the flow of information is controlled by the internal gate structure of the double-layer LSTM, and weights are dynamically assigned to the key features through the attention mechanism Attention; among them, the gate structure includes a forget gate, an input gate, and an output gate, and the flow of information is controlled by the sigmoid function and the tanh function; specifically, the input gate determines which new information is stored in the cell state, and the input gate includes two parts: one part is the sigmoid layer that determines which values will update the cell state, and the other part is the tanh layer that creates a new candidate value vector to be added to the state, and the calculation formula is:

[0102] i t = σ(W i X t + U i h t-1 + b i )

[0103]

[0104] where: i t is the output of the input gate, is the new candidate cell state, X t is the data of the current time step of the input, U i , UC is the weight matrix of the hidden state at the previous time step, with shape U×U, σ is the sigmoid function, tanh is the hyperbolic tangent function, and h t-1 is the output at the previous time step, and W i , W C and the corresponding bias b i , b C are the weight matrix and bias vector of the input gate.

[0105] The forget gate determines which information is discarded from the cell. The output ranges from 0 to 1, where 0 means completely forgotten and 1 means completely retained. The calculation formula is:

[0106] f t = σ(W f X t + U f h t-1 + b f )

[0107] In the formula: f t is the output of the forget gate, σ is the sigmoid function, W f and b f are the weight matrix and bias vector of the forget gate, and U f is the weight matrix of the hidden state at the previous time step, with shape U×U.

[0108] The cell state C t is updated by the outputs of the forget gate f t and the input gate i t . The update calculation process is:

[0109]

[0110] The output gate determines the hidden state h t at the current moment, which is based on the current input X t and the updated cell state C t . The calculation formula is:

[0111] o t = σ(W o X t + U o h t-1 + b o )

[0112] h t = o t · tanh(C t )

[0113] In the formula: o t is the output of the output gate, tanh is the hyperbolic tangent function, and Wo and b o are the weight matrix and bias vector of the output gate, and U o is the weight matrix of the hidden state at the previous time step, with a shape of U×U.

[0114] In this embodiment, according to the new activity range of cattle and sheep, combined with the individual identification of cattle and sheep, through the electronic fence technology, it is judged whether the cattle and sheep exceed the new activity range. If the cattle and sheep exceed the new activity range, a low-frequency electric pulse signal is sent to the corresponding cattle and sheep positioning device to guide the cattle and sheep back to the new activity range.

[0115] Obtain the individual identification information of cattle and sheep and the corresponding initial activity range data, construct an associated database of the individual identification of cattle and sheep and the initial activity range, and bind the individual identification of cattle and sheep to the electronic fence. According to the positioning data uploaded by the cattle and sheep positioning device, obtain the real-time position information of the cattle and sheep. Combine the individual identification of cattle and sheep, retrieve the associated database of the individual identification of cattle and sheep and the initial activity range, and determine the activity range corresponding to the individual identification of cattle and sheep. Compare the real-time position information of the cattle and sheep with the activity range data to perform an out-of-bounds judgment. If the real-time position of the cattle and sheep exceeds the activity range, trigger an out-of-bounds alarm, and record the individual identification of the cattle and sheep and the out-of-bounds time. For the individual identification of cattle and sheep that triggers the out-of-bounds alarm, retrieve the binding information of the individual identification of cattle and sheep to the electronic fence, obtain the corresponding electronic fence number, and generate a low-frequency electric pulse sending instruction containing the individual identification of the cattle and sheep and the electronic fence number. Send the low-frequency electric pulse sending instruction to the corresponding electronic fence through the wireless communication module. After receiving the low-frequency electric pulse sending instruction, the electronic fence extracts the individual identification of the cattle and sheep and the electronic fence number, and determines the corresponding cattle and sheep positioning device. According to the types of cattle and sheep, obtain the associated data of the types of cattle and sheep, the pulse intensity, and the pulse frequency, determine the corresponding pulse intensity and pulse frequency, and the electronic fence sends a low-frequency electric pulse signal with a specific pulse intensity and pulse frequency to the corresponding cattle and sheep positioning device. The cattle and sheep positioning device receives the low-frequency electric pulse signal, generates a corresponding stimulus. After the cattle and sheep receive the stimulus, they judge the direction of the stimulus source, generate an avoidance behavior, return to the activity range, the cattle and sheep positioning device uploads the position again, and judges whether the cattle and sheep return to the range.

[0116] It should be noted that the method of the embodiments of the present disclosure can be executed by a single device, such as a computer or a server. The method of this embodiment can also be applied to a distributed scenario and completed by multiple devices cooperating with each other. In the case of such a distributed scenario, one of the multiple devices can only execute one or more steps of the method of the embodiments of the present disclosure, and these multiple devices will interact with each other to complete the described method.

[0117] It should be noted that some embodiments of the present disclosure have been described above. Other embodiments are within the scope of the appended claims. In some cases, it should be understood that the magnitudes of the sequence numbers of the steps in the above embodiments do not imply the order of execution. The execution order of each process should be determined according to its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention. The actions or steps recited in the claims may be executed in a different order than in the above embodiments and still achieve the desired results. Additionally, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0118] Embodiment 2

[0119] Based on the same inventive concept, corresponding to the method of any of the above embodiments, the present invention further provides a digital management system for cattle and sheep pastures. The system is used to implement the method of any one of the above, and the system includes: an initial position acquisition module, a position prediction module, a positioning correction module, an abnormal behavior alarm module, a group distribution determination module, and a pasture pressure assessment module;

[0120] The initial position acquisition module is used to obtain the pre-established rasterized model of the pasture geographical information, combine the positioning data with positioning tags and time tags uploaded by the sparse cattle and sheep positioning devices, and obtain the initial position information of the cattle and sheep at the moment according to the moment when the positioning data is uploaded;

[0121] The position prediction module is used to predict the position of the cattle and sheep in the rasterized model at the next moment according to the initial position information of the cattle and sheep, fuse the movement speed and movement direction of the cattle and sheep, and obtain the predicted position of the cattle and sheep;

[0122] The positioning correction module is used to obtain the positioning signal strength uploaded by the positioning device, judge the positioning signal strength of the predicted position of the cattle and sheep according to the signal propagation model. If the positioning signal strength is lower than the preset threshold, the positioning algorithm based on the time difference of arrival is adopted, and the time difference of receiving signals by multiple positioning devices is used to calculate the position of the cattle and sheep, and the corrected position of the cattle and sheep is obtained;

[0123] The abnormal behavior alarm module is used to match the corrected position of the cattle and sheep with the historical activity trajectory by using the nearest neighbor algorithm according to the corrected position of the cattle and sheep and the initial position of the cattle and sheep. If the deviation of the matched trajectory is greater than the preset threshold, an abnormal behavior alarm is triggered;

[0124] The group distribution determination module is used to determine the distribution state of the cattle and sheep group in the rasterized model by using the density clustering algorithm according to the corrected positions of all the cattle and sheep, and obtain the central position and range of the cattle and sheep group distribution;

[0125] The grassland pressure assessment module is used to judge the grassland pressure of a local area according to the central position and range of the distribution of cattle and sheep groups, in combination with a pre-established grassland carrying capacity model. If the grassland pressure of the local area exceeds a preset threshold, a new activity range for cattle and sheep is generated according to the historical activity trajectories of cattle and sheep.

[0126] The system of the above embodiment is used to implement the corresponding method for digital management of cattle and sheep grasslands in any of the foregoing embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be elaborated here.

[0127] It should be noted that the above digital management system for cattle and sheep grasslands is embodied in the form of functional units. The term "module" here can be implemented in the form of software and / or hardware, and no specific limitation is made thereto.

[0128] For example, the "module" can be a software program, a hardware circuit, or a combination of both that implements the above functions. The hardware circuit may include an application specific integrated circuit (ASIC), an electronic circuit, a processor (such as a shared processor, a dedicated processor, or a group of processors, etc.) for executing one or more software or firmware programs, and a memory, a merged logic circuit, and / or other suitable components that support the described functions.

[0129] The embodiments of the present disclosure are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of the appended claims. Therefore, any omission, modification, equivalent substitution, improvement, etc., made within the spirit and principle of the embodiments of the present disclosure shall be included within the protection scope of the present disclosure.

Claims

1. A digital management method for cattle and sheep pastures, characterized in that: The method comprises: Obtain the pre-established grassland geographic information raster model, combine the positioning data with positioning tags and time tags uploaded by the sparse cattle and sheep positioning device, and obtain the initial position information of the cattle and sheep at that moment according to the time of uploading the positioning data; According to the initial position information of cattle and sheep, the movement speed and direction of cattle and sheep are integrated to predict the position of cattle and sheep in the rasterized model at the next moment and obtain the predicted position of cattle and sheep; Obtain the positioning signal strength uploaded by the positioning device, and determine the positioning signal strength of the predicted position of the cattle and sheep according to the signal propagation model. If the positioning signal strength is lower than the preset threshold, use the positioning algorithm based on arrival time difference to calculate the position of the cattle and sheep using the time difference of receiving signals from multiple positioning devices to obtain the corrected position of the cattle and sheep; According to the corrected position of cattle and sheep and the initial position of cattle and sheep, the nearest neighbor algorithm is used to match the corrected position of cattle and sheep with the historical activity trajectory. If the deviation of the matched trajectory is greater than the preset threshold, an abnormal behavior alarm is triggered; According to the corrected positions of all cattle and sheep, the density clustering algorithm is used to determine the distribution status of the cattle and sheep groups in the rasterization model, and the central position and range of the distribution of the cattle and sheep groups are obtained; Based on the central location and range of the distribution of cattle and sheep populations, combined with the pre-established grassland carrying capacity model, the grassland pressure in the local area is judged. If the grassland pressure in the local area exceeds the preset threshold, a new cattle and sheep activity range is generated based on the historical activity trajectories of cattle and sheep.

2. The method according to claim 1, characterized in that The pre-established grassland geographic information rasterization model is obtained, combined with the positioning data with positioning tags and time tags uploaded by the sparse cattle and sheep positioning device, and according to the time of uploading the positioning data, the initial position information of the cattle and sheep at that moment is obtained, including: Obtain a rasterized model of grassland geographic information, determine the data time tag based on the data uploaded by the cattle and sheep positioning device, and determine the time corresponding to the data time tag; According to the time corresponding to the data time tag, the grassland geographic information at that time is extracted from the grassland geographic information rasterization model, the positioning tag distribution in the grassland geographic information at that time is determined, and the positioning tag distribution information at that time is obtained; Using the distribution information of the positioning tag at that moment, according to the positioning tag in the data uploaded by the cattle and sheep positioning device, it is determined that the positioning tag corresponds to the cattle and sheep, and the cattle and sheep information corresponding to the positioning tag is obtained; According to the cattle and sheep information corresponding to the positioning tag, combined with the grassland geographic information at that moment, if the cattle and sheep information matches the positioning tag distribution information in the grassland geographic information, the positioning tag distribution information in the grassland geographic information is used to obtain the initial position of the cattle and sheep; According to the distribution information of the positioning tags in the grassland geographic information at that moment, if the cattle and sheep information does not match the distribution information of the positioning tags in the grassland geographic information, the positioning tags in the data uploaded by the cattle and sheep positioning device are used to obtain the initial positions of the cattle and sheep.

3. The method according to claim 1, characterized in that According to the initial position information of cattle and sheep, the movement speed and direction of cattle and sheep are integrated to predict the position of cattle and sheep in the rasterized model at the next moment. The predicted position of cattle and sheep is obtained by: Step 1: Collect the cattle position, sheep position, cattle speed, sheep speed, cattle direction and sheep direction information at the initial moment of the cattle and sheep, divide the area where the cattle and sheep are located into a rasterized model, and record the current timestamp; Step 2: According to the cattle and sheep positions at the initial moment, determine the initial grid coordinates of cattle and sheep in the grid model. According to the cattle speed, sheep speed, cattle direction and sheep direction of cattle and sheep, combined with the Kalman filter algorithm, obtain the motion state vector and state transfer matrix of cattle and sheep. The state vector contains the grid coordinates and speed of cattle and sheep, and the state transfer matrix describes the transfer probability of cattle and sheep between grids. Step 3: Through the Kalman filter algorithm, the motion state vector and state transfer matrix of cattle and sheep are integrated to obtain the predicted value, which includes the grid coordinates and speed prediction of cattle and sheep at the next moment; Step 4: When the system obtains the observation values ​​of cattle and sheep, the observation values ​​include the grid coordinates and speeds actually measured by cattle and sheep at the next moment, and calculates the difference between the predicted value and the observed value to obtain the residual vector. The residual vector and the covariance matrix are combined to update the state estimation and covariance matrix in the Kalman filter algorithm. Step 5: According to the updated state estimation and covariance matrix, the predicted position of the cattle and sheep at the next moment is determined to obtain more accurate grid coordinates and speeds of the cattle and sheep at the next moment; Step 6: If the timestamp changes, the updated grid coordinates and speed of cattle and sheep are used as the new initial state, and steps 3 to 5 are repeated for cyclic iterative prediction; Step 7: Based on the grid coordinates of cattle and sheep obtained in each iteration, the continuous movement trajectories of cattle and sheep are generated in the rasterized model to determine the activity range of cattle and sheep.

4. The method according to claim 1, characterized in that: Obtain the positioning signal strength uploaded by the positioning device, and determine the positioning signal strength of the predicted position of cattle and sheep according to the signal propagation model. If the positioning signal strength is lower than the preset threshold, use the positioning algorithm based on arrival time difference to calculate the position of cattle and sheep using the time difference of receiving signals from multiple positioning devices. Obtaining the corrected position of cattle and sheep includes: Obtain periodic positioning signal strength data uploaded by multiple positioning devices, and determine the functional relationship between signal strength and distance based on the attenuation model of signal propagation in space; Based on the historical location data of cattle and sheep and the functional relationship between signal strength and distance, the signal propagation model is used to estimate the positioning signal strength value of the predicted location of cattle and sheep. If the positioning signal strength value is lower than the preset threshold, the positioning algorithm based on arrival time difference is activated; The positioning device obtains the reception time of the positioning signals sent by the cattle and sheep to multiple positioning devices, and calculates the time difference of the positioning signals of the cattle and sheep arriving at different positioning devices according to the reception time of the multiple positioning devices; According to the time difference between the cattle and sheep positioning signals arriving at different positioning devices, a hyperbolic positioning model is used to calculate multiple hyperbolic equations; According to multiple hyperbolic equations, the initial positions of cattle and sheep are solved iteratively using the least square method to obtain the initial position coordinates of cattle and sheep; According to the preliminary position coordinates of cattle and sheep, combined with the position information of the positioning device, the Kalman filter algorithm is used to fuse the preliminary position coordinates of cattle and sheep to obtain the fused position coordinates of cattle and sheep; According to the fusion position coordinates of cattle and sheep, combined with the space vector finite element method, it is judged whether the fusion position coordinates of cattle and sheep produce space vector distortion. If distortion occurs, the fusion position coordinates of cattle and sheep are corrected to obtain the corrected position of cattle and sheep.

5. The method according to claim 1, characterized in that According to the corrected position of cattle and sheep and the initial position of cattle and sheep, the nearest neighbor algorithm is used to match the corrected position of cattle and sheep with the historical activity trajectory. If the deviation of the matched trajectory is greater than the preset threshold, the abnormal behavior alarm is triggered, including: Obtaining the initial position coordinates and time stamps of the cattle and sheep, and determining the initial position and revised position of each cattle and sheep according to the revised position coordinates and time stamps of the cattle and sheep; For each cattle or sheep, the displacement vector of the cattle or sheep is calculated based on the initial position and the corrected position, and the displacement direction and distance of the cattle or sheep are obtained; Obtain the historical activity trajectory data of cattle and sheep, which includes historical position coordinates and time tags, and use the nearest neighbor algorithm to calculate the Euclidean distance between the corrected position and the position coordinates in the historical trajectory to determine the historical trajectory point corresponding to the minimum distance; If the minimum distance is greater than the preset distance threshold, it is determined that the cattle and sheep have a trajectory jump, and the time mark of the trajectory jump is recorded; For cattle and sheep with trajectory jumps, the time difference between the time mark and the current time is calculated to obtain the time interval of the trajectory jump; If the time interval is less than the preset time threshold, it is judged that the cattle and sheep have multiple trajectory jumps in a short period of time, triggering the abnormal behavior determination process; According to the abnormal behavior judgment process, the corresponding video surveillance data of the cattle and sheep is obtained, the cattle and sheep targets in the video are detected through the target detection algorithm, and the optical flow method is used to track the movement status of the cattle and sheep to determine whether the cattle and sheep have abnormal behavior, and obtain the final abnormal behavior alarm result.

6. The method according to claim 1, characterized in that According to the corrected positions of all cattle and sheep, the density clustering algorithm is used to determine the distribution status of the cattle and sheep population in the rasterization model, and the central position and range of the cattle and sheep population distribution are obtained, including: Obtain the spatial coordinates of the corrected positions of all cattle and sheep, and assign the positions of cattle and sheep to corresponding grid cells according to the spatial coordinates; According to the number of cattle and sheep in each grid unit, the DBSCAN density clustering algorithm is used to obtain the cluster center of the distribution of cattle and sheep groups; If the distance of the spatial coordinates of the cluster centers is less than the preset threshold, multiple cluster centers are merged to determine the final center position of the cattle and sheep group; According to each cluster center and the number of cattle and sheep it contains, we determine whether each cluster is a herd of cattle or sheep, and get the number of cattle and sheep; According to the number of cattle and sheep in each grid unit and the area of ​​the grid unit, the density of cattle and sheep populations in each grid unit is calculated; According to the number of cattle and sheep, the number of herds, and the population density, the population size is obtained, and the distribution status of cattle and sheep populations in the grid model is obtained; According to the distribution status of cattle and sheep herds in the grid model, the spatial range of the distribution of cattle and sheep herds is determined.

7. The method according to claim 1, characterized in that According to the central location and range of the distribution of cattle and sheep groups, combined with the pre-established grassland carrying capacity model, the grassland pressure in the local area is judged. If the grassland pressure in the local area exceeds the preset threshold, a new cattle and sheep activity range is generated based on the historical activity trajectory of cattle and sheep, including: Obtain the location information of cattle and sheep, periodically collect the center location data of cattle and sheep groups through the Beidou navigation system, and simultaneously collect the distribution range data of cattle and sheep groups within a predetermined time period to obtain the density distribution map of cattle and sheep groups; The rainfall and temperature data are obtained through meteorological satellite remote sensing data of a specific area. Combined with the grassland area and grassland type data collected by drone aerial photography, the theoretical livestock carrying capacity of the area is calculated based on the pre-established grassland carrying capacity model. According to the location information of cattle and sheep, the number of cattle and sheep and the area of ​​grassland, the number of cattle and sheep per unit area of ​​grassland is calculated, and it is determined whether the number of cattle and sheep per unit area of ​​grassland exceeds the preset threshold. If it exceeds, it is determined that the grassland pressure in the area is too high; If the grassland pressure does not exceed the preset threshold, the current activity range of cattle and sheep is maintained according to the location information of cattle and sheep, combined with electronic fence technology, to obtain a new density distribution map of cattle and sheep groups; If the grassland pressure exceeds the preset threshold, the historical trajectory data of cattle and sheep in the area is retrieved, and the foraging behavior of cattle and sheep is simulated according to the pre-established cattle and sheep behavior model to obtain the predicted activity range of cattle and sheep in the next stage; Obtain the predicted data of the next stage of cattle and sheep activity range, and determine whether the grassland pressure in the new range exceeds the preset threshold through the grassland carrying capacity model. If not, determine the range as the new cattle and sheep activity range; If the grassland pressure in the new range still exceeds the preset threshold, the linear regression algorithm is used to predict the multiple new activity ranges of the cattle and sheep groups after they are dispersed, based on the historical migration patterns of cattle and sheep, combined with the number of cattle and sheep and the carrying capacity of the grassland, to obtain a new density distribution map of the cattle and sheep groups. The particle swarm algorithm is used for range optimization, and the long short-term memory network algorithm is used to predict the long-term grassland pressure change trend. The final cattle and sheep activity range is determined based on the grassland change trend.

8. A digital management system for cattle and sheep pastures, the system being used to implement the method according to any one of claims 1 to 7, characterized in that: The system includes: an initial position acquisition module, a position prediction module, a positioning correction module, an abnormal behavior alarm module, a group distribution determination module and a grassland pressure assessment module; The initial position acquisition module is used to obtain a pre-established grassland geographic information rasterization model, combine the positioning data with positioning tags and time tags uploaded by the sparse cattle and sheep positioning device, and obtain the initial position information of the cattle and sheep at that moment according to the time of uploading the positioning data; The position prediction module is used to predict the position of the cattle and sheep in the rasterization model at the next moment according to the initial position information of the cattle and sheep, integrating the movement speed and movement direction of the cattle and sheep, and obtaining the predicted position of the cattle and sheep; The positioning correction module is used to obtain the positioning signal strength uploaded by the positioning device, and judge the positioning signal strength of the predicted position of the cattle and sheep according to the signal propagation model. If the positioning signal strength is lower than the preset threshold, a positioning algorithm based on arrival time difference is adopted, and the time difference of receiving signals by multiple positioning devices is used to calculate the position of the cattle and sheep to obtain the corrected position of the cattle and sheep; The abnormal behavior alarm module is used to match the corrected positions of cattle and sheep with historical activity trajectories according to the corrected positions of cattle and sheep and the initial positions of cattle and sheep using the nearest neighbor algorithm. If the deviation of the matched trajectories is greater than a preset threshold, an abnormal behavior alarm is triggered; The group distribution determination module is used to determine the distribution state of the cattle and sheep group in the rasterization model according to the corrected positions of all cattle and sheep, and obtain the central position and range of the cattle and sheep group distribution by using a density clustering algorithm; The grassland pressure assessment module is used to judge the grassland pressure in the local area according to the central location and range of the distribution of cattle and sheep populations, combined with a pre-established grassland carrying capacity model. If the grassland pressure in the local area exceeds a preset threshold, a new cattle and sheep activity range is generated based on the historical activity trajectories of the cattle and sheep.

Images