Sheep flock physiological data monitoring and management system for animal husbandry

Abstract

The invention relates to the field of sheep flock health management, discloses a sheep flock physiological data monitoring and management system for animal husbandry, and aims to overcome the defects of single monitoring index, lack of intelligent early warning, insufficient data processing and the like in the prior art. The system comprises an electronic ear tag installed on a sheep ear and used for monitoring heartbeat, exercise amount and body temperature data in real time and transmitting the data to a rear-end management system through wireless communication. The rear-end system is provided with a data processing module, an early warning module, a display module, a data comparison module and an LSTM prediction module, and the functions of fine data preprocessing, abnormal early warning, real-time display, historical and predicted data comparison and future physiological state prediction are achieved. When the physiological data of the sheep are abnormal, the system gives an alarm immediately, changes the color of the electronic ear tag, generates a hyperbola comparison graph, visually displays the difference between real-time data and predicted data, provides scientific and timely health management basis for breeding personnel, and effectively improves the health condition and production performance of the sheep flock.

Description

Technical Field

[0001] The present invention relates to the field of sheep flock health management, and in particular to a sheep flock physiological data monitoring and management system for animal husbandry. Background Art

[0002] In the livestock industry, sheep health management is critical to ensuring production efficiency and animal welfare. Traditionally, herders or breeders rely on manual observation to monitor sheep health. This method is not only time-consuming and labor-intensive, but also often fails to detect early health issues. With the development of Internet of Things (IoT) technology and big data analytics, intelligent and automated monitoring methods are gradually becoming a new trend in livestock management. Existing sheep monitoring systems mostly focus on monitoring single indicators, such as body temperature or exercise volume, and lack comprehensive physiological data monitoring and analysis capabilities. These systems often lack intelligent early warning mechanisms and are unable to issue timely alerts when sheep physiological indicators show abnormalities. As a result, potential health issues remain unaddressed, which in turn affects the health and production performance of the entire flock.

[0003] Some systems have attempted to integrate multiple physiological data sets, but they still face deficiencies in data processing and predictive analysis. For example, insufficient data preprocessing limits the accuracy of analytical results; a lack of effective data comparison methods makes it difficult for breeders to intuitively understand the differences between the sheep's current physiological state and its historical or predicted state; and inaccurate predictive models fail to provide strong support for breeding decisions. Summary of the Invention

[0004] The object of the present invention is to provide a sheep herd physiological data monitoring and management system for animal husbandry to solve the problems raised in the above background technology.

[0005] To achieve the above-mentioned object, the present invention provides the following technical solution: a sheep herd physiological data monitoring and management system for animal husbandry, the system comprising: Electronic ear tags are installed on the sheep's ears to monitor the sheep's heart rate, exercise volume and body temperature in real time; A back-end management system is wirelessly connected to the electronic ear tag and is used to receive, store and analyze physiological data transmitted by the electronic ear tag; the back-end management system further includes: A data processing module, used for pre-processing the received physiological data; The early warning module is used to determine whether physiological data is abnormal based on preset thresholds and issue an alarm when an abnormality occurs; The display module is used to display real-time curves of physiological data and alarm information. When a certain physiological data of the sheep exceeds the preset threshold, the early warning module will issue a prompt alarm in the back-end system, and the electronic ear tag will change color at the same time. A data comparison module is used to compare real-time physiological data with historical data or model prediction data to generate a hyperbola comparison graph; LSTM prediction model, used to predict the future physiological state of sheep based on historical physiological data; The data comparison module plots the real-time physiological data into a curve that changes over time, aligns the timestamps with the predicted data curve output by the LSTM prediction model, and places the two curves in the same coordinate system for comparison.

[0006] Preferably, the electronic ear tag further comprises: Sensors for collecting heart rate, movement, and body temperature data; The microprocessor is used to process the data collected by the sensor and send it to the back-end management system through the wireless communication module; the microprocessor compresses the data before sending it.

[0007] Preferably, the electronic ear tag further comprises a unique identifier for identification.

[0008] Preferably, the network structure of the LSTM prediction model is: Input layer, receiving historical physiological data sequence within the time window; Network layer: Contains two LSTM hidden layers, the first hidden layer contains 128 LSTM units, and the second hidden layer contains 64 LSTM units. Each LSTM unit has a forget gate, input gate, and output gate to control the updating, transmission, and forgetting of information. The ReLU activation function is used to perform nonlinear transformation on the output of each LSTM unit. The fully connected layer maps the output of the LSTM layer to the predicted physiological data value; The output layer outputs the predicted physiological data values, including the predicted values ​​of heart rate, exercise volume and body temperature.

[0009] Preferably, the steps of constructing the LSTM prediction model include: S1: Collect historical physiological data, including heart rate, exercise volume, and body temperature data, and randomly divide the collected feature data set into training set, validation set, and test set. The training set is used for model training, the validation set is used for model verification and parameter adjustment, and the test set is used to evaluate the final model performance. S2: Divide the training dataset into multiple batches, each batch contains a certain amount of time series data; S3: Use the training data set to iteratively train the LSTM model, calculate the gradient of the loss function through the backpropagation algorithm, and use the gradient descent method to update the model parameters. The parameter update formula is: ; in, represents the model parameters, including the weights and biases of the convolutional layers, is the learning rate, which is used to control the step size of parameter updates. is the loss function with respect to the parameters The gradient of ; the loss function uses the mean square error MSE, and the calculation formula is: ; Where n is the number of data points, is the model prediction value, is the actual value; S4: Use the validation dataset to regularly evaluate model performance, adjust the learning rate based on the evaluation results, and use early stopping to prevent overfitting; S5: After the model training is completed, the model is finally evaluated using the test dataset data.

[0010] Preferably, the implementation steps of the early warning module include: Step 1: Set normal threshold ranges for physiological data, including upper and lower limits for heart rate, exercise volume, and body temperature; Step 2: Analyze the received real-time physiological data and trigger an alarm if the data exceeds the normal threshold range; Step 3: After the alarm is triggered, the back-end management system displays the alarm information, including the abnormal physiological data type, the sheep number and the alarm time; at the same time, the electronic ear tag changes color, with red indicating an abnormal state and green indicating a healthy state.

[0011] Preferably, the step of generating a hyperbola comparison graph of the data comparison module includes: Step 1: Plot real-time physiological data into a curve that changes over time; Step 2: Plot the predicted data output by the LSTM prediction model into a curve that changes over time; Step 3: Align the timestamps of the two curves to ensure that the data at each time point are comparable; Step 4: Place the two curves in the same coordinate system and set different colors or line types to distinguish them; Step 5: Mark the significant differences between the two curves in the comparison chart as the focus of management attention.

[0012] Preferably, the step of marking the significant differences includes: Step 1: Calculate the difference between real-time physiological data and predicted data at each time point; Step 2: Set the difference threshold based on historical data and experience; Step 3: Mark the time points where the difference exceeds the threshold as significant difference points; Step 4: Use special marks to show significant differences in the comparison chart, and provide specific numerical values ​​and timestamp information of the differences.

[0013] Preferably, the data processing module further includes: Data cleaning unit, used to process missing values ​​and outliers in collected data; The data standardization unit is used to standardize the data and eliminate the dimensional differences in the distribution range of different data.

[0014] Preferably, the method for processing missing values ​​by the data cleaning unit includes: For numerical features, the median of the non-missing values ​​of the feature is used to fill the missing values; For categorical features, the mode of the non-missing values ​​of the feature is used to fill the missing values; For missing values ​​in time series data, the average value of the adjacent values ​​is used to fill them; For features with obvious business significance and few missing values, they are manually filled according to business logic; The method for processing abnormal values ​​by the data cleaning unit includes: For numerical features, the 3σ principle is used to identify and process outliers. Specifically, the mean and standard deviation of the feature are calculated, and values ​​exceeding the mean ± 3 times the standard deviation are considered outliers.

[0015] Compared with the prior art, the present invention has the following beneficial effects: This system uses electronic ear tags to monitor a variety of physiological data, including heart rate, activity level, and body temperature, in real time, providing comprehensive, real-time insights into the flock's health. Compared to traditional manual observation methods, this system significantly improves monitoring efficiency and accuracy, enabling early detection of health issues in sheep, providing a valuable window for timely intervention and treatment. The system's built-in early warning module intelligently analyzes physiological data based on preset thresholds. Upon detecting abnormal data, an alert is immediately issued in the back-end management system, and an on-site notification is provided through a color change in the electronic ear tag. This dual early warning mechanism ensures that breeders can respond quickly, isolating, inspecting, and treating abnormal sheep promptly, effectively preventing the spread and progression of disease and safeguarding the overall health of the flock.

[0016] The data comparison module compares real-time physiological data with historical data or model predictions, generating a hyperbolic comparison chart. This allows breeders to intuitively and clearly understand the differences between the sheep's current physiological state and the historical or predicted state. This visual data presentation simplifies data interpretation and improves the scientific nature and accuracy of breeding decisions. The LSTM prediction model predicts the sheep's future physiological state based on historical physiological data, providing breeders with forward-looking health management recommendations. By comparing real-time and predicted data, breeders can more scientifically formulate feeding plans and adjust the feeding environment, thereby optimizing the production performance and health of the flock. This data-based prediction and decision-making support contributes to the refined management and intelligent development of animal husbandry.

[0017] The application of this system has significantly improved livestock production efficiency and reduced sheep losses and medical expenses due to health issues. Through real-time monitoring and intelligent early warning, the system ensures that sheep receive timely and effective care, improving animal welfare and aligning with the concept of sustainable development in modern livestock farming. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is a working principle diagram of a sheep herd physiological data monitoring and management system for animal husbandry according to the present invention; Figure 2 This is a flow chart for connecting the electronic ear tag with the back-end management system; Figure 3 This is a diagram of the working principle of the LSTM prediction model. DETAILED DESCRIPTION

[0019] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0020] See also Figure 1-3 The present invention provides a technical solution: a sheep herd physiological data monitoring and management system for animal husbandry, the system comprising: Data Processing Module: This module first pre-processes the received physiological data, including data cleaning, noise removal, and format conversion, to ensure data accuracy and usability. This meticulous data pre-processing lays a solid foundation for subsequent data analysis and prediction.

[0021] Early Warning Module: This module monitors and assesses physiological data in real time based on preset thresholds. If a physiological metric exceeds a preset threshold, the module immediately issues an alert in the backend system and, via wireless communication, causes the electronic ear tag to change color, allowing farmers to quickly identify and address the abnormality.

[0022] Display Module: The display module intuitively displays real-time physiological data curves and alarm information. Through the graphical interface, farmers can clearly see the changing trends of various physiological data, as well as the specific time and value of abnormal data, so they can respond promptly.

[0023] Data Comparison Module: This module plots real-time physiological data as a time-varying curve and aligns the timestamps with the predicted data curve output by the LSTM prediction model. By comparing the two curves in the same coordinate system, breeders can intuitively understand the difference between the sheep's current physiological state and the historical or predicted state, providing a basis for breeding decisions.

[0024] LSTM prediction model: This prediction model, built on a long short-term memory (LSTM) neural network, can use historical physiological data to learn the changing patterns of sheep's physiological state and predict their physiological state in the future. Through continuous training and optimization, the LSTM prediction model can provide more accurate prediction results and provide a scientific basis for breeding management. The present invention will be further described below in conjunction with Examples 1 to 4: Example 1: As the front-end equipment of the sheep health monitoring system, the electronic ear tag includes the following key parts: A01.Sensor: Heartbeat sensor: Using a highly sensitive piezoelectric sensor, it can monitor the sheep's heartbeat frequency in real time to ensure the accuracy and real-time nature of the data.

[0025] Movement volume sensor: Using an acceleration sensor, the frequency and intensity of sheep's activities are detected to assess their movement volume, providing data support for analyzing the sheep's activity status.

[0026] Body temperature sensor: Use a thermistor or infrared sensor, close to the sheep's ear skin, to monitor body temperature changes in real time and reflect the health status of the sheep.

[0027] A02. Microprocessor: The microprocessor receives raw data from the sensor and performs preliminary processing, such as filtering and noise removal, to improve data quality. Before transmission, the microprocessor compresses the data using efficient compression algorithms (such as LZ77 and LZ78) to reduce data transmission volume and energy consumption. The microprocessor also controls the overall workflow of the electronic ear tag, including regular data collection, data processing, and communication with back-end management systems.

[0028] A03. Wireless Communication Module: This module utilizes wireless communication technologies such as Bluetooth Low Energy (BLE) or LoRa to achieve long-distance, stable communication with the back-end management system. The module is responsible for transmitting compressed physiological data to the back-end management system and receiving commands from the back-end management system, such as configuring parameters and querying data.

[0029] A04. Unique Identifier: Each electronic ear tag has a built-in unique identifier (such as a UID) that is used to distinguish data from different sheep in the back-end management system. The unique identifier is set at the time the electronic ear tag leaves the factory to ensure global uniqueness and facilitate data management and traceability.

[0030] The data processing module is the core part of the back-end management system. It is responsible for pre-processing the received physiological data, such as cleaning and standardization, to provide an accurate and reliable data basis for subsequent data analysis and early warning.

[0031] B01.Data Cleaning Unit: Missing value handling: For numerical features (such as heart rate, body temperature, etc.), the median of the non-missing values ​​of the feature is used to fill the missing values ​​to ensure the continuity and stability of the data.

[0032] For categorical features (such as sheep breed and sex), the mode of the non-missing values ​​of the feature is used to fill the missing values ​​to reflect the characteristics of most sheep.

[0033] For missing values ​​in time series data (such as missing exercise data within a certain period), the average value of the previous and next adjacent values ​​is used to fill in the missing values ​​to maintain the consistency of the data.

[0034] For features with obvious business significance and few missing values ​​(such as specific disease history), they are manually filled in according to business logic to ensure data integrity.

[0035] Outlier handling: For numerical features, the 3σ principle is used to identify and address outliers. Specifically, the mean and standard deviation of the feature are calculated. Values ​​exceeding ±3 times the standard deviation of the mean are considered outliers and replaced or deleted based on business logic. For example, data showing an abnormally high heart rate may indicate that the sheep is stressed or ill, requiring further verification and action.

[0036] B02.Data Standardization Unit: The data standardization unit is responsible for standardizing the cleaned data to eliminate the dimensional differences in the distribution range of different data.

[0037] Commonly used standardization methods (such as Z-score standardization, Min-Max standardization, etc.) are used to convert the original data into standard scores or normalized values, so that the data between different features are comparable.

[0038] The standardized data is easier to use for subsequent data analysis and model training, thus improving the accuracy of prediction results.

[0039] Example 2: This embodiment describes in detail the network structure of the LSTM prediction model and its construction steps, aiming to provide a model that can accurately predict the physiological state of a flock and provide a scientific basis for flock health management.

[0040] The LSTM prediction model uses a deep learning long short-term memory (LSTM) network structure, which specifically includes the following parts: A01. Input Layer: The input layer receives historical physiological data sequences within a time window as model input. The size of the time window is set based on actual needs; for example, data from the past 24 hours, 48 ​​hours, or even longer can be selected. Input data includes physiological indicators such as heart rate, exercise volume, and body temperature, which constitute the model's feature dimensions.

[0041] A02. Network Layer: The network layer consists of two LSTM hidden layers to capture long-term dependencies in time series data. The first hidden layer contains 128 LSTM units, and the second hidden layer contains 64 LSTM units. Each LSTM unit has a forget gate, an input gate, and an output gate. These gating mechanisms control the updating, transfer, and forgetting of information, effectively processing temporal dependencies and long-term memory in time series data.

[0042] The ReLU (Rectified Linear Unit) activation function is used to perform nonlinear transformation on the output of each LSTM unit to increase the expressive power of the model.

[0043] A03. Fully connected layer: The fully connected layer maps the output of the LSTM layer to the predicted physiological data value. The fully connected layer converts the output of the LSTM layer into the final predicted value through linear transformation and activation function (such as ReLU or Sigmoid).

[0044] A04. Output layer: The output layer outputs predicted physiological data values, including heart rate, exercise volume, and body temperature. These predicted values ​​are used to compare with actual physiological data to evaluate the model's predictive performance.

[0045] The steps to build the LSTM prediction model include: S1. Data preparation: Collect historical physiological data, including heart rate, exercise volume and body temperature. These data can be monitored in real time through electronic ear tags and stored in the back-end management system.

[0046] The collected feature dataset is randomly divided into a training set, a validation set, and a test set. The training set is used for model training, the validation set is used for model verification and parameter adjustment, and the test set is used to evaluate the final model performance. The division ratio can be set according to actual needs, for example, 70% for training, 15% for validation, and 15% for test.

[0047] S2. Data batch division: The training dataset is divided into multiple batches, each containing a certain amount of time series data. The batch size can be adjusted based on the complexity of the model and computing resources to ensure efficient and stable model training.

[0048] S3. Model training: The LSTM model is iteratively trained using the training dataset. The gradient of the loss function is calculated using the backpropagation algorithm, and the model parameters are updated using the gradient descent method. The parameter update formula is: ; in, represents the model parameters, including the weights and biases of the convolutional layers, is the learning rate, which is used to control the step size of parameter updates. is the loss function with respect to the parameters gradient.

[0049] The loss function uses the mean square error (MSE), and the calculation formula is: ; Where n is the number of data points, is the model prediction value, is the actual value, and MSE is used to measure the difference between the model prediction value and the actual value, guiding the optimization of model parameters.

[0050] S4. Model Validation and Parameter Adjustment: Use the validation dataset to regularly evaluate model performance and adjust hyperparameters such as learning rate based on the evaluation results to improve the generalization ability of the model.

[0051] Use early stopping to prevent overfitting. When the performance on the validation set stops improving or starts to degrade, stop model training to avoid overfitting the model on the training set.

[0052] S5. Model Evaluation: After model training is complete, the model is evaluated using the test dataset. Evaluation metrics can include mean squared error (MSE) and mean absolute error (MAE) to comprehensively assess the model's predictive performance.

[0053] Based on the evaluation results, the model structure or parameters can be further adjusted to optimize the model's prediction effect.

[0054] Example 3: This embodiment describes in detail the implementation steps of the early warning module, which aims to monitor the physiological data of the sheep in real time, detect abnormalities in a timely manner and trigger alarms, thereby ensuring the health and safety of the sheep.

[0055] The implementation steps of the early warning module include: Step 1. Set the normal threshold range of physiological data: Normal threshold ranges for heart rate, exercise volume, and body temperature are set based on the sheep's physiological characteristics and breeding experience. For example, the normal range for heart rate might be set between 60 and 100 beats per minute, the normal range for exercise volume might be set based on the sheep's activity habits, and the normal range for body temperature might be set between 38°C and 40°C.

[0056] These normal threshold ranges can be appropriately adjusted according to factors such as season, age, and gender to ensure the accuracy and effectiveness of the early warning.

[0057] After the normal threshold range is set, it is stored in the database of the back-end management system and serves as the basis for subsequent judgment of whether the physiological data is abnormal.

[0058] Step 2. Judge the received real-time physiological data: The back-end management system receives physiological data from the electronic ear tag in real time, including heart rate, exercise volume and body temperature.

[0059] The system pre-processes these data, such as denoising and filtering, to improve the accuracy and reliability of the data.

[0060] The system then compares the pre-processed data with the set normal threshold range. If the data exceeds the normal threshold range, it is judged as abnormal data and the alarm mechanism is triggered.

[0061] Step 3. Processing after the alarm is triggered: When an alarm is triggered, the backend management system immediately displays the alarm information. The alarm information includes the abnormal physiological data type (such as abnormal heart rate, abnormal exercise volume, or abnormal body temperature), the sheep number (to identify the specific sheep with the abnormality), and the alarm time (recording the time when the abnormality occurred).

[0062] At the same time, the system sends alarm notifications to the breeders, which can be achieved through SMS, email or APP push, so that the breeders can take timely measures.

[0063] The electronic ear tag also displays color changes to indicate the sheep's health. When physiological data is abnormal, the tag turns red, indicating an abnormal state. When physiological data is within the normal range, the tag remains green, indicating a healthy state. This intuitive color change helps farmers quickly identify abnormal sheep.

[0064] Suppose a sheep's heart rate suddenly rises to 120 beats per minute, exceeding the set normal threshold range (60 to 100 beats per minute). Upon receiving this abnormal data, the back-end management system immediately triggers an alarm. The system displays the alarm message: "Abnormal heart rate, sheep number: 001, alarm time: 2023-04-01 10:30:00". At the same time, the farmer receives the alarm notification and goes to check on the sheep. The electronic ear tag is now red, clearly indicating that the sheep is in an abnormal state. Based on the alarm information and on-site observations, the farmer takes timely measures, such as contacting a veterinarian for diagnosis and treatment, thereby effectively ensuring the health and safety of the sheep. Example 4: This example describes in detail the steps for generating a hyperbola comparison graph in the data comparison module and the method for marking significant differences. This aims to visually compare real-time physiological data with the predicted data from the LSTM prediction model, helping managers quickly identify and focus on significant differences between the data, thereby enabling timely implementation of appropriate measures to safeguard the health of the flock. The steps for generating a hyperbola comparison graph include: Step 1. Draw real-time physiological data curve: Obtain real-time sheep physiological data from the back-end management system, including heart rate, exercise volume, and body temperature.

[0065] With time as the horizontal axis and physiological data value as the vertical axis, use chart drawing tools (such as Matplotlib, ECharts, etc.) to draw real-time physiological data into a curve that changes over time.

[0066] The curve color or line type can be set according to actual needs to facilitate distinction and observation.

[0067] Step 2. Draw the LSTM prediction data curve: Using the trained LSTM prediction model, we predict physiological data for a period of time in the future based on historical physiological data. We also plot the predicted data as a time-varying curve, again with time as the horizontal axis and the predicted value as the vertical axis.

[0068] In order to distinguish it from the real-time physiological data curve, the predicted data curve should use a different color or line type.

[0069] Step 3. Timestamp alignment: Ensure that the real-time physiological data curve and the predicted data curve are aligned on the time axis, that is, the data at each time point are comparable.

[0070] If the time intervals between data collection and prediction are inconsistent, timestamp conversion and alignment are required.

[0071] Step 4. Place them in the same coordinate system: Placing the real-time physiological data curve and the predicted data curve in the same coordinate system facilitates intuitive comparison.

[0072] Adjust the scale, labels, and titles of the coordinate system to make the chart clearer and easier to read.

[0073] Step 5. Mark significant differences: By calculating the difference between real-time physiological data and predicted data at each time point and setting the difference threshold, the significant difference points between the two curves are marked.

[0074] Significant differences should be shown in the comparison chart with special marks (such as circles, asterisks, etc.), and the specific numerical value and time stamp information of the difference should be provided.

[0075] The steps for marking significant differences include: Step 1. Calculate the difference: The real-time physiological data and the predicted data at each time point are subtracted to obtain a difference sequence.

[0076] The difference can be positive or negative, indicating that the real-time data is higher or lower than the forecasted data.

[0077] Step 2. Set the difference threshold: Set a reasonable difference threshold based on historical data and experience.

[0078] The difference threshold can be adjusted according to data fluctuations, model prediction accuracy, and management needs.

[0079] Step 3. Mark significant differences: Traverse the difference sequence and mark the time points where the difference exceeds the threshold as significant difference points.

[0080] One or more thresholds can be set to distinguish between different levels of significant differences.

[0081] Step 4. Show significant differences in the comparison chart: In the hyperbola comparison chart, significant differences are shown with special marks (such as red circles, bold lines, etc.).

[0082] The specific value and timestamp information of the difference are provided next to or below the significant difference point, so that managers can quickly locate and understand the abnormal situation.

[0083] Suppose that on a given day, the real-time heart rate data collected for a particular sheep differs significantly from the predicted data output by the LSTM prediction model. Using the hyperbolic curve comparison chart, managers can clearly see that the real-time heart rate curve (solid blue line) deviates significantly from the predicted heart rate curve (dashed red line) at certain time points (such as 10:00, 14:00, and 18:00). Following the steps for labeling significant difference points, these time points are marked as significant and displayed as red circles in the chart. The specific difference values ​​(such as +10, -15, +8) and timestamp information are also provided. Managers can immediately monitor the data at these time points, analyze possible causes (such as frightened sheep or illness), and take appropriate measures (such as checking the sheep's condition and adjusting the feeding environment) to ensure the health of the flock.

[0084] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0085] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A sheep herd physiological data monitoring and management system for animal husbandry, characterized in that: The system comprises: Electronic ear tags are installed on the sheep's ears to monitor the sheep's heart rate, exercise volume and body temperature in real time; A back-end management system is wirelessly connected to the electronic ear tag and is used to receive, store and analyze physiological data transmitted by the electronic ear tag; the back-end management system further includes: A data processing module, used for pre-processing the received physiological data; The early warning module is used to determine whether physiological data is abnormal based on preset thresholds and issue an alarm when an abnormality occurs; The display module is used to display real-time curves of physiological data and alarm information. When a certain physiological data of the sheep exceeds the preset threshold, the early warning module will issue a prompt alarm in the back-end system, and the electronic ear tag will change color at the same time. A data comparison module is used to compare real-time physiological data with historical data or model prediction data to generate a hyperbola comparison graph; LSTM prediction model, used to predict the future physiological state of sheep based on historical physiological data; The data comparison module plots the real-time physiological data into a time-varying curve, aligns the timestamps with the predicted data curve output by the LSTM prediction model, and places the two curves in the same coordinate system for comparison; The steps for building the LSTM prediction model include: S1: Collect historical physiological data, including heart rate, exercise volume, and body temperature data, and randomly divide the collected feature data set into training set, validation set, and test set. The training set is used for model training, the validation set is used for model verification and parameter adjustment, and the test set is used to evaluate the final model performance. S2: Divide the training dataset into multiple batches, each batch contains a certain amount of time series data; S3: Use the training data set to iteratively train the LSTM model, calculate the gradient of the loss function through the backpropagation algorithm, and use the gradient descent method to update the model parameters. The parameter update formula is: ； in, represents the model parameters, including the weights and biases of the convolutional layers, is the learning rate, which is used to control the step size of parameter updates. is the loss function with respect to the parameters The gradient of ; the loss function uses the mean square error MSE, and the calculation formula is: ； Where n is the number of data points, is the model prediction value, is the actual value; S4: Use the validation dataset to regularly evaluate model performance, adjust the learning rate based on the evaluation results, and use early stopping to prevent overfitting; S5: After the model training is completed, the model is finally evaluated using the test dataset data.

2. A sheep herd physiological data monitoring and management system for animal husbandry according to claim 1, characterized in that: The electronic ear tag further comprises: Sensors for collecting heart rate, movement, and body temperature data; The microprocessor is used to process the data collected by the sensor and send it to the back-end management system through the wireless communication module; the microprocessor compresses the data before sending it.

3. A sheep herd physiological data monitoring and management system for animal husbandry according to claim 2, characterized in that: The electronic ear tag also includes a unique identifier for identification.

4. The sheep herd physiological data monitoring and management system for animal husbandry according to claim 1, characterized in that: The network structure of the LSTM prediction model is: Input layer, receiving historical physiological data sequence within the time window; Network layer: Contains two LSTM hidden layers, the first hidden layer contains 128 LSTM units, and the second hidden layer contains 64 LSTM units. Each LSTM unit has a forget gate, input gate, and output gate to control the updating, transmission, and forgetting of information. The ReLU activation function is used to perform nonlinear transformation on the output of each LSTM unit. The fully connected layer maps the output of the LSTM layer to the predicted physiological data value; The output layer outputs the predicted physiological data values, including the predicted values ​​of heart rate, exercise volume and body temperature.

5. The sheep herd physiological data monitoring and management system for animal husbandry according to claim 1, characterized in that: The implementation steps of the early warning module include: Step 1: Set normal threshold ranges for physiological data, including upper and lower limits for heart rate, exercise volume, and body temperature; Step 2: Analyze the received real-time physiological data and trigger an alarm if the data exceeds the normal threshold range; Step 3: After the alarm is triggered, the back-end management system displays the alarm information, including the abnormal physiological data type, the sheep number and the alarm time; at the same time, the electronic ear tag changes color, with red indicating an abnormal state and green indicating a healthy state.

6. The sheep herd physiological data monitoring and management system for animal husbandry according to claim 1, characterized in that: The step of generating a hyperbola comparison graph of the data comparison module includes: Step 1: Plot real-time physiological data into a curve that changes over time; Step 2: Plot the predicted data output by the LSTM prediction model into a curve that changes over time; Step 3: Align the timestamps of the two curves to ensure that the data at each time point are comparable; Step 4: Place the two curves in the same coordinate system and set different colors or line types to distinguish them; Step 5: Mark the significant differences between the two curves in the comparison chart as the focus of management attention.

7. A sheep herd physiological data monitoring and management system for animal husbandry according to claim 6, characterized in that: The step of marking the significant differences includes: Step 1: Calculate the difference between real-time physiological data and predicted data at each time point; Step 2: Set the difference threshold based on historical data and experience; Step 3: Mark the time points where the difference exceeds the threshold as significant difference points; Step 4: Use special marks to show significant differences in the comparison chart, and provide specific numerical values ​​and timestamp information of the differences.

8. The sheep herd physiological data monitoring and management system for animal husbandry according to claim 1, characterized in that: The data processing module further includes: Data cleaning unit, used to process missing values ​​and outliers in collected data; The data standardization unit is used to standardize the data and eliminate the dimensional differences in the distribution range of different data.

9. The sheep herd physiological data monitoring and management system for animal husbandry according to claim 7, characterized in that: The method for processing missing values ​​by the data cleaning unit includes: For numerical features, the median of the non-missing values ​​of the feature is used to fill the missing values; For categorical features, the mode of the non-missing values ​​of the feature is used to fill the missing values; For missing values ​​in time series data, the average value of the adjacent values ​​is used to fill them; For features with obvious business significance and few missing values, they are manually filled according to business logic; The method for processing abnormal values ​​by the data cleaning unit includes: For numerical features, the 3σ principle is used to identify and process outliers. Specifically, the mean and standard deviation of the feature are calculated, and values ​​exceeding the mean ± 3 times the standard deviation are considered outliers.

Images