Intelligent yak breeding monitoring system

Abstract

The invention relates to the technical field of modern agriculture, and discloses an intelligent yak breeding monitoring system which comprises a sensor network module, a data processing center module, an intelligent analysis module and a remote monitoring terminal module. The sensor network module integrates various sensors and is used for monitoring physical signs, behaviors and breeding environment parameters of yaks in real time; the data processing center module is responsible for receiving, processing and storing original data transmitted by the sensor network; the intelligent analysis module establishes a health assessment model and an environment optimization algorithm based on the processed data, and realizes real-time assessment of the health condition of the yaks and intelligent optimization of the breeding environment; the remote monitoring terminal module provides a user interface and supports remote control and early warning notification functions. The intelligent yak breeding monitoring system aims at solving the problems that an existing yak breeding environment monitoring technology is single in data, insufficient in intelligent degree and difficult to meet the requirements of precise breeding and scientific management.

Description

Technical Field

[0001] The present invention relates to the field of modern agricultural technologies, and particularly to an intelligent yak breeding monitoring system. Background Art

[0002] With the booming development of the yak breeding industry, the precise monitoring and scientific management of its breeding environment have become key factors in promoting the sustainable development of the industry. As a unique and rare livestock species on the plateau, yaks have a unique and complex growth environment and are extremely sensitive to conditions such as temperature, humidity, and grassland quality. Through the organic combination of precise monitoring and scientific management, the yak breeding industry will be able to better meet market demands, achieve a win-win situation of ecological and economic benefits, and make greater contributions to the economic development and social stability of the plateau region.

[0003] However, there are obvious defects in the current yak breeding environment monitoring technology, which are difficult to meet the actual breeding needs. On the one hand, the monitoring data is single-dimensional. Existing systems mostly focus on basic indicators such as temperature and humidity. These single data dimensions not only limit the breeders' comprehensive understanding of the environment but also hinder the formulation of precise breeding strategies. On the other hand, the lack of intelligence has become a limiting factor. The existing monitoring equipment has a low level of automation, and data collection and analysis rely on manual labor, resulting in low efficiency and easy errors. At the same time, there is a lack of an intelligent early warning and decision support system, making it difficult to respond in a timely manner and take effective measures in the face of sudden environmental changes or disease risks. Summary of the Invention

[0004] The purpose of the present invention is to solve the problems of single-dimensional data and lack of intelligence in the existing yak breeding environment monitoring technology, which are difficult to meet the needs of precise breeding and scientific management, and to propose an intelligent yak breeding monitoring system.

[0005] The technical solution of the present invention to solve the above technical problems is as follows:

[0006] An intelligent yak breeding monitoring system, comprising: a sensor network module, a data processing center module, an intelligent analysis module, and a remote monitoring terminal module;

[0007] The sensor network module integrates a variety of sensors for real-time monitoring of yak body signs, behaviors, and breeding environment parameters;

[0008] The data processing center module is responsible for receiving, processing, and storing the original data transmitted by the sensor network;

[0009] The intelligent analysis module, based on the processed data, establishes a health assessment model and an environmental optimization algorithm to achieve real-time assessment of yak health status and intelligent optimization of the breeding environment;

[0010] The remote monitoring terminal module provides a user interface and supports remote control and early warning notification functions.

[0011] On the basis of the above technical solutions, the present invention can also be improved as follows.

[0012] Furthermore, the sensor network module includes: a physical sign sensor sub-module, a behavior sensor sub-module, and an environmental sensor sub-module;

[0013] The physical sign sensor sub-module is an intelligent ear tag installed on the yak's ear and integrated with a heart rate sensor, a body temperature sensor, and an acceleration sensor, which is used to monitor the heart rate, body temperature, and activity status of the yak in real time;

[0014] The behavior sensor sub-module is to install a video camera and an infrared thermal imager in the breeding area, and combine deep learning algorithms to identify the behavior patterns of yaks, including but not limited to eating, resting, and moving;

[0015] The environmental sensor sub-module is to distribute and install temperature and humidity sensors, light sensors, and ammonia sensors in the breeding area, which are used to monitor the breeding environment parameters in real time.

[0016] Furthermore, the data processing center module includes: a data acquisition sub-module, a data transmission sub-module, and a data storage sub-module;

[0017] The data acquisition sub-module is responsible for receiving the raw data from the sensor network and performing preprocessing operations such as denoising and filtering;

[0018] The data transmission sub-module uses LoRa wireless communication technology to achieve low-power and long-distance data transmission between the sensor network and the data processing center;

[0019] The data storage sub-module stores the preprocessed data in the database for subsequent analysis and query.

[0020] Furthermore, the intelligent analysis module includes: a health assessment model, an environmental optimization algorithm, and an anomaly detection algorithm;

[0021] The health assessment model is based on the physical sign parameters and behavior patterns of yaks to establish a health assessment model and evaluate the health status of yaks in real time;

[0022] The environmental optimization algorithm is to formulate a breeding environment optimization strategy according to the environmental sensor data and in combination with the physiological needs of yaks, including but not limited to automatically adjusting the temperature and humidity, and light;

[0023] The anomaly detection algorithm is to use machine learning algorithms to perform anomaly detection on the physical signs and behavior data of yaks, and timely discover and warn of potential health problems.

[0024] Furthermore, the health assessment model adopts the following dynamic adaptive algorithm formula:

[0025] Health Index

[0026] w i It is the basic weight coefficient, calculated by the CRITIC method, taking into account the comparison intensity and conflict of indicators;

[0027] is the standardized processing item, p i is the real-time physiological parameter value, μ i is the historical mean, σ i is the standard deviation;

[0028] σ i To dynamically adjust the coefficient, an adaptive weighted algorithm is used:

[0029]

[0030] where δ ij is the environmental factor influence coefficient, γ j Real-time monitoring value of environmental factors;

[0031] β is the comprehensive environmental impact coefficient, which is calculated by the entropy weight method;

[0032] E is the environmental pressure index, which is synthesized by reducing the dimensions of multi-dimensional environmental parameters such as temperature, humidity, altitude, and air pressure through principal component analysis.

[0033] Furthermore, the anomaly detection algorithm adopts a machine learning architecture that integrates support vector machine and random forest. Specifically, a feature vector X = [heart rate, body temperature, activity level, ...] is constructed for the yak's physical signs and behavior data, and a single-class classifier is constructed through SVM, whose decision function is The kernel function K(·,·) uses the radial basis kernel. At the same time, a random forest is used to build an integrated model of multiple decision trees, and each tree splits the node based on the Gini coefficient. Finally, a weighted fusion strategy is used. Output abnormal score, when the score exceeds the dynamic threshold θ(t) and the heart rate is higher than μ for m consecutive sampling points HR +3σ HR When the warning mechanism is triggered, a notification containing the abnormal level and recommended measures is automatically sent to the breeding staff, where the weight ω SVM ,ω RE Dynamically optimize based on real-time performance of the model.

[0034] Furthermore, the remote monitoring terminal includes: a user interface, remote control and early warning notification;

[0035] The user interface provides an intuitive user interface to display yak physical signs, behaviors, environmental parameters and health assessment results;

[0036] The remote control supports remote control of breeding equipment through a mobile phone APP or a web-based terminal, including but not limited to automatic feeders, waterers, and ventilation equipment;

[0037] The warning notification automatically sends a warning notification to the breeding personnel when abnormal data is detected, reminding them to process it in a timely manner.

[0038] Furthermore, the user interface uses Web technology to develop a responsive web interface, which adapts to different device screens, integrates a data visualization library, and displays the dynamic changes of yak physical signs, behaviors, and environmental parameters in real time. The remote control communicates with the data processing center through RESTful API to send remote control instructions. The warning notification is that when abnormal data is detected, the system automatically generates a warning message and sends it to the breeding personnel through SMS, email, or APP push.

[0039] Furthermore, the instructions include: adjusting the feeding amount, controlling the switch of the waterer, and adjusting the rotation speed of the ventilation equipment. The warning information includes the type of abnormal data, the occurrence time, the possible cause, and the recommended handling measures.

[0040] Compared with the prior art, the technical solution of the present application has the following beneficial technical effects:

[0041] In the present invention, a physical sign and environment sensor are integrated through a sensor network module to construct a two-dimensional monitoring system of "yak - environment", realizing multi-modal data fusion. Through an intelligent analysis module, a health assessment model and an environment optimization algorithm are established, which can not only accurately warn of disease risks, but also dynamically analyze the impact of the environment on yak behavior, providing a decision-making basis beyond a single indicator for breeders. At the level of data processing and decision-making efficiency, the system converts raw data into actionable decision-making suggestions through automated data cleaning, feature extraction, and model training. This intelligent processing method improves the data utilization efficiency by several times and significantly reduces human errors. In addition, through real-time warning and remote monitoring terminal modules, the system realizes millisecond-level response to risks and mobile management, dynamically adjusts breeding strategies, transforms passive response into active prevention and control, and promotes the implementation of precision breeding through a scientific decision-making support system. By quantitatively analyzing the relationship between yak feed intake and grassland quality, and between environmental parameters and reproductive rate, a quantifiable decision-making basis is provided for feeding plans and breeding times. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a structural block diagram of an intelligent yak breeding monitoring system of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0043] 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 of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without making creative efforts fall within the protection scope of the present invention.

[0044] An intelligent yak breeding monitoring system of the present invention includes: a sensor network module, a data processing center module, an intelligent analysis module, and a remote monitoring terminal module;

[0045] The sensor network module integrates a variety of sensors for real-time monitoring of yak physical signs, behaviors, and breeding environment parameters;

[0046] The data processing center module is responsible for receiving, processing, and storing the original data transmitted by the sensor network;

[0047] The intelligent analysis module, based on the processed data, establishes a health assessment model and an environment optimization algorithm to realize the real-time assessment of the yak's health status and the intelligent optimization of the breeding environment;

[0048] The remote monitoring terminal module provides a user interface and supports remote control and early warning notification functions.

[0049] In a preferred embodiment of the present invention, it can be further configured that: the sensor network module includes: a physical sign sensor sub-module, a behavior sensor sub-module, and an environment sensor sub-module;

[0050] The physical sign sensor sub-module is an intelligent ear tag installed on the yak's ear and integrated with a heart rate sensor, a body temperature sensor, and an acceleration sensor for real-time monitoring of the yak's heart rate, body temperature, and activity status;

[0051] The behavior sensor sub-module is to install a video camera and an infrared thermal imager in the breeding area, and combine deep learning algorithms to identify the behavior patterns of yaks, including but not limited to eating, resting, and moving;

[0052] The environment sensor sub-module is to distribute and install temperature and humidity sensors, light sensors, and ammonia sensors in the breeding area for real-time monitoring of breeding environment parameters. The physical sign sensor sub-module realizes continuous monitoring of yak physical signs through the intelligent ear tag. The behavior sensor sub-module combines deep learning algorithms to automatically identify yak behaviors, breaking through the limitation that traditional video monitoring can only provide visual images and realizing structured analysis of behavior patterns; the environment sensor sub-module provides multi-dimensional data support for the regulation of the breeding environment through multi-parameter monitoring and distributed layout.

[0053] The intelligent ear tag installed on the yak's ear not only integrates a heart rate sensor, a body temperature sensor, and an acceleration sensor, but also incorporates a low-power wireless communication chip and a micro energy storage device. It realizes low-power long-distance data transmission through the LoRaWAN protocol. The surface of the ear tag uses biocompatible materials and reduces the foreign body sensation of the yak through bionic design. The system uses an adaptive sampling algorithm to increase the sampling frequency of the acceleration sensor (up to 200Hz) when the yak is moving, and reduces the sampling frequency to 1Hz in the stationary state to balance data accuracy and battery life. In addition, the ear tag is built with an abnormal state detection mechanism. When continuously abnormal heart rate or body temperature is detected, it can actively wake up the communication module to send an emergency warning signal to ensure that the breeder responds in a timely manner. The video cameras installed in the breeding area use 360° panoramic stitching technology and combine with an infrared thermal imager to achieve non-stop monitoring day and night. The deep learning algorithm is based on a hybrid model of convolutional neural network (CNN) and long short-term memory network (LSTM) to classify and identify yak behaviors. The system realizes accurate identification of 12 typical behaviors such as eating, resting, and moving by constructing a yak behavior database containing more than 100,000 hours of labeled video data, and the average recognition accuracy exceeds 92%. In addition, the system can combine the individual characteristics of the yak (such as body size, age) and historical behavior patterns to give early warnings for abnormal behaviors (such as being away from the herd for a long time, a sudden decrease in activity), and generate a behavior heat map to assist the breeder in analyzing the group behavior characteristics. The temperature and humidity sensors, light sensors, and ammonia sensors are arranged in a hierarchical and distributed manner in the breeding area. Among them, the temperature and humidity sensors are deployed in a 50m×50m grid, the light sensors are installed on the top of the shed and around the sports ground, and the ammonia sensors are mainly deployed in the feeding area and the excretion area. The sensors are networked through the ZigBee protocol to achieve real-time data synchronization and self-network repair. The system can combine environmental parameters and yak behavior data to establish an environment-behavior correlation model. For example, when the ammonia concentration exceeds the threshold, the system can combine the rest time and movement frequency of the yak to evaluate its tolerance to the breeding environment and automatically trigger the ventilation equipment control instruction. In addition, the environmental sensor sub-module also integrates a weather prediction interface, which can predict extreme weather such as rainfall and temperature drop 48 hours in advance and provide disaster prevention suggestions for breeders.

[0054] In a preferred embodiment of the present invention, it can be further configured as follows: The data processing center module includes: a data acquisition sub-module, a data transmission sub-module, and a data storage sub-module;

[0055] The data acquisition sub-module is responsible for receiving the raw data from the sensor network and performing preprocessing operations such as denoising and filtering;

[0056] The data transmission sub-module uses LoRa wireless communication technology to achieve low-power, long-distance data transmission between the sensor network and the data processing center;

[0057] The data storage sub-module stores the preprocessed data in the database for subsequent analysis and query. The data acquisition sub-module eliminates the abnormal data generated by environmental interference or equipment failures in the sensor network through real-time denoising and filtering, ensuring the accuracy of the original data. The data transmission sub-module uses LoRa wireless communication technology, breaking through the limitations of traditional technologies that rely on wired networks or short-distance wireless communication, achieving low-power, long-distance data transmission covering the entire breeding area, and significantly reducing the network deployment cost and maintenance difficulty. The data storage sub-module classifies and stores the preprocessed data in multiple dimensions such as time, type, and individual yaks through a structured database design, providing efficient data retrieval capabilities for subsequent health assessment, environmental analysis, and decision optimization.

[0058] The data acquisition sub-module is built with an adaptive filtering algorithm that can dynamically adjust the denoising parameters according to the sensor type. For example, for heart rate sensor data, a dynamic denoising model based on Kalman filtering is used to effectively suppress the random noise generated during the movement of yaks. For temperature and humidity sensor data, wavelet transform denoising technology is used to eliminate high-frequency environmental interference. In addition, the module integrates a data quality assessment mechanism that automatically marks abnormal data and triggers an artificial review process by comparing the deviation between historical data and real-time data. The system supports multi-source data fusion, aligning the data of different sensors according to timestamps to generate data packets in a unified format, providing a standardized input for subsequent analysis. The LoRa wireless communication technology adopts a star network topology structure, realizing the interconnection between sensor nodes and the data processing center through gateway devices. To improve transmission reliability, the module is built with a redundant backup mechanism that automatically switches to the backup channel when the main channel fails. The data transmission uses the AES-128 encryption algorithm to ensure data security. In addition, the module supports dynamic adjustment of the transmission power, adaptively adjusting the transmission rate according to the signal strength, reducing power consumption while ensuring data integrity. The system can combine the terrain characteristics of the breeding area and optimize the transmission path through the distributed deployment of LoRa gateways. For example, relay nodes are used to expand the coverage area in mountainous breeding areas, and the battery life is extended by reducing the transmission power in plain breeding areas. The database adopts a distributed storage architecture, supporting multi-dimensional indexing such as individual yaks, breeding areas, and time periods. The module is built with data cleaning rules that can automatically identify and correct duplicate or incorrect data. The system supports data compression storage, improving the original data compression rate to over 70% through the LZ77 algorithm while ensuring the lossless restoration of the decompressed data. In addition, the data storage sub-module integrates a data visualization interface that can generate analysis charts such as trend charts of yak health status and heat maps of environmental parameters in real time, providing intuitive data insights for breeders. For example, the system can automatically compare the weight growth curves of different batches of yaks, identify individuals with abnormal growth, and associate their physical signs with environmental data to assist breeders in optimizing feeding strategies.

[0059] In a preferred embodiment, the present invention can be further configured as follows: The intelligent analysis module includes: a health assessment model, an environment optimization algorithm, and an anomaly detection algorithm;

[0060] The health assessment model is based on yak physical signs parameters and behavior patterns to establish a health assessment model, and real-time assesses the health status of yaks;

[0061] The environment optimization algorithm formulates a breeding environment optimization strategy according to environmental sensor data and in combination with the physiological needs of yaks, including but not limited to automatically adjusting temperature, humidity, and lighting;

[0062] The anomaly detection algorithm uses machine learning algorithms to detect anomalies in yak physical signs and behavior data, and timely discovers and warns of potential health problems. The health assessment model realizes the dynamic quantitative assessment of the health status of yaks by constructing a correlation network of physical signs parameters and behavior patterns, breaking through the limitation of traditional technologies relying on manual experience judgment; The environment optimization algorithm realizes the refined control of the breeding environment based on the intelligent matching of environmental sensor data and the physiological needs of yaks, and solves the problem of blindness in adjusting parameters such as temperature, humidity, and lighting in traditional technologies; The anomaly detection algorithm can identify potential health risks in advance through the real-time analysis of physical signs and behavior data by machine learning technology, significantly shortening the anomaly response time.

[0063] The health assessment model uses multi-modal data fusion technology to spatio-temporally align the physical signs parameters such as the heart rate, body temperature, and number of steps of yaks with the behavior patterns such as feeding frequency, rest duration, and social behavior to construct a dynamic health index (DHI). By introducing a long short-term memory network (LSTM) and an attention mechanism, the model captures the long-term trends and mutation characteristics of physical signs and behavior data. For example, when the number of steps of a yak decreases by more than 30% and the body temperature fluctuates by more than 0.5°C within 3 consecutive days, the model can automatically determine its health status as "sub-healthy" and associate its recent environmental parameters (such as ammonia concentration, light intensity) to generate a diagnostic report containing potential causes. In addition, the model supports individualized calibration and can dynamically adjust the evaluation threshold according to the characteristics of the yak's breed, age, gender, etc. to improve the accuracy of the evaluation. The environmental optimization algorithm is based on the physiological requirement database of yaks, which contains the tolerance thresholds of temperature, humidity, light, and ammonia concentration at different growth stages (such as cub stage, fattening stage). Through a reinforcement learning framework, the algorithm real-time optimizes the control strategy of the breeding environment. For example, when the detected ammonia concentration exceeds the threshold, the algorithm not only starts the ventilation equipment but also adjusts the ventilation intensity and frequency in combination with the yak's behavior pattern (such as a decrease in feeding frequency) to avoid energy waste caused by excessive ventilation. In addition, the algorithm integrates a meteorological prediction interface, which can predict extreme weather (such as heavy rain, high temperature) 48 hours in advance and dynamically adjust the control priority of environmental parameters. For example, under a high temperature warning, it gives priority to ensuring the activation of the ventilation and sunshade facilities in the shed rather than humidity adjustment. The anomaly detection algorithm uses a hybrid machine learning model, combining the Isolation Forest and Autoencoder technologies to perform multi-level anomaly recognition on the physical signs and behavior data of yaks. The Isolation Forest is used to quickly locate global anomaly points (such as a sudden drop in heart rate, abnormal movement trajectory), and the Autoencoder captures local subtle changes through the reconstruction error (such as body temperature fluctuations, shortened feeding duration). The system can dynamically adjust the sensitivity of anomaly detection in combination with the individual health records of yaks. For example, for old yaks, the threshold for heart rate fluctuations is appropriately lowered to avoid false alarms; for yaks in the fattening stage, the anomaly determination standard for the number of steps is raised to prevent misjudgment caused by excessive exercise. In addition, the algorithm supports the traceability analysis of anomaly events and generates a report containing the cause of the anomaly, the scope of influence, and disposal suggestions by associating environmental parameters with historical data.

[0064] In a preferred embodiment of the present invention, it can be further configured as follows: The health assessment model adopts the following dynamic adaptive algorithm formula:

[0065] Health index

[0066] w i Is the basic weight coefficient, calculated by the CRITIC method, comprehensively considering the index comparison intensity and conflict;

[0067] is the standardized processing item, p i is the real-time physiological parameter value, μ i is the historical mean value, σ i is the standard deviation;

[0068] σ i is the dynamic adjustment coefficient, adopting an adaptive weighted algorithm:

[0069]

[0070] where δ ij is the environmental factor influence coefficient, γ j is the real-time monitoring value of the environmental factor;

[0071] β is the comprehensive environmental influence coefficient, calculated by the entropy weight method;

[0072] E is the environmental stress index, synthesized by dimensionality reduction of multi-dimensional environmental parameters such as temperature, humidity, altitude, and air pressure through the principal component analysis method, and the basic weight coefficient is calculated by the CRITIC method, realizing the objective weighting of the evaluation index and avoiding the limitations of subjective weight setting in traditional technologies; the standardized processing item combines real-time physiological parameters with historical data, effectively eliminating the influence of individual differences and environmental fluctuations on the evaluation results; the dynamic adjustment coefficient couples the real-time monitoring values of environmental factors (such as temperature, humidity, altitude, and air pressure) with the comprehensive influence coefficient through an adaptive weighted algorithm, enabling the model to dynamically adjust the evaluation threshold according to environmental changes.

[0073] In the core formula of the health assessment model, the basic weight coefficient is calculated by the CRITIC method, which comprehensively considers the contrast intensity (reflected by the standard deviation) and conflict (reflected by the correlation coefficient matrix) of each physiological index (such as heart rate, body temperature, and number of exercise steps), ensuring the scientificity and rationality of weight allocation. For example, there may be a high conflict between heart rate variability and body temperature fluctuations in health assessment. The CRITIC method calculates their conflict coefficient and dynamically adjusts the weight ratio of the two to avoid a single index dominating the assessment result. The standardization processing item uses the Z-score standardization method to compare the real-time physiological parameter values with the historical mean and standard deviation to generate a dimensionless health index, making the assessment results comparable among different yak individuals. The environmental factor influence coefficient in the dynamic adjustment coefficient is obtained by fitting a linear regression model. For example, when the environmental temperature exceeds the comfortable temperature range of yaks (5℃ - 25℃), the environmental factor influence coefficient will linearly increase with the degree of temperature deviation, thereby enhancing the adjustment range of the health index. The environmental stress index is synthesized by dimensionality reduction of multi-dimensional environmental parameters such as temperature, humidity, altitude, and air pressure through the principal component analysis method (PCA). Specifically, the system first normalizes each environmental parameter to eliminate the dimension difference; then extracts the principal components through PCA, calculates the variance contribution rate of each principal component, and determines its weight in the environmental stress index. For example, in the plateau breeding scenario, the altitude parameter may have a higher variance contribution rate, so its weight in the environmental stress index is larger. The environmental comprehensive influence coefficient is calculated by the entropy weight method, which automatically allocates weights based on the variation degree of environmental parameters, avoiding the subjectivity of artificially setting weights. For example, when a certain environmental parameter (such as ammonia concentration) fluctuates violently in the short term, the entropy weight method will increase the weight of this parameter, thus more sensitively reflecting the impact of environmental changes on health. The dynamic adjustment coefficient is not only used for the calculation of the health index but also can be extended to the formulation of environmental optimization strategies. For example, when the environmental stress index exceeds the threshold, the system can automatically trigger an environmental control plan, such as starting ventilation equipment and adjusting the opening and closing angle of sunshades. In addition, the model supports individual calibration, dynamically adjusting the sensitivity of the environmental stress index according to the characteristics of the yak, such as breed, age, and gender. For example, for old yaks, appropriately reduce the influence weight of temperature and humidity fluctuations on their health index to avoid stress reactions caused by excessive regulation; for yaks in the cub stage, increase the influence weight of light intensity changes on their health index to ensure the light stability of their growth environment. The system can also generate a report containing environmental optimization suggestions by combining the correlation analysis of historical health data and environmental parameters. For example, when the health index of a certain batch of yaks generally decreases, the system can associate their recent environmental parameters (such as ammonia concentration, temperature and humidity fluctuations) to generate optimization suggestions such as "increase ventilation frequency" or "adjust light duration".

[0074] In a preferred embodiment of the present invention, it can be further configured that the anomaly detection algorithm adopts a machine learning architecture integrating support vector machine and random forest. Specifically, a feature vector X = [heart rate, body temperature, activity level,...] is constructed for yak physical signs and behavior data, and a one-class classifier is constructed through SVM, and its decision function is where the kernel function K(·,·) adopts a radial basis kernel; at the same time, a multi-tree decision tree ensemble model is constructed using random forest, and each tree splits nodes based on the Gini coefficient; finally, through a weighted fusion strategy outputs an anomaly score. When the score exceeds the dynamic threshold θ(t) and the heart rate is higher than μ for m consecutive sampling points HR +3σ HR a warning mechanism is triggered to automatically send a notice containing the anomaly level and recommended measures to the breeding personnel, where the weights ω SVM , ω RE are dynamically optimized according to the real-time performance of the model. The SVM one-class classifier captures the non-linear boundary of physical sign data through the radial basis kernel function (RBF), effectively identifying tiny anomaly fluctuations; the random forest ensemble model uses the Gini coefficient splitting mechanism of multiple decision trees to enhance the generalization ability for complex behavior patterns; the weighted fusion strategy realizes the optimal matching of detection performance in different scenarios by dynamically adjusting the model weights; the dynamic threshold mechanism combines the number of consecutive abnormal sampling points of the heart rate to avoid false alarms caused by occasional noise interference; finally, through the individualized customization of warning notifications, decision-making support including anomaly level, recommended measures and real-time environmental parameters is provided for the breeding personnel.

[0075] In a preferred embodiment of the present invention, it can be further configured that the remote monitoring terminal includes: a user interface, remote control and warning notification;

[0076] The user interface provides an intuitive user interface to display yak physical signs, behaviors, environmental parameters and health assessment results;

[0077] The remote control supports remote control of breeding equipment through a mobile APP or a web page, including but not limited to automatic feeders, waterers, and ventilation equipment;

[0078] The warning notification automatically sends a warning notification to the breeding personnel when abnormal data is detected, reminding them to handle it in time. The intuitive user interface realizes the dynamic monitoring of individual and group yaks through multi-dimensional data visualization (such as physical sign trend charts, environmental parameter heat maps, health assessment radar charts); the remote control function realizes the precise regulation of the breeding environment through cross-platform compatibility (mobile APP and web page) and real-time feedback of device status; the warning notification mechanism significantly reduces the response time for anomaly handling through multi-level warning thresholds (such as mild, moderate, severe) and personalized push strategies (such as hierarchical notifications based on the permissions of breeding personnel).

[0079] The user interface adopts a hierarchical architecture design, displaying yak physical signs (such as heart rate, body temperature, activity level), behaviors (such as feeding frequency, rumination duration), environmental parameters (such as temperature and humidity, light intensity), and health assessment results (such as comprehensive health index, disease risk level). The specific implementation methods are as follows: Physical sign trend chart: Displays the physical sign data of individual or group yaks in the form of a line chart, supporting time dimension switching by hour, day, week, and month. For example, the heart rate data of a certain yak is sampled at a frequency of 1Hz, and the user interface can display the heart rate fluctuation curve of the past 24 hours in real time, and mark the abnormal interval with colors (such as red indicating that the heart rate exceeds the threshold). Environmental parameter heat map: Displays the distribution of environmental parameters in each area of the farm in the form of a grid map, supporting the superposition display of multiple parameters (such as temperature and humidity, ammonia concentration). For example, when the temperature in a certain area exceeds 30°C, the grid color of that area automatically turns orange and triggers the linkage of ventilation equipment. Health assessment radar chart: Displays the comprehensive health index of yaks in the form of a radar chart, covering multi-dimensional indicators such as physiology, behavior, and environment. For example, if the body temperature, activity level, and environmental ammonia concentration of a certain yak are all within the normal range, its health assessment radar chart will present a regular polygon; if a certain indicator is abnormal, the corresponding sector area will turn red.

[0080] In a preferred embodiment of the present invention, it can be further configured as follows: The user interface uses Web technology to develop a responsive web interface, adapts to different device screens, integrates a data visualization library, and displays the dynamic changes of yak physical signs, behaviors, and environmental parameters in real time. Remote control communicates with the data processing center through the RESTful API to send remote control instructions. The early warning notification is that when abnormal data is detected, the system automatically generates early warning information and sends it to the breeding personnel through SMS, email, or APP push. The responsive web interface realizes the full adaptation to different device screens (such as mobile phones, tablets, computers) through the integration of dynamic layout and the data visualization library, ensuring that the breeding personnel can obtain the real-time data of yaks anytime and anywhere; the RESTful API communication mechanism interacts with the data processing center through a standardized interface, simplifies the device control process, and reduces communication latency; the multi-channel early warning notification strategy ensures that abnormal information can be conveyed to the breeding personnel in the first time through the multi-dimensional coverage of SMS, email, and APP push, thus shortening the risk handling time.

[0081] In a preferred embodiment of the present invention, it can be further configured as follows: The instructions include: adjusting the feeding amount, controlling the switch of the water dispenser, and adjusting the rotation speed of the ventilation equipment. The early warning information includes the type of abnormal data, the occurrence time, the possible reasons, and the recommended handling measures. The refined instruction control enables the breeding personnel to dynamically adjust the environmental parameters and feeding strategies according to the real-time needs of yaks, avoiding resource waste and health risks; the structured early warning information provides a scientific decision-making basis for the breeding personnel through multi-dimensional data correlation analysis, shortening the abnormal handling time.

[0082] Dynamic adjustment of feeding amount. The "adjust feeding amount" function in the instruction is dynamically calculated based on the weight, growth stage, health status of yaks and environmental parameters (such as temperature, humidity). The specific implementation methods are as follows: Intelligent algorithm model: Use machine learning algorithms (such as random forest or neural network) to construct a feeding amount prediction model. Input the real-time data of yaks (such as weight, feeding frequency) and historical data (such as growth curve, disease record), and output the optimal feeding amount suggestion. For example, if a yak's weight has been growing slowly recently and its feeding frequency has decreased, the system will automatically increase its feeding amount and send the instruction to the automatic feeder. Phased regulation strategy: Set feeding amount regulation rules according to the growth stage of yaks (such as juvenile, adult, pregnant). For example, the feeding amount of pregnant yaks needs to be gradually increased and reduced to the basic level one week before giving birth to reduce the risk of dystocia.

[0083] This intelligent yak breeding monitoring system constructs an intelligent management system covering the whole process through the collaborative work of multiple modules. The system takes the sensor network module as the "perception layer". Through integrated physical sign sensors (such as heart rate ear tags), behavior sensors (such as video cameras) and environmental sensors (such as temperature and humidity sensors), it real-time collects the physiological parameters, behavior patterns and breeding environment data of yaks. After preprocessing such as denoising and filtering, it is transmitted to the data processing center module through LoRa wireless communication technology to form a structured data set. The intelligent analysis module is based on these data. Through a health assessment model (dynamically adjusting the health index by combining the CRITIC method and the entropy weight method), an environmental optimization algorithm (formulating a regulation strategy based on environmental data and yak needs) and an anomaly detection algorithm (realizing anomaly recognition by integrating the SVM and random forest architectures), it realizes intelligent decision-making. The remote monitoring terminal module realizes human-computer interaction and precise regulation through the user interface, remote control and warning notification functions. The system continuously optimizes the breeding management strategy through the closed-loop mechanism of "data collection - intelligent analysis - remote execution - feedback optimization", significantly improving the breeding efficiency and the health level of yaks, and providing technical support for large-scale breeding.

[0084] It should be noted that, in this document, relational terms such as first and second are only used 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 "comprising", "including" or any other variant thereof are intended to cover non-exclusive inclusion, such that a process, method, article or device comprising a series of elements not only includes those elements but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the phrase "comprising a..." does not exclude the presence of additional identical elements in the process, method, article or device comprising said element.

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

Claims

1. An intelligent yak breeding monitoring system, characterized in that, It includes: A sensor network module, a data processing center module, an intelligent analysis module, and a remote monitoring terminal module; The sensor network module integrates multiple sensors and is used to monitor yak physical signs, behaviors, and breeding environment parameters in real time; The data processing center module is responsible for receiving, processing, and storing the raw data transmitted by the sensor network; Based on the processed data, the intelligent analysis module establishes a health assessment model and an environment optimization algorithm to achieve real-time assessment of the yak's health status and intelligent optimization of the breeding environment; The remote monitoring terminal module provides a user interface and supports remote control and early warning notification functions.

2. The intelligent yak breeding monitoring system according to claim 1, characterized in that, The sensor network module includes: a physical sign sensor sub-module, a behavior sensor sub-module, and an environment sensor sub-module; The physical sign sensor sub-module is an intelligent ear tag installed on the yak's ear, which integrates a heart rate sensor, a body temperature sensor, and an acceleration sensor, and is used to monitor the yak's heart rate, body temperature, and activity status in real time; The behavior sensor sub-module is to install a video camera and an infrared thermal imager in the breeding area, and combine deep learning algorithms to identify the behavior patterns of yaks, including but not limited to eating, resting, and exercising; The environment sensor sub-module is to distribute and install temperature and humidity sensors, light sensors, and ammonia sensors in the breeding area, and is used to monitor the breeding environment parameters in real time.

3. An intelligent yak breeding monitoring system according to claim 1, characterized in that, The data processing center module includes: a data acquisition sub-module, a data transmission sub-module, and a data storage sub-module; The data acquisition sub-module is responsible for receiving the raw data from the sensor network and performing preprocessing operations such as denoising and filtering; The data transmission sub-module uses LoRa wireless communication technology to achieve low-power and long-distance data transmission between the sensor network and the data processing center; The data storage sub-module stores the preprocessed data in a database for subsequent analysis and query.

4. The intelligent yak breeding monitoring system according to claim 1, characterized in that, The intelligent analysis module includes: a health assessment model, an environment optimization algorithm, and an anomaly detection algorithm; The health assessment model is to establish a health assessment model based on yak physical sign parameters and behavior patterns, and to evaluate the health status of yaks in real time; The environment optimization algorithm is to formulate a breeding environment optimization strategy according to the environment sensor data and in combination with the physiological needs of yaks, including but not limited to automatically adjusting temperature and humidity, and light; The anomaly detection algorithm is to use machine learning algorithms to detect anomalies in yak physical signs and behavior data, and to timely discover and give early warnings of potential health problems.

5. The intelligent yak breeding monitoring system according to claim 4, characterized in that, The health assessment model adopts the following dynamic adaptive algorithm formula: Health Index w i is the basic weight coefficient, calculated by the CRITIC method, comprehensively considering the comparison intensity and conflict of indicators; is a standardized processing item, p i is a real-time physiological parameter value, μ i is a historical mean, σ i is the standard deviation; σ i is a dynamic adjustment coefficient, and an adaptive weighted algorithm is adopted: where δ ij is the environmental factor influence coefficient, and γ j is the real-time monitoring value of the environmental factor; β is the comprehensive environmental impact coefficient, which is calculated by the entropy weight method; E is the environmental stress index, which is synthesized by reducing the dimension of multi-dimensional environmental parameters such as temperature and humidity, altitude, and air pressure through the principal component analysis method.

6. The intelligent yak breeding monitoring system according to claim 4, characterized in that, The abnormal detection algorithm adopts a machine learning architecture that fuses support vector machines and random forests. Specifically, a feature vector X = [heart rate, body temperature, activity level,...] is constructed for yak physical signs and behavior data, and a one-class classifier is constructed through SVM, and its decision function is where the kernel function K(·,·) adopts a radial basis kernel; at the same time, a multi-tree decision tree ensemble model is constructed using random forests, and each tree splits nodes based on the Gini coefficient; finally, through a weighted fusion strategy an abnormal score is output. When the score exceeds the dynamic threshold θ(t) and the heart rate is higher than μ for m consecutive sampling points HR +3σ HR a warning mechanism is triggered, and a notification containing the abnormal level and recommended measures is automatically sent to the breeding personnel, where the weights ω SVM , ω RE are dynamically optimized according to the real-time performance of the model.

7. The intelligent yak breeding monitoring system according to claim 1, characterized in that, The remote monitoring terminal includes: a user interface, remote control, and early warning notification; The user interface provides an intuitive user interface to display yak physical signs, behaviors, environmental parameters, and health assessment results; The remote control supports remote control of breeding equipment through a mobile phone APP or a web page, including but not limited to automatic feeders, waterers, and ventilation equipment; The early warning notification automatically sends an early warning notification to the breeding personnel when abnormal data is detected, reminding them to handle it in time.

8. An intelligent yak breeding monitoring system according to claim 7, characterized in that, The user interface uses web technology to develop a responsive web interface that adapts to different device screens, integrates a data visualization library, and displays the dynamic changes of yak physical signs, behaviors, and environmental parameters in real time. The remote control communicates with the data processing center through the RESTful API to send remote control instructions. The warning notification is that when abnormal data is detected, the system automatically generates a warning message and sends it to the breeding personnel via SMS, email, or APP push.

9. The intelligent yak breeding monitoring system according to claim 8, wherein, The instructions include: adjusting the feeding amount, controlling the water dispenser switch, and adjusting the ventilation equipment speed. The warning message includes the abnormal data type, occurrence time, possible cause, and recommended handling measures.

Images