Artificial assistance and cattle mating method

Abstract

The invention relates to an artificial assistance and cattle mating method. The artificial assistance and cattle mating method comprises the steps that genital tract microenvironment parameters of to-be-mated cattle are monitored in real time, and the microenvironment parameters at least comprise the pH value, the temperature, the oxygen content and the metabolite concentration; collecting multi-dimensional data of the physiological state of the cattle to be hybridized, and preprocessing the multi-dimensional data and the microenvironment parameters through an edge calculation unit; analyzing the embryonic development state based on metabonomics and evaluating the quality; a micro-fluidic technology is combined to realize accurate regulation and control of a microenvironment; a hierarchical time memory network model and a Bayesian network are applied to construct an early pregnancy state prediction model, and high-precision pregnancy result prediction within 7-10 days after hybridization is achieved; according to the artificial assistance and cattle mating method, through multi-dimensional monitoring and accurate regulation and control, the pregnancy success rate is remarkably increased, and the breeding management cost is reduced.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of livestock breeding, and particularly relates to a method for artificially assisting and breeding cattle. BACKGROUND

[0002] In the livestock industry, reproductive efficiency is one of the key factors determining the economic benefits of breeding. Artificial insemination technology, as an important means of modern breeding, is widely used in the breeding process of cattle. However, the current artificial insemination technology still faces many challenges, especially in the early embryo development quality monitoring and pregnancy status prediction.

[0003] In traditional cattle breeding management, breeders often rely on external environmental control (such as adjusting the temperature and humidity of the shed) to affect the breeding process, lacking the ability to accurately monitor and regulate the metabolic level of the embryo development microenvironment. Existing research has shown that the pH value, temperature, oxygen content, and glucose concentration of the reproductive tract microenvironment are closely related to the quality of early embryo development, but existing monitoring methods mainly rely on indirect measurement methods, which cannot capture the dynamic changes of these key parameters in real time, resulting in poor embryo development environment and affecting the pregnancy rate.

[0004] In addition, traditional pregnancy status confirmation mainly relies on the appearance of obvious signs, such as observing whether the cow returns to estrus within 21 days after insemination, or performing a B-ultrasound examination 35-60 days after insemination. This late confirmation method leads to delayed management decisions and resource waste, and cannot achieve early intervention and precise management. Existing technology also lacks comprehensive analysis of the changes in multiple physiological indicators of the cow, making it difficult to accurately predict pregnancy outcomes at an early stage.

[0005] Although related research has attempted to predict pregnancy status early by detecting blood hormone levels and analyzing behavior patterns, these methods often focus on only a single or a few indicators, with limited accuracy and prediction timeliness, making it difficult to meet the needs of modern precision breeding for early and accurate prediction.

[0006] Therefore, there is an urgent need for a method and system that can accurately monitor and regulate the reproductive tract microenvironment, comprehensively analyze multiple physiological indicators, and accurately predict early pregnancy status, in order to improve the success rate of insemination, reduce breeding costs, and improve breeding economic benefits. SUMMARY

[0007] The purpose of the present application is to provide a method for artificially assisting and breeding cattle with simple structure and reasonable design, which solves the technical problems of inaccurate monitoring and regulation of embryo development microenvironment and delayed prediction of pregnancy status in traditional technology.

[0008] The present application achieves the above-mentioned purposes through the following technical solutions: A method for artificial assisted and cattle breeding, comprising the following steps: Deploying a micro multi-parameter sensor array in the reproductive tract of a cow to monitor microenvironment parameters including pH value, temperature, oxygen content and trace metabolite concentration in real time; Integrating wearable physiological sensors, behavior monitoring systems and diet monitoring devices to collect multi-dimensional data of the physiological state of the cow, and pre-processing through an edge computing unit; Applying metabolomics database and systems biology methods to analyze the activity of key metabolic pathways of embryo development, construct an embryo metabolic state evaluation model, and evaluate the quality of embryo development; According to the metabolic evaluation results, combining microfluidic technology to realize accurate regulation of the microenvironment parameters of the reproductive tract, and providing individualized nutritional intervention programs; Applying hierarchical temporal memory network model to analyze the change pattern of physiological indicators, combining Bayesian network for multi-source information fusion, constructing an early pregnancy state prediction model, and generating a pregnancy state evaluation report within 7-10 days after breeding.

[0009] As a further optimization scheme of the present application, the micro multi-parameter sensor array includes a pH sensor, a temperature sensor, a dissolved oxygen sensor and a metabolite detection sensor for detecting metabolite concentrations including lactic acid and glucose.

[0010] As a further optimization scheme of the present application, it further includes the steps of noise filtering and data standardization processing of microenvironment parameter data, the noise filtering adopts the method of combining wavelet transform and median filtering, and the data standardization processing adopts Z-score method.

[0011] As a further optimization scheme of the present application, the wearable physiological sensor system includes a neck activity monitor, a body surface temperature sensor and a heart rate monitor, the behavior monitoring system includes a video monitoring device and a location tracking system, and the diet monitoring device includes an automatic feeding station and an intelligent waterer.

[0012] As a further optimization scheme of the present application, the step of constructing an embryo metabolic state evaluation model includes: establishing an embryo development metabolomics knowledge base; using a non-targeted metabolomics analysis method to process metabolite data; using a pathway enrichment analysis method to calculate the activity level of key metabolic pathways; using a random forest algorithm to construct an embryo development quality evaluation model; and calculating the deviation degree of the current metabolic state from the ideal development trajectory.

[0013] As a further optimization scheme of the present application, the step of realizing accurate regulation of the microenvironment parameters of the reproductive tract in combination with microfluidic technology comprises: configuring a microfluidic drug delivery system; determining the target value of the microenvironment parameters to be regulated; calculating the optimal intervention scheme using a model predictive control algorithm; executing the microenvironment parameter regulation measures; and generating a personalized nutritional intervention scheme.

[0014] As a further optimization scheme of the present application, the microenvironment parameter regulation measures include micro-drug release, buffer injection and nutrient supplementation, and the parameter range that can be regulated by the system includes: pH value 6.5-8.0, oxygen content 3-8%, glucose concentration 2-6mmol / L.

[0015] As a further optimization scheme of the present application, in the step of applying a hierarchical time memory network model to analyze the change pattern of the physiological indicators, the training of the hierarchical time memory network model adopts an end-to-end supervised learning method, and the loss function includes a reconstruction loss, a prediction loss and a regularization term.

[0016] As a further optimization scheme of the present application, in the step of constructing an early pregnancy state prediction model, the nodes of the Bayesian network include microenvironment parameter nodes, metabolic state nodes, physiological indicator nodes and hidden nodes, the network structure is determined by a structure learning algorithm, and the conditional probability table is estimated by a parameter learning algorithm.

[0017] The present application also discloses an artificial assisted and cattle breeding system, comprising: A microenvironment monitoring module is configured to deploy a micro multi-parameter sensor array to monitor the microenvironment parameters of the reproductive tract in real time. A physiological monitoring module is configured to integrate a wearable physiological sensor, a behavior monitoring system and a diet monitoring device to collect multi-dimensional data of the physiological state of the cow. A metabolic analysis module is configured to analyze the activity of key metabolic pathways of embryo development and evaluate the quality of early embryo development. A microenvironment regulation module is configured to realize accurate regulation of the local microenvironment parameters of the reproductive tract according to the metabolic evaluation results. A pregnancy prediction module is configured to apply a hierarchical time memory network model to analyze the change pattern of the physiological indicators and predict the early pregnancy state.

[0018] The present application has the following advantages: The present application realizes accurate regulation by monitoring the microenvironment parameters of the reproductive tract in real time through a micro multi-parameter sensor array and combining microfluidic technology, significantly improves the early development environment of embryos, and improves the implantation rate of fertilized eggs.

[0019] The present application can give a pregnancy state prediction within 7-10 days after mating by hierarchical time memory network and multi-source data fusion analysis of Bayesian network, 15-20 days earlier than traditional methods, providing a scientific basis for breeding management decision.

[0020] Based on real-time monitoring of microenvironment parameters and metabolic state data, the present application can generate individualized environment regulation and nutrition intervention programs for each cow, providing customized management strategies for cows of different breeds, different lactation stages and different body conditions, making management more refined and intelligent.

[0021] By accurately predicting the pregnancy state early, the breeding farm can adjust the management strategy for non-pregnant cows early, reducing unnecessary feeding investment and economic losses caused by prolonged empty time.

[0022] The present application innovatively integrates micro-sensing technology, microfluidic technology, metabolomics analysis and deep learning, etc. Frontier technologies, realizing the overall monitoring and evaluation from the metabolic level of embryos to the overall physiological state, breaking through the technical limitations of traditional single sign observation. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 is the overall method flowchart of the artificial assisted and cow mating method of the present application. DETAILED DESCRIPTION

[0024] The following detailed description of the present application will be described in conjunction with the accompanying drawings, it is necessary to point out here that the following detailed description is only used to further illustrate the present application, and cannot be understood as limiting the scope of protection of the present application, and those skilled in the art can make some non-essential improvements and adjustments to the present application according to the above application content.

[0025] Before describing the present application in detail, in order to help understanding the technical scheme of the present application, the following first explains the terms used in the present application: Microenvironment parameters: refer to the local environmental conditions in the reproductive tract that affect embryo development, including but not limited to pH, temperature, oxygen content, carbon dioxide concentration, glucose concentration, etc. Physical and chemical parameters.

[0026] Metabolomics: the science of studying all metabolites in a specific biological system, revealing the overall state of metabolic activity in the body by analyzing the types, concentrations and change patterns of metabolites.

[0027] Systems biology method: a method of integrating multi-level biological information and studying the overall function and characteristics of biological systems, emphasizing the analysis of interactions between components in the body.

[0028] Microfluidics: A technology that controls the flow of microfluids using microscale channels, enabling precise control over microenvironment parameters.

[0029] Hierarchical Temporal Memory Network (HTMN): An improved neural network model capable of handling data with temporal continuity and hierarchical structure, particularly suitable for analyzing long-term physiological indicator changes.

[0030] Bayesian Network: A probabilistic graph model that represents conditional dependencies between random variables, effectively handling uncertainty and incomplete information.

[0031] In existing artificial assisted and cattle breeding techniques, there are the following key technical problems: Firstly, the traditional environmental control method lacks the ability to accurately monitor and regulate the metabolic level of the embryo development microenvironment. Existing technologies mainly rely on the control of macro-environmental parameters such as shed temperature, humidity, etc., and cannot directly monitor and regulate the metabolic state of the embryo development microenvironment in the cow's reproductive tract, making it difficult to ensure the quality of early embryo development.

[0032] Secondly, existing monitoring methods cannot capture real-time changes in reproductive tract microenvironment parameters. Traditional monitoring techniques such as ultrasound examination and blood hormone level detection lack real-time and continuous nature, and cannot timely reflect the dynamic changes of microenvironment parameters, leading to the inability to timely intervene and optimize the embryo development environment, ultimately affecting the conception rate.

[0033] Thirdly, the confirmation of pregnancy status after breeding is heavily dependent on the appearance of obvious signs. In existing technologies, the confirmation of pregnancy status usually requires observation of estrus behavior after 21 days or B-ultrasound examination after 35-60 days, resulting in delayed breeding management decisions and resource waste due to late confirmation.

[0034] Fourthly, existing technologies lack comprehensive analysis capabilities for multi-dimensional physiological indicator changes in cows. Traditional methods often focus on a single or a few physiological indicators, failing to establish multi-dimensional and multi-scale data analysis models, which cannot accurately predict pregnancy outcomes in the early stage, limiting the realization of precision breeding management.

[0035] The technical solution of the present application is mainly applied to the artificial assisted breeding process in modern large-scale dairy and beef cattle farms. In these scenarios, the farm usually has perfect infrastructure and certain technical conditions, including basic network communication equipment, data processing equipment, and a professional veterinary team. Specific application scenarios are as follows: Dairy farm: The technical solution can be applied to the breeding management of various scale dairy farms, especially for high-yield dairy cows with decreased reproductive performance due to high metabolic load. By accurately monitoring and regulating the reproductive tract microenvironment, the breeding success rate can be improved.

[0036] Beef cattle breeding base: For the directional breeding of high-value beef cattle breeds, this technical solution can optimize reproductive efficiency, reduce reproductive interval, and improve economic efficiency through early pregnancy state prediction.

[0037] Embryo transfer center: In the application process of embryo transfer technology, this technical solution can be used to optimize the reproductive tract microenvironment of recipient cows, improve the implantation rate and development quality of transferred embryos.

[0038] Precision breeding management system: As an important part of the precision breeding management system, this technical solution integrates with existing feeding management, health monitoring, and other systems to realize data-driven decision-making throughout the breeding cycle.

[0039] In these application scenarios, this technical solution will effectively solve key technical challenges such as early embryonic development microenvironment monitoring, regulation, and early pregnancy state prediction, significantly improving the success rate of artificial assisted breeding and breeding efficiency.

[0040] Reference Figure 1 According to one embodiment of the present application, a method for artificial assisted and cattle breeding is provided, which comprises the following steps: Step S01: Deploy a micro multi-parameter sensor array to monitor the microenvironment parameters in real time; In this step, a specially designed micro multi-parameter sensor array is deployed in the cow's reproductive tract to monitor the microenvironment parameters in real time. The specific implementation is as follows: Step S01-1: Use biocompatible materials to prepare a micro multi-parameter sensor array. The sensor array is wrapped with medical-grade silicone, and internally integrated with pH sensors, temperature sensors, dissolved oxygen sensors, and specific metabolite detection sensors (such as lactic acid, glucose, etc.). The sensor diameter is not more than 5mm, and the length is not more than 20mm, ensuring non-invasive to the cow's reproductive tract.

[0041] Step S01-2: The micro multi-parameter sensor array is fixed near the cervix of the cow through the vagina, and the monitoring data is transmitted in real time through a wireless communication module. The fixing method uses a biocompatible bracket with elasticity to prevent the sensor from falling off or shifting, while not affecting the normal physiological activities of the cow.

[0042] Step S01-3: The sensor array collects microenvironment parameter data at intervals of 5-10 minutes, including but not limited to: pH value (accuracy ±0.05), temperature (accuracy ±0.1℃), dissolved oxygen content (accuracy ±0.1mg / L), lactic acid concentration (accuracy ±0.2mmol / L), and glucose concentration (accuracy ±0.2mmol / L). The raw data collected is transmitted to a data relay device worn on the surface of the cow through a low-power Bluetooth protocol, and then transmitted to a central data processing system through a wireless network.

[0043] Step S01-4: Perform noise filtering and data standardization on the collected microenvironment parameter data. Noise filtering uses a combination of wavelet transform and median filtering to effectively remove abnormal data caused by cow activity or sensor vibration. Data standardization uses the Z-score method to enable unified analysis and processing of parameter data with different dimensions.

[0044] Step S01-5: Based on time series analysis, calculate the changing trends and fluctuation characteristics of each microenvironment parameter. Specifically, the exponentially weighted moving average method is used to calculate the parameter changing trend, and the moving standard deviation method is used to calculate the degree of parameter fluctuation, forming a parameter change feature vector to provide basic data for subsequent analysis. The mathematical expression is as follows: For any parameter Its exponentially weighted moving average The calculation is as follows:

[0045] in, for Time parameters The measured value, This is the smoothing coefficient (generally taken as 0.1-0.3).

[0046] Parameter fluctuation degree The calculation is as follows: ; in, for arrive Time parameters The average value, This represents the size of the time window.

[0047] Step S02: Integrate multi-source physiological data acquisition systems to construct a monitoring network for the physiological state of cows; This step integrates various physiological monitoring devices to construct a comprehensive monitoring network for the physiological state of cows, collecting multi-dimensional physiological indicator data. The specific implementation method is as follows: Step S02-1: Configure a wearable physiological sensor system, including a neck movement monitor, a body surface temperature sensor, and a heart rate monitor. The neck movement monitor uses a triaxial accelerometer to detect the intensity and frequency of the cow's activity; the body surface temperature sensor is attached behind the cow's ear to monitor changes in body surface temperature; the heart rate monitor uses a non-invasive electrocardiogram sensor to detect heart rate and its variability.

[0048] Step S02-2: Install the behavior monitoring system, including video monitoring equipment and location tracking system. Video monitoring equipment is deployed in the cow activity area to collect behavior data such as feeding, drinking, and lying. The location tracking system based on UWB (Ultra-Wideband) technology records the location information and movement trajectory of the cow in the farm in real time, with an accuracy of ±0.5m.

[0049] Step S02-3: Deploy intelligent diet monitoring equipment, including automatic feeding station and intelligent waterer. The automatic feeding station records the amount of each feeding, feeding duration and frequency; the intelligent waterer records the amount of drinking and drinking frequency.

[0050] Step S02-4: Use edge computing unit to preprocess physiological data, and perform preliminary processing and feature extraction on the original monitoring data. The edge computing unit uses a low-power embedded processor to directly process the data collected by the sensor and extract key feature parameters such as activity intensity integral value, diurnal fluctuation amplitude of body temperature, heart rate variability index, etc., reducing data transmission volume and central server computing burden.

[0051] Step S02-5: Build a data fusion model to integrate multi-source heterogeneous data collected by each subsystem. The data fusion uses a multi-level fusion architecture, first synchronizing time and aligning space at the feature level, then integrating multi-source information at the decision level to form a complete cow physiological state feature vector. The fusion model uses a dynamic Bayesian network, which can be expressed as follows: Given multi-source data , where represents the observation value of the th data source, and the posterior probability of the cow physiological state feature vector is calculated as follows:

[0052] Considering the conditional independence between different data sources, the calculation is simplified as follows:

[0053] where is the prior probability, is the likelihood function, which is obtained by training historical data.

[0054] Step S03: Analyze the embryo development state based on metabolomics and construct an early embryo metabolic state evaluation model; This step uses metabolomics and systems biology methods to analyze the activity of key metabolic pathways in embryo development and construct an early embryo metabolic state evaluation model. The specific implementation is as follows: Step S03-1: Establishing an embryo development metabolomics knowledge base. Based on existing research data and literature, a knowledge base containing metabolic characteristics and key metabolite markers at each stage of early embryonic development is constructed. The knowledge base includes the metabolic changes, metabolic pathway activity change patterns, and key regulatory node information of the embryo from fertilization to implantation.

[0055] Step S03-2: Using non-targeted metabolomics analysis methods, process the metabolite data collected in step S01. Based on the analysis principles of liquid chromatography-mass spectrometry (LC-MS) and gas chromatography-mass spectrometry (GC-MS) technology, the metabolite data detected by the sensor is pattern recognized and classified to identify key metabolites and their concentration change patterns.

[0056] Step S03-3: Constructing a metabolic pathway activity evaluation model to analyze the activity level of key metabolic pathways. Using pathway enrichment analysis (Pathway Enrichment Analysis) method, based on the detected metabolite change pattern, the activity level of glycolysis pathway, tricarboxylic acid cycle, amino acid metabolism and other key metabolic pathways is calculated. The formula for calculating the activity of the pathway is as follows:

[0057] Wherein, is the activity score of the th metabolic pathway, is the number of detected metabolites contained in the pathway, is the weight of the th metabolite in the pathway (based on topological analysis), is the current concentration of the th metabolite, is the reference concentration of the metabolite.

[0058] Step S03-4: Applying machine learning algorithms to construct an embryo development quality evaluation model based on metabolic pathway activity characteristics. Random forest algorithm is used to integrate the prediction results of multiple decision trees, with the input being the metabolic pathway activity score vector and the output being the embryo development quality score (0-100 points). Model training uses historical data sets containing metabolic characteristic data and final pregnancy results corresponding to different embryo development quality levels.

[0059] Step S03-5: Real-time calculation of the deviation degree of the current metabolic state from the ideal development trajectory to generate a metabolic state evaluation report. The deviation degree calculation uses the dynamic time warping (Dynamic Time Warping, DTW) algorithm to calculate the distance between the current metabolic trajectory and the ideal development trajectory:

[0060] Wherein, for the current metabolic state time series, for the ideal metabolic trajectory, for the single-point distance metric.

[0061] The evaluation report includes the overall development quality score, the activity level of each key metabolic pathway, abnormal metabolic indicator alerts, and regulation suggestions.

[0062] Step S04: Precise regulation of the reproductive tract microenvironment using microfluidic technology to generate personalized nutritional intervention programs; This step is based on the evaluation results described above, and uses microfluidic technology to precisely regulate the local microenvironment parameters of the reproductive tract to optimize the early embryonic development environment. The specific implementation is as follows: Step S04-1: Construct a microfluidic drug delivery system, including a micro-drug storage unit, a precision pumping device, and a controllable release valve. The microfluidic system is prepared using MEMS technology and integrated into the sensor array in step S01, forming a closed-loop control system that integrates "monitoring-analysis-intervention" as one.

[0063] Step S04-2: Based on the metabolic state evaluation results in step S03, determine the target value of the microenvironment parameters that need to be regulated. The system calculates the optimal parameter control target based on the deviation of the ideal embryonic development trajectory and the current metabolic state, including pH adjustment range, oxygen content optimization level, glucose supply concentration, etc.

[0064] Step S04-3: Use the model predictive control (MPC) algorithm to calculate the optimal intervention program. The MPC controller predicts the parameter change trend in the future based on the system model, and obtains the optimal control sequence by solving the optimization problem:

[0065] ; where, is the predicted system output (microenvironment parameter) at time , is the reference trajectory (target value), is the control input (intervention measure), and is the weight matrix, is the system model, , , , , are control and state constraints.

[0066] Step S04-4: Perform microenvironment parameter regulation measures, including micro-drug release, buffer injection, nutrient supplementation, etc. The microfluidic system accurately controls the working parameters of each execution unit according to the optimal control sequence calculated, to realize fine regulation of the reproductive tract microenvironment. The system can regulate the following parameters: pH value 6.5-8.0 (accuracy ±0.1), oxygen content 3-8% (accuracy ±0.2%), glucose concentration 2-6 mmol / L (accuracy ±0.2 mmol / L).

[0067] Step S04-5: Generate individualized nutritional intervention plan, including daily ration adjustment suggestions and nutritional additive formula. Based on the metabolic state evaluation and microenvironment monitoring results, the system analyzes the overall metabolic demand of the cow and the embryo development demand to generate an individualized nutritional intervention plan. The plan content includes: energy and protein balance ratio, key trace element supplementation plan, functional additive recommendation and dosage. The nutritional plan is customized for cows of different breeds, different lactation stages and different body condition scores.

[0068] Step S05: Apply hierarchical temporal memory network model to construct early pregnancy state prediction system; This step applies hierarchical temporal memory network model and Bayesian network for multi-source information fusion based on the multi-source data collected in the previous steps to construct an early pregnancy state prediction model. The specific implementation is as follows: Step S05-1: Construct a hierarchical temporal memory network (HTMN) model for analyzing physiological index time series data. The HTMN model contains three layers of hierarchical structure: the input layer receives the original physiological index time series data; the hidden layer contains multiple LSTM (Long Short-Term Memory) units that capture variation patterns at different time scales; the output layer generates a feature vector representing the intrinsic pattern of the time series data. The HTMN model training uses an end-to-end supervised learning method, and the loss function is defined as:

[0069] wherein, is the reconstruction loss, which measures the reconstruction accuracy of the model for the input sequence; is the prediction loss, which measures the prediction accuracy of the model for the future sequence; is the regularization term to prevent overfitting; , , is the weight coefficient.

[0070] Step S05-2: Apply the HTMN model to analyze the physiological index variation pattern and extract features from multiple aspects of microenvironment parameters, metabolic state and overall physiological state. For each type of physiological index ( HTMN model outputs a feature vector characterizing the change pattern of the indicator.

[0071] Step S05-3: Construct a Bayesian network model to realize multi-source information fusion. The nodes of the Bayesian network include: microenvironment parameter nodes (pH value, temperature, oxygen content, etc.), metabolic state nodes (key metabolic pathway activity), physiological indicator nodes (body temperature, heart rate, activity, etc.), and hidden nodes (pregnancy status). The network structure is determined by a structure learning algorithm, and the conditional probability table is estimated by a parameter learning algorithm.

[0072] Step S05-4: Based on the Bayesian network model, calculate the posterior probability of the pregnancy status after mating. Given the observation evidence (including all monitored microenvironment parameters, metabolic state indicators, and physiological indicators), calculate the posterior probability of the pregnancy status :

[0073] where the joint probability is calculated according to the factorization of the Bayesian network:

[0074] where is the i-th node in the Bayesian network, is the parent node set of .

[0075] Step S05-5: Generate an early pregnancy status evaluation report, including the pregnancy probability, confidence interval, and key influencing factor analysis. The evaluation report is generated within 7-10 days after mating, providing a prediction result of the pregnancy status 15-20 days earlier than traditional methods, and providing a scientific basis for breeding management decisions. The report also contains a time series trend chart of the pregnancy status evaluation, showing the change pattern of the pregnancy probability over time, and the prediction of future development trends.

[0076] Device embodiment For the above method, the present application also provides a device for artificial assistance and cattle mating, which comprises the following modules: Microenvironment monitoring module: including a micro multi-parameter sensor array and a data acquisition unit, for real-time monitoring of reproductive tract microenvironment parameters (pH value, temperature, oxygen content, etc.) and trace metabolite changes, and collecting embryo development environment data.

[0077] ​​Physiological monitoring module: including wearable physiological sensor system, behavior monitoring system and intelligent diet monitoring device, used for collecting multi-dimensional data of the physiological state of the cow, and constructing a comprehensive physiological index monitoring network.

[0078] Metabolic analysis module: including metabolomics data processing unit and metabolic state evaluation model, based on metabolomics and systems biology methods to analyze the activity of key metabolic pathways of embryo development, and evaluate the quality of early embryo development.

[0079] Microenvironment regulation module: including microfluidic drug delivery system and environment regulation execution unit, according to the metabolic evaluation results to realize the accurate regulation of the local microenvironment parameters of the reproductive tract, and optimize the early development environment of the embryo.

[0080] Pregnancy prediction module: including hierarchical time memory network model and Bayesian information fusion unit, analyzing the change pattern of physiological indicators, realizing multi-source information fusion, and constructing an early pregnancy state prediction model.

[0081] Central processing module: including edge computing unit and cloud server, responsible for data processing, model running and result generation, and coordinating the work of each functional module.

[0082] User interaction module: including mobile application and data visualization interface, providing intuitive data display and decision support tools for farm managers.

[0083] The above modules work together to realize the complete function of the artificial assisted and cattle breeding method proposed in the application. Among them, the microenvironment monitoring module, the physiological monitoring module and the microenvironment regulation module are deployed on the cow side, and communicate with the central processing module through a wireless network; the metabolic analysis module, the pregnancy prediction module and the user interaction module are deployed on the management side, and receive and process the data transmitted from the cow side through the central processing module, to generate evaluation results and intervention schemes.

[0084] Computer device embodiment The method of the application can also be realized by a computer device, which includes: Processor: responsible for performing various computing tasks, including data processing, model training and inference, etc.

[0085] Memory: including ROM and RAM, used for storing operating system, application program and various data. The memory stores a computer program for executing the above method steps.

[0086] Communication interface: including wired and wireless communication interface, used for data exchange with various sensor devices, execution units and user terminals.

[0087] Data processing module: including signal processing unit, feature extraction unit and pattern recognition unit, responsible for pre-processing and feature extraction of sensor data.

[0088] Model running module: including machine learning model library and inference engine, responsible for running metabolic state evaluation model, microenvironment regulation algorithm and conception state prediction model.

[0089] When the computer device is running, the processor executes the computer program stored in the memory to realize each step of the artificial assistance and cattle breeding method described above. The monitoring data from the sensor device is received through the communication interface, processed by the data processing module, input into the model running module for analysis and prediction, and finally the evaluation results and intervention scheme are generated and sent to the execution unit and user terminal through the communication interface.

[0090] The computer device can be a standalone server device, or multiple cooperative computing nodes in a distributed system, or a hybrid architecture system deployed in edge computing devices and cloud servers. Regardless of the form, the computer device can implement the artificial assistance and cattle breeding method proposed in the present application, significantly improving the breeding success rate and economic benefits.

Claims

1. A method of artificial assisted and cattle breeding, characterized in that, The method comprises, S1, real-time monitoring of the microenvironment parameters of the reproductive tract of the cow to be bred, the microenvironment parameters at least including pH value, temperature, oxygen content and metabolite concentration; S2, collecting multi-dimensional data of the physiological state of the cow to be bred, and preprocessing the multi-dimensional data and the microenvironment parameters through an edge computing unit; S3, based on a pre-constructed metabolomics database and a systems biology method, analyzing the activity of key metabolic pathways of embryo development, constructing an embryo metabolic state evaluation model, and evaluating the embryo development quality using the model; S4, according to the embryo metabolic state evaluation result obtained in step S3, combining microfluidic technology to accurately regulate the microenvironment parameters of the reproductive tract, and providing an individualized nutritional intervention scheme; S5, applying a hierarchical time memory network model to analyze the change pattern of the preprocessed physiological indicators, and using a Bayesian network to perform multi-source information fusion on the physiological indicator change pattern, the microenvironment parameters and the embryo metabolic state evaluation result, to construct an early pregnancy state prediction model, and generate a pregnancy state evaluation report within 7-10 days after breeding.

2. A method of artificial insemination of cattle as claimed in claim 1 wherein: A preset micro multi-parameter sensor array is used to collect and monitor the microenvironment parameters of the reproductive tract of the cow to be bred; the micro multi-parameter sensor array includes a pH value sensor, a temperature sensor, a dissolved oxygen sensor and a metabolite detection sensor for detecting lactic acid and glucose concentration.

3. A method of artificial insemination of cattle as claimed in claim 1 wherein: The preprocessing in step (S2) includes noise filtering and data standardization processing of the microenvironment parameter data and the physiological state multi-dimensional data; wherein the noise filtering adopts a method combining wavelet transform and median filtering, and the data standardization processing adopts a Z-score method.

4. A method of artificial insemination of cattle as claimed in claim 1 wherein: The multi-dimensional data of the physiological state includes physiological data, behavior data and diet data, which are collected based on a wearable physiological sensor, a behavior monitoring system and a diet monitoring device; wherein the wearable physiological sensor includes a neck activity monitor, a body surface temperature sensor and a heart rate monitor; the behavior monitoring system includes a video monitoring device and a location tracking system; and the diet monitoring device includes an automatic feeding station and an intelligent waterer.

5. A method of artificial insemination of cattle as claimed in claim 1, wherein: The construction of the embryo metabolic state evaluation model in step S3 specifically includes: S3a, establishing an embryo development metabolomics knowledge base; S3b, processing the metabolite data preprocessed in step S2 by using a non-targeted metabolomics analysis method; S3c, calculating the activity level of key metabolic pathways by using a pathway enrichment analysis method; S3d, constructing an embryo development quality evaluation model based on the key metabolic pathway activity level by using a random forest algorithm; S3e, calculating the deviation degree of the current metabolic state from the ideal development trajectory to evaluate the embryo development quality by using the embryo development quality evaluation model.

6. A method of artificial insemination of cattle as claimed in claim 1 wherein: The accurate regulation of the microenvironment parameters of the reproductive tract in step S4 in combination with the microfluidic technology specifically includes: S4a, configuring a microfluidic drug delivery system; S4b, determining the target value of the microenvironment parameters to be regulated according to the embryo metabolic state evaluation result; S4c, calculating the optimal intervention scheme by using a model predictive control algorithm; S4d, executing the microenvironment parameter regulation measures based on the optimal intervention scheme; S4e generating a personalized nutritional intervention plan based on the regulatory measures.

7. A method of artificial insemination of cattle as claimed in claim 6 wherein: The microenvironment parameter regulatory measures in step S4d include one or more of micro-drug release, buffer injection, and nutrient supplementation. The adjustable range of the microenvironment parameters includes: pH value 6.5-8.0, oxygen content 3-8%, glucose concentration 2-6mmol / L.

8. A method of artificial insemination of cattle as claimed in claim 1 wherein: The hierarchical temporal memory network model applied in step S5 contains an input layer, a hidden layer, and an output layer, wherein the hidden layer contains a plurality of long short-term memory (LSTM) units for capturing physiological indicator change patterns of different time scales.

9. A method of artificial insemination of cattle as claimed in claim 8 wherein: The hierarchical temporal memory network model is trained in an end-to-end supervised learning manner, and its loss function includes a reconstruction loss, a prediction loss, and a regularization term.

10. A method of artificial insemination and breeding of cattle as claimed in claim 1, 8 or 9, wherein: The Bayesian network in step S5 contains microenvironment parameter nodes, metabolic state nodes, physiological indicator nodes, and hidden nodes; the structure of the Bayesian network is determined by a structure learning algorithm, and the conditional probability table is estimated by a parameter learning algorithm.

Images