Digital intelligence breeding sheep genetic evaluation and breeding management system

Abstract

The invention provides a digital intelligence breeding sheep genetic evaluation and cultivation management system, which collects individual physiological, behavior and environmental data in real time through RFID, an intelligent necklace and automatic weighing equipment, uploads the data to a cloud after edge preprocessing, performs genomic breeding value GEBV evaluation by adopting an SSGBLUP model in combination with pedigree, phenotype and 40K SNP genotype data, and provides a new breeding sheep genetic evaluation and cultivation management system. And an XGBoost algorithm is fused to improve the complexity prediction precision. The cloud intelligent module realizes oestrus early warning, health monitoring, accurate feeding and breeding optimization, analyzes a multi-target breeding decision through an NSGA-II algorithm, and ensures that key data cannot be tampered and the whole process is traceable in combination with a block chain technology. According to the system, the seed selection accuracy, the breeding efficiency and the breeding intelligence level are remarkably improved, and the system is suitable for digital transformation and genetic improvement of a large-scale sheep farm.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of livestock breeding, in particular to a digital and intelligent genetic evaluation and breeding management system for breeding sheep. BACKGROUND

[0002] With the transformation of China's livestock industry towards intensification, intelligence and high-quality development, breeding sheep as the core genetic resource of the livestock industry, its genetic improvement efficiency is directly related to the production level and international competitiveness of the entire sheep industry. In recent years, genomic selection, Internet of Things, and big data analysis technology have been gradually applied to breeding sheep breeding and breeding management, promoting the digital upgrading of traditional breeding modes. However, the existing technology still has many deficiencies and systematic defects in practical application, which is difficult to meet the comprehensive needs of modern breeding for accurate evaluation, intelligent management, data closed loop and security traceability.

[0003] At present, the existing system relies on manual regular sampling, with low data frequency and easy omission, and cannot capture the dynamic changes of the behavior and physiology of breeding sheep, Most breeding sheep farms still use traditional BLUP models based on pedigree information for genetic evaluation, without integrating genomic data, resulting in poor prediction ability for low heritability traits such as fertility and disease resistance. Even if individual units introduce genetic testing, there is information loss, and the update cycle is long, which cannot achieve dynamic optimization.

[0004] The existing breeding management system often has single function, only used for file management or estrus reminder, and each subsystem of weighing, feeding and disease monitoring runs independently, with non-uniform data format, closed interface, and cannot realize cross-platform sharing.

[0005] Most current systems use simple threshold alarm, lack of multi-source data fusion analysis capability, and have high false alarm rate. The recognition of complex behavior still relies on manual experience, without introducing machine learning or deep learning model, resulting in untimely and inaccurate early warning, and it is difficult to achieve early intervention.

[0006] Breeding sheep as an important national livestock genetic resource, its breeding data has high sensitivity. However, the existing system generally lacks a perfect security mechanism, with chaotic user permissions, unencrypted data transmission, and non-traceable operation logs, there is a risk of data leakage and tampering. Therefore, a digital and intelligent genetic evaluation and breeding management system for breeding sheep is proposed. SUMMARY

[0007] The present application aims at the problems existing in the prior art. In order to achieve the above-mentioned application purposes, the present application provides the following technical solutions: a digital and intelligent genetic evaluation and breeding management system for sheep, comprising a data acquisition and perception module, a data storage and management module, a genetic evaluation and breeding analysis module, an intelligent breeding management module, a decision support and optimization module, and a system security and permission management module; The data acquisition and perception module is deployed at a breeding site, collects individual and environmental data of breeding sheep through a wireless sensor network and a wired interface, and transmits the data to a local edge gateway via a LoRa or NB-IoT communication protocol; the edge gateway runs a lightweight data preprocessing program, performs data compression, abnormality preliminary screening and caching, and then uploads the cleaned data to a cloud server through a 4G / 5G or optical fiber network; The cloud server adopts a micro-service architecture based on Kubernetes, and each functional module runs in an independent containerized service, and the modules are asynchronously communicated and loosely coupled integrated through a RESTful API and a message middleware RabbitMQ; The data acquisition and perception module pushes the original data stream to the data storage and management module in real time through an MQTT protocol, the latter receives and stores the data in a structured manner and labels the metadata, and provides a unified data access interface for other modules; The genetic evaluation and breeding analysis module obtains pedigree information, body weight, daily weight gain, lambing number, wool fiber diameter, wool yield, wool length, genotype data SNP chip detection results and health records of breeding sheep by calling the API of the data storage and management module, calculates a genomic breeding value GEBV once every quarter based on an SSGBLUP model, writes the evaluation results back to the data storage module, and triggers the intelligent breeding management module to update individual management strategies; The intelligent breeding management module subscribes to dynamic data of body temperature, exercise amount and feeding behavior of the data acquisition and perception module in real time, combines the GEBV ranking and individual production target output by the genetic evaluation module, uses an LSTM neural network and a rule engine to perform estrus prediction, health warning and feeding scheme analysis, and pushes execution instructions to on-site execution equipment or a mobile terminal APP; The decision support and optimization module periodically pulls overall genetic parameters of a breeding population, including an average inbreeding coefficient, a genetic gain rate and a trait variance component, from the data storage and management module, analyzes breeding sheep selection, elimination and mating combination suggestions by combining an economic weight model and an NSGA-II multi-objective optimization algorithm, and uses the output results as prior information for the next round of model input of the genetic evaluation and breeding analysis module to form a closed-loop optimization; The system security and permission management module runs through all modules, realizes user identity authentication by adopting an OAuth 2.0 protocol, allocates operation permissions based on an RBAC role-based access control model, and encrypts data transmission between all modules by using SSL / TLS, and meanwhile, utilizes a blockchain node to record key operation logs, and performs hash chain storage for GEBV publishing. The system as a whole supports daily data processing capacity ≥ 1 million, average communication delay between modules ≤ 800 milliseconds, system availability ≥ 99.9%, and realizes full-link automatic closed-loop management from data sensing to intelligent decision-making.

[0008] As a preferred technical solution of the present application, the data acquisition and sensing module comprises: The RFID ear tag has a working frequency of 134.2 kHz, and an identification distance of 0.5-1.2 meters, The three-axis acceleration motion monitoring necklace has a sampling frequency ≥ 50 Hz, and is used to record daily activity for ≥ 8 hours, The automatic weighing platform has an accuracy of ±0.2 kg, and each sheep is automatically weighed not less than 2 times per week, The infrared body temperature sensor has a measurement range of 35-42℃, an accuracy of ±0.1℃, and is collected every 6 hours, The environmental sensor group comprises a temperature and humidity sensor, an ammonia concentration sensor with a range of 0-100 ppm and a resolution of 0.1 ppm, and a fiber sensor, All sensors upload data to a local gateway through a LoRa wireless communication protocol, and the gateway to the cloud adopts a 4G / 5G or optical fiber network.

[0009] As a preferred technical solution of the present application, the data storage and management module adopts a MongoDB and Hadoop HDFS hybrid architecture, Among them, the structured data pedigree and breeding record are stored in a MongoDB cluster, and the unstructured data video and sensor original stream are stored in a HDFS, The data retention period is not less than 10 years, supports daily new data volume ≥ 50 GB, and the data cleaning rules include missing value interpolation by using a linear interpolation method, and abnormal value detection based on a 3σ principle and unit standardization processing.

[0010] As a preferred technical solution of the present application, the genetic evaluation and breeding analysis module adopts a single-step genomic BLUP SSGBLUP model to calculate individual breeding value EBV and genomic breeding value GEBV, Among them, the genotype data is derived from fine wool sheep 40K SNP chip detection, and the pedigree tracing depth is not less than 5 generations, The model fixed effect includes field, birth year and season, and gender, and the random effect includes individual additive genetic effect and residual, GEBV is calculated every quarter, and the traits include daily gain target increase ≥8%, lambing number target increase ≥0.3 lambs / female, and backfat thickness target reduction ≥0.2 mm.

[0011] As a preferred technical solution of the present application, the genetic evaluation and breeding analysis module further introduces an XGBoost machine learning model to perform nonlinear prediction on reproductive performance and disease resistance, The input features include historical lambing records, duration of body temperature fluctuation rate > 39.5°C, white blood cell count trend, and environmental ammonia concentration > 20 ppm, The model training adopts 5-fold cross-validation, and the prediction accuracy is not less than 85%, and the AUC value is ≥0.88.

[0012] As a preferred technical solution of the present application, the intelligent breeding management module realizes behavior recognition and abnormal early warning based on a rule engine and an LSTM neural network, Among them, the estrus behavior recognition is jointly determined by the increase of motion amount by ≥150% compared with the baseline and the standing time > 30 minutes / day, The disease early warning threshold is set as: continuous 24-hour body temperature > 40.0°C or feeding time reduction ≥40%, The system automatically analyzes and warns the information and pushes it to the administrator through the APP, and the response time is ≤30 seconds.

[0013] As a preferred technical solution of the present application, the intelligent breeding management module provides a personalized feeding scheme analysis function, According to the individual daily gain target setting range 150-300 g / day, body condition score BCS 1-5 points, and dynamic adjustment of energy intake 5% for every 5°C decrease in environmental temperature, The recommendation error is controlled within ±5%, and the feeding plan supports the linkage execution with the automatic feeder.

[0014] As a preferred technical solution of the present application, the decision support and optimization module adopts a multi-objective genetic algorithm NSGA-II to optimize the selection and matching of sheep, The optimization targets include: maximizing the comprehensive breeding index CTI of offspring, minimizing the inbreeding coefficient target <6.25%, and balancing the trait weight daily gain 40%, lambing number 30%, and health 30%, The top 10 optimal mating combinations are recommended each time, and genetic progress simulation prediction is provided for 3-5 generations.

[0015] As a preferred technical solution of the present application, the system security and permission management module adopts blockchain technology to chain and store key breeding data, The private chain is built using a Hyperledger Fabric framework, and on-chain data includes: mating records, genotype detection reports, GEBV evaluation results, The data hash value is synchronized on the chain every hour, ensuring that it is tamper-proof, and the traceability response time is less than or equal to 2 seconds.

[0016] As a preferred technical solution of the application, the system supports the RESTful API interface and the external system, Including the data format consistent with the national livestock and poultry genetic resource platform data format NY / T 2521-2013 standard, the third-party gene detection company supports FASTA, VCF format, The data interaction of the intelligent feeding equipment Modbus TCP protocol has an interface calling success rate of greater than or equal to 99.9%, and daily processing of external requests is greater than or equal to 10,000 times.

[0017] Compared with the prior art, the application has the following beneficial effects: The application adopts a single-step genomic BLUP SSGBLUP model, fuses pedigree, phenotype and 50K SNP genotype data, overcomes the limitation of the traditional BLUP model which only relies on pedigree, and improves the selection accuracy by more than 30%; the individual genomic breeding value GEBV is accurately predicted, and the prediction ability of low heritability traits such as lambing number and disease resistance is significantly enhanced, and the genetic progress speed is increased by 25-40%; the breeding value is dynamically updated every quarter, long-term genetic planning is supported, inbreeding cumulative inbreeding coefficient is controlled to be less than 6.25%, and population genetic diversity is ensured.

[0018] The application realizes 7*24 hours unattended data acquisition through RFID, intelligent collar and automatic weighing table equipment, and the data acquisition frequency reaches once every 15 minutes, which is much higher than the frequency of manual recording; The system optimizes the mating scheme, combines the multi-objective genetic algorithm NSGA-II, avoids inbreeding risk while maximizing genetic gain, the test group has a conception rate of 89.7%, which is 11.4 percentage points higher than that of the control group; the average number of lambs per pregnancy is increased by 0.3, the total number of lambs per year is increased by 17.9%, the breeding capacity of the breeding sheep farm is significantly enhanced; the average daily weight gain of individuals is increased by 18g / d+8.6%, the feed conversion efficiency is increased by 10.8%, and the economic benefits of breeding are significantly improved.

[0019] The system realizes automatic data acquisition, intelligent analysis and task pushing, reduces the manual recording, statistics and judgment links, and reduces the management working hours by 37.8%; supports mobile APP to realize real-time viewing, early warning and task execution, improves the management response speed and collaborative efficiency; the system availability is more than 99.9% per year, the operation is stable, the maintenance cost is low, and the system is suitable for long-term deployment in large-scale and intensive breeding farms.

[0020] All module communication of the application adopts SSL / TLS encrypted transmission to prevent data leakage; through HyperledgerFabric private chain storage, data is realized to be non-tamperable and traceable, meeting the requirements of national livestock genetic resource management standards. BRIEF DESCRIPTION OF DRAWINGS Figure 1 A system core module data block diagram is provided for the application; Figure 2 A data acquisition and perception module data block diagram is provided for the application; Figure 3 A system performance data block diagram is provided for the application; Figure 4 A system closed-loop logic block diagram is provided for the application. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical solutions and advantages of the embodiments of the application clearer, the technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings. Obviously, the described embodiments are a specific implementation of the application, and are not limited to all embodiments.

[0022] Therefore, the following detailed description of the embodiments of the application is not intended to limit the scope of the claimed application, but only represents some embodiments of the application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the application without creative labor are within the scope of protection of the application.

[0023] It should be noted that the embodiments in the application and the features and technical solutions in the embodiments can be combined with each other without conflict, and attention should be paid to the fact that similar reference numbers and letters represent similar items in the following drawings, so that once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.

[0024] As a preferred technical solution of the application, a digital and intelligent breeding sheep genetic evaluation and breeding management system, a digital and intelligent breeding sheep genetic evaluation and breeding management system, includes a data acquisition and perception module, a data storage and management module, a genetic evaluation and breeding analysis module, an intelligent breeding management module, a decision support and optimization module, and a system security and permission management module; The data acquisition and perception module is deployed in the breeding site, collects individual and environmental data of breeding sheep through a wireless sensor network and a wired interface, and transmits the data to a local edge gateway via a LoRa or NB-IoT communication protocol; the edge gateway runs a lightweight data preprocessing program, performs data compression, abnormality preliminary screening and caching, and then uploads the cleaned data to a cloud server through a 4G / 5G or optical fiber network; The cloud server adopts a Kubernetes-based microservice architecture, and each functional module runs as an independent containerized service. The modules communicate with each other asynchronously through RESTful API and message middleware RabbitMQ for loose coupling integration. The data acquisition and perception module pushes the raw data stream to the data storage and management module in real time through the MQTT protocol. The latter receives and stores the data in a structured manner and annotates the metadata, and provides a unified data access interface for other modules. The genetic evaluation and breeding analysis module obtains the pedigree information, phenotype data weight, daily gain, lambing number, genotype data SNP chip detection results, and health records of the breeding ewe by calling the API of the data storage and management module. Based on the SSGBLUP model, it calculates the genomic breeding value (GEBV) once a quarter and writes the evaluation results back to the data storage module, triggering the intelligent breeding management module to update the individual management strategy. The intelligent breeding management module subscribes to the dynamic data of body temperature, exercise, and feeding behavior of the data acquisition and perception module in real time. Combined with the GEBV ranking and individual production target output by the genetic evaluation module, it uses LSTM neural network and rule engine for estrus prediction, health warning, and feeding scheme analysis, and pushes the execution instructions to the on-site execution equipment or mobile APP. The decision support and optimization module pulls the overall genetic parameters of the population, including average inbreeding coefficient, genetic gain rate, and trait variance component, from the data storage and management module every 30 days. Combined with the economic weight model and NSGA-II multi-objective optimization algorithm, it analyzes the breeding ewe selection, culling, and mating combination suggestions. The output results serve as prior information for the next round of model input of the genetic evaluation and breeding analysis module, forming a closed-loop optimization. The system security and permission management module runs through all modules. It uses OAuth 2.0 protocol for user identity authentication, assigns operation permissions based on RBAC role-based access control model, and encrypts all data transmission between modules through SSL / TLS. It also uses blockchain nodes to hash and chain store the operation logs of key operations such as mating confirmation and GEBV release. The overall system supports daily data processing capacity ≥1 million, average communication delay between modules ≤800 ms, system availability ≥99.9%, and realizes full-link automated closed-loop management from data perception to intelligent decision-making.

[0025] The data acquisition and perception module includes: RFID ear tag operating frequency 134.2 kHz, identification distance 0.5-1.2 meters, Three-axis acceleration motion monitoring necklace sampling frequency ≥50 Hz, used to record daily activity for ≥8 hours, Automatic weighing platform accuracy ±0.2 kg, each sheep is automatically weighed not less than 2 times per week, Infrared body temperature sensor measurement range 35-42℃, accuracy ±0.1℃, every 6 hours, The environmental sensor group includes temperature and humidity sensors, ammonia concentration sensors with a range of 0-100 ppm and a resolution of 0.1 ppm, and optical fiber sensors, All sensors upload data to the local gateway through the LoRa wireless communication protocol, and the gateway to the cloud uses 4G / 5G or optical fiber network.

[0026] The data storage and management module uses a hybrid architecture of MongoDB and Hadoop HDFS, Where structured data pedigree and breeding records are stored in MongoDB clusters, and unstructured data videos and sensor raw streams are stored in HDFS, Data retention period is not less than 10 years, supporting daily new data volume ≥50 GB, data cleaning rules include missing value interpolation using linear interpolation method, and outlier detection based on 3σ principle and unit standardization.

[0027] The genetic evaluation and breeding analysis module uses a single-step genomic BLUP (SSGBLUP) model to calculate individual breeding value (EBV) and genomic breeding value (GEBV), Where genotype data comes from 50K SNP chip detection of fine wool sheep, and pedigree tracing depth is not less than 5 generations, Model fixed effects include session, birth year season, and gender, and random effects include individual additive genetic effects and residual, GEBV is calculated every quarter, and traits include target improvement of daily gain ≥8%, target improvement of lambing number ≥0.3 lambs per female, and target reduction of backfat thickness ≥0.2 mm.

[0028] The genetic evaluation and breeding analysis module further introduces an XGBoost machine learning model for nonlinear prediction of fertility and disease resistance, Input features include historical lambing records, duration of body temperature fluctuation rate >39.5℃, white blood cell count trend, and environmental ammonia concentration >20 ppm, Model training uses 5-fold cross-validation, with a prediction accuracy of not less than 85% and an AUC value ≥0.88.

[0029] The intelligent breeding management module realizes behavior recognition and abnormal early warning based on rule engine and LSTM neural network, Where estrus behavior recognition is determined by a joint determination of a sudden increase in exercise volume by ≥150% compared to baseline and standing time >30 minutes / day, Disease warning threshold is set as: continuous 24-hour body temperature >40.0℃ or feeding time reduction ≥40%, The system automatically analyzes the early warning information and pushes it to the administrator through the APP, and the response time is ≤30 seconds.

[0030] The intelligent breeding management module provides personalized feeding scheme analysis function, According to the individual daily weight gain target setting range 150-300 g / day, body condition score BCS 1-5 points and environmental temperature, the energy intake is increased by 5% for every 5℃ drop, the concentrate supplement amount is dynamically adjusted, The recommended error is controlled within ±5%, and the feeding plan supports linkage execution with automatic feeders.

[0031] The decision support and optimization module uses multi-objective genetic algorithm NSGA-II for sheep selection and optimization, The optimization targets include: maximizing the comprehensive breeding index CTI of offspring, minimizing the inbreeding coefficient target <6.25%, balancing the weight of traits such as daily weight gain 40%, lambing number 30%, and health 30%, The top 10 optimal mating combinations are recommended each time, and genetic progress simulation prediction is provided for a period of 3-5 generations.

[0032] The system security and permission management module uses blockchain technology to chain and store key breeding data, A private chain is built using the Hyperledger Fabric framework, and the chained data includes: mating records, genotype detection reports, and GEBV evaluation results, The data hash value is synchronized and chained once every hour to ensure that it cannot be tampered with, and the traceability response time is ≤2 seconds.

[0033] The system supports RESTful API interface and external system docking, Including data format consistent with NY / T 2521-2013 standard with national livestock and poultry genetic resource platform, supporting FASTA, VCF format with third-party gene detection company, The data interaction of intelligent feeding equipment Modbus TCP protocol has an interface call success rate ≥99.9%, and processes external requests ≥10,000 times per day on average.

[0034] Test example: In order to verify the effectiveness, stability and economic value of the "digital and intelligent sheep breeding genetic evaluation and breeding management system" in actual production, a test group system management group and a control group traditional management group are used for parallel control, and the system evaluates the improvement effect in genetic evaluation accuracy, reproductive efficiency, growth performance and management cost.

[0035] I. Test conditions and grouping

[0036] System deployment test group: Each sheep wears RFID ear tags and smart collars; Install 2 sets of automatic weighing platform, 12 groups of environmental sensors, 8 channels of video monitoring; Edge gateway 2, 4G networking, cloud system deployment on private cloud platform; Management personnel equipped with mobile APP, real-time receive early warning and task.

[0037] II. Test process 1. Data collection and system operation The system started on September 1, 2023, and ran continuously for 365 days; Upload data every 15 minutes, with a total of 120 million data collected; Automatic weighing completed 36,000 effective records, estrus behavior recognition triggered 427 times, health abnormality warning 89 times confirmed by veterinarians, accuracy 86.5%; All mating, lambing, disease treatment records are automatically synchronized to the cloud, and the key operation chain storage rate reaches 100%.

[0038] 2. Genetic evaluation implementation 4 times of genomic detection 50K SNP chip in December 2023, March, June and September 2024, data for SSGBLUP model update; GEBV ranking is released once every quarter, and the system automatically recommends the top 20% individuals as the core breeding population; Inbreeding coefficient real-time monitoring, system automatically avoids inbreeding combination target <6.25%, actual average inbreeding coefficient from initial 5.1% to 5.8% below, no >6.25% mating.

[0039] 3. Intelligent management implementation Estrus warning average 14.2 hours in advance, artificial insemination timely rate from 68% of the control group to 91.3%; The system dynamically adjusts the concentrate supplement according to the daily gain target of 220 g / d, with an error control of ±4.7%; In the health warning, 77 cases of diseases including pneumonia, hoof disease, and metritis were confirmed after 89 alarms, and early intervention reduced treatment cost by 31.2%.

[0040] 4. Mating and reproductive management Decision support module analyzes mating recommendations every month, a total of 186 optimized mating combinations are recommended; Test group conception rate 89.7% 161 / 180, significantly higher than the control group 78.3% 141 / 180; The total number of lambs was 356 in the test group and 302 in the control group, and the average number of lambs per pregnancy was 1.98 vs 1.68, an increase of 0.3 lambs per pregnancy.

[0041] III. Test results and data analysis

[0042] IV. System stability and safety verification The system has no major failures throughout the year, with an availability of 99.92%; The edge gateway network interruption and continuous transmission success rate is 100%, and the data loss rate is less than 0.1%; The blockchain module has a total of 1,243 key operation records on the chain, and all hash verifications pass, with no data tampering events; In the multi-user concurrent access test, 200 users operate simultaneously, and the average response time is ≤2.3 seconds.

[0043] V. Conclusion Through this test verification, the system of the present application has the following significant advantages in practical application: Significant improvement in genetic evaluation accuracy: The selection accuracy is improved by 32% through the fusion of genomic data by SSGBLUP, and the genetic progress of the core group is accelerated; Significant improvement in reproductive efficiency: The estrus detection rate and conception rate are significantly improved, and the number of lambs per pregnancy is increased by 0.3 lambs; Optimization of breeding management: Realize precise feeding, early disease warning, reduce the incidence and feed waste; Reduce the dependence on manpower: Reduce the management time by 37.8%, realize the large-scale and standardized operation; Ensure data security and traceability: Blockchain technology ensures the authenticity and credibility of breeding data, and meets the requirements of seed industry supervision.

[0044] VI. Test significance This test fully proves the feasibility, stability and economic value of the system in the actual production of sheep breeding farms, and can provide a replicable and popularized technical paradigm for the digital transformation of modern seed industry in China. It is recommended to be popularized and applied in national core breeding farms, stud rams stations and large-scale breeding enterprises.

[0045] The above examples are only used to illustrate the present application and do not limit the technical solutions described in the present application. Although the present application has been described in detail with reference to the above embodiments, the present application is not limited to the above specific embodiments, and any modification or substitution of the present application; all technical solutions and improvements within the spirit and scope of the present application are covered in the scope of the claims of the present application.

Claims

1. A digitalized breeding sheep genetic evaluation and management system, characterized in that, include: The system includes modules for data acquisition and sensing, data storage and management, genetic evaluation and breeding analysis, intelligent aquaculture management, decision support and optimization, and system security and access control. The data acquisition and sensing module is deployed at the breeding site. It collects data on individual breeding sheep and the environment through a wireless sensor network and a wired interface, and transmits the data to the local edge gateway via LoRa or NB-IoT communication protocols. The edge gateway runs a lightweight data preprocessing program to perform data compression, anomaly screening and caching, and then uploads the cleaned data to the cloud server via 4G / 5G or fiber optic network. The cloud server adopts a microservice architecture based on Kubernetes. Each functional module runs as an independent containerized service, and the modules communicate asynchronously and are loosely coupled through RESTful APIs and the message middleware RabbitMQ. The data acquisition and sensing module pushes the raw data stream to the data storage and management module in real time via the MQTT protocol. After receiving the data, the latter performs structured storage and metadata annotation, and provides a unified data access interface for other modules. The genetic evaluation and breeding analysis module obtains pedigree information, phenotypic data, body weight, daily weight gain, number of lambs, genotypic data, SNP chip detection results, and health records of breeding sheep by calling the API of the data storage and management module. Based on the SSGBLUP model, it calculates the genome breeding value (GEBV) every quarter and writes the evaluation results back to the data storage module, while triggering the intelligent breeding management module to update the individual management strategy. The intelligent breeding management module subscribes to the dynamic data of body temperature, activity level and feeding behavior of the data acquisition and sensing module in real time. Combined with the GEBV ranking and individual production target output by the genetic assessment module, it uses LSTM neural network and rule engine to predict estrus, provide health warning and feed plan analysis, and pushes the execution instructions to the on-site execution equipment or mobile APP. The decision support and optimization module periodically retrieves the average inbreeding coefficient, genetic gain rate, and trait variance components of the overall genetic parameters of the population from the data storage and management module every 30 days. It then combines the economic weight model with the NSGA-II multi-objective optimization algorithm to analyze suggestions for the selection, culling, and mating combinations of breeding sheep. The output results serve as prior information for the next round of model input in the genetic evaluation and breeding analysis module, forming a closed-loop optimization. The system security and access control module runs through all modules. It uses the OAuth 2.0 protocol to authenticate users, assigns operation permissions based on the RBAC role-based access control model, and encrypts data transmission between all modules through SSL / TLS. At the same time, it uses blockchain nodes to hash and store evidence of key operation logs, including confirmation and GEBV publication. The system supports a daily data processing volume of ≥1 million records, with an average communication latency of ≤800 milliseconds between modules.

2. The intelligent breeding sheep genetic evaluation and management system according to claim 1, characterized in that, The data acquisition and sensing module includes: an RFID ear tag with a working frequency of 134.2 kHz and an identification distance of 0.5–1.2 meters; a triaxial acceleration motion monitoring collar with a sampling frequency of ≥50 Hz for recording daily activity of ≥8 hours; an automatic weighing platform with an accuracy of ±0.2 kg, automatically weighing each sheep at least twice a week; and an infrared body temperature sensor with a measurement range of 35–42℃ and an accuracy of ±0.1℃, collecting data every 6 hours. The detection of wool traits involves collecting wool samples from approximately one palm's width behind the scapula on the side of the forelimb during the one-year-old assessment. A portable, all-weather rapid wool fiber analyzer is used for timely testing (achieving objective, accurate, and rapid detection of key indicators such as wool traits; solving problems such as cumbersome traditional identification methods, large errors, and poor timeliness; overcoming limitations in testing environment and technical personnel, allowing even graduate students to perform the tests after training). Traditional fine-wool sheep breeding stock identification relies on subjective judgment by experienced technicians. Objective testing requires sending the sheep to a standard laboratory with constant temperature and humidity, conducted by dedicated personnel, which demands high standards. Furthermore, the traditional BLUP (Breast Plus Upbringing) method is used for breeding stock evaluation only after phenotypic index determination, resulting in a long timeframe and potentially delaying the selection and mating of the selected sheep. Our system effectively utilizes the rapid analyzer to complete objective detection of wool traits promptly and accurately, improving breeding efficiency. It is recommended that these advantages be included. Although this system has been applied at our sheep farm, with the need for transformation and upgrading of grassland animal husbandry, our initial intention, or rather, the primary application of this system, is to large-scale fine-wool sheep farms. The environmental sensor suite includes a temperature and humidity sensor, an ammonia concentration sensor with a range of 0–100 ppm and a resolution of 0.1 ppm, and a fiber optic sensor. All sensors upload data to the local gateway via the LoRa wireless communication protocol, and the gateway connects to the cloud via 4G / 5G or fiber optic networks.

3. The intelligent breeding sheep genetic evaluation and management system according to claim 1, characterized in that, The data storage and management module adopts a hybrid architecture of MongoDB and Hadoop HDFS; Structured data, including pedigrees and reproductive records, is stored in a MongoDB cluster, while unstructured data, including videos and raw sensor streams, is stored in HDFS. The data retention period is no less than 10 years, and it supports daily new data volume of ≥50 GB. The data cleaning rules include linear interpolation for missing value imputation, outlier detection based on the 3σ principle and unit standardization.

4. The intelligent breeding sheep genetic evaluation and management system according to claim 1, characterized in that, The genetic evaluation and breeding analysis module uses the PBLUP and single-step genome SSGGBLUP models to calculate the individual breeding value EBV and the genome breeding value GEBV, respectively. The genotype data were obtained from the 40K SNP chip detection of fine wool sheep, the pedigree tracing depth was no less than 5 generations, the model fixed effects included farm number, birth season and sex, and the random effects included individual additive genetic effects and residuals. GEBV calculations are updated quarterly, and traits include a daily weight gain target increase of ≥8%, a lambing count target increase of ≥0.3 lambs / fetus, and a backfat thickness target decrease of ≥0.2 mm.

5. The digitalized breeding sheep genetic evaluation and management system according to claim 4, characterized in that, The genetic evaluation and breeding analysis module further introduces the XGBoost machine learning model to perform nonlinear prediction of fertility and disease resistance; Input features include historical lambing records, duration of body temperature fluctuation >39.5℃, white blood cell count trend, and environmental ammonia concentration >20 ppm as high risk; The model was trained using 5-fold cross-validation, with a prediction accuracy of no less than 85% and an AUC value of ≥0.

88.

6. The digitalized breeding sheep genetic evaluation and management system according to claim 1, characterized in that, The intelligent aquaculture management module uses a rule engine and an LSTM neural network to achieve behavior recognition and anomaly warning. Among them, estrus behavior is identified by a combination of a sudden increase in physical activity ≥150% from baseline and standing time >30 minutes / day. The disease warning threshold is set as follows: body temperature >40.0℃ for 24 consecutive hours or feeding time reduced by ≥40%. The system automatically analyzes the early warning information and pushes it to the administrator via the APP, with a response time of ≤30 seconds.

7. The digitalized breeding sheep genetic evaluation and management system according to claim 1, characterized in that, The intelligent aquaculture management module provides personalized feeding plan analysis functions; The amount of concentrate supplementation is dynamically adjusted based on the individual's daily weight gain target range of 150–300 g / day, body condition score of 1–5, and energy intake of 5% for every 5°C decrease in ambient temperature. The recommended error is within ±5%, and the feeding plan supports linkage with automatic feeders.

8. The digitalized breeding sheep genetic evaluation and management system according to claim 1, characterized in that, The decision support and optimization module uses the multi-objective genetic algorithm NSGA-II to optimize the selection and mating of breeding sheep. The optimization objectives include: maximizing the Comprehensive Breeding Index (CTI) of offspring, minimizing the inbreeding coefficient target of <6.25%, balancing the weights of daily weight gain (40%), litter size (30%), and health (30%). Each time, the top 10 optimal breeding combinations are recommended, and a genetic progression simulation prediction cycle of 3–5 generations is provided.

9. The digitalized breeding sheep genetic evaluation and management system according to claim 1, characterized in that, The system security and access control module uses blockchain technology to store key breeding data on the blockchain. A private chain is built using the Hyperledger Fabric framework. The data stored on the chain includes: mating records, genotype testing reports, and GEBV evaluation results. The data hash value is synchronized to the chain once per hour, and the traceability response time is ≤2 seconds.

10. The digitalized breeding sheep genetic evaluation and management system according to claim 1, characterized in that, The system supports RESTful API interfaces for interfacing with external systems; the intelligent feeding equipment uses Modbus TCP protocol for data interaction, with an interface call success rate of ≥99.9% and an average daily processing of ≥10,000 external requests.

Images