Breeding method for improving continuous fertilization ability of chickens

Abstract

The invention relates to the technical field of poultry genetic breeding, and particularly discloses a breeding method for improving the continuous fertilization ability of chickens. The core of the method comprises the following steps: constructing a hen individualized physiological baseline and a cock semen quality baseline; dynamically evaluating the reproductive steady state of the hens in the laying period, regulating and controlling the microenvironment as required, and synchronously updating the seminal fluid quality of the cocks; determining a hybridization window period based on the fertilization potential index, and combining historical data to intelligently match and hybridize; early evaluating the fertilization ability of the candidate offspring by using the prediction model, and screening and establishing a new generation of breeding group; key data is extracted every day to generate a data block hash evidence so as to guarantee traceability. The method can be used for precise breeding of good chicken varieties, and has the advantages of adapting to individual physiological differences, improving mating and breeding efficiency and guaranteeing credibility of breeding data; meanwhile, the standardization degree of the breeding process is high, and breed improvement and upgrading of the large-scale chicken raising industry can be remarkably promoted.

Description

Technical Field

[0001] This application relates to the field of poultry genetics and breeding technology, and more specifically, it relates to a breeding method for improving the continuous fertilization ability of chickens. Background Technology

[0002] In large-scale chicken farming, the chicken's ability to continuously fertilize directly affects the fertilization rate, hatchability, and economic benefits of hatching eggs, making it one of the core objectives of breed selection. Existing breeding methods for improving the continuous fertilization ability of chickens are mostly based on screening using uniform indicators across the flock. This involves monitoring macroscopic data such as overall semen quality, egg production, and fertilization rate to determine the breeding direction and conduct generational iterations. The core function of these methods is to initially screen individuals with relatively superior fertilization abilities, ensuring the basic needs of poultry production, and are suitable for breed improvement under traditional extensive farming models.

[0003] As the aquaculture industry upgrades towards precision and intelligence, the shortcomings of existing breeding methods are becoming increasingly apparent: they fail to establish specific evaluation benchmarks for individual physiological differences, relying solely on uniform group standards, which can easily lead to misjudgment of superior individuals or the omission of inferior ones; at the same time, they lack dynamic monitoring and proactive control of individual reproductive status, making it difficult to cope with the impact of environmental fluctuations on reproductive function, and the entire breeding process lacks a reliable data storage mechanism, making it impossible to guarantee the authenticity and traceability of the data. The core technical problem arising from this is the insufficient precision and stability of existing breeding methods, which cannot meet the requirements of precision breeding for individual difference adaptation, dynamic control, and data reliability, thus hindering the improvement of breeding efficiency in chickens' continuous fertilization ability. Summary of the Invention

[0004] To address the issues of insufficient accuracy and stability in the breeding of chickens' continuous fertilization ability in existing technologies, this application provides a breeding method to improve the continuous fertilization ability of chickens.

[0005] A breeding method for improving the continuous fertilization ability of chickens includes the following steps:

[0006] S1. Construction of individualized physiological baseline: After the flock is transferred to the laying house and before mating begins, a continuous period of environmental stability is selected, and multidimensional physiological time-series data of individual hens in the breeding flock are continuously collected. Based on the data, an individualized multidimensional physiological baseline is established for each hen. At the same time, the semen quality of individual roosters in the breeding flock is periodically monitored to establish a semen quality baseline for individual roosters.

[0007] S2. Dynamic Physiological Assessment and Homeostasis Regulation: During the laying period, the physiological data of individual hens are monitored in real time, and their real-time deviation from the physiological baseline is calculated. Based on the real-time deviation, the reproductive system homeostasis score and continuous fertilization potential index of the individual are output through the first prediction model, and the parameters of the first prediction model are recorded. When the reproductive system homeostasis score decreases and the environmental parameters deviate from the set range, adaptive environmental fine-tuning for the individual is triggered. In parallel, the real-time semen quality index of individual roosters is updated according to periodic monitoring.

[0008] S3. Intelligent Mating Decision and Execution: Based on the fertilization potential index of individual hens, dynamically determine when they enter the mating window; when an individual hen is in the mating window, combine the real-time semen quality index of individual roosters, and based on historical mating performance data, recommend roosters or semen for the current hen, and execute the mating operation; record the mating combination information and all-dimensional contextual data of this mating.

[0009] S4. Reverse selection and generation iteration: Based on historically accumulated individual full-dimensional data, construct a second predictive model that associates genomic information, physiological patterns, environmental data, and continuous fertilization capacity phenotype; save the parameters of the second predictive model; use the second predictive model to perform early prediction and simulation of the continuous fertilization capacity of candidate offspring; based on the early prediction results and combined with genetic background, select individuals with high continuous fertilization potential to form a new generation of breeding population, and repeat steps S1 to S4.

[0010] S5. Trusted Data Storage: Key feature values ​​generated during the breeding process are extracted daily, packaged into data blocks, and their hash values ​​are calculated for storage; when it is necessary to verify the authenticity of the data, the hash values ​​are compared.

[0011] By adopting the above technical solution, individualized physiological baselines and rooster semen quality baselines are established before the flock is transferred to the laying house for breeding. Wearable sensors and environmental sensors work together to collect multidimensional physiological time-series data. Data is transmitted in real time through edge computing nodes, establishing a unique assessment benchmark for each hen. During the laying period, physiological data of hens is collected in real time, and a 24-hour sliding time window is used to calculate the real-time deviation from the baseline. The deviation data is input into the first prediction model, which outputs a reproductive system homeostasis score and a continuous fertilization potential index. When the homeostasis score decreases and environmental parameters are abnormal, a directional micro-wind device is activated to fine-tune the individual microenvironment. Simultaneously, the real-time semen quality index of roosters is updated through periodic sampling and testing. The breeding window is dynamically determined based on the continuous fertilization potential index, combined with historical breeding results. The data is used to calculate the similarity of physiological state vectors to screen high-quality mating records, match currently available high-quality roosters for mating, and record full-dimensional data through the breeding management system. A second prediction model is constructed by integrating full-dimensional data such as physiological feature vectors and genotype data during the peak egg production period. When the candidate offspring are 8 weeks old, the genome is simulated based on pedigree information. The parental second prediction model is called to simulate physiological response characteristics under multiple environmental stress scenarios. The two are combined to predict the continuous fertilization ability of the candidate offspring. Individuals are selected to form a new generation of breeding population based on genetic background. Key feature values ​​such as core physiological data and model parameters are extracted daily, packaged into data blocks in JSON format, hashed using the SHA-256 algorithm, and uploaded to the blockchain node for notarization. Data authenticity is verified by hash value comparison. The overall technical solution breaks through the traditional extensive breeding logic of populations. It achieves refined control over the breeding process of chickens’ continuous fertilization ability through the coordinated linkage of various technical links. Its core innovation lies in incorporating individual physiological differences into the core consideration of breeding. It establishes a personalized regulation mechanism through baseline construction and dynamic deviation analysis, integrates genomic information and prediction models to achieve pre-optimization of the breeding cycle, and uses hash value evidence to improve the credibility of breeding data. All technical means are interconnected to form a closed loop.

[0012] Preferably, in step S1, the multidimensional physiological time-series data includes at least the core body temperature data collected at 10-15 min sampling intervals, three-dimensional activity data aggregated at 1 min sampling intervals and through a 5 min window, courtship acceptance posture frequency and active feather preening frequency data obtained based on image recognition and behavior analysis, and heart rate variability index extracted by collecting signals for 3-5 min daily through a photoplethysmography sensor; the continuous environmental stability period is at least 7 consecutive days, with the ambient temperature maintained at 18-22℃ and the relative humidity maintained at 50-65%.

[0013] By adopting the above technical solution, multidimensional physiological time-series data are collected using a combination of differentiated sampling intervals and multi-source detection technology: core body temperature data is collected at 10-15 minute sampling intervals using an implanted temperature sensor; three-dimensional activity data is collected at 1 minute sampling intervals using a wearable sensor with an integrated IMU, and then aggregated using a 5-minute sliding window mean; courtship posture frequency and active feather preening frequency data are collected from images captured by a high-definition camera, and the YOLOv5 algorithm is used for behavior recognition and frequency statistics; heart rate variability indicators are collected daily for 3-5 minutes using a photoplethysmography (PPG) sensor, and time-domain and frequency-domain features are extracted using wavelet transform. A continuous stable environmental period is maintained by an intelligent chicken house environmental control system. During the 7-day stable period, temperature and humidity sensors are used for real-time monitoring and feedback control to ensure an ambient temperature of 18-22℃ and a relative humidity of 50-65℃. Under these conditions, periodic sampling and testing of rooster semen quality are carried out simultaneously. The collected data are standardized and then used to construct personalized multidimensional physiological baselines for hens and semen quality baselines for roosters.

[0014] Preferably, in step S2, the calculation of the real-time deviation is specifically as follows: using 24h as a sliding time window, the standardized deviation score of the observed value of each physiological indicator within the window relative to its individual baseline is calculated, and the deviation scores of all indicators constitute the real-time deviation vector; the reproductive system homeostasis score is a continuous value between 0 and 1; the continuous fertilization potential index is a continuous value between 0 and 1.

[0015] By adopting the above technical solution, the real-time deviation calculation uses a 24-hour sliding time window combined with the Z-score standardization algorithm. The standardized deviation score of each physiological indicator is calculated using the formula (observed value - baseline mean) / baseline standard deviation. The deviation scores of all indicators are sorted according to a preset dimension to form a real-time deviation vector, which serves as the input data for the first prediction model. The reproductive system homeostasis score and the continuous fertilization potential index are both output as continuous values ​​between 0 and 1, generated through mapping in the fully connected layer of the first prediction model. The reproductive system homeostasis score serves as the triggering parameter for homeostasis regulation, while the continuous fertilization potential index serves as a quantitative parameter for determining the mating window and prioritizing. Together, they constitute a quantitative data system for individual reproductive status assessment, achieving data integration between dynamic physiological assessment and intelligent mating decision-making.

[0016] Preferably, in step S2, the triggering condition is: the individual's reproductive system homeostasis score decreases by more than 20% from its highest value in the past 24 hours within 2 hours, and its microenvironment temperature is higher than 24℃ or lower than 16℃; the fine-tuning includes initiating a directional breeze with a wind speed of 0.5-1.5m / s for a duration of 30-60min.

[0017] By adopting the above technical solution, the triggering condition employs a dual-logic judgment: the data processing module compares the individual's reproductive system steady-state score in real time, calculating the decrease in value within 2 hours relative to the highest value of the past 24 hours; simultaneously, a micro-environmental temperature sensor deployed 30cm away from the chicken's activity area collects temperature data. When both conditions are met simultaneously, adaptive environmental fine-tuning is triggered. Fine-tuning is performed by a directional micro-wind device, which starts at a preset wind speed of 0.5 to 1.5 meters per second. The PLC controller sets the continuous adjustment duration to 30 to 60 minutes, directionally acting on the area where the abnormal individual is located. During the adjustment process, micro-environmental temperature data is collected in real time and fed back to the controller until the temperature returns to a suitable range.

[0018] Preferably, in step S3, the dynamic determination rule for the mating window period is as follows: when an individual's continuous fertilization potential index remains at a level greater than 0.8 for more than 3 consecutive hours, it is determined that it has entered the mating window period; the mating tasks are prioritized and scheduled according to the continuous fertilization potential index value of individuals within the window period.

[0019] By adopting the above technical solution, the determination of the mating window period is verified by both data thresholds and duration: the continuous fertilization potential index is output in real time by the first prediction model, and the mating window period is determined by the duration of the cumulative index greater than 0.8 by a timer. When the cumulative duration reaches 3 hours, the mating window period is determined. The 3-hour duration is set based on the physiological rhythm data of chickens during the egg-laying period to ensure coverage of the stable physiological stages before and after ovulation. The mating task scheduling is realized through the breeding management system. The system generates a task list by sorting the individuals' continuous fertilization potential index values ​​in descending order within the window period, prioritizing the allocation of high-quality roosters or semen resources to individuals with higher index values, and pushing scheduling instructions to breeders through the system terminal to ensure that mating operations are executed according to priority.

[0020] Preferably, in step S3, the step of recommending roosters or semen for the current hen based on historical mating effectiveness data specifically includes: calculating the similarity between the current hen's physiological state vector and the hen state vector in the mating records in the historical database, and filtering out historical records with a similarity greater than 70%; selecting the top 20% of records with the best continuous fertilization days from the filtered records, and analyzing the characteristics of the roosters or semen used in them; and recommending 1-3 roosters based on the real-time semen quality index of the roosters when recommending the current hen.

[0021] By adopting the above technical solution, the mating matching recommendation process follows a three-step procedure: First, a cosine similarity algorithm is used to calculate the similarity between the current hen's physiological state vector and the hen state vectors in historical mating records, setting a 70% similarity threshold to filter valid historical records. Second, valid records are sorted in descending order by the number of days of continuous fertilization, and the semen motility, genotype, and other characteristic parameters of roosters in the top 20% of records are extracted to construct a high-quality rooster feature dataset. Third, real-time semen quality index data of available roosters is used to calculate the matching degree with the high-quality rooster feature dataset, outputting the 1 to 3 roosters with the highest matching degree as the recommendation results. The recommendation results are displayed through the breeding management system, providing alternative solutions for mating execution and avoiding the impact of unforeseen circumstances of a single recommended individual on the mating progress.

[0022] Preferably, in step S4, the individual full-dimensional data includes the physiological feature vector aggregated weekly during the peak egg production period, genotype data based on single nucleotide polymorphism chips, environmental data, and the phenotypic value of continuous fertilization days determined by egg candling sampling; the second prediction model is a generative model that can generate predicted physiological patterns and continuous fertilization capacity phenotypic values ​​based on the input genome, historical physiological patterns, and assumed environment.

[0023] By adopting the above technical solution, individual full-dimensional data is integrated from four data sources: the physiological characteristic vector of peak egg production is generated by aggregating daily physiological data through weekly averaging; genotype data is obtained through Illumina chicken 60K SNP chip detection, selecting SNP loci associated with reproductive traits; environmental data is collected in real time by chicken house environmental sensors and stored synchronously; and the phenotypic value of continuous fertilization days is obtained through sampling egg candling detection, with 30 eggs randomly selected from each hen, incubated, recorded, and averaged. The second prediction model is constructed using a generative adversarial network. The model input layer receives three types of data: genomic data, historical physiological patterns, and hypothetical environment data. The generator simulates changes in physiological patterns under different environments, and the discriminator verifies and optimizes the simulation results. Finally, it outputs the predicted physiological patterns and phenotypic values ​​of continuous fertilization ability, realizing the correlation modeling of genetic, physiological, and environmental factors.

[0024] Preferably, in step S4, the early prediction and simulation of the candidate offspring's ability to sustain fertilization specifically involves: for the candidate offspring, at 8 weeks of age, simulating its genome based on its pedigree information; calling the second prediction model of its parents to simulate at least three hypothetical environmental stress scenarios to obtain the predicted physiological responses of the parents; and combining the simulated genome of the candidate offspring with the features extracted from the predicted responses of the parents to predict its sustained fertilization potential index curve characteristics and sustained fertilization days under the hypothetical environmental scenarios.

[0025] By adopting the above technical solution, early prediction and simulation are carried out in three steps: First, at 8 weeks of age, AlphaSimR software is used to simulate the genome sequence of the parent offspring based on the parental pedigree information and genotype data; Second, three hypothetical environmental stress scenarios are set: high temperature 32-35℃, low temperature 10-15℃, and high humidity 75-85%. The second prediction model of the parent is called and the parameters of the three scenarios are input respectively to obtain the predicted physiological response data of the parent and extract core features such as the lowest value of the homeostasis score and the recovery time; Third, the simulated genome of the candidate offspring is fused with the core feature data of the parent and input into the second prediction model to output the continuous fertilization potential index curve features and continuous fertilization days under different hypothetical environmental scenarios.

[0026] Preferably, in step S4, the individual selection criteria based on early prediction results are: in more than 50% of the hypothetical environmental scenarios, the predicted average continuous fertilization potential index is greater than 0.75, and the predicted continuous fertilization days are more than 10% higher than the average of the candidate generation group to which it belongs.

[0027] By adopting the above technical solution, individual selection employs a dual-condition verification technique: First, the predicted results of candidate offspring under three hypothetical environmental scenarios are statistically analyzed, and the predicted average continuous fertilization potential index for more than 50% of scenarios is calculated and verified to be greater than 0.75; simultaneously, the average predicted continuous fertilization days of the candidate generation population is calculated, and the individual predicted value is verified to be more than 10% higher than the average. Pedigree database data is simultaneously accessed, and the inbreeding coefficient within three generations of the individual is calculated using a common parentage coefficient algorithm. Individuals with a coefficient greater than 6.25% are excluded, and the finally selected individuals form a new generation breeding population, ensuring the genetic diversity and continuous fertilization potential of the breeding population.

[0028] Preferably, in step S5, the key feature values ​​include at least the hen's core body temperature, heart rate variability, rooster's real-time semen quality index, first prediction model parameters, second prediction model parameters, mating combination information, and chicken house environmental temperature and humidity data.

[0029] By adopting the above technical solution, key feature values ​​are automatically extracted daily at 24:00 by the data acquisition module and integrated into five data subsets according to a preset data structure: hen physiological data, rooster semen data, model parameter data, mating information data, and environmental data. These subsets are then packaged into data blocks using JSON format. The SHA-256 algorithm is used to calculate the hash value of each data block. The hash value is then associated with the corresponding date and generation number and uploaded to a permissioned blockchain network. Network nodes include servers at the breeding base, third-party testing institutions, and patent holders. When data authenticity needs to be verified, the data block to be verified is extracted, the hash value is recalculated, and compared with the hash value stored on the blockchain to verify the data's authenticity.

[0030] In summary, this application has the following beneficial effects:

[0031] 1. The method of this application, by constructing individualized physiological baselines and rooster semen quality baselines, combined with a full-chain design of dynamic physiological assessment and homeostasis regulation, intelligent mating decision-making, reverse selection iteration, and credible data storage, achieves a leap from group screening to individual precision breeding, effectively improving the accuracy and stability of continuous fertilization ability breeding, while ensuring the traceability of breeding data.

[0032] 2. In this application, the preferred method is to collect individualized multidimensional physiological time-series data and calculate the deviation in real time. By triggering adaptive environmental fine-tuning through the reproductive system homeostasis score, the accurate assessment and active homeostasis maintenance of the hen's reproductive status are achieved, reducing the impact of environmental fluctuations on reproductive function.

[0033] 3. The method of this application dynamically determines the mating window period by using the continuous fertilization potential index and recommends male and female matching by combining historical mating performance data. This achieves precise scheduling and optimal combination matching of mating tasks, effectively improving the mating success rate and subsequent continuous fertilization results.

[0034] 4. This application adopts a generative second prediction model, which combines pedigree information to complete the early prediction of the continuous fertilization ability of candidate offspring at 8 weeks of age, which greatly shortens the breeding cycle, reduces the wrong elimination of superior individuals, and improves the breeding efficiency of generation iteration.

[0035] 5. This application achieves reliable data storage and authenticity verification of the entire breeding process by extracting key feature values ​​daily to generate data blocks and calculating hash values ​​for data storage. This solves the problem of difficulty in tampering with and tracing breeding data, and provides a reliable guarantee for the protection of plant variety rights and data sharing. Attached Figure Description

[0036] Figure 1 This is a flowchart of a breeding method for improving the continuous fertilization ability of chickens, as provided in this application. Detailed Implementation

[0037] The present application will be further described in detail below with reference to embodiments and comparative examples. Unless otherwise specified, the experimental methods used below are conventional methods. Unless otherwise specified, the materials, reagents, methods and instruments used are all conventional materials, reagents, methods and instruments in the art, which can be obtained by those skilled in the art through commercial channels or prepared according to literature methods.

[0038] Technical concept:

[0039] In the current field of poultry genetics and breeding, traditional techniques for selecting chickens with sustained fertilization capacity generally employ a broad-based screening method using uniform population standards, resulting in insufficient precision and stability in selection. The root cause lies in the failure to fully consider individual physiological differences, relying solely on macro-population data to establish screening criteria, leading to misjudgments of superior individuals or the omission of inferior ones. Furthermore, the lack of dynamic monitoring and proactive control mechanisms for individual reproductive status makes it impossible to address the adverse effects of environmental fluctuations on reproductive function. Additionally, the lack of reliable data storage and traceability throughout the entire selection process makes it difficult to guarantee data authenticity. Coupled with the absence of predictive models linking genetic, physiological, and environmental factors, this results in long selection cycles and low efficiency.

[0040] This technical solution addresses the aforementioned issues by constructing a comprehensive technical system encompassing individualized precision, dynamic regulation, intelligent matching, early prediction, and reliable data. First, it establishes individualized physiological baselines for hens and semen quality baselines for roosters through multi-source sensing and differentiated sampling, compensating for insufficient individual adaptation. Next, it achieves dynamic physiological assessment by calculating deviations using a sliding time window and combining this with a predictive model, while proactively maintaining reproductive homeostasis through a directional micro-adjustment mechanism. It dynamically determines the mating window based on the fertilization potential index and achieves precise matching by reusing historical high-quality mating experience using a cosine similarity algorithm. Furthermore, it utilizes generative predictive models and pedigree simulation to achieve early evaluation of candidate offspring, shortening the breeding cycle. Finally, it ensures data reliability through hash algorithms and blockchain-based evidence storage. These collaborative technologies systematically improve the accuracy and efficiency of breeding.

[0041] This embodiment uses white-feathered laying hens as the breeding target. Two thousand 1-day-old healthy white-feathered laying hens were selected, with a female-to-male ratio of 1:10, and raised in a standardized intelligent breeding base. The method of this invention was used to conduct continuous fertilization selection across three generations. The experimental site was a closed intelligent chicken house equipped with core equipment such as an environmental control system, a multi-dimensional physiological monitoring system, an image recognition system, and a blockchain evidence storage server. The feed used in the experiment was a complete compound feed conforming to NY / T 33-2004 "Chicken Feeding Standards". All testing equipment used was calibrated to ensure data acquisition accuracy. The specific implementation steps are as follows:

[0042] S1. Individualized Physiological Baseline Construction: This step aims to establish a personalized, multidimensional physiological baseline for each hen, while simultaneously establishing a semen quality baseline for roosters. This provides a benchmark for subsequent dynamic assessment and breeding decisions. The specific implementation process is as follows:

[0043] 1. Experimental Preparation: After the chickens were transferred to the laying house at 18 weeks of age, the environment of the chicken house was pre-controlled to enter a 7-day environmental stabilization period. During this period, the environmental control system was used to maintain the ambient temperature at 18-22℃, the relative humidity at 50-65%, the light cycle at 16 hours of light and 8 hours of darkness, the light intensity at 20-30 lux, and the feed intake at 120g / chicken / day to ensure stable environmental parameters without fluctuations. At the same time, each hen was fitted with a smart leg band integrating a photoplethysmography pulse wave sensor and a three-dimensional accelerometer, and each rooster was fitted with an identification leg band. Ten high-definition infrared cameras and five millimeter-wave radars were evenly distributed in the chicken house, covering the entire area. An automatic semen collection device was set up in the rooster feeding area.

[0044] 2. Multidimensional physiological time-series data collection: During the environmental stabilization period, all monitoring equipment will be activated to continuously collect multidimensional physiological time-series data for each hen in the breeding flock. The specific collection plan is as follows:

[0045] Core body temperature data: collected via the built-in temperature sensor of the smart ankle bracelet, with a sampling interval of 10-15 minutes, recording one set of data each time, and a data accuracy of ±0.1℃;

[0046] Three-dimensional activity data: collected by the three-dimensional accelerometer built into the smart ankle bracelet, with a sampling interval of 1 minute and an aggregation window of 5 minutes. The average and peak acceleration values ​​within the window are calculated as three-dimensional activity data.

[0047] Specific behavioral frequency data: Image data was collected using a high-definition infrared camera, combined with image recognition algorithms and behavioral analysis models to identify and statistically analyze the frequency of courtship acceptance postures for each hen, in units of times / day; the frequency of active feather preening, in units of times / day; the behavioral recognition accuracy of the image recognition algorithm is ≥95%;

[0048] Heart rate variability indicators: Signals are collected by the photoplethysmography (PPG) sensor built into the smart ankle bracelet. The signal is collected for 3-5 minutes every day from 10:00 to 10:05 AM at a sampling frequency of 100Hz. The heart rate variability coefficient, RR interval standard deviation and other indicators are then extracted through signal processing.

[0049] 3. Periodic monitoring of rooster semen quality: During the environmental stability period, semen was collected and tested from each rooster every 3 days. Specifically, 0.3-0.5 mL of semen was collected from each rooster using an automated semen collection device. A computer-aided semen analysis system was used to test indicators such as semen motility and semen density. Three sets of parallel data were recorded for each test, and the average value was taken as the semen quality data for that test. Finally, based on all semen quality data during the environmental stability period, a semen quality baseline for each rooster was established.

[0050] 4. Individualized baseline construction: After the environmental stabilization period, the multidimensional physiological time-series data of each hen are preprocessed to remove abnormal data, specifically data that exceeds the normal physiological range, such as core body temperature >43℃ or <38℃. The statistical characteristics of each indicator during the environmental stabilization period, such as mean, standard deviation, and median, are calculated. Based on this, a personalized multidimensional physiological baseline is established for each hen. The baseline data is stored in a local database and associated with the corresponding hen's identity information.

[0051] S2. Dynamic Physiological Assessment and Homeostasis Regulation: This step aims to monitor the physiological state of hens in real time and assess the homeostasis of the reproductive system, dynamically update the semen quality of roosters, and make environmental adjustments for abnormal individuals to ensure stable individual reproductive function. The specific implementation process is as follows:

[0052] 1. Real-time data monitoring and deviation calculation: After the environmental stabilization period ends and the egg-laying period begins, the monitoring equipment is kept running continuously to collect physiological data and chicken house environmental parameters for each hen in real time. The sliding time window method is used to calculate the real-time deviation. Specifically, a 24-hour sliding time window is used, and the data within the window is updated every hour. The standardized deviation score of each physiological indicator observation value relative to its individual baseline is calculated. The deviation scores of all indicators are integrated into a real-time deviation vector, with the vector dimension consistent with the number of physiological indicators. The deviation score is calculated as (observed value - baseline mean) / baseline standard deviation.

[0053] 2. The reproductive system homeostasis score and continuous fertilization potential index are output by inputting the real-time deviation vector into the pre-trained first prediction model. The model uses an LSTM neural network, and the training data consists of historical physiological data and reproductive status correlation data of white-feathered laying hens. The model outputs a reproductive system homeostasis score for each hen, which is a continuous value between 0 and 1. The closer the score is to 1, the more stable the reproductive system is. The continuous fertilization potential index is also a continuous value between 0 and 1. The closer the index is to 1, the higher the potential for continuous fertilization. The score and index data are recorded every hour and stored in the database. At the same time, the current parameters of the first prediction model are recorded.

[0054] 3. Adaptive Environmental Fine-Tuning Triggering and Execution: Real-time monitoring of the reproductive system homeostasis score and microenvironmental temperature for each hen. When trigger conditions are met: the individual's reproductive system homeostasis score decreases by more than 20% from its highest value in the past 24 hours within 2 hours, and its microenvironmental temperature is above 24℃ or below 16℃. The fine-tuning operation is as follows: Activate the directional ventilation device in the corresponding area, setting the wind speed to 0.5-1.5 m / s, and ventilate continuously for 30-60 minutes. Monitor the microenvironmental temperature in real time during ventilation. Stop ventilation once the temperature returns to the range of 18-22℃. Record the trigger time, fine-tuning parameters, temperature change curve, and other data for each fine-tuning operation.

[0055] 4. Real-time semen quality index of roosters is updated. During the egg-laying period, the semen quality of each rooster is tested every 5 days. The test items and methods are the same as in step S1. The real-time semen quality index of each rooster is updated based on the test results and synchronized to the breeding decision system in real time.

[0056] S3. Intelligent Mating Decision and Execution: This step aims to accurately determine the mating window for hens and combine rooster semen quality and historical data to achieve optimal mating matching, thereby improving mating effectiveness. The specific implementation process is as follows:

[0057] 1. Dynamic determination of the breeding window: The system reads the continuous fertilization potential index of each hen in real time and dynamically determines the breeding window based on the following rules: When the continuous fertilization potential index of an individual remains above 0.8 for more than 3 consecutive hours, the hen is determined to have entered the breeding window. The system automatically records the start and end times of the window for each hen and prioritizes individuals within the window based on their continuous fertilization potential index values, with higher indices indicating higher priority. A breeding task list is then generated and pushed to the breeding management terminal.

[0058] 2. Mating and Matching Recommendation: When hens are in the mating window, the system recommends roosters based on historical mating performance data. The specific process is as follows:

[0059] Data retrieval and similarity calculation: The local historical breeding database is called to extract the physiological state vectors of all hens in the historical breeding records. The cosine similarity algorithm is used to calculate the similarity between the current hen's physiological state vector and the hen's state vectors in the historical records.

[0060] Historical record filtering: Filter out historical mating records with a similarity greater than 70% to ensure that the filtered records have reference value;

[0061] Analysis of high-quality records: From the selected historical records, the top 20% of records with the best performance in continuous fertilization days were selected, and the baseline characteristics of semen quality and identity information of the roosters used in these records were extracted;

[0062] Recommendation results generation: Based on the real-time semen quality index of currently available roosters, excluding roosters with a semen quality index <0.7, and matching the characteristics of roosters in historical high-quality records, 1-3 roosters are recommended for the current hen. At the same time, the recommendation reason is generated, such as "Recommended rooster A, whose semen quality index is 0.92, and its similarity with the characteristics of historically matched high-quality roosters is 85%".

[0063] 3. Mating Execution and Data Recording: Breeding personnel execute mating operations according to the mating task list and recommended results, using artificial insemination. Mating times are selected daily from 6:00-8:00 AM or 5:00-7:00 PM. During mating, information on the mating combination and comprehensive situational data are recorded, including mating time, the hen's sustained fertilization potential index, the rooster's real-time semen quality index, chicken house ambient temperature, and relative humidity. All data is uploaded to the database in real time.

[0064] S4. Reverse selection and generation iteration: This step aims to use historical data to build a predictive model, achieve early prediction of the continuous fertilization ability of candidate offspring, select high-quality individuals to form a new generation of breeding population, and complete generation iteration. The specific implementation process is as follows:

[0065] 1. Individual comprehensive data organization: Collect comprehensive individual data accumulated during the previous generation's breeding process, organize and standardize the data, including the following data types:

[0066] Physiological characteristic vector: The weekly aggregated physiological characteristic vector of each chicken during its peak egg production period of 35-40 weeks of age, including the weekly average core body temperature, average activity level, average behavior frequency, etc.

[0067] Genotype data: Single nucleotide polymorphism (SNP) chips were used to perform genetic testing on each chicken to obtain genotype data;

[0068] Environmental data: Time-series data on chicken house environment temperature, humidity, light, feed nutrition, etc. during the breeding process;

[0069] The phenotypic value of continuous fertilization days was determined by sampling and candling. 30 eggs were randomly selected from each hen, and the number of continuous fertilization days was recorded after hatching. The average value was taken as the phenotypic value of continuous fertilization days for that individual.

[0070] 2. Construction and Training of the Second Predictive Model: Based on the processed, comprehensive individual data, a second predictive model is constructed. This model employs a generative adversarial network (GAN) and can generate predicted physiological patterns and phenotypic values ​​for continued fertilization capacity based on the input genomic information, historical physiological patterns, and hypothetical environmental parameters. 70% of the comprehensive data is used as the training set, and 30% as the test set for model training. During training, model parameters are optimized until the model's prediction accuracy is ≥85%. The trained second predictive model parameters are then saved to the database.

[0071] 3. Early prediction and simulation of candidate offspring: When the offspring reach 8 weeks of age, early prediction and simulation of their continued fertilization capacity are conducted. The specific process is as follows:

[0072] Genome simulation: Based on the pedigree information of candidate progeny, a pedigree reconstruction algorithm is used to simulate their genome information;

[0073] Obtaining predicted physiological responses of parents: The second prediction model corresponding to the parents is called, and three hypothetical environmental stress scenarios are set: Scenario 1: high temperature stress, temperature 32-35℃; Scenario 2: low temperature stress, temperature 10-15℃; Scenario 3: humidity stress, humidity 75-85%. The genomic information, historical physiological patterns and hypothetical environmental parameters of the parents are input into the model to obtain the predicted physiological responses of the parents under the three scenarios, including physiological index change curves, homeostasis score changes, etc.

[0074] Candidate offspring capability prediction: By combining simulated genomic information of candidate offspring with features extracted from parental predicted responses, such as the lowest homeostatic score and recovery time under stress, the data are input into the second prediction model to predict the characteristics of the continuous fertilization potential index curve and the number of days of continuous fertilization of candidate offspring under three hypothetical environmental scenarios.

[0075] 4. Selection of High-Quality Individuals and Establishment of a New Generation Breeding Population: Based on early prediction results, individuals with high continuous fertilization potential are selected. The selection criteria are: under more than 50% of the hypothetical environmental scenarios, their predicted average continuous fertilization potential index is greater than 0.75, and their predicted continuous fertilization days are more than 10% higher than the average of their candidate generation population. Simultaneously, considering the genetic background of the candidate offspring, inbred individuals within three generations are excluded. Finally, 200 high-quality individuals with a female-to-male ratio of 1:10 are selected to form a new generation breeding population. Steps S1 to S4 are repeated to conduct the next generation of breeding. This embodiment completes three generations of iterative breeding.

[0076] S5. Trusted Data Storage: This step aims to ensure the authenticity and integrity of data during the breeding process and to achieve data traceability. The specific implementation process is as follows:

[0077] 1. Key Feature Extraction and Data Block Generation: Every day at 24:00, the system automatically extracts key feature values ​​generated during the breeding process from the database. These include: the daily average core body temperature, heart rate variability core index, reproductive system homeostasis score, and continuous fertilization potential index for each hen; the real-time semen quality index for each rooster; parameters of the first and second prediction models; information on all mating combinations for the day; and the daily average and extreme values ​​of temperature and humidity in the chicken house. The extracted key feature values ​​are packaged into data blocks according to a fixed format, with each data block associated with the date and generation number.

[0078] 2. Hash value calculation and notarization: The SHA-256 hash algorithm is used to calculate the hash value of the generated data block to obtain a unique hash value. This hash value is then uploaded to a permissioned blockchain network for notarization. The blockchain network nodes include the breeding base server, the third-party testing agency server, and the patent holder server to ensure that the notarized data cannot be tampered with.

[0079] 3. Data authenticity verification: When it is necessary to verify the authenticity of data, extract the data block to be verified, recalculate its hash value, and compare the calculated hash value with the corresponding hash value stored in the blockchain network. If the two are consistent, it proves that the original data is authentic, complete, and has not been tampered with; if the two are inconsistent, it indicates that the original data has been tampered with or damaged.

[0080] Through the above three generations of breeding, a white-feathered laying hen strain with stable continuous fertilization ability was successfully bred. The entire breeding process was traceable and the operation was standardized, which effectively improved the breeding efficiency and accuracy, and verified the feasibility and superiority of the method of this invention.

[0081] To further verify the effectiveness and superiority of the breeding method of this invention, a performance test experiment was set up. By comparing it with traditional breeding methods, tests were conducted on core dimensions such as continuous fertilization ability, environmental adaptability, and breeding efficiency. The specific experimental contents are as follows:

[0082] The test subjects included an experimental group and a control group. The experimental group consisted of 100 healthy hens and 10 roosters selected from three generations of white-feathered laying hens bred using the method of this invention. The control group consisted of 100 healthy hens and 10 roosters selected from the same generation of white-feathered laying hens bred using traditional breeding methods based on uniform population indicators without individualized baselines or dynamic regulation. Both experimental and control subjects were 40 weeks old and had undergone the same basic immunization program, with no history of major infectious diseases.

[0083] The test site was a closed intelligent chicken house, divided into an experimental group feeding area and a control group feeding area. The environmental parameters of the two areas could be controlled independently, and they were equipped with the same environmental monitoring equipment, physiological monitoring equipment, semen testing equipment, and egg candling equipment. The specific environmental settings were as follows: during the routine test phase, the ambient temperature was 20-25℃ and the relative humidity was 50-60%; during the stress test phase, the environment was controlled according to the preset stress scenarios, with each stress scenario lasting for 7 days and a routine environmental recovery period of 10 days between scenarios; the specific implementation of feed and management was as follows: both groups were fed complete compound feed that met the NY / T 33-2004 "Chicken Feeding Standards", with a feeding amount of 120g / bird / day, a light cycle of 16h light / 8h darkness, and a light intensity of 20-30 lux.

[0084] The routine performance test for days 1-7 was conducted as follows: Under normal environmental conditions, both groups were inseminated using artificial insemination. Insemination took place from 6:00 to 8:00 AM daily for a period of 7 days. After insemination, hatching eggs were continuously collected, and the number of hatching eggs was recorded daily. The fertilization status was checked using the candling method until no fertilized eggs were found for 3 consecutive days. The number of days of continuous fertilization was recorded. The average fertilization rate and the coefficient of variation of the fertilization rate were calculated.

[0085] The environmental stress performance test from day 18 to day 56 consisted of the following tests: high temperature stress, low temperature stress, and high humidity stress. The test procedure for each stress scenario was as follows: Scenario control: The environmental parameters of the corresponding breeding area were adjusted to the stress conditions and maintained for 7 days; Mating and data collection: Artificial insemination was carried out on day 3 of the stress scenario, and the mating method was the same as that of the routine test; Fertilization detection: After mating, hatching eggs were continuously collected and candled, and the number of days of continuous fertilization and the average fertilization rate under each stress scenario were recorded.

[0086] The specific content of the breeding efficiency statistics is as follows: the total breeding cycle of the two groups for three generations is counted; the consistency between the early screening results at 8 weeks of age and the actual continuous fertilization ability at 40 weeks of age in the experimental group is verified, and the accuracy rate of early screening is calculated; the elimination rate of superior individuals in each generation of the two groups is counted, and the criteria for judging superior individuals are: continuous fertilization days ≥ 12 days and fertilization rate ≥ 85%.

[0087] All test data are expressed as mean ± standard deviation. Independent samples t-tests were used to compare differences between the two groups, and P < 0.05 was considered statistically significant.

[0088] Table 1 shows the comparison data of core performance indicators between the experimental group and the control group under normal conditions.

[0089] Table 1:

[0090]

[0091] Table 2 shows the comparison data of performance indicators between the experimental group and the control group under different environmental stresses.

[0092] Table 2:

[0093]

[0094] Table 3 shows the comparison data of breeding efficiency indicators between the experimental group and the control group.

[0095] Table 3:

[0096]

[0097] Analysis of test results: As shown in the table above, the experimental group had a 64.7% higher average continuous fertilization period and a 10.6% higher average fertilization rate under normal conditions compared to the control group, while the coefficient of variation for fertilization rate decreased by 42.9%, all of which were statistically significant. Under various environmental stress scenarios, the experimental group had significantly higher continuous fertilization period and fertilization rate than the control group, demonstrating stronger environmental adaptability. In terms of breeding efficiency, the experimental group had a 19.4% shorter total breeding cycle across three generations compared to the control group, a 37.3% higher early screening accuracy, and a 37.9% lower average elimination rate of superior individuals per generation, indicating that the method of this invention can significantly improve breeding efficiency and reduce the waste of superior individuals.

[0098] This specific embodiment is merely an explanation of this application and is not intended to limit it. After reading this specification, those skilled in the art can make modifications to this embodiment without contributing any inventive step, but such modifications are protected by patent law as long as they fall within the scope of the claims of this application.

Claims

1. A breeding method for improving the continuous fertilization ability of chickens, characterized in that: Includes the following steps: S1. Construction of individualized physiological baseline: After the flock is transferred to the laying house and before mating begins, a continuous period of environmental stability is selected, and multidimensional physiological time-series data of individual hens in the breeding flock are continuously collected. Based on the data, an individualized multidimensional physiological baseline is established for each hen. At the same time, the semen quality of individual roosters in the breeding flock is periodically monitored to establish a semen quality baseline for individual roosters. S2. Dynamic physiological assessment and homeostasis regulation: During the laying period, the physiological data of individual hens are monitored in real time, and their real-time deviation from the physiological baseline is calculated. Based on the real-time deviation, the reproductive system homeostasis score and continuous fertilization potential index of an individual are output through the first prediction model, and the parameters of the first prediction model are recorded; when the reproductive system homeostasis score decreases and the environmental parameters deviate from the set range, adaptive environmental fine-tuning for that individual is triggered; in parallel, the real-time semen quality index of the rooster individual is updated according to periodic monitoring. S3. Intelligent Mating Decision and Execution: Based on the fertilization potential index of individual hens, dynamically determine when they enter the mating window; when an individual hen is in the mating window, combine the real-time semen quality index of individual roosters, and based on historical mating performance data, recommend roosters or semen for the current hen, and execute the mating operation; record the mating combination information and all-dimensional contextual data of this mating. S4. Reverse selection and generation iteration: Based on historically accumulated individual full-dimensional data, construct a second predictive model that associates genomic information, physiological patterns, environmental data, and continuous fertilization capacity phenotype; save the parameters of the second predictive model; use the second predictive model to perform early prediction and simulation of the continuous fertilization capacity of candidate offspring; based on the early prediction results and combined with genetic background, select individuals with high continuous fertilization potential to form a new generation of breeding population, and repeat steps S1 to S4. S5. Trusted Data Storage: Key feature values ​​generated during the breeding process are extracted daily, packaged into data blocks, and their hash values ​​are calculated for storage; when it is necessary to verify the authenticity of the data, the hash values ​​are compared.

2. The breeding method for improving the continuous fertilization ability of chickens according to claim 1, characterized in that: In step S1, the multidimensional physiological time-series data includes at least the core body temperature data collected at 10-15 min sampling intervals, three-dimensional activity data aggregated at 1 min sampling intervals and through a 5 min window, courtship acceptance posture frequency and active feather preening frequency data obtained based on image recognition and behavior analysis, and heart rate variability index extracted by collecting signals for 3-5 min daily through a photoplethysmography sensor; the continuous environmental stability period is at least 7 consecutive days, with the ambient temperature maintained at 18-22℃ and the relative humidity maintained at 50-65%.

3. The breeding method for improving the continuous fertilization ability of chickens according to claim 1, characterized in that: In step S2, the calculation of the real-time deviation is specifically as follows: using 24 hours as a sliding time window, the standardized deviation score of the observed value of each physiological indicator within the window relative to its individual baseline is calculated, and the deviation scores of all indicators constitute the real-time deviation vector; the reproductive system homeostasis score is a continuous value between 0 and 1; the continuous fertilization potential index is a continuous value between 0 and 1.

4. The breeding method for improving the continuous fertilization ability of chickens according to claim 3, characterized in that: In step S2, the triggering conditions are: the individual's reproductive system homeostasis score decreases by more than 20% from its highest value in the past 24 hours within 2 hours, and its microenvironment temperature is higher than 24℃ or lower than 16℃; the fine-tuning includes initiating a directional breeze with a wind speed of 0.5-1.5m / s for a duration of 30-60min.

5. The breeding method for improving the continuous fertilization ability of chickens according to claim 1, characterized in that: In step S3, the dynamic determination rule for the mating window period is as follows: when an individual's continuous fertilization potential index remains at a level greater than 0.8 for more than 3 consecutive hours, it is determined that it has entered the mating window period; the mating tasks are prioritized and scheduled according to the continuous fertilization potential index value of individuals within the window period.

6. The breeding method for improving the continuous fertilization ability of chickens according to claim 1, characterized in that: In step S3, the step of recommending roosters or semen for the current hen based on historical mating effectiveness data specifically includes: calculating the similarity between the current hen's physiological state vector and the hen state vector in the mating records in the historical database, and filtering out historical records with a similarity greater than 70%; from the filtered records, selecting the top 20% of records with the best continuous fertilization days, and analyzing the characteristics of the roosters or semen used; and combining the current hen's matching recommendation, using the rooster's real-time semen quality index to recommend 1-3 roosters.

7. The breeding method for improving the continuous fertilization ability of chickens according to claim 1, characterized in that: In step S4, the individual full-dimensional data includes the physiological feature vector aggregated weekly during the peak egg production period, genotype data based on single nucleotide polymorphism chips, environmental data, and the phenotypic value of continuous fertilization days determined by egg candling through sampling; the second prediction model is a generative model that can generate predicted physiological patterns and continuous fertilization capacity phenotypic values ​​based on the input genome, historical physiological patterns, and assumed environment.

8. The breeding method for improving the continuous fertilization ability of chickens according to claim 1, characterized in that: In step S4, the early prediction and simulation of the continuous fertilization capacity of candidate offspring specifically involves: for candidate offspring, at 8 weeks of age, simulating their genome based on their pedigree information; calling the second prediction model of their parents to simulate at least three hypothetical environmental stress scenarios to obtain the predicted physiological responses of the parents; and combining the simulated genome of the candidate offspring with the features extracted from the predicted responses of the parents to predict the characteristics of their continuous fertilization potential index curve and the number of days of continuous fertilization under the hypothetical environmental scenarios.

9. The breeding method for improving the continuous fertilization ability of chickens according to claim 8, characterized in that: In step S4, the criteria for selecting individuals based on early prediction results are: in more than 50% of the hypothetical environmental scenarios, their predicted average continuous fertilization potential index is greater than 0.75, and their predicted continuous fertilization days are more than 10% higher than the average of their candidate generation group.

10. The breeding method for improving the continuous fertilization ability of chickens according to claim 1, characterized in that: In step S5, the key feature values ​​include at least the hen's core body temperature, heart rate variability, rooster's real-time semen quality index, first prediction model parameters, second prediction model parameters, mating combination information, and chicken house environmental temperature and humidity data.

Images