Livestock veterinary hybridization data analysis system

Abstract

The invention relates to the technical field of animal husbandry data processing, in particular to an animal husbandry veterinary hybridization data analysis system, which comprises a state transition analysis module for collecting data of a female animal breeding cycle, integrating data of all female animals to construct a transition probability matrix, and predicting a future state transition path of the female animals according to the transition probability matrix, and generating a female animal state transition evaluation result. According to the method and the system, the state change paths of the female livestock in different stages are determined by analyzing the data of each state of the female livestock breeding cycle, accurate prediction and quantitative evaluation of the states in the breeding cycle are realized, and making of the breeding plan is more accurate. By evaluating the residence time and the conversion trend of each stage, the optimal hybridization opportunity is predicted, and the overall breeding efficiency is improved. Based on similarity analysis of behavior data such as activity frequency, body temperature and reproduction times, individual female animals are divided into different behavior groups, and a more targeted reproduction management strategy is made for different groups.

Description

Technical Field

[0001] The present invention relates to the technical field of livestock data processing, and in particular to a livestock veterinary breeding data analysis system. Background Art

[0002] Livestock data processing collects, processes, stores, and analyzes livestock production data to improve the production efficiency, economic benefits, and sustainability of the livestock industry. It includes animal health monitoring, breeding data analysis, nutritional formula optimization, production process tracking, and environmental monitoring, etc. With the development of Internet of Things, big data, and artificial intelligence technologies, livestock data processing can achieve automated data collection and analysis, helping livestock industry managers to grasp the production situation in real time.

[0003] Among them, the livestock veterinary breeding data analysis system collects, integrates, and analyzes breeding-related data, such as the estrus period of female livestock, breeding times, breeding success rate, veterinary records, etc., to help managers better formulate breeding plans and optimize breeding efficiency. The system can not only achieve precise management of the reproductive information of individual animals, but also improve the breeding rate of the overall herd, reduce resource waste, and is an important means for the livestock industry to achieve scientific and modern management.

[0004] In the existing system for managing the reproductive process of female livestock, although it can collect and integrate reproductive data, it is difficult to conduct detailed analysis and prediction of the state within the reproductive cycle, which easily leads to inaccurate judgment of the breeding timing, thereby reducing the breeding success rate. In addition, in terms of paying attention to the individual differences and behavioral characteristics among different female livestock in the prior art, it is difficult to formulate a more targeted breeding plan. Although the system can record basic health information, it is not conducive to the continuous monitoring of physiological parameters and real-time abnormal identification, thus missing some hidden health risks and increasing the possibility of breeding failure. At the same time, due to the difficulty of the system in processing data to capture dynamic changes in a timely manner, the flexibility of breeding management is insufficient, which affects the effective utilization of resources and also reduces the health level and breeding efficiency of female livestock. Summary of the Invention

[0005] The purpose of the present invention is to solve the deficiencies existing in the prior art, and to propose a livestock veterinary breeding data analysis system.

[0006] To achieve the above purpose, the present invention adopts the following technical solution: A livestock veterinary breeding data analysis system includes:

[0007] The state transition analysis module collects data on the reproductive cycle of female livestock, integrates the data of all female livestock to construct a transition probability matrix, predicts the future state transition path of female livestock according to the transition probability matrix, and generates an evaluation result of the female livestock state transition;

[0008] Based on the evaluation results of the female livestock state transition and the behavioral characteristic data of the female livestock, the behavior clustering and grouping module unifies the behavioral characteristic data into feature vectors. By analyzing the distances between the feature vectors, it divides the population according to similarity and obtains the result of the behavioral characteristic population division.

[0009] Based on the physiological information of the female livestock in the result of the behavioral characteristic population division, the fluctuation and anomaly detection module continuously records the physiological data of the female livestock in a specific behavioral group, analyzes the basic fluctuation of the physiological information, and simultaneously identifies the abnormally fluctuating physiological information to generate the analysis result of abnormal physiological fluctuations.

[0010] The breeding data analysis module combines the evaluation results of the female livestock state transition, the result of the behavioral characteristic population division, and the analysis result of abnormal physiological fluctuations, integrates and analyzes the breeding data of the female livestock, selects the optimal breeding time according to the previous breeding success rate data, and generates the recommended breeding information.

[0011] As a further solution of the present invention, the obtaining steps of the constructed transition probability matrix are specifically as follows:

[0012] Based on the estrus, pregnancy, lactation, and recovery period state data of the female livestock in the breeding cycle, record the time and state identifier of the state change of each female livestock, and sort the data in chronological order to obtain the female livestock state sequence data.

[0013] According to the female livestock state sequence data, use the formula:

[0014]

[0015] Calculate the probability P of transitioning from state i to state j ij to obtain the transition probability information between states.

[0016] where N ij is the total number of times the female livestock transitions from state i to state j, N ik is the number of times the female livestock transitions from state i to state k, T k is the average residence time of state k, k is an index variable, m is the total number of states, and σ T is the standard deviation of the residence time of the female livestock in state i.

[0017] According to the transition probability information between states, count the number of times the female livestock transitions from each state to other states, integrate the state transition frequencies, and construct a transition probability matrix based on the state transition data of the female livestock within the breeding cycle.

[0018] As a further solution of the present invention, the obtaining steps of the predicted future state transition path of the female livestock are specifically as follows:

[0019] Based on the said transition probability matrix, analyze the residence time of the female livestock in each state during the reproductive cycle and the possibility of state transition, extract the probability values of each state in the matrix in sequence and conduct data association to obtain the evaluation result of the state transition trend;

[0020] According to the evaluation result of the state transition trend, take the node of the current state as the starting point of the path, analyze the transition probability of the next state, and gradually add it to the transition path node to generate the evaluation result of the female livestock state transition.

[0021] As a further solution of the present invention, the specific steps for obtaining the distance between the analysis feature vectors are as follows:

[0022] According to the evaluation result of the female livestock state transition and the behavioral characteristic data, extract the data of the activity frequency, body temperature, body weight and breeding times of the female livestock, standardize and integrate the data into feature vectors to obtain the feature vector set of the female livestock;

[0023] According to the feature vector set of the female livestock, analyze the distance between the feature vectors, and adopt the formula:

[0024]

[0025] Calculate the Euclidean distance d between female livestock q and female livestock r in all behavioral characteristics qr , and establish a group classification of the behavioral characteristics of the female livestock;

[0026] where, z qt′ and z rt′ are the numerical values of female livestock q and female livestock r on the t'-th behavioral characteristic respectively, s is the number of feature vector dimensions, representing the number of behavioral characteristics, which are currently the activity frequency, body temperature, body weight and breeding times.

[0027] As a further solution of the present invention, the specific steps for obtaining the result of the behavioral characteristic group division are as follows:

[0028] According to the group classification of the behavioral characteristics of the female livestock, extract the behavioral characteristic data in the female livestock group, including the activity frequency, body temperature, body weight, breeding times, conduct statistical analysis on the data, and judge the characteristic differences between the groups to obtain the analysis result of the characteristic differences;

[0029] According to the analysis result of the characteristic differences, determine the behavioral characteristic range of each group, further optimize the grouping, merge the groups with close characteristic ranges into type groups, adjust the boundary characteristics of each type group, and optimize the consistency of the characteristics of the female livestock within the same type group to obtain the result of the behavioral characteristic group division.

[0030] As a further solution of the present invention, the specific steps for obtaining the basic fluctuation condition of the analysis physiological information are as follows:

[0031] Based on the result of the grouping of the behavioral characteristics, extract the physiological information of the body temperature, heart rate, and activity frequency of the female livestock from the group, sort the data in chronological order, monitor the basic fluctuations of the physiological information, and obtain the physiological fluctuation record of the group;

[0032] According to the physiological fluctuation record of the group, use the formula:

[0033]

[0034] Calculate the change value ΔP′ of the y-th item of physiological information y , and obtain the analysis result of the change of the physiological parameter;

[0035] wherein, P′ y,h represents the value of the y-th item of physiological parameter at the current moment h, P′ y,h-1 represents the value of the y-th item of physiological parameter at the previous moment h - 1, D represents the number of days difference between two measurements, W y is the weighting coefficient of the y-th item of physiological parameter, and S y is the seasonal adjustment value of the y-th item of physiological parameter.

[0036] As a further solution of the present invention, the steps for obtaining the physiological information for identifying abnormal fluctuations are specifically as follows:

[0037] According to the analysis result of the change of the physiological parameter, compare the changes of the body temperature, heart rate, and activity frequency recorded daily with the conventional range statistically obtained in the past, judge whether each item of data exceeds the normal fluctuation range, and generate the screening result of abnormal data;

[0038] According to the screening result of the abnormal data, record all the data marked as abnormal in chronological order, count the frequency and time distribution of the occurrence of abnormal fluctuations, check and analyze whether there are periodic characteristics, and generate the analysis result of abnormal physiological fluctuations.

[0039] As a further solution of the present invention, the steps for obtaining the selected optimal breeding timing are specifically as follows:

[0040] Based on the evaluation result of the conversion of the female livestock state, the result of the grouping of the behavioral characteristics, and the analysis result of abnormal physiological fluctuations, match the reproductive state information, behavioral characteristics, and physiological condition data of the female livestock according to individuals and time series, sort out the data within the time point, and analyze the change trends of the body temperature, heart rate, and activity frequency during the reproductive cycle to obtain the reproductive correlation information of the female livestock;

[0041] According to the reproductive correlation information of the female livestock, use the formula:

[0042]

[0043] Calculate the breeding success probability GS of the female livestock population to obtain recommended breeding information;

[0044] Among them, R f is the breeding status score of the f-th female livestock, A f is the health status score of the f-th female livestock, N' represents the total number of female livestock participating in the analysis, G f is the behavioral characteristic adaptability score of the f-th female livestock.

[0045] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0046] In the present invention, by analyzing the data of each state in the breeding cycle of female livestock, the state change path of female livestock in different stages is clarified, accurate prediction and quantitative evaluation of the state within the breeding cycle are realized, and the formulation of the breeding plan is made more precise. By evaluating the residence time and conversion trend of each stage, the best breeding time is predicted, and the overall breeding efficiency is improved. Based on the similarity analysis of behavioral data such as activity frequency, body temperature, and breeding times, individual female livestock are divided into different behavioral groups, and more targeted breeding management strategies are formulated for different groups. By continuously recording the fluctuations of physiological parameters and monitoring in combination with historical thresholds, accurate identification and early warning of abnormal physiological fluctuations are realized. Health problems can be detected early, breeding risks can be reduced, and the stability of the breeding process can be ensured. In addition, by integrating information on breeding status, behavioral characteristics, and physiological conditions, analyzing breeding adaptability and breeding success rate, a decision-making basis is provided for managers, and the overall breeding efficiency of the livestock herd is also improved. Description of the Drawings

[0047] Figure 1 is the system flow chart of the present invention;

[0048] Figure 2 is the flow chart for constructing the transition probability matrix of the present invention;

[0049] Figure 3 is the flow chart for predicting the future state transition path of female livestock of the present invention;

[0050] Figure 4 is the flow chart for analyzing the distance between eigenvectors of the present invention;

[0051] Figure 5 is the flow chart for obtaining the division result of behavioral characteristic groups of the present invention;

[0052] Figure 6 is the flow chart for analyzing the basic fluctuation situation of physiological information of the present invention;

[0053] Figure 7 is the flow chart for identifying abnormal physiological information of fluctuations of the present invention;

[0054] Figure 8 Flow chart for selecting the optimal breeding time for the present invention. Specific embodiments

[0055] In order to make the objectives, technical solutions and advantages of the present invention more clear and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0056] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. are based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as a limitation of the present invention. In addition, in the description of the present invention, the meaning of "a plurality" is two or more unless otherwise specifically defined.

[0057] Please refer to Figure 1 , a breeding data analysis system for animal husbandry and veterinary medicine includes:

[0058] The state transition analysis module collects the estrus, pregnancy, lactation, and recovery state data of the female livestock's reproductive cycle, arranges the state data of each female livestock in the entire reproductive cycle in chronological order as a state sequence, calculates the transition probability between states by analyzing the sequence data of state changes, integrates the data of all female livestock to construct a transition probability matrix, evaluates the residence time and transition possibility of each state according to the transition probability matrix, predicts the future state transition path of the female livestock, and generates the evaluation result of the female livestock state transition;

[0059] The behavior clustering and grouping module unifies the behavioral characteristic data into feature vectors based on the evaluation result of the female livestock state transition and the behavioral characteristic data of the female livestock's activity frequency, body temperature, body weight, and breeding times, calculates the similarity between female livestock by analyzing the distance between feature vectors, divides the groups according to the similarity, divides the female livestock into several groups, analyzes the feature differences of each group, establishes the typed grouping of the female livestock, and obtains the result of the behavioral characteristic group division;

[0060] The fluctuation and anomaly detection module is based on the physiological information of the body temperature, heart rate, and activity frequency of the female livestock in the result of the behavioral characteristic group division, continuously records the physiological data of the female livestock in a specific behavioral group, analyzes the basic fluctuation of the physiological information, calculates the change value of the parameter, compares the change value with the normal range, and refers to the threshold range statistically recorded from past data to identify the abnormally fluctuating physiological information, records the time and frequency of the abnormal fluctuation, and generates the analysis result of the abnormal physiological fluctuation;

[0061] The breeding data analysis module combines the evaluation results of the female livestock status conversion, the classification results of the behavioral characteristic groups, and the analysis results of abnormal physiological fluctuations, integrates and analyzes the breeding data of the female livestock, extracts the key information in the breeding status, behavioral characteristics, and physiological conditions, calculates the suitability and breeding success probability of the female livestock group according to the previous breeding success rate data, selects the optimal breeding time, and generates the recommended breeding information;

[0062] The evaluation results of the female livestock status conversion include the average residence time of each status, the conversion probability between statuses, and the predicted future status conversion path. The classification results of the behavioral characteristic groups include the average activity frequency of each group, the body temperature change range, and the weight fluctuation range. The analysis results of abnormal physiological fluctuations include the physiological parameters of abnormal fluctuations, the occurrence time, and the fluctuation frequency. The recommended breeding information includes the suitability score, the breeding success probability, and the optimal breeding time.

[0063] Please refer to Figure 2 , and the specific steps for obtaining the transition probability matrix are as follows:

[0064] Based on the estrus, pregnancy, lactation, and recovery period status data of the female livestock in the breeding cycle, record the time and status identifier of the status change of each female livestock, sort the data in chronological order, and obtain the female livestock status sequence data;

[0065] According to the different statuses of the female livestock in the breeding cycle, including estrus, pregnancy, lactation, and recovery period data, use data collection devices (such as RFID tags or biosensors) to monitor the status changes of each female livestock in real time, record the specific time and status identifier of each status change, collect all the status data and import it into the Pandas library or SQL database of Python, and sort it in chronological order to generate the complete status sequence of each female livestock. First, clean the imported data to remove duplicate values or missing items to ensure the accuracy and consistency of the data, and then sort the data according to the time stamp to form the time sequence of the female livestock in the entire breeding cycle. For records with abnormal data (such as significantly abnormal status time spans), identify and clean them through anomaly detection methods based on time intervals to ensure the rationality of the status sequence.

[0066] According to the female livestock status sequence data, use the formula:

[0067]

[0068] Calculate the probability P of transitioning from state i to state j ij , and obtain the transition probability information between states;

[0069] Among them, P ijReflects the conversion tendency of female livestock among different reproductive states (such as estrus, pregnancy, lactation, or recovery period), N ij is the total number of times a female livestock converts from state i to state j, representing the conversion frequency of female livestock from state i to state j, used to quantify the actual occurrence frequency of each state conversion. The statistical analysis of the female livestock state data sequence is carried out through the Pandas library in Python, N ik is the number of times a female livestock converts from state i to state k, which is statistically obtained from the state sequence data, T k is the average residence time of state k, where k is an index variable representing all possible state numbers, and m is the total number of possible states;

[0070] T i and T j respectively represent the average residence times of state i and state j, reflecting the average residence duration of female livestock in each reproductive state (such as estrus period, pregnancy period, etc.), helping to analyze the stability and conversion time distribution of female livestock in different states, T i The calculation method is as follows:

[0071]

[0072] In the formula, t in is the residence time of the female livestock in state i for the nth time, and the residence duration of the female livestock in a specific state is captured by the sensor, N i is the total number of times the female livestock appears in state i, which is statistically obtained from the sequence data. n represents the observation number of each residence, used to calculate the sum of all residence times. The average residence time can be obtained using data analysis tools such as SPSS and Excel;

[0073] T j The calculation method is as follows:

[0074]

[0075] In the formula, t jn is the residence time of the female livestock in state j for the nth time, N j is the total number of times the female livestock appears in state j;

[0076] σ T is the standard deviation of the residence time of the female livestock in state i, representing the fluctuation of the residence time between different states, used to measure the consistency of the residence time of the female livestock between different states, reflecting the degree of change in the time distribution during the state conversion of the female livestock. The calculation method is as follows:

[0077]

[0078] For example, in the dataset of the reproductive cycle of female livestock: If the number of conversions from the estrus period (state i) to the pregnancy period (state j) is Nij = 8, the average residence time T in the estrus period (state i) i = 5 days, the average residence time T in the pregnancy period (state j) j = 10 days, the standard deviation σ of the state residence time in the estrus period T = 3, the sum of the residence times of each state multiplied by the total number of times the estrus period transitions to all states

[0079] Calculate the numerator part:

[0080] N ij ×(T i +T j ) = 8×(5 + 10) = 8×15 = 120

[0081] Calculate the denominator part:

[0082]

[0083] Calculate the final probability P ij :

[0084]

[0085] The calculated result P ij ≈0.9756 indicates that after the female animal is in the estrus period, there is approximately a 97.56% probability of transitioning to the pregnancy period.

[0086] According to the transition probability information between states, count the number of times the female animal transitions from each state to other states, integrate the state transition frequencies, and construct a transition probability matrix based on the state transition data of the female animal during the reproductive cycle;

[0087] After integrating the state sequence data of all female animals, a transition probability matrix is ​​constructed to accurately analyze the conversion relationship between each state in the reproductive cycle of female animals. The specific steps are as follows: First, the state sequence of each female animal is analyzed one by one, and each instance of the female animal's conversion from one state to another is recorded. For example, when the female animal enters the pregnancy period from the estrus period, it is recorded as a conversion from "estrus to pregnancy period". The number of times each female animal converts from the estrus period to the pregnancy period, lactation period, or recovery period in the entire reproductive cycle is counted in sequence, and the same information is recorded for the conversion of other states until the complete state conversion data distribution of the female animal in the entire reproductive cycle is obtained. Repeat this process for the state sequence of each female animal, and accumulate the number of conversions between each state one by one. Then, the number of conversions of all female animals is summarized to obtain a total state conversion number matrix. The rows and columns of this matrix represent the four states in the reproductive cycle (such as estrus, pregnancy, lactation, and recovery period), and each element value in the matrix represents the total number of conversions from the corresponding row state to the column state. For example, the first row and first column of the matrix represents the cumulative number of transitions from estrus to pregnancy for all female animals, and the second row and second column represents the cumulative number of transitions from pregnancy to lactation, and so on, filling in all elements in the matrix until the construction of the entire state transition number matrix is ​​completed. Finally, bring in the transition probability P calculated in paragraph 2 ij , convert the transition frequency of each state into the corresponding probability matrix element. For example, if the transition probability P from estrus to pregnancy has been calculated in paragraph 2 ij =0.9756, then fill this value into the corresponding position of the transition probability matrix to represent the probability of transitioning from estrus to pregnancy. Fill in all the calculated transition probabilities in sequence so that each element of the matrix represents the specific transition probability from one state to another. The resulting transition probability matrix provides a detailed structure of the state transitions in the reproductive cycle of female livestock, in which each element quantifies the probability of transitioning from one state to another.

[0088] See also Figure 3 , the specific steps for obtaining the predicted future state transition path of the female livestock are:

[0089] Based on the transition probability matrix, the residence time of each state and the possibility of state transition in the reproductive cycle of the female livestock are analyzed, and the probability value of each state in the matrix is ​​extracted in turn and data association is performed to obtain the state transition trend evaluation result;

[0090] By constructing the transition probability matrix, further analyze the average residence time and conversion possibility of the female livestock in each state during the reproductive cycle. First, extract the self-transition probability of each state from the matrix, that is, the probability that the female livestock continues to maintain the current state and does not convert to other states within one cycle. For the average residence time of each state, it can be calculated based on the self-transition probability. Specifically, the higher the self-transition probability, the longer the residence time of the female livestock in this state; when the self-transition probability is lower, the residence time is shorter. According to this relationship between probability and time, analyze the self-transition probability of each state in turn to calculate its average residence time. After obtaining the average residence time of each state, combine it with the transition probability between states to further refine the conversion possibility of the female livestock. Specifically, obtain the probability value of each state in the transition probability matrix converting to other states, and combine it with the average residence time of the current state, that is, the residence time of each state corresponds to the magnitude of its transition probability. For example, if the residence time in the estrus period is 5 days and the transition probability is 0.2, then there is a 20% probability of transitioning to the next state within 5 days; and so on, evaluate the residence time and transition probability of other states item by item, and finally obtain the distribution data of the residence time and conversion possibility of each state within the cycle.

[0091] According to the evaluation results of the state transition trend, take the node of the current state as the starting point of the path, analyze the transition probability of the next state, and gradually add it to the transition path node to generate the evaluation result of the female livestock state transition;

[0092] Based on the transition probabilities obtained from the transition probability matrix and the residence times in each state, predict the state transition path of the female livestock in the future reproductive cycle. First, select the current state as the starting point, read the row of this state from the transition probability matrix to obtain the transition probability distribution to other states. For each possible transition state, extract its corresponding probability value and record it, and at the same time establish a path node. Then, in the next step, use the current node state as the new starting state and search for the transition probability distribution of this state in the matrix again. By means of cumulative chain probability, gradually calculate the probability of transferring to the next state at each step and update the transition path of each state node. To predict the multi-step transition path, introduce a set prediction step length for recursion. Each time, update the state result of the previous step and substitute it into the transition matrix to generate new transition probabilities. During the entire path prediction process, according to the level of cumulative probability, screen out the path with the highest transition possibility, and combine the residence times in each state to judge the duration of the female livestock in each state. Continuously predict the next step of each state by recursively calling the rows and columns of the transition matrix, and finally generate a path containing multiple transition nodes. This path is the transition prediction sequence of the female livestock in the future reproductive cycle, and provides probabilistic evaluation data on the future state changes of the female livestock in the reproductive cycle through the cumulative probabilities of multiple state nodes, providing a reliable prediction basis for the reproductive management and state monitoring of the female livestock.

[0093] Please refer to Figure 4 , and the specific steps for obtaining the distance between eigenvectors are as follows:

[0094] According to the female livestock state transition evaluation results and behavioral characteristic data, extract the data of the activity frequency, body temperature, body weight, and number of reproductive times of the female livestock, standardize and integrate the data into eigenvectors to obtain the eigenvector set of the female livestock;

[0095] Combine the behavioral characteristic data of female livestock (including activity frequency, body temperature, body weight, and number of reproductions) to establish a feature vector. Convert each item of characteristic data into a unified numerical representation. First, extract the behavioral characteristic data from the data acquisition system. Record the activity frequency data according to the number of daily activities, ensure the data is clear and without duplicates, and extract the activity mean value. The body temperature data is obtained through a body temperature sensing device. Represent the daily body temperature by the average of the maximum and minimum body temperatures per day, and further average the body temperature records within the cycle. The body weight data uses the data set of each weighing, and represents the characteristic value by the monthly average body weight of each female livestock. The number of reproductions data is directly extracted from the historical records as the cumulative value. To unify these characteristic values, input the data into a standardization processing system and adopt the maximum-minimum normalization method to map each feature to the range between 0 and 1. For example, use the MinMaxScaler in Python for normalization processing. In this way, the obtained feature vector includes the standardized activity frequency, body temperature, body weight, and number of reproductions, forming a unified feature vector format.

[0096] According to the feature vector set of female livestock, analyze the distance between feature vectors, using the formula:

[0097]

[0098] Calculate the Euclidean distance d between female livestock q and female livestock r in all behavioral characteristics (activity frequency, body temperature, body weight, number of reproductions) qr , and establish a group classification of the behavioral characteristics of female livestock;

[0099] where d qr is used to quantify the similarity between two female livestock in these behavioral characteristics. The smaller the distance value, the closer the two are in each behavioral characteristic. z qt′ and z rt′ are the numerical values representing female livestock q and female livestock r in the t'-th behavioral characteristic respectively, including activity frequency, body temperature, body weight, and number of reproductions. The activity frequency is recorded by an activity monitoring device, and the mean value of the daily activity times is used as the characteristic value. The body temperature is extracted from the body temperature monitoring device and averaged over the cycle after taking the average of the daily maximum and minimum values. The body weight is regularly recorded by a weighing device, and the average body weight within the cycle is used as the characteristic. The number of reproductions is extracted from the breeding records as the cumulative value. s is the number of feature vector dimensions, representing the number of behavioral characteristics. Currently, there are four characteristics: activity frequency, body temperature, body weight, and number of reproductions.

[0100] For example, if the standardized feature vectors of female livestock q and female livestock r are respectively and

[0101] Calculate the square of the difference of each feature:

[0102] Activity frequency: (0.6 - 0.5) 2 = 0.01; Body temperature: (0.7 - 0.6) 2 = 0.01; Body weight: (0.5 - 0.4) 2 = 0.01; Reproduction times: (0.8 - 0.7) 2 = 0.01.

[0103]

[0104] The results show that the similarity distance of female livestock q and r in behavioral characteristics such as activity frequency, body temperature, body weight, and reproduction times is 0.2. Based on the distance d qr , using population division methods such as hierarchical clustering or Kmeans clustering, group female livestock with high similarity into the same population. First, by setting a distance threshold, compare all female livestock pairwise, record each d qr value, and then divide female livestock with smaller distances into the same initial group according to the distance threshold. In hierarchical clustering, starting from the initial groups, gradually merge groups with high similarity until an appropriate number of populations are formed. In Kmeans clustering, based on the feature vectors of each female livestock, select the center points and perform iterative optimization, and assign the female livestock to the population represented by the center point with the closest distance. The final division result forms several populations, each population consisting of female livestock with relatively high similarity in behavioral characteristics, providing a clear grouping structure for subsequent feature difference analysis.

[0105] Please refer to Figure 5 , the specific steps for obtaining the division result of the behavioral characteristic population are as follows:

[0106] According to the population grouping of female livestock behavioral characteristics, extract the behavioral characteristic data in the female livestock population, including activity frequency, body temperature, body weight, and reproduction times, perform statistical analysis and processing on the data, and judge the characteristic differences between populations to obtain the characteristic difference analysis result;

[0107] First, extract the behavioral characteristic data of each female animal from the divided groups, including activity frequency, body temperature, weight, and number of reproductions, and import these data into statistical analysis software, such as SPSS or the pandas library in Python, to generate descriptive statistical information of each characteristic, including mean, standard deviation, and distribution, so as to clarify the overall characteristic profile of each group; then, use variance analysis to test the significant difference of the behavioral characteristic data of different groups. The specific method is to analyze the mean difference between groups for each characteristic item, and record the significance index of each characteristic between groups to verify whether there is a significant difference between each behavioral characteristic between groups. For the correlation analysis between characteristics, use the Pearson correlation coefficient calculation method to test the correlation between each characteristic combination one by one, so as to understand the degree of correlation between activity frequency and body temperature and weight, and determine the dependence between characteristics. Finally, summarize the results of variance analysis and correlation analysis to form a characteristic difference analysis report for each group, and clarify the unique characteristic structure of each group.

[0108] According to the results of feature difference analysis, the range of behavioral features of each group is determined, and the grouping is further optimized by merging groups with similar feature ranges into type groups, adjusting the boundary features of each type group, and optimizing the consistency of female livestock features within the same type group to obtain the results of behavioral feature group division;

[0109] First, according to the characteristic differences of each group obtained from the above analysis results, the numerical range of each characteristic is set, and the groups with similar characteristic ranges are merged into one type group to ensure that the female livestock in the same type group maintains a high degree of consistency in behavioral characteristics; then, the hierarchical clustering method or Kmeans clustering method is used to further optimize the group, by selecting the center point of the type group and performing multiple iterations of optimization, setting clustering parameters in Python's sklearn library, using the initial type center point to perform secondary distribution on the existing group, and gradually assigning female livestock with closer characteristic values ​​to the corresponding type group; in each round of iteration, the center point of each type group is updated, and the range of boundary characteristics is adjusted to ensure that each female livestock matches the core characteristics of the type group to which it belongs to the greatest extent. Finally, the typed grouping structure of female livestock is formed, so that the structure of each group is clear and the characteristics are consistent.

[0110] See also Figure 6 ,The specific steps for obtaining the basic fluctuation of physiological information are as follows:

[0111] Based on the results of behavioral characteristic group division, the physiological information of body temperature, heart rate and activity frequency of female animals is extracted from the group, the data is sorted in chronological order, the basic fluctuation of physiological information is observed, and the physiological fluctuation record of the group is obtained;

[0112] First, continuously collect the body temperature, heart rate, and activity frequency of each female livestock through an electronic monitoring device and automatically upload them. Ensure that the data collected each time is bound to the unique identifier of the female livestock for long-term tracking. When exporting data through the system interface, organize it in chronological order to generate a continuous time series data table. Subsequently, the platform processes the data, including data cleaning to eliminate outliers and missing data, and grouping and classifying the physiological fluctuations in different time periods. Through the comparison of physiological characteristics with historical benchmark values, the system screens out individuals or groups with obvious fluctuations and marks the female livestock individuals with potential health risks. Finally, based on the daily and periodic fluctuation analysis, visually present the monitoring results.

[0113] According to the physiological fluctuation records of the group, use the formula:

[0114]

[0115] Calculate the change value ΔP′ of the y-th physiological information y , and obtain the analysis result of the change of physiological parameters;

[0116] Among them, ΔP′ y is used to capture the change of a specific physiological parameter between different time nodes, P′ y,h represents the value of the y-th physiological parameter at the current moment h, collected by the monitoring device, P′ y,h-1 represents the value of the y-th physiological parameter at the previous moment h - 1, D represents the number of days difference between two measurements, directly calculated by comparing the date stamps of the two sampling times, W y is the weighting coefficient of the y-th physiological parameter, used to balance the importance of different parameters, obtained from historical data analysis. The specific process includes performing multiple regression analysis on each physiological parameter or using a correlation analysis tool (such as SPSS) to calculate its importance under different conditions, and setting corresponding weights according to the influence of the parameter on the success of breeding. For parameters with higher correlation, such as the importance of heart rate in estrus cycle prediction, a higher weight W y can be given to improve the accuracy of the result, S y is the seasonal adjustment value of the y-th physiological parameter, obtained from historical data analysis. Use time series analysis methods or historical mean comparison (such as Python's time series analysis tool) to determine the benchmark levels of each parameter in different seasons, and quantify these benchmark differences as the adjustment value S y , to ensure that the calculation result can reflect the normal fluctuations in different seasons and reflect the influence of seasonality on physiological data.

[0117] For example, if the recorded body temperature, heart rate, and activity frequency of the female livestock at the current moment are P′ temp,h = 39.0 °C, P′ hr,h = 80 bpm, P′act,h = 50 times per day, the previous record was P' temp,h-1 = 37.8 °C, P' hr,h-1 = 75 bpm, P' act,h-1 = 48 times per day, measurement interval D = 2 days, weight parameter is W temp = 1.1, W hr = 1.05, W act = 1.2, seasonal adjustment value is S temp = 0.2, S hr = 0.1, S act = 0.15.

[0118] Calculation of body temperature change:

[0119]

[0120] Calculation of heart rate change:

[0121]

[0122] Calculation of activity frequency change:

[0123]

[0124] The results show that the body temperature change is 0.86 °C, the heart rate change is 2.725 bpm, and the activity frequency change is 1.35 times per day.

[0125] Please refer to Figure 7 , and the specific steps for obtaining physiological information to identify abnormal fluctuations are as follows:

[0126] Based on the analysis results of the changes in physiological parameters, compare the daily recorded changes in body temperature, heart rate, and activity frequency with the conventional ranges statistically obtained in the past to determine whether each data item exceeds the normal fluctuation range, and generate a screening result for abnormal data;

[0127] Based on the change value ΔP' of physiological information y , compare it with the conventional range and perform abnormal detection with reference to the threshold range statistically obtained from past data. Specifically, the normal range of body temperature is set to 37.5 °C to 39.0 °C, the normal range of heart rate is 60 bpm to 80 bpm, and the daily fluctuation value of activity frequency is between 45 times per day and 55 times per day. The reference values are obtained from the data statistics of the past 30 days. For example, the reference average value of body temperature is 38.2 °C, the reference average value of heart rate is 70 bpm, and the reference average value of activity frequency is 50 times per day. Use the mean of each physiological parameter plus or minus 3 times the standard deviation as the detection threshold range for determining abnormalities. During the analysis process, ΔP' temp = 0.86, ΔP' hr = 2.725 and ΔP'act The change value of 1.35 is compared with the corresponding physiological reference value. Body temperature analysis: The current body temperature change is 0.86 °C. When the reference body temperature is 38.2 °C, this change raises the body temperature to 39.06 °C, slightly higher than the upper limit of the normal range (39.0 °C), which is determined to be a slight abnormal fluctuation. Heart rate analysis: The heart rate change is 2.725 bpm, raising the reference heart rate from 70 bpm to 72.725 bpm, still within the normal range and not determined to be abnormal. Activity frequency analysis: The activity frequency change is 1.35 times per day, increasing the reference activity frequency from 50 times per day to 51.35 times per day, within the normal fluctuation range. During the comparison process, when the result caused by the change value exceeds the normal range, the system marks this data as abnormal data and records the reason for the abnormality. Based on the above results, the change in body temperature exceeds the threshold range and should be further observed, while other physiological parameters are within the normal range.

[0128] According to the screening results of abnormal data, all data marked as abnormal are recorded in chronological order, the frequency and time distribution of abnormal fluctuations are statistically analyzed, and whether there are periodic characteristics is checked and analyzed to generate the analysis results of abnormal physiological fluctuations;

[0129] After identifying the physiological information of abnormal fluctuations, record the time and frequency of abnormal fluctuations to generate the final analysis results of abnormal physiological fluctuations. First, record each body temperature data marked as abnormal in chronological order and indicate the specific timestamp. For example, record that the body temperature at 10:00 on October 27 is 39.06 °C, and this time point is the marked point of abnormal fluctuations. Then, conduct a frequency statistics on abnormal fluctuations over consecutive days to analyze whether it has periodic characteristics. For example, if body temperature fluctuations occur at the same time period every day, it may be related to specific behaviors or environmental factors. In addition, accumulate the number of times the daily body temperature exceeds the normal range to evaluate the severity and frequency trend of the abnormality. For example, if abnormal body temperatures are recorded every day for 3 consecutive days, on-site intervention or inspection may be required. The final generated abnormal analysis results will include the time series, frequency distribution, and potential periodic patterns of body temperature abnormalities, providing a decision-making basis for veterinarians to ensure that the health status of female livestock is within an appropriate range and avoid affecting the success rate of breeding.

[0130] Please refer to Figure 8 for the specific steps to obtain the optimal breeding time:

[0131] Based on the evaluation results of the female livestock state transition, the results of the behavioral characteristic group division, and the analysis results of abnormal physiological fluctuations, match the reproductive state information, behavioral characteristics, and physiological condition data of the female livestock according to individuals and time series, organize the data within the time point, and analyze the change trends of body temperature, heart rate, and activity frequency during the reproductive cycle to obtain the reproductive-related information of the female livestock;

[0132] First, arrange the behavioral characteristic data of female livestock (such as activity frequency, food intake, movement trajectory, etc.) in chronological order of daily records, and establish an aligned sequence with physiological data such as body temperature and heart rate. Then, use time series analysis methods, such as the sliding window technique, to calculate the change trends between behavioral and physiological data in intervals of 7 days or 30 days respectively, and observe their synchronization or lag relationship. For example, when the body temperature rises, whether there is an increase in heart rate fluctuation or a decrease in activity frequency, etc. Next, use the dynamic time warping method to compare the fluctuation patterns of behavioral characteristics and physiological data, find similar patterns among different individuals, and judge the degree of influence of health status on reproductive behavior. The similar patterns among different individuals include: Body temperature and heart rate synchronous fluctuation pattern: During the estrus cycle, the time period of rising body temperature is accompanied by the same change trend of rising heart rate. These individuals usually reach the appropriate breeding state within 23 days of the body temperature peak, and individuals matching this pattern are more suitable for centralized breeding arrangements. Reverse correlation pattern between activity frequency and reproductive cycle: Some individuals have a decrease in activity frequency around the ovulation period and an increase in activity frequency during the non-reproductive period. Identifying individuals with similar behavioral patterns helps monitor the progress of reproductive status based on the activity level. Abnormal physiological fluctuation pattern: Some individuals also show abnormal fluctuations in body temperature or heart rate outside the estrus period, such as a long-term increase in body temperature or an abnormal acceleration of heart rate. This type of pattern indicates that health status may affect reproductive behavior and requires further health assessment. Pattern of consistent reproductive behavior and environmental response: Different female livestock individuals show consistent physiological and behavioral changes under the same environmental changes (such as temperature changes, water intake fluctuations), for example, high temperature weather causes the body temperature and heart rate of multiple female livestock to rise synchronously and activity to decrease. The discovery of this type of pattern helps evaluate the impact of environmental factors on reproductive behavior. For the time periods with abnormal fluctuations, further analyze whether the behavioral pattern does not match the reproductive status, such as a decrease in activity frequency during the body temperature rising period, to reveal potential health risks. Finally, based on the analysis results of these interaction patterns, provide accurate reference data for predicting reproductive success rate and selecting breeding timing.

[0133] According to the reproductive correlation information of female livestock, use the formula:

[0134]

[0135] Calculate the reproductive success probability GS of the female livestock group to obtain the recommended breeding information;

[0136] Among them, GS is used to evaluate whether the group is at an appropriate breeding timing, and R f is the reproductive status score of the f-th female livestock, which is determined by the monitoring results of body temperature, heart rate, and ovulation time. For example, when the body temperature reaches the peak and the heart rate rises, it indicates that the ovulation period has arrived. This score is obtained through logical regression analysis of historical data and monitoring tools such as Excel. A fis the health status score of the f-th female livestock, determined based on the analysis of abnormal physiological fluctuations. For example, higher scores are obtained when body temperature and heart rate fluctuations are within the healthy range. N' represents the total number of female livestock participating in the analysis, G f is the behavioral characteristic adaptability score of the f-th female livestock, reflecting whether the behavioral characteristics match those in the estrus period. These behaviors are correlated with the reproductive success rate through multiple regression analysis and calculation, mainly including daily activity frequency, changes in food intake, movement trajectories, and mating performance. Behavioral data are collected daily through monitoring devices.

[0137] To calculate the behavioral characteristic adaptability score G f for each female livestock, a regression model needs to be established to correlate the behavioral data with the reproductive success rate. The specific operations are as follows:

[0138] Collect the daily behavioral data of all female livestock, such as daily activity frequency, food intake, and movement trajectories, and pair these data with the historical reproductive success rate.

[0139] Use the statsmodels in Python or SPSS tools to establish a multiple regression model, setting the target variable as the reproductive success rate and the independent variables as the various behavioral data.

[0140] Regression model setting:

[0141] Y = β 0 + β 1 X 1 + β 2 X 2 + β 3 X 3 + ε

[0142] where Y represents the reproductive success rate, X 1 , X 2 , X 3 represent the activity frequency, food intake, and movement trajectories respectively, β 0 is the intercept, β 1 , β 2 , β 3 are the regression coefficients of their respective behavioral characteristics, and ε is the error term.

[0143] Model fitting and coefficient calculation: Substitute the collected behavioral data into the regression model for fitting, and calculate the regression coefficients of each behavioral characteristic through the least squares method. The magnitude of the regression coefficient reflects the degree of influence of the behavioral characteristic on the reproductive success rate.

[0144] Calculation of behavioral adaptability score: According to the fitting results, calculate the behavioral adaptability score G f for each female livestock. The formula is as follows:

[0145] G f = β1 X 1f +β 2 X 2f +β 3 X 3f

[0146] Among them, X 1f , X 2f , X 3f respectively represent the data of the activity frequency, food intake and movement trajectory of the f-th female livestock.

[0147] Result verification and application: According to the results of the regression model, calculate the behavioral characteristic score G of each female livestock f and then use it for the calculation of overall reproductive suitability.

[0148] If there are 3 female livestock in the group, the scores are as follows: The first female livestock: R 1 = 0.8, G 1 = 0.9, A 1 = 0.95; The second female livestock: R 2 = 0.7, G 2 = 0.85, A 2 = 0.9; The third female livestock: R 3 = 0.9, G 3 = 0.8, A 3 = 0.85; Total number of female livestock: N' = 3.

[0149] Substitute into the formula:

[0150]

[0151] This result shows that the overall reproductive success probability of the current female livestock group is 61%. Assuming that this result is compared with the historical reproductive success rate (such as 55%), it can be judged whether the current group is in the best reproductive state. If it is higher than the benchmark value, it is prompted that the current group is suitable for breeding operations; if it is lower than the benchmark value, there may be health or behavioral abnormalities that require further intervention. This result will provide support for reproductive management decisions to ensure breeding at the best time, improve the reproductive success rate and optimize the reproductive management strategy.

[0152] The above is only a preferred embodiment of the present invention, and it does not limit the present invention in other forms. Any person skilled in the art may use the disclosed technical content to make changes or modifications into equivalent embodiments with equivalent changes and apply them to other fields. However, as long as it does not depart from the technical content of the technical solution of the present invention, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present invention still belong to the protection scope of the technical solution of the present invention.

Claims

1. A livestock and veterinary breeding data analysis system, characterized in that: The system comprises: The state transition analysis module collects data on the reproductive cycle of the female livestock, integrates the data of all the female livestock to construct a transition probability matrix, predicts the future state transition path of the female livestock based on the transition probability matrix, and generates the evaluation results of the state transition of the female livestock; The behavior clustering grouping module unifies the behavior characteristic data into a feature vector based on the state transition evaluation result of the maternal livestock and the behavior characteristic data of the maternal livestock, analyzes the distance between the feature vectors, divides the groups according to the similarity, and obtains the behavior characteristic group division result; The fluctuation abnormality detection module continuously records the physiological data of the female animals in the specific behavior group based on the physiological information of the female animals in the behavior characteristic group division result, analyzes the basic fluctuation of the physiological information, and identifies the physiological information of abnormal fluctuation, and generates abnormal physiological fluctuation analysis results; The breeding data analysis module combines the results of the mother animal's state transition assessment, the results of the behavioral characteristic group division and the abnormal physiological fluctuation analysis to integrate and analyze the breeding data of the mother animal, selects the best breeding time based on the previous breeding success rate data, and generates recommended breeding information.

2. The livestock and veterinary breeding data analysis system according to claim 1, characterized in that: The acquisition steps of constructing the transition probability matrix are specifically as follows: Based on the status data of estrus, pregnancy, lactation and recovery period of the female livestock in the reproductive cycle, the time and status identification of each female livestock's status change are recorded, and the data are sorted in chronological order to obtain the status sequence data of the female livestock; According to the state sequence data of the maternal animal, the formula is used: Calculate the probability P of transitioning from state i to state j ij , get the transition probability information between states; Among them, N ij is the total number of times the female animal changes from state i to state j, N ik is the number of transitions from state i to state k of the female animal, T k is the average stay time in state k, k is the index variable, m is the total number of states, σ T is the standard deviation of the time the female animal stays in state i; According to the transition probability information between the states, the number of times the female animal transitions from each state to other states is counted, and the state transition frequencies are integrated. According to the state transition data of the female animal during the breeding cycle, a transition probability matrix is ​​constructed.

3. The livestock and veterinary breeding data analysis system according to claim 2, characterized in that: The steps for obtaining the predicted future state transition path of the female livestock are specifically as follows: Based on the transition probability matrix, the residence time of each state of the female livestock during the breeding cycle and the possibility of state transition are analyzed, the probability value of each state in the matrix is ​​extracted in turn and data association is performed to obtain the state transition trend evaluation result; According to the state transition trend evaluation result, the node of the current state is used as the path starting point, the transition probability of the next state is analyzed, and the transition path nodes are gradually added to generate the mother animal state transition evaluation result.

4. The livestock and veterinary breeding data analysis system according to claim 3, characterized in that: The steps of obtaining the distance between the analysis feature vectors are specifically as follows: Extracting the activity frequency, body temperature, weight and reproduction number data of the female livestock according to the state transition evaluation results and behavioral characteristic data of the female livestock, standardizing the data and integrating them into feature vectors to obtain a feature vector set of the female livestock; According to the characteristic vector set of the female livestock, the distance between the characteristic vectors is analyzed, and the formula is used: Calculate the Euclidean distance d between all behavioral characteristics of female animal q and female animal r qr , establish groupings based on behavioral characteristics of female animals; Among them, z qt′ and z rt′ are the values ​​of the behavioral characteristics of the female livestock q and female livestock r in the t′th item, respectively, and s is the dimension of the feature vector, which represents the number of behavioral characteristics, currently activity frequency, body temperature, weight and number of reproductions.

5. The livestock and veterinary breeding data analysis system according to claim 4, characterized in that: The steps for obtaining the behavior characteristic group division result are specifically as follows: According to the grouping of the maternal livestock behavioral characteristics, the behavioral characteristic data in the maternal livestock group are extracted, including activity frequency, body temperature, weight, and number of reproductions, and the data are statistically analyzed and processed, and the characteristic differences between the groups are determined to obtain characteristic difference analysis results; Based on the results of the characteristic difference analysis, the behavioral characteristic range of each group is determined, and the grouping is further optimized by merging groups with similar characteristic ranges into type groups, adjusting the boundary characteristics of each type group, and optimizing the consistency of the characteristics of the female animals within the same type group to obtain the behavioral characteristic group division results.

6. The livestock and veterinary breeding data analysis system according to claim 5, characterized in that: The steps for obtaining the basic fluctuation of the physiological information are specifically as follows: Based on the behavioral characteristic group division results, extract the physiological information of body temperature, heart rate and activity frequency of the female animals from the group, organize the data in chronological order, monitor the basic fluctuation of the physiological information, and obtain the physiological fluctuation record of the group; Based on the physiological fluctuation records of the group, the formula is used: Calculate the change value ΔP′ of the yth item of physiological information y , obtain the analysis results of changes in physiological parameters; Among them, P′ y,h represents the value of the yth physiological parameter at the current time h, P′ y,h-1 represents the value of the yth physiological parameter at the previous moment h-1, D represents the difference in days between the two measurements, and W y is the weighting coefficient of the yth physiological parameter, S y is the seasonally adjusted value of the yth physiological parameter.

7. The livestock and veterinary breeding data analysis system according to claim 6, characterized in that: The step of obtaining the physiological information for identifying abnormal fluctuations is specifically as follows: According to the analysis results of the changes in the physiological parameters, the changes in the daily recorded body temperature, heart rate and activity frequency are compared with the normal range of previous statistics to determine whether each data exceeds the normal fluctuation range and generate a screening result for abnormal data; According to the screening results of the abnormal data, all data marked as abnormal are recorded in chronological order, the frequency and time distribution of abnormal fluctuations are counted, and whether there are periodic characteristics is checked and analyzed to generate abnormal physiological fluctuation analysis results.

8. The livestock and veterinary breeding data analysis system according to claim 7, characterized in that: The steps for obtaining the optimal breeding time are specifically as follows: Based on the results of the maternal livestock state transition assessment, the results of the behavioral characteristic group division, and the results of the abnormal physiological fluctuation analysis, the reproductive state information, behavioral characteristics, and physiological condition data of the maternal livestock are matched by individuals and time series, the data within the time point are sorted, and the changing trends of body temperature, heart rate, and activity frequency during the reproductive cycle are analyzed to obtain the maternal livestock reproduction-related information; According to the breeding related information of the female livestock, the formula is used: Calculate the reproductive success probability GS of the female livestock group and obtain the recommended breeding information; Among them, R f is the reproductive status score of the f-th female animal, A f is the health status score of the f-th female animal, N′ represents the total number of female animals involved in the analysis, G f is the adaptability score of the behavioral characteristics of the f-th female animal.

Images