CN-122022576-A - Intelligent breeding management method and system based on animal vital sign data

CN122022576ACN 122022576 ACN122022576 ACN 122022576ACN-122022576-A

Abstract

The invention provides an intelligent breeding management method and system based on animal vital sign data, and relates to the field of intelligent breeding management. The method comprises the steps of receiving and analyzing animal electronic ear tag signals through a breeding management system, marking data abnormal events, tracking animal individual growth performance influenced by the abnormal events, calculating deviation from expected growth performance, and carrying out association analysis with group benefit indexes to generate an economic benefit report by combining breeding resource consumption, performance deviation and preset economic parameters to quantify economic influence caused by the abnormal events. The invention establishes a complete quantitative evaluation system from abnormal events to individual performance deviation and then to economic influence. The invention can provide visual and operable decision basis for the manager, pertinently take measures and optimize the cultivation management strategy, thereby realizing accurate resource management and remarkably improving the overall production efficiency and economic benefit of the farm.

Inventors

PAN HAIHAI
LUO YUANMING
CHEN XIQI

Assignees

无锡富华物联科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260128

Claims (10)

1. An intelligent breeding management method based on animal vital sign data is characterized by comprising the following steps: Receiving and analyzing animal electronic ear tag signals through a breeding management system, marking animal electronic ear tag signal data abnormal events, and recording breeding resource consumption related to the animal electronic ear tag signal data abnormal events, wherein the animal electronic ear tag signal data abnormal events comprise electronic ear tag signal intensity abnormality, missed reading, reading delay or repeated recording; Tracking individual growth performance of animals corresponding to the animal electronic ear tag signal data abnormal event based on the animal electronic ear tag signal data abnormal event, and calculating performance deviation between the individual growth performance and expected growth performance; determining economic influence caused by abnormal events of the animal electronic ear tag signal data based on the consumption of the culture resources, the performance deviation and preset economic parameters; And carrying out association analysis on economic influence caused by the abnormal event of the animal electronic ear tag signal data and the group benefit index to generate an economic benefit report.
2. The method of claim 1, wherein the step of receiving and analyzing the animal electronic ear tag signal via the farming management system, and marking the animal electronic ear tag signal data for abnormal events, comprises: receiving the animal electronic ear tag signals through a radio frequency identification reader arranged at the resource points in the animal breeding stations, and monitoring the access condition of the animals at the resource points in the animal breeding stations through a presence sensor arranged at the entrance of the resource points in each animal breeding station; Analyzing the received animal electronic ear tag signal, analyzing signal intensity data of the animal electronic ear tag signal, and if the signal intensity data is lower than a preset signal intensity threshold value in a first preset time range, marking an animal electronic ear tag signal data abnormal event with abnormal electronic ear tag signal intensity; If the presence sensor detects that an animal enters a resource point in an animal culturing station and the residence time exceeds the preset minimum interaction time, the radio frequency identification reader does not receive an animal electronic ear tag signal within the animal residence time range, and an animal electronic ear tag signal data abnormal event which is missed is marked; If the radio frequency identification reader receives the animal electronic ear tag signal and the presence sensor detects that an animal enters a resource point in an animal breeding station, a time interval exists between the animal electronic ear tag signal and the resource point, and the time interval is larger than a first preset time interval, an animal electronic ear tag signal data abnormal event with delayed reading is marked; If the radio frequency identification reader receives animal electronic ear tag signals corresponding to the same animal twice or more in a second preset time interval, or the presence sensor senses sensing signals corresponding to the same animal entering a resource point in the same animal breeding station twice or more in a third preset time interval, the repeatedly recorded animal electronic ear tag signal data abnormal event is marked.
3. The method for intelligent farming management based on animal vital sign data according to claim 2, wherein recording the farming resource consumption associated with the animal electronic ear tag signal data anomaly event comprises: for each marked animal electronic ear tag signal data abnormal event, synchronously counting and recording the corresponding resource point-allocated culture resource consumption in the animal culturing station within a second preset time range before and after the occurrence of the animal electronic ear tag signal data abnormal event, The cultivation resource consumption comprises feed resource consumption and water resource consumption; the feed resource consumption is obtained by measuring the weight of the fed feed through a weighing sensor integrated on the feed distributor; the water resource consumption is obtained by measuring the drinking volume by means of a flowmeter integrated in the drinking device.
4. The intelligent farming management method based on animal vital sign data of claim 1, wherein individual growth performance of animals corresponding to abnormal events of the animal electronic ear tag signal data includes weight gain performance and feed conversion efficiency performance; The weight gain performance is obtained by periodically measuring and recording the weight of animals through an automatic weighing platform; The feed conversion efficiency performance is obtained by acquiring the total feed consumption of each animal in a third preset time range and the weight increase of each animal in the third preset time range and calculating the total feed consumption; The expected growth performance includes an expected weight gain performance and an expected feed conversion efficiency performance; the expected weight gain performance is obtained based on a Gompertz growth curve constructed by animal varieties and age of day, and the Gompertz growth curve represents daily weight change conditions of animals of corresponding varieties under ideal culture conditions; the expected feed conversion efficiency performance is obtained by calculating the average feed conversion efficiency performance of all animals of the same batch and variety in real time.
5. The intelligent farming management method according to claim 4, wherein the performance deviation comprises a weight performance deviation and a feed conversion efficiency performance deviation, wherein, The weight performance bias is a difference between the weight gain performance and the expected weight gain performance; the feed conversion efficiency performance bias is the difference between the feed conversion efficiency performance and the expected feed conversion efficiency performance.
6. The method for intelligent farming management based on animal vital sign data according to claim 1, wherein in the step of analyzing animal electronic ear tag signals, if a problem of failure in verification of rfid reader information or data format error occurs, the following steps are performed to confirm the identity of the animal: activating a visual sensor arranged at a resource point in an animal breeding station, and acquiring animal image data entering the current resource point through the visual sensor; performing data preprocessing and feature extraction on the animal image data acquired by the vision sensor; Comparing the extracted visual feature vector with the visual feature reference vectors of all animals in the same class in the visual identity feature reference library set for all animals, calculating the similarity between the current animal and all animals, and sorting according to the similarity data to generate a possible identity list; The method comprises the steps of fusing the presence sensing data acquired by a presence sensor with the situation information of each homonym in the culture management system, and carrying out weighted fusion calculation on the calculated similarity and fusion information data obtained based on the fusion of the presence sensing data and the situation information to obtain the visual auxiliary identity confidence coefficient of each animal in the possible identity list, wherein the situation information of each homonym in the culture management system is obtained based on animal daily activity data statistics, and the animal daily activity data comprises access records of the animals at resource points and activity contents at the resource points; And selecting an animal individual with the highest visual auxiliary identity confidence value from the possible identity list, if the visual auxiliary identity confidence value exceeds a preset confidence threshold, correlating the current animal identity with the highest visual auxiliary identity confidence value, otherwise, performing manual rechecking.
7. The intelligent farming management method according to claim 6, wherein the data preprocessing and feature extraction of the animal image data collected by the vision sensor comprises: background elimination, namely eliminating the background of the animal image data by using a Gaussian mixture model, and separating an animal foreground area in the animal image data; Contour extraction and normalization, namely performing binarization treatment on the animal foreground region to extract an animal external contour feature vector; The key feature point detection comprises the steps of extracting feature vectors of animal body type key points in the animal foreground area through a convolutional neural network, wherein the animal body type key points at least comprise animal shoulder, hip and head vertexes; and Mao Seban, carrying out color information analysis on the animal foreground region, extracting a color histogram or texture characteristics, and encoding the color histogram or texture characteristics into a hairline pattern characteristic vector.
8. The intelligent farming management method according to claim 7, wherein in the step of comparing the extracted visual feature vector with the visual feature reference vectors in the visual identity reference library set for all animals, and calculating the similarity between the current animal and all animals, Calculating the similarity between the visual feature vector of the current animal and the visual feature reference vector of each other similar animal in the visual identity feature reference library by adopting a cosine similarity calculation formula, wherein the visual feature vector comprises an animal external contour feature vector, a body type key point feature vector and a hairline pattern feature vector; the calculation formula of the similarity is expressed as follows: Similarity(A,B)=(A·B)/(||A|| ||B||) Wherein A represents a visual feature vector extracted by a current animal, B is a visual feature reference vector of any one of the same animal in the visual identity feature reference library, A.B represents a vector dot product, and A and B represent L2 norms of the visual feature vector A extracted by the current animal and the visual feature reference vector B of any one of the same animal in the visual identity feature reference library.
9. The intelligent farming management method based on animal vital sign data of claim 6, wherein the visual auxiliary identity confidence is calculated by the formula: Confidence _assist=w_visual× similarity_vision+w_behavior x similarity_behavior +w/u context x similarity_context Wherein w_vision represents a visual feature weight coefficient, w_behavior represents a behavior pattern weight coefficient, w_context represents a context information weight coefficient, and w_vision+w_behavior+w_context=1, similarity_vision represents Similarity calculated by a visual feature vector, similarity_behavior represents matching degree of a behavior pattern of a current animal with a history of behavior patterns of each of the same animals, and similarity_context represents matching degree of context information of the current animal with known individual habits of each of the same animals.
10. An intelligent farming management system based on animal vital sign data, comprising: The animal electronic ear tag signal data abnormal event recording module is used for receiving and analyzing animal electronic ear tag signals through the breeding management system, marking animal electronic ear tag signal data abnormal events and recording breeding resource consumption related to the animal electronic ear tag signal data abnormal events, wherein the animal electronic ear tag signal data abnormal events comprise electronic ear tag signal intensity abnormality, missing reading, reading delay or repeated recording; The performance deviation calculation module is used for tracking individual growth performance of animals corresponding to the animal electronic ear tag signal data abnormal event based on the animal electronic ear tag signal data abnormal event, and calculating performance deviation between the individual growth performance and expected growth performance; the economic influence determining module is used for determining economic influence caused by abnormal events of the animal electronic ear tag signal data based on the consumption of the culture resources, the performance deviation and preset economic parameters; and the economic benefit report generation module is used for carrying out association analysis on economic influence caused by the abnormal event of the animal electronic ear tag signal data and the group benefit index to generate an economic benefit report.

Description

Intelligent breeding management method and system based on animal vital sign data Technical Field The application relates to the field of intelligent culture management, in particular to an intelligent culture management method and system based on animal vital sign data. Background In modern large-scale farms, in order to achieve fine management of individual animals, vital sign sensors and electronic ear tags are usually provided for each animal, so as to ensure accuracy and stability of data acquisition. However, physical factors in the cultivation environment, such as dirt, dust and water vapor, may affect the long-term stable operation of these devices, especially the radio frequency signal transmission of the electronic ear tag may become unstable, resulting in intermittent errors in individual identification, further affecting the correct association of vital sign data and individual identification, and finally negatively affecting the accurate allocation and overall benefit of cultivation resources. The design work logic of the signal receiving antenna and the matched identification module of the data acquisition base station in the culture management system is usually based on stable and clear radio frequency signals. In the face of such unstable, time-intense and time-weak signals caused by the fouling layer, systems can be challenged when decoding. When the signal strength is below the decoding threshold, the system may fail to identify the ear tag, resulting in the individual identity information being missed. If the signal experiences a short decay in the identification process, the system may take longer to decode successfully, causing a read delay. Further complicating matters, if an individual animal passes through an identification area with its ear tag signal briefly interrupted by fluctuations and then recovered in the same pass, the system may incorrectly determine that it is a double independent pass, thereby producing a repeated record of the identity of the same individual. These inaccurate individual presence information directly affects the accuracy of the internal core data structures and computing programs of the farming management system. If duplicate identification occurs, the system may create temporary duplicate identity records for the same individual, resulting in subsequent vital sign data being erroneously associated with those duplicate records, thereby causing temporary confusion of individual data with identity. In addition, the calculation program for adjusting the feed supply amount according to the animal activity amount is also affected, resulting in a deviation in the feed supply amount calculation. Based on the deviated identity recognition and data association, resource allocation instructions sent by the cultivation management system to the discharge pipeline of the automatic feeding equipment and the valves and nozzles of the drinking water system are deviated along with the deviation. Such long term resource allocation disturbances, which occur around individual identification uncertainties, can have a cumulative negative impact on the overall animal population growth performance and breeding benefits. The uniformity of animal population growth can be significantly reduced, and the weight difference between individuals is out of normal range, resulting in difficult management of the marketing lot. The system cannot accurately track the relationship between the actual consumption of each feed and the growth of the individual. This efficiency degradation is not caused by disease or feed quality, but rather results from persistent errors in underlying data correlations. Finally, when the manager of the farm performs batch benefit accounting, obvious profit drop can be found, but the existing scoring model for evaluating the individual health level of animals and the early disease early warning model do not report serious abnormal conditions. These models typically rely on the accuracy of vital sign data and the correct association with the identity of the individual. Since the errors in identification are intermittent, non-thorough, and vital sign data itself may still be valid, but are erroneously associated, these upper-level models cannot capture the root cause of the problem. The manager is faced with the situation that the system looks normal operation, but the actual production result is continuously not ideal, and the root cause of the problem cannot be located, so that a new technical method capable of effectively treating the intermittent identification uncertainty caused by the physical environment factors and realizing true and accurate resource management on the basis is urgently needed. Disclosure of Invention Therefore, the invention provides an intelligent culture management method and system based on animal vital sign data, and aims to solve the problems that in the existing culture management, electronic ear tag signals are unstable due to physica