WO2023165141A1 - Poultry health assessment method and system based on group motion amount statistical characteristics - Google Patents

Poultry health assessment method and system based on group motion amount statistical characteristics Download PDF

Info

Publication number
WO2023165141A1
WO2023165141A1 PCT/CN2022/126504 CN2022126504W WO2023165141A1 WO 2023165141 A1 WO2023165141 A1 WO 2023165141A1 CN 2022126504 W CN2022126504 W CN 2022126504W WO 2023165141 A1 WO2023165141 A1 WO 2023165141A1
Authority
WO
WIPO (PCT)
Prior art keywords
poultry
data
group
exercise
individual
Prior art date
Application number
PCT/CN2022/126504
Other languages
French (fr)
Chinese (zh)
Inventor
肖德琴
刘啸虎
黄一桂
招胜秋
卞智逸
林探宇
冯健昭
Original Assignee
华南农业大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 华南农业大学 filed Critical 华南农业大学
Publication of WO2023165141A1 publication Critical patent/WO2023165141A1/en

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A01—AGRICULTURE; FORESTRY; ANIMAL HUSBANDRY; HUNTING; TRAPPING; FISHING
    • A01K—ANIMAL HUSBANDRY; AVICULTURE; APICULTURE; PISCICULTURE; FISHING; REARING OR BREEDING ANIMALS, NOT OTHERWISE PROVIDED FOR; NEW BREEDS OF ANIMALS
    • A01K35/00—Marking poultry or other birds
    • A—HUMAN NECESSITIES
    • A01—AGRICULTURE; FORESTRY; ANIMAL HUSBANDRY; HUNTING; TRAPPING; FISHING
    • A01K—ANIMAL HUSBANDRY; AVICULTURE; APICULTURE; PISCICULTURE; FISHING; REARING OR BREEDING ANIMALS, NOT OTHERWISE PROVIDED FOR; NEW BREEDS OF ANIMALS
    • A01K45/00—Other aviculture appliances, e.g. devices for determining whether a bird is about to lay
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/30—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
    • Y02A40/00—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production
    • Y02A40/70—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production in livestock or poultry

Definitions

  • the present invention relates to the technical field of artificial intelligence, more specifically, to a poultry health assessment method and system based on statistical characteristics of group exercise.
  • the existing detection method using foot rings to detect the activity track data of poultry running can effectively reduce the cost of transformation, and can complete the classification of health status according to the activity track data of poultry, but there is a problem of low detection accuracy in the classification of poultry frailty and disease. defect.
  • the existing methods often directly give early warning information, or issue unhealthy warnings, without giving a health score, so they can only monitor abnormal behavior at a certain point in time, and cannot combine past data to monitor poultry individual or health status.
  • the health status assessment of the flock cannot describe the health status of poultry.
  • the present invention provides a poultry health assessment method based on statistical characteristics of group exercise, and a method based on statistical characteristics of group exercise poultry health assessment system.
  • a poultry health assessment method based on group exercise statistical characteristics comprising the following steps:
  • poultry is taken as the research object, and the movement volume of the poultry is collected through the three-axis sensor configured on the foot ring as estimated data, and the collected three-axis sensor data is preprocessed to obtain high-precision poultry movement data; combining the same individual at different times
  • individual scoring results and group scoring results you can set the corresponding Through the comparison of the obtained individual scores and group scores, high-accuracy and multi-angle poultry health assessment results such as defective, diseased, dead, and healthy poultry individuals can be obtained.
  • the present invention also proposes a poultry health assessment system based on statistical characteristics of group exercise, which is applied to the above-mentioned poultry health assessment method based on statistical characteristics of group exercise.
  • a poultry health assessment system based on statistical characteristics of group exercise, which is applied to the above-mentioned poultry health assessment method based on statistical characteristics of group exercise.
  • a foot ring is configured with a three-axis sensor and a communication module, and the foot ring is worn on the foot of the poultry;
  • the data receiving module is used to receive the poultry data returned by the foot ring at a certain time interval
  • the data preprocessing module is used to preprocess the received poultry data to obtain poultry exercise data
  • An abnormality detection and correction module is used for abnormal data detection and correction of the poultry exercise data to obtain corrected poultry exercise data;
  • the individual scoring module is used to perform statistics and data distribution on the poultry exercise data of the same individual in different time periods according to the corrected poultry exercise data, and perform individual scoring;
  • the group scoring module is used to perform statistics and data distribution on the poultry exercise data of groups belonging to the same individual in different time periods according to the corrected poultry exercise data, and perform group scoring;
  • the evaluation module is used to evaluate the poultry health status by combining the poultry exercise data, individual scoring results and group scoring results, and generate poultry health evaluation results.
  • the beneficial effect of the technical solution of the present invention is that the present invention uses a three-axis sensor to obtain poultry data for score calculation and evaluation of poultry health status, and at the same time detects and corrects abnormalities of the collected poultry data, which can effectively Improve the precision and accuracy of poultry health detection; the present invention conducts statistics and data distribution on the poultry exercise data of the same individual and group in different time periods, and then performs individual scoring and group scoring on poultry, effectively avoiding estimation deviation caused by single data, and further improving Poultry health detection accuracy, and can intuitively reflect the health status of poultry.
  • FIG. 1 is a flow chart of the poultry health assessment method based on statistical characteristics of group exercise in embodiment 1.
  • FIG. 2 is a flow chart of preprocessing poultry data in Example 2.
  • FIG. 3 is a flow chart of individual scoring of poultry in Example 3.
  • FIG. 3 is a flow chart of individual scoring of poultry in Example 3.
  • FIG. 4 is a flow chart of performing group scoring on poultry in Example 3.
  • FIG. 4 is a flow chart of performing group scoring on poultry in Example 3.
  • FIG. 5 is a flow chart of evaluating the health status of poultry in Example 3.
  • Fig. 6 is a structure diagram of the poultry health assessment system based on the statistical characteristics of group exercise in embodiment 4.
  • This embodiment proposes a poultry health assessment method based on group exercise statistical characteristics, as shown in FIG. 1 , which is a flow chart of the poultry health assessment method based on group exercise statistical characteristics in this embodiment.
  • poultry data is obtained by using a foot ring equipped with a three-axis sensor, which includes the displacement of the individual poultry in the three-axis direction.
  • high-precision poultry movement data is further obtained, and it has certain adaptability , while adapting to the detection of body temperature and other data.
  • the poultry data including the displacement of the individual poultry in the three-axis direction can further calculate the cumulative displacement and the cumulative rotation angle, which are used to calculate the movement data of the poultry.
  • this embodiment adopts the method of data comparison in small time neighborhoods, and combines the distribution of the same individual in different time periods and groups in different time periods to make linear estimation and distribution similarity estimation, and obtain the poultry individual Scoring and group scoring can directly reflect the activity of poultry.
  • the individual scores and group scores of poultry obtained through statistical analysis can be evaluated based on threshold calculations.
  • the state is obtained based on the distribution of group exercise volume based on the threshold value.
  • the group with the highest exercise amount is regarded as healthy poultry, and the group with the least exercise amount is regarded as defective poultry.
  • Health status such as illness and death can be expressed by short-term score changes, and the score change is extremely low. was assessed as death, and a lower change in score was assessed as disease.
  • the health status of poultry such as defective, sick, dead and healthy can be distinguished.
  • triaxial sensors are used to obtain poultry data for scoring calculation and evaluation of poultry health status, and at the same time, abnormal detection and correction are performed on the collected poultry data, which can effectively improve the accuracy and accuracy of poultry health detection.
  • this embodiment conducts statistics and data distribution on the poultry exercise data of the same individual and group in different time periods, and then performs individual scoring and group scoring on poultry, effectively avoiding single data from causing estimation. Deviation can further improve the accuracy of poultry health detection, and at the same time, it can intuitively reflect the activity of poultry groups.
  • This embodiment makes improvements on the basis of the poultry health assessment method based on the statistical characteristics of group exercise proposed in embodiment 1.
  • the data of the poultry acquired in the S1 step includes the displacement v of the poultry, wherein the displacement v includes the displacement components x, y, z.
  • the steps of preprocessing the acquired data in this embodiment include:
  • V ⁇ n v n
  • v n represents the displacement of the poultry detected for the nth time within the data return time interval ⁇ t of the foot ring.
  • the data return time interval ⁇ t of the ankle ring is determined by the internal parameters of the ankle ring. Since the setting of ⁇ t has an impact on the estimation, in one embodiment, the initial setting is 0.5h, and then adjusted according to the specific situation.
  • x n , yn , and z n represent the displacement components in the three-axis directions of the poultry detected for the nth time.
  • ⁇ is a constant close to 0, and in one embodiment, it is set to 0.0001.
  • ⁇ is a parameter used to adjust the exponential scaling, so that the range of the accumulated amount of rotation angle and the accumulated amount of displacement is similar. Considering factors such as different species and regions, it needs to be adjusted according to the actual situation.
  • the initial value of ⁇ is set to 1, and then adjusted according to the actual situation.
  • age is used as the grouping condition, which can ensure effective group estimation.
  • the poultry movement data in this embodiment is composed of displacement accumulation and rotation angle accumulation. There is a slight difference in calculation according to the difference in value range and movement characteristics of the two data: the displacement accumulation is directly accumulated by the displacement change; The value range of the cumulative amount of rotation angle is smaller than that of the cumulative amount of displacement, and at the same time, the value difference under different motions is also small, so the exponential operation is used to increase the gap, and the scaling factor is used to adjust the range and then accumulated.
  • step S2 when performing abnormal data detection and correction on the poultry exercise data, specifically, the second-order difference method is used to linearly fit the poultry exercise m t in the small time neighborhood at time t, and the expression formula is as follows:
  • any of the following methods can be used to obtain:
  • ⁇ is an adjustment parameter whose value is greater than 1, and ⁇ m represents the second-order difference value of poultry exercise at each moment within a certain period of time. In an embodiment, ⁇ takes a value of 1.1.
  • ⁇ is an adjustment parameter, and its value is set according to specific situations.
  • the initial value of the adjustment parameter ⁇ is 3.
  • the steps of correcting abnormal data include:
  • m t,1 2m t-1 -m t-2
  • m t,1 represents the fitting value of poultry exercise at the tth time.
  • m t ⁇ m t,1 +(1- ⁇ )m t,2
  • ⁇ is an adjustment parameter, and the value range is [0,1]. In an embodiment, the value of the adjustment parameter ⁇ is 0.5.
  • this embodiment assumes that the obtained abnormal data only contains peak data, that is, in a small time domain, the group The data of only one ankle ring is abnormal. According to the characteristics of its peak, the second order difference of the data is calculated, and the threshold is used to detect data anomalies. For this anomaly, two correction methods are adopted: individual linear correction for different time series and mode correction for group exercise. The two correction methods are linearly combined to obtain a final correction result for abnormal data.
  • This embodiment makes improvements on the basis of the poultry health assessment method based on the statistical characteristics of group exercise proposed in embodiment 2.
  • the data of poultry exercise data judged as normal and/or the corrected poultry exercise data is data structured and stored In the cache queue structure of the corresponding group; the cache queue structure stores the individual poultry exercise data m i,j of N different poultry individuals belonging to the same group in the past d days, and the group exercise data of the group Among them, group exercise data is the average value of poultry exercise in the group in the last d days, i ⁇ [0,N], j ⁇ [0,d ⁇ 24/ ⁇ t], and ⁇ t is the data return time interval of the ankle ring.
  • the poultry exercise data after abnormality detection and correction will be stored in two categories: individual data and group data.
  • individual data When storing, only the data within a period of time is saved, and all the data is saved as a queue structure.
  • the data When the data is obtained in a new day, it is temporarily stored. After the data of one day is completely stored, the original earliest data is replaced to ensure the timeliness of the data. , so that the features used for analogy can be adapted to various changes, complete the adaptive function in timing, and ensure the robustness of the algorithm in the time span.
  • group data is divided into short-term group data and long-term group data to obtain results in different time spans.
  • step S3 the step of individually scoring the poultry includes:
  • p i,j,k represents the normalization result of the historical exercise data of poultry individual i in the same time neighborhood
  • m i,j-24*k represents the physical activity of poultry individual i in the same time neighborhood before the kth day Historical exercise data.
  • the neighborhood parameter ⁇ is initialized as That is, round up 2/ ⁇ t.
  • ⁇ is an adjustment parameter, and its value is a constant close to 0.
  • the adjustment parameter ⁇ is initially set to 0.0000000001.
  • the adjustment parameter ⁇ is initially set to 3.
  • q j represents the normalized result of the exercise data of the group to which the poultry belongs
  • ⁇ is an adjustment parameter, and the value range is [0,1]; Represents the round-up function.
  • the initial value of the adjustment parameter ⁇ is set to 0.3.
  • the scores are compressed nonlinearly and enlarged to [100,0]. According to the calculation results of the individual scores a i above, 100 is the healthiest and 0 is the unhealthiest. The calculated poultry scoring results are stored for subsequent judgment.
  • FIG. 3 it is a flowchart of individual scoring of poultry in this embodiment.
  • the KL divergence estimation method is mainly used, and the individual score is calculated in combination with the individual contemporaneous score and the group contemporaneous score.
  • the individual score at the same time is mainly by comparing the data of a single individual at the same time on different days.
  • this embodiment uses the Gaussian distribution to convolve the comparison results of different days to obtain the individual score.
  • the group contemporaneous score directly compares individual data with short-term group data, and finally linearly fuses the individual contemporaneous score and the group contemporaneous score to obtain the final individual score.
  • Method 1 Calculate the individual score a i of N poultry individuals in the group group corresponding to the poultry, and take the mean value to obtain the average score o i of the group group corresponding to the poultry individual i, and then obtain the group score t i through nonlinear compression:
  • l , j, k represent the normalization results of the historical exercise data of poultry groups in the same time neighborhood, Represents the historical exercise data of the poultry group in the same time range before the kth day.
  • ⁇ is an adjustment parameter
  • FIG. 4 it is a flow chart of group scoring of poultry by method 2 in this embodiment.
  • Method 1 is obtained by directly using the mean value of the individual scores a i of N poultry individuals in the group group and performing nonlinear compression, which is suitable for application scenarios such as a large number of poultry individuals in the group group and limited calculation conditions.
  • Method 2 uses the comparison method of small temporal neighborhoods to divide group scoring into long-term group evaluation and short-term group evaluation.
  • the long-term group evaluation compares the distribution of the same time neighborhood on different dates, and then uses Gaussian convolution to obtain scores for different dates;
  • Short-term group assessment utilizes the continuity of transformation, uses short-term data to estimate a theoretical linear forecast value, and compares the forecasted value with the actual value to obtain a score. After a linear fusion of the two scores, the final score is obtained.
  • Method 2 can more intuitively reflect the activity of poultry in group groups.
  • poultry health status is evaluated in combination with poultry exercise data, individual scoring results and group scoring results, and the specific steps include:
  • the poultry i is evaluated as normal.
  • first threshold ⁇ 1 is smaller than the second threshold ⁇ 2 , which is specifically set according to actual application conditions.
  • the first threshold ⁇ 1 is set to 25, and the second threshold ⁇ 2 is set to 100.
  • m i, z is the exercise data of poultry i.
  • poultry i is assessed as ordinary poultry
  • FIG. 5 it is a flow chart of evaluating the health status of poultry in this embodiment.
  • the classification of defective, diseased, dead and healthy poultry is divided into two classification problems.
  • the first category is the classification problem of illness, death and normal.
  • Step S4.1 is used to evaluate the health status of poultry such as sickness, death and normal, and the obtained individual health score is calculated based on the threshold value, wherein the sickness and death can be expressed by the change of the score in a short period of time, and the score becomes extremely low when it is Death, lower score is disease.
  • the second category is the defective, robust and ordinary poultry, which are evaluated through steps S4.2 and S4.3. Among them, the group exercise distribution of defective and robust is obtained based on the threshold value. One batch as defective poultry.
  • This embodiment proposes a poultry health assessment system based on statistical characteristics of group exercise, which is applied to any of the poultry health assessment methods proposed in Examples 1-3.
  • FIG. 6 it is a structure diagram of the poultry health assessment system of this embodiment.
  • a foot ring 1 the foot ring 1 is configured with a three-axis sensor 101 and a communication module 102, and the foot ring 1 is worn on the foot of a poultry;
  • the data receiving module 2 is used to receive the poultry data returned by the foot ring 1 at a certain time interval
  • the data preprocessing module 3 is used to preprocess the received poultry data to obtain poultry exercise data
  • Abnormality detection and correction module 4 used for abnormal data detection and correction of the poultry exercise data, to obtain corrected poultry exercise data
  • the individual scoring module 5 is used to perform statistics and data distribution on the poultry exercise data of the same individual in different time periods according to the corrected poultry exercise data, and perform individual scoring;
  • the group scoring module 6 is used to perform statistics and data distribution on the poultry exercise data of groups belonging to the same individual in different time periods according to the corrected poultry exercise data, and perform group scoring;
  • the evaluation module 7 is used to evaluate the poultry health status by combining the poultry exercise data, individual scoring results and group scoring results, and generate poultry health evaluation results.
  • the foot ring 1 configured with the three-axis sensor 101 and the communication module 102 is worn on the poultry individual to be evaluated, wherein the three-axis sensor 101 collects the displacement of the poultry individual in the three-axis direction in real time, and then uses the preset The data return time interval ⁇ t, and the poultry data collected at several moments in the current time interval ⁇ t period are sent to the data receiving module 2 through the communication module 102 .
  • the data receiving module 2 transmits the poultry data sent by the foot ring 1 to the data preprocessing module 3 for processing.
  • the data preprocessing module 3 calculates the accumulated displacement, the rotation angle component and the accumulated amount of rotation angle of the poultry detected in the data currently returned by the foot ring 1, and further calculates the movement amount of the poultry in the currently returned data, and finally The calculated poultry movement is transmitted to the anomaly detection and correction module 4 .
  • the second order difference method is used to linearly fit the poultry movement quantity at the tth time in the small time neighborhood, and then use the preset threshold th to Abnormal data detection is performed on the second-order difference value of poultry exercise at time t, and the individual linear correction for different time series and the mode correction method for group exercise are respectively used for the data detected as abnormal, and then an abnormal data is obtained through linear combination The final correction result of .
  • the abnormality detection and correction module 4 will detect the normal data and the corrected data, and perform group grouping according to the age of the corresponding poultry individual.
  • the individual scoring module 5 reads the normal data and corrected data of poultry individuals, performs statistics and data distribution on the poultry exercise data of the same individual in different time periods, and performs individual scoring.
  • the individual scoring module 5 uses an estimation method of KL divergence, and combines individual contemporaneity scores and group contemporaneity scores to calculate individual scores.
  • the individual score at the same time is mainly obtained by comparing the data of a single individual at the same time on different dates, and the Gaussian distribution is used to convolve the comparison results on different days to obtain the individual score; the group score at the same time directly compares the individual data with the short-term group data, and finally the individual score at the same time and the short-term group data
  • the group contemporaneous scores are linearly fused to obtain the final individual scores, which are sent to the evaluation module 7 .
  • the group scoring module 6 adopts the mean value of the individual scores a i of N poultry individuals in the group grouping and performs nonlinear compression to obtain the group scoring result; or adopts a small time neighborhood comparison method to divide the group scoring into long-term group evaluation and Short-term group evaluation, in which long-term group evaluation compares the distribution of the same temporal neighborhood on different dates, and then uses Gaussian convolution to obtain scores for different dates; short-term group evaluation uses the continuity of the transformation to estimate the theoretical linear prediction with short-term data Value, compare the estimated value with the actual value to get a score. After a linear fusion of the two scores, the final group score is obtained and sent to the evaluation module 7 .
  • the evaluation module 7 evaluates the poultry health status in combination with the poultry exercise data, individual score results and group score results.
  • the evaluation module 7 first evaluates the sickness, death and normal state of the poultry, and the obtained individual health score is calculated based on the threshold value, wherein the sickness and death can be represented by the score change in a short period of time, and the score becomes extremely low. for death, lower score for disease.
  • the evaluation module 7 further evaluates the defective, robust and normal states of the poultry, wherein the defective and robust use group exercise distribution is obtained based on a threshold, the batch with the highest exercise amount is regarded as healthy poultry, and the batch with the smallest exercise amount is regarded as defective poultry , otherwise it is assessed as ordinary poultry.

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Public Health (AREA)
  • Biomedical Technology (AREA)
  • Data Mining & Analysis (AREA)
  • Environmental Sciences (AREA)
  • Epidemiology (AREA)
  • General Health & Medical Sciences (AREA)
  • Primary Health Care (AREA)
  • Pathology (AREA)
  • Databases & Information Systems (AREA)
  • Birds (AREA)
  • Animal Husbandry (AREA)
  • Biodiversity & Conservation Biology (AREA)
  • Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
  • Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
  • Measuring And Recording Apparatus For Diagnosis (AREA)

Abstract

The present invention relates to the technical field of artificial intelligence. Provided are a poultry health assessment method and system based on group motion amount statistical characteristics. The method comprises the following steps: acquiring poultry data from foot rings which are worn by poultry and are provided with three-axis sensors, and preprocessing the acquired data to obtain poultry motion amount data; performing abnormal data detection and correction on the poultry motion amount data, so as to obtain corrected poultry motion amount data; compiling statistics and data distribution on the poultry motion amount data of the same individual and the same group in different time periods according to the corrected poultry motion amount data, and performing individual scoring and group scoring on the poultry according to statistical and data distribution results of the poultry motion amount data; and assessing the health state of the poultry in view of the poultry motion amount data, an individual scoring result and a group scoring result, so as to obtain a poultry health assessment result. By means of the present invention, the poultry health testing precision and accuracy can be effectively improved, and the poultry health condition can be visually reflected.

Description

一种基于群体运动量统计特征的家禽健康评估方法及系统A poultry health assessment method and system based on statistical characteristics of group exercise 技术领域technical field
本发明涉及人工智能技术领域，更具体地，涉及一种基于群体运动量统计特征的家禽健康评估方法及系统。The present invention relates to the technical field of artificial intelligence, more specifically, to a poultry health assessment method and system based on statistical characteristics of group exercise.
背景技术Background technique
随着人民生活水平的不断提高，人们对于各类家禽的需求量也越发增多。在规模较大的家禽(鸡、鸭和鹅)的养殖过程中，较为弱小的个体突发意外，例如抢食等行为而受伤，或发生个体打架受伤的情况较为无法避免，要及时救护，或及时清理，以免造成进一步损失是非常必要的。因此，对于运动量的合理监测以及转化，得到用于健康评估的健康评分对于保证家禽(鸡、鸭和鹅)的稳定养殖和提高产值具有重要意义。传统家禽养殖中，家禽的健康状况都由人工观测，由于家禽疾病发生和发展迅速，且时间发生较为随机，人工观测的方法费时、费力且容易出错。随着计算机视觉技术的成熟与普及，越来越多的学者通过摄像头获取动物(鸡、鸭和鹅)生活图像或视频，并根据图像特征对动物(鸡、鸭和鹅)行为进行检测，但这些方法都需要部署摄像头等设备，不论对于笼养情况还是平养情况，都需要部署大量摄像头才能达到拍摄范围的要求或拍摄清晰度的要求，否则就会影响到后期判断的效果。With the continuous improvement of people's living standards, people's demand for all kinds of poultry is also increasing. In the breeding process of large-scale poultry (chickens, ducks and geese), it is unavoidable for weaker individuals to be injured by unexpected accidents, such as grabbing food, or individual fights and injuries, so timely rescue is required, or It is very necessary to clean up in time to avoid further damage. Therefore, for the reasonable monitoring and conversion of exercise, obtaining the health score for health assessment is of great significance to ensure the stable breeding of poultry (chicken, duck and goose) and increase the output value. In traditional poultry farming, the health status of poultry is observed manually. Due to the rapid occurrence and development of poultry diseases, and the time of occurrence is relatively random, the method of manual observation is time-consuming, laborious and error-prone. With the maturity and popularization of computer vision technology, more and more scholars obtain live images or videos of animals (chickens, ducks and geese) through cameras, and detect the behavior of animals (chickens, ducks and geese) according to image features. These methods all require the deployment of cameras and other equipment. Regardless of whether it is caged or flat, it is necessary to deploy a large number of cameras to meet the requirements of the shooting range or shooting clarity, otherwise it will affect the effect of later judgments.
现有采用脚环检测家禽运行的活动轨迹数据的检测方法，能够有效降低改造成本，能够根据家禽的活动轨迹数据完成健康状态的分类，但针对家禽的体弱和疾病的分类存在检测精度低的缺陷。此外，已有的方法往往都直接给出预警信息，或者发出不健康的警告，没有给出一个健康状况的评分，因此只能达到某时刻点异常行为的监测，并不能结合过往数据进行家禽个体或群体的健康状况评估，无法描述家禽健康状况的优劣。The existing detection method using foot rings to detect the activity track data of poultry running can effectively reduce the cost of transformation, and can complete the classification of health status according to the activity track data of poultry, but there is a problem of low detection accuracy in the classification of poultry frailty and disease. defect. In addition, the existing methods often directly give early warning information, or issue unhealthy warnings, without giving a health score, so they can only monitor abnormal behavior at a certain point in time, and cannot combine past data to monitor poultry individual or health status. The health status assessment of the flock cannot describe the health status of poultry.
发明内容Contents of the invention
本发明为克服上述现有技术所述的家禽健康检测精度低、无法描述家禽健康状况的优劣的缺陷，提供一种基于群体运动量统计特征的家禽健康评估方法，以及一种基于群体运动量统计特征的家禽健康评估系统。In order to overcome the defects of low poultry health detection accuracy and inability to describe the health status of poultry described in the prior art, the present invention provides a poultry health assessment method based on statistical characteristics of group exercise, and a method based on statistical characteristics of group exercise poultry health assessment system.
为解决上述技术问题，本发明的技术方案如下：In order to solve the problems of the technologies described above, the technical solution of the present invention is as follows:
一种基于群体运动量统计特征的家禽健康评估方法，包括以下步骤：A poultry health assessment method based on group exercise statistical characteristics, comprising the following steps:
S1、从家禽所佩戴的带三轴传感器的脚环获取家禽数据，并对获取的数据进行预处理，得到家禽运动量数据；S1. Obtain poultry data from foot rings with three-axis sensors worn by poultry, and preprocess the acquired data to obtain poultry exercise data;
S2、对所述家禽运动量数据进行异常数据检测和修正，得到经过修正的家禽运动量数据；S2. Perform abnormal data detection and correction on the poultry exercise data to obtain corrected poultry exercise data;
S3、根据所述经过修正的家禽运动量数据，对同一个体以及群体在不同时间段的家禽运动量数据进行统计及数据分布，并根据家禽运动量数据的统计及数据分布结果对家禽进行个体评分及群体评分；S3. According to the corrected poultry exercise data, perform statistics and data distribution on the poultry exercise data of the same individual and group in different time periods, and perform individual scoring and group scoring on the poultry according to the statistics and data distribution results of the poultry exercise data ;
S4、结合家禽运动量数据、个体评分结果和群体评分结果对家禽健康状态进行评估，得到家禽健康评估结果。S4. Combining the poultry exercise data, the individual score results and the group score results to evaluate the poultry health status, and obtain the poultry health assessment results.
本技术方案中，以家禽作为研究对象，通过脚环配置的三轴传感器采集家禽的运动量作为估计数据，对采集的三轴传感器数据进行预处理得到高精度的家禽运动量数据；结合同一个体不同时间段以及群体不同时间段的分布进行统计及数据分布，进而得到相应的健康评分；进一步结合家禽运动量数据、个体评分结果和群体评分结果对家禽健康状态进行评估的过程中，可以根据实际应用设置相应的阈值，通过所得的个体评分及群体评分比较得到家禽个体的残次、患病、死亡、健壮等高准确率且多角度的家禽健康评估结果。In this technical solution, poultry is taken as the research object, and the movement volume of the poultry is collected through the three-axis sensor configured on the foot ring as estimated data, and the collected three-axis sensor data is preprocessed to obtain high-precision poultry movement data; combining the same individual at different times In the process of evaluating poultry health status by combining poultry exercise data, individual scoring results and group scoring results, you can set the corresponding Through the comparison of the obtained individual scores and group scores, high-accuracy and multi-angle poultry health assessment results such as defective, diseased, dead, and healthy poultry individuals can be obtained.
进一步的，本发明还提出了一种基于群体运动量统计特征的家禽健康评估系统，应用于上述基于群体运动量统计特征的家禽健康评估方法。其中包括：Furthermore, the present invention also proposes a poultry health assessment system based on statistical characteristics of group exercise, which is applied to the above-mentioned poultry health assessment method based on statistical characteristics of group exercise. These include:
脚环，所述脚环配置有三轴传感器和通信模块，且所述脚环佩戴设置在家禽的脚部；A foot ring, the foot ring is configured with a three-axis sensor and a communication module, and the foot ring is worn on the foot of the poultry;
数据接收模块，用于接收由脚环以一定时间间隔回传的家禽数据；The data receiving module is used to receive the poultry data returned by the foot ring at a certain time interval;
数据预处理模块，用于对接收的家禽数据进行预处理，得到家禽运动量数据；The data preprocessing module is used to preprocess the received poultry data to obtain poultry exercise data;
异常检测和修正模块，用于对所述家禽运动量数据进行异常数据检测和修正，得到经过修正的家禽运动量数据；An abnormality detection and correction module is used for abnormal data detection and correction of the poultry exercise data to obtain corrected poultry exercise data;
个体评分模块，用于根据所述经过修正的家禽运动量数据，对同一个体不同时间段的家禽运动量数据进行统计及数据分布，并进行个体评分；The individual scoring module is used to perform statistics and data distribution on the poultry exercise data of the same individual in different time periods according to the corrected poultry exercise data, and perform individual scoring;
群体评分模块，用于根据所述经过修正的家禽运动量数据，对同一个体所属群体分组在不同时间段的家禽运动量数据进行统计及数据分布，并进行群体评 分；The group scoring module is used to perform statistics and data distribution on the poultry exercise data of groups belonging to the same individual in different time periods according to the corrected poultry exercise data, and perform group scoring;
评估模块，用于结合家禽运动量数据、个体评分结果和群体评分结果对家禽健康状态进行评估，生成家禽健康评估结果。The evaluation module is used to evaluate the poultry health status by combining the poultry exercise data, individual scoring results and group scoring results, and generate poultry health evaluation results.
与现有技术相比，本发明技术方案的有益效果是：本发明采用三轴传感器获取家禽数据用于家禽健康状态的评分计算及评估，同时对采集的家禽数据进行异常检测和修正，能够有效提高家禽健康检测精度、准确率；本发明对同一个体以及群体在不同时间段的家禽运动量数据进行统计及数据分布，进而对家禽进行个体评分及群体评分，有效避免单个数据造成估计偏差，进一步提高家禽健康检测精度，同时能够直观地反映家禽健康状况。Compared with the prior art, the beneficial effect of the technical solution of the present invention is that the present invention uses a three-axis sensor to obtain poultry data for score calculation and evaluation of poultry health status, and at the same time detects and corrects abnormalities of the collected poultry data, which can effectively Improve the precision and accuracy of poultry health detection; the present invention conducts statistics and data distribution on the poultry exercise data of the same individual and group in different time periods, and then performs individual scoring and group scoring on poultry, effectively avoiding estimation deviation caused by single data, and further improving Poultry health detection accuracy, and can intuitively reflect the health status of poultry.
附图说明Description of drawings
图1为实施例1的基于群体运动量统计特征的家禽健康评估方法的流程图。FIG. 1 is a flow chart of the poultry health assessment method based on statistical characteristics of group exercise in embodiment 1.
图2为实施例2中对家禽数据进行预处理的流程图。FIG. 2 is a flow chart of preprocessing poultry data in Example 2.
图3为实施例3中对家禽进行个体评分的流程图。FIG. 3 is a flow chart of individual scoring of poultry in Example 3. FIG.
图4为实施例3中对家禽进行群体评分的流程图。FIG. 4 is a flow chart of performing group scoring on poultry in Example 3. FIG.
图5为实施例3中对家禽健康状态进行评估的流程图。FIG. 5 is a flow chart of evaluating the health status of poultry in Example 3.
图6为实施例4的基于群体运动量统计特征的家禽健康评估系统的架构图。Fig. 6 is a structure diagram of the poultry health assessment system based on the statistical characteristics of group exercise in embodiment 4.
具体实施方式Detailed ways
附图仅用于示例性说明，不能理解为对本专利的限制；The accompanying drawings are for illustrative purposes only and cannot be construed as limiting the patent;
对于本领域技术人员来说，附图中某些公知结构及其说明可能省略是可以理解的。For those skilled in the art, it is understandable that some well-known structures and descriptions thereof may be omitted in the drawings.
下面结合附图和实施例对本发明的技术方案做进一步的说明。The technical solutions of the present invention will be further described below in conjunction with the accompanying drawings and embodiments.
实施例1Example 1
本实施例提出一种基于群体运动量统计特征的家禽健康评估方法，如图1所示，为本实施例的基于群体运动量统计特征的家禽健康评估方法的流程图。This embodiment proposes a poultry health assessment method based on group exercise statistical characteristics, as shown in FIG. 1 , which is a flow chart of the poultry health assessment method based on group exercise statistical characteristics in this embodiment.
本实施例提出的基于群体运动量统计特征的家禽健康评估方法中，包括以下步骤：In the poultry health assessment method based on group exercise statistical characteristics proposed by the present embodiment, comprise the following steps:
S1、从家禽所佩戴的带三轴传感器的脚环获取家禽数据，并对获取的数据进行预处理，得到家禽运动量数据。S1. Obtain poultry data from foot rings with triaxial sensors worn by poultry, and preprocess the acquired data to obtain poultry exercise data.
本实施例中，采用配置有三轴传感器的脚环获取家禽数据，其中包含家禽个体在三轴方向上的位移量，经过数据的与处理后进一步得到高精度的家禽运动量 数据，且具有一定适应性，同时适配于体温等数据的检测。In this embodiment, poultry data is obtained by using a foot ring equipped with a three-axis sensor, which includes the displacement of the individual poultry in the three-axis direction. After data processing, high-precision poultry movement data is further obtained, and it has certain adaptability , while adapting to the detection of body temperature and other data.
本实施例中含家禽个体在三轴方向上的位移量的家禽数据，可进一步地计算位移累加量和转动角累加量，用于计算家禽运动量数据。In this embodiment, the poultry data including the displacement of the individual poultry in the three-axis direction can further calculate the cumulative displacement and the cumulative rotation angle, which are used to calculate the movement data of the poultry.
S2、对所述家禽运动量数据进行异常数据检测和修正，得到经过修正的家禽运动量数据。S2. Perform abnormal data detection and correction on the poultry exercise data to obtain corrected poultry exercise data.
本实施例中，考虑到现今硬件基础较好，大部分脚环包含异常处理模块，本实施例假定所得到的异常数据只包含尖峰数据，即在一个小的时间领域内，群体中只有一个脚环的数据发生异常。针对尖峰数据的特点，在数据修正过程中可采用线性修正、众数修正等方式，得到经过修正的家禽运动量数据。In this embodiment, considering that the current hardware foundation is relatively good, most foot rings contain exception processing modules. This embodiment assumes that the obtained abnormal data only contains peak data, that is, in a small time domain, there is only one The ring data is abnormal. According to the characteristics of peak data, in the data correction process, linear correction, mode correction and other methods can be used to obtain the corrected poultry exercise data.
S3、根据所述经过修正的家禽运动量数据，对同一个体以及群体在不同时间段的家禽运动量数据进行统计及数据分布，并根据家禽运动量数据的统计及数据分布结果对家禽进行个体评分及群体评分。S3. According to the corrected poultry exercise data, perform statistics and data distribution on the poultry exercise data of the same individual and group in different time periods, and perform individual scoring and group scoring on the poultry according to the statistics and data distribution results of the poultry exercise data .
考虑到避免单个数据造成估计偏差，本实施例采取小时间邻域的数据比较的方式，结合同一个体不同时间段以及群体不同时间段的分布做出线性估计和分布相似度估计，得到家禽的个体评分及群体评分，可以直观反映家禽的活跃情况。In consideration of avoiding estimation bias caused by single data, this embodiment adopts the method of data comparison in small time neighborhoods, and combines the distribution of the same individual in different time periods and groups in different time periods to make linear estimation and distribution similarity estimation, and obtain the poultry individual Scoring and group scoring can directly reflect the activity of poultry.
S4、结合家禽运动量数据、个体评分结果和群体评分结果对家禽健康状态进行评估，得到家禽健康评估结果。S4. Combining the poultry exercise data, the individual score results and the group score results to evaluate the poultry health status, and obtain the poultry health assessment results.
在具体实施过程中，针对家禽个体的残次、患病、死亡、健壮等健康状态，可以通过统计分析得到的家禽的个体评分及群体评分并基于阈值计算进行评估，其中残次、健壮等健康状态使用群体运动量分布基于阈值取得，运动量最高的一批作为健壮家禽，运动量最小的一批作为残次家禽；而患病、死亡等健康状态可以通过短期内的评分变化表示，评分变化为极低则评估为死亡，评分变化较低则评估为患病。通过以上方式就可以区分家禽的残次、患病、死亡以及健壮等健康状态。In the specific implementation process, for the health status of poultry individuals, such as defectiveness, disease, death, and robustness, the individual scores and group scores of poultry obtained through statistical analysis can be evaluated based on threshold calculations. The state is obtained based on the distribution of group exercise volume based on the threshold value. The group with the highest exercise amount is regarded as healthy poultry, and the group with the least exercise amount is regarded as defective poultry. Health status such as illness and death can be expressed by short-term score changes, and the score change is extremely low. was assessed as death, and a lower change in score was assessed as disease. Through the above methods, the health status of poultry such as defective, sick, dead and healthy can be distinguished.
本实施例采用三轴传感器获取家禽数据用于家禽健康状态的评分计算及评估，同时对采集的家禽数据进行异常检测和修正，能够有效提高家禽健康检测精度、准确率。此外，本实施例根据所述经过修正的家禽运动量数据，对同一个体以及群体在不同时间段的家禽运动量数据进行统计及数据分布，进而对家禽进行个体评分及群体评分，有效避免单个数据造成估计偏差，进一步提高家禽健康检测精度，同时能够直观地反映家禽群体的活跃情况。最后结合家禽的运动量数据 和家禽的个体评分、群体评分进行家禽健康状态评估，根据家禽的不同健康状态下所反映的情况设置相关阈值，进而得到多种家禽健康评估结果，同时能够直观表示家禽健康状况的优劣。In this embodiment, triaxial sensors are used to obtain poultry data for scoring calculation and evaluation of poultry health status, and at the same time, abnormal detection and correction are performed on the collected poultry data, which can effectively improve the accuracy and accuracy of poultry health detection. In addition, according to the corrected poultry exercise data, this embodiment conducts statistics and data distribution on the poultry exercise data of the same individual and group in different time periods, and then performs individual scoring and group scoring on poultry, effectively avoiding single data from causing estimation. Deviation can further improve the accuracy of poultry health detection, and at the same time, it can intuitively reflect the activity of poultry groups. Finally, combine the poultry's exercise data with the poultry's individual score and group score to evaluate the poultry's health status, set relevant thresholds according to the conditions reflected in different health status of the poultry, and then obtain a variety of poultry health assessment results, and at the same time, it can visually represent the poultry health The pros and cons of the situation.
实施例2Example 2
本实施例在实施例1提出的基于群体运动量统计特征的家禽健康评估方法的基础上作出改进。This embodiment makes improvements on the basis of the poultry health assessment method based on the statistical characteristics of group exercise proposed in embodiment 1.
本实施例提出的基于群体运动量统计特征的家禽健康评估方法中，在S1步骤中获取的家禽的数据包括家禽的位移量v，其中位移量v包括在三轴方向上的位移分量x、y、z。In the poultry health assessment method based on the statistical characteristics of group exercise proposed in this embodiment, the data of the poultry acquired in the S1 step includes the displacement v of the poultry, wherein the displacement v includes the displacement components x, y, z.
进一步的，如图2所示，本实施例中对获取的数据进行预处理的步骤包括：Further, as shown in Figure 2, the steps of preprocessing the acquired data in this embodiment include:
S1.1、计算所述脚环当前回传的数据中检测的家禽的位移累加量V，其表达公式如下：S1.1. Calculate the cumulative displacement V of the poultry detected in the data currently returned by the foot ring, and its expression formula is as follows:
V＝∑ nv n V＝∑ n v n
式中，v n表示脚环的数据回传时间间隔Δt内第n次检测的家禽的位移量。 In the formula, v n represents the displacement of the poultry detected for the nth time within the data return time interval Δt of the foot ring.
其中，脚环的数据回传时间间隔Δt由脚环内部参数确定，由于Δt的设置对估计存在影响，在一实施例中，初始设置为0.5h，而后根据具体情况调整。The data return time interval Δt of the ankle ring is determined by the internal parameters of the ankle ring. Since the setting of Δt has an impact on the estimation, in one embodiment, the initial setting is 0.5h, and then adjusted according to the specific situation.
S1.2、计算数据回传时间间隔Δt内每一次检测的家禽的转动角分量w，再采用指数缩放方式计算数据回传时间间隔Δt内家禽的转动角累加量W，其表达公式如下：S1.2. Calculate the rotation angle component w of each detected poultry within the data return time interval Δt, and then use the exponential scaling method to calculate the cumulative amount W of the poultry’s rotation angle within the data return time interval Δt. The expression formula is as follows:
Figure PCTCN2022126504-appb-000001
Figure PCTCN2022126504-appb-000001
式中，x n、y n、z n表示第n次检测的家禽的三轴方向上的位移分量。τ为接近0的常数，在一实施例中，设置为0.0001。 In the formula, x n , yn , and z n represent the displacement components in the three-axis directions of the poultry detected for the nth time. τ is a constant close to 0, and in one embodiment, it is set to 0.0001.
α为用于调整指数缩放比例的参数，使得转动角累加量和位移累加量范围相近。考虑到物种、地域不同等因素，需根据实际情况调整，在一实施例中，将α初始设置值为1，而后根据实际情况调整。α is a parameter used to adjust the exponential scaling, so that the range of the accumulated amount of rotation angle and the accumulated amount of displacement is similar. Considering factors such as different species and regions, it needs to be adjusted according to the actual situation. In one embodiment, the initial value of α is set to 1, and then adjusted according to the actual situation.
S1.3、根据家禽的位移累加量V和转动角累加量W，计算得到当前回传的数据中的家禽运动量m＝V+W。S1.3. According to the cumulative displacement V and the cumulative rotation W of the poultry, calculate the poultry movement m=V+W in the currently returned data.
S1.4、对家禽运动量数据进行群体分组，将日龄相同的家禽对应的运动量作为一组群体数据。S1.4. Carry out group grouping on the poultry exercise volume data, and use the corresponding exercise volume of the same day-old poultry as a group of group data.
本实施例中采用日龄作为分组条件，能够保证有效的群体估计。In this embodiment, age is used as the grouping condition, which can ensure effective group estimation.
本实施例中的家禽运动量数据由位移累加量和转动角累加量构成，针对两种数据取值范围和运动特点的不同，计算上略有差异：位移累加量直接由位移变化量累加而成；转动角累加量取值范围相较位移累加量较小，同时不同运动量下数值差异也较小，所以采用指数运算加大差距，利用缩放因子调整其范围而后累加。The poultry movement data in this embodiment is composed of displacement accumulation and rotation angle accumulation. There is a slight difference in calculation according to the difference in value range and movement characteristics of the two data: the displacement accumulation is directly accumulated by the displacement change; The value range of the cumulative amount of rotation angle is smaller than that of the cumulative amount of displacement, and at the same time, the value difference under different motions is also small, so the exponential operation is used to increase the gap, and the scaling factor is used to adjust the range and then accumulated.
在S2步骤中，在对家禽运动量数据进行异常数据检测和修正时，具体的，采用二阶差分方法将第t时刻家禽运动量m t在小时间邻域内进行线性拟合，其表达公式如下： In step S2, when performing abnormal data detection and correction on the poultry exercise data, specifically, the second-order difference method is used to linearly fit the poultry exercise m t in the small time neighborhood at time t, and the expression formula is as follows:
Δm t＝m t+1-2m t+m t-1 Δm t ＝m t+1 -2m t +m t-1
利用预设的阈值th对第t时刻家禽运动量二阶差分值进行异常数据检测：Use the preset threshold th to detect the abnormal data of the second-order difference value of the poultry exercise amount at the tth moment:
若第t时刻家禽运动量二阶差分值Δm t大于或等于预设的阈值th，则判断为存在异常数据，并对第t时刻家禽运动量m t进行修正； If the second-order difference value Δm t of poultry movement at the t-th moment is greater than or equal to the preset threshold th, it is judged that there is abnormal data, and the poultry movement m t at the t-th moment is corrected;
若第t时刻家禽运动量二阶差分值Δm t小于预设的阈值th，则判断为正常数据。 If the second-order difference value Δm t of poultry movement amount at the tth moment is smaller than the preset threshold th, it is judged as normal data.
进一步的，在设置阈值th的过程中，可采用以下任一方法得到：Further, in the process of setting the threshold th, any of the following methods can be used to obtain:
(1)统计一定时间段内的家禽运动量二阶差分值，取该时间段内家禽运动量二阶差分值的最大值作为阈值th，其表达公式如下：(1) Count the second-order difference of poultry exercise in a certain period of time, and take the maximum value of the second-order difference of poultry exercise in this period of time as the threshold th, and its expression formula is as follows:
th＝βmaxΔmth=βmaxΔm
式中，β为取值大于1的调节参数，Δm表示一定时间段内各时刻的家禽运动量二阶差分值。在一实施例中，β取值为1.1。In the formula, β is an adjustment parameter whose value is greater than 1, and Δm represents the second-order difference value of poultry exercise at each moment within a certain period of time. In an embodiment, β takes a value of 1.1.
(2)统计一定时间段内的家禽运动量二阶差分值，根据该时间段内家禽运动量二阶差分值的均值
Figure PCTCN2022126504-appb-000002
和方差Δm 2计算阈值th，其表达公式如下： (2) Statistics of the second-order difference value of poultry exercise amount in a certain period of time, according to the mean value of the second-order difference value of poultry exercise amount in this time period
Figure PCTCN2022126504-appb-000002
and the variance Δm 2 to calculate the threshold th, the expression formula is as follows:
Figure PCTCN2022126504-appb-000003
Figure PCTCN2022126504-appb-000003
式中，θ为调节参数，取值根据具体情况设定，在一实施例中，调节参数θ的初始值取值为3。In the formula, θ is an adjustment parameter, and its value is set according to specific situations. In one embodiment, the initial value of the adjustment parameter θ is 3.
在确定异常数据后，针对该时刻的数据进行修正。本实施例中，对异常数据进行修正的步骤包括：After determining the abnormal data, correct the data at that moment. In this embodiment, the steps of correcting abnormal data include:
对于第t时刻家禽运动量m t，采用基于可线性拟合的线性假设进行修正，其表达公式如下： For the poultry movement m t at the tth moment, the linear assumption based on linear fitting is used for correction, and the expression formula is as follows:
m t,1＝2m t-1-m t-2 m t,1 = 2m t-1 -m t-2
式中，m t,1表示第t时刻家禽运动量的拟合值。 In the formula, m t,1 represents the fitting value of poultry exercise at the tth time.
获取同一组家禽运动量数据，计算该分组的家禽群体在第t时刻的众数m t,2，根据第t时刻的家禽运动量的拟合值m t,1和众数m t,2进行修正，其表达公式如下： Obtain the same group of poultry exercise data, calculate the mode m t,2 of the poultry population in this group at time t, and make corrections according to the fitted value m t,1 and mode m t,2 of poultry exercise at time t, Its expression formula is as follows:
m t＝γ·m t,1+(1-γ)m t,2 m t ＝γ·m t,1 +(1-γ)m t,2
式中，γ为调节参数，取值范围为[0,1]。在一实施例中，调节参数γ取值为0.5。In the formula, γ is an adjustment parameter, and the value range is [0,1]. In an embodiment, the value of the adjustment parameter γ is 0.5.
本实施例中，考虑到现今硬件设备较为稳定，数据时间上相邻且运动为一连续过程，本实施例假定所得到的异常数据只包含尖峰数据，即在一个小的时间领域内，群体中只有一个脚环的数据发生异常。针对其尖峰的特点，计算数据的二阶差分，利用阈值检测数据异常。针对该异常，采取两种修正方法：针对不同时间序列的个体线性修正和针对群体运动量的众数修正。将两种修正方法线性结合，得到一个异常数据的最终修正结果。In this embodiment, considering that the current hardware equipment is relatively stable, the data are adjacent in time and the movement is a continuous process, this embodiment assumes that the obtained abnormal data only contains peak data, that is, in a small time domain, the group The data of only one ankle ring is abnormal. According to the characteristics of its peak, the second order difference of the data is calculated, and the threshold is used to detect data anomalies. For this anomaly, two correction methods are adopted: individual linear correction for different time series and mode correction for group exercise. The two correction methods are linearly combined to obtain a final correction result for abnormal data.
实施例3Example 3
本实施例在实施例2提出的基于群体运动量统计特征的家禽健康评估方法的基础上作出改进。This embodiment makes improvements on the basis of the poultry health assessment method based on the statistical characteristics of group exercise proposed in embodiment 2.
本实施例提出的基于群体运动量统计特征的家禽健康评估方法中，在完成异常数据的检测和修正后，将判断为正常数据和/或完成修正的家禽运动量数据进行数据结构化，并将其存储在相应分组的缓存队列结构中；所述缓存队列结构存储有属于同一分组的N个不同家禽个体近d天的家禽个体运动量数据m i,j，以及该分组的群体运动量数据
Figure PCTCN2022126504-appb-000004
其中群体运动量数据
Figure PCTCN2022126504-appb-000005
为该分组内近d天的家禽运动量的均值，i∈[0,N]，j∈[0,d×24/Δt]，Δt为所述脚环的数据回传时间间隔。 In the poultry health assessment method based on the statistical characteristics of group exercise proposed in this embodiment, after the detection and correction of abnormal data is completed, the data of poultry exercise data judged as normal and/or the corrected poultry exercise data is data structured and stored In the cache queue structure of the corresponding group; the cache queue structure stores the individual poultry exercise data m i,j of N different poultry individuals belonging to the same group in the past d days, and the group exercise data of the group
Figure PCTCN2022126504-appb-000004
Among them, group exercise data
Figure PCTCN2022126504-appb-000005
is the average value of poultry exercise in the group in the last d days, i∈[0,N], j∈[0,d×24/Δt], and Δt is the data return time interval of the ankle ring.
本实施例中，经过异常检测及修正的家禽运动量数据将被分两类储存：个体数据和群体数据。在储存时都只保存一段时间跨度内的数据，全体数据保存为队列结构，在新一天获得数据时，将其临时存储，完整储存一天的数据后，顶替原有最早数据，保证数据的时效性，以求用于类比的特征可适合各种变化，完成时序上的自适应功能，保证算法在时间跨度上的鲁棒性。在计算的运用上，群体数据分为短期群体数据和长期群体数据两种数据使用，以求得不同时间跨度上的结果。In this embodiment, the poultry exercise data after abnormality detection and correction will be stored in two categories: individual data and group data. When storing, only the data within a period of time is saved, and all the data is saved as a queue structure. When the data is obtained in a new day, it is temporarily stored. After the data of one day is completely stored, the original earliest data is replaced to ensure the timeliness of the data. , so that the features used for analogy can be adapted to various changes, complete the adaptive function in timing, and ensure the robustness of the algorithm in the time span. In the application of calculation, group data is divided into short-term group data and long-term group data to obtain results in different time spans.
进一步的，在S3步骤中，对家禽进行个体评分的步骤包括：Further, in step S3, the step of individually scoring the poultry includes:
S3.1、选取当前时刻j 0的一个时间邻域的家禽个体运动量数据m i,j，其中j∈[j 0-δ,j 0]，δ为邻域参数；对同一时间邻域内的家禽个体历史运动量数据进行归一化操作，其表达公式如下： S3.1. Select poultry individual exercise data m i,j in a time neighborhood of the current moment j 0 , where j∈[j 0 -δ,j 0 ], δ is the neighborhood parameter; for poultry in the same time neighborhood The individual historical exercise volume data is normalized, and the expression formula is as follows:
Figure PCTCN2022126504-appb-000006
Figure PCTCN2022126504-appb-000006
式中，p i,j,k表示同一时间邻域内的家禽个体i的历史运动量数据的归一化结果，m i,j-24*k表示第k天前同一时间邻域内的家禽个体i的历史运动量数据。 In the formula, p i,j,k represents the normalization result of the historical exercise data of poultry individual i in the same time neighborhood, m i,j-24*k represents the physical activity of poultry individual i in the same time neighborhood before the kth day Historical exercise data.
在一实施例中，邻域参数δ初始化为
Figure PCTCN2022126504-appb-000007
即对2/Δt取上整。 In one embodiment, the neighborhood parameter δ is initialized as
Figure PCTCN2022126504-appb-000007
That is, round up 2/Δt.
S3.2、根据家禽运动量数据的归一化结果p i,j,k计算第k天前的个体同期比较结果s i,k，其表达公式如下： S3.2. According to the normalized results p i,j,k of poultry exercise data, calculate the individual comparison results s i,k before the kth day. The expression formula is as follows:
Figure PCTCN2022126504-appb-000008
Figure PCTCN2022126504-appb-000008
式中，ω为调节参数，取值为一接近0的常数。In the formula, ω is an adjustment parameter, and its value is a constant close to 0.
在一实施例中，调节参数ω初始化设置为0.0000000001。In one embodiment, the adjustment parameter ω is initially set to 0.0000000001.
S3.3、采用高斯卷积对k个个体同期比较结果进行融合，得到家禽个体i的个体同期评分c i，其表达公式如下： S3.3. Gaussian convolution is used to fuse the comparison results of k individuals over the same period to obtain the individual period score c i of poultry individual i, and its expression formula is as follows:
Figure PCTCN2022126504-appb-000009
Figure PCTCN2022126504-appb-000009
式中，
Figure PCTCN2022126504-appb-000010
为均值为0标准差为1的正态分布概率密度函数；σ为调节参数，且σ∈(0,5]。 In the formula,
Figure PCTCN2022126504-appb-000010
is a normal distribution probability density function with mean 0 and standard deviation 1; σ is an adjustment parameter, and σ∈(0,5].
在一实施例中，调节参数σ初始化设置为3。In an embodiment, the adjustment parameter σ is initially set to 3.
S3.4、选取当前时刻j 0的半径为δ的时间邻域上的家禽群体运动量数据
Figure PCTCN2022126504-appb-000011
然后对该时间邻域内的家禽群体运动量数据
Figure PCTCN2022126504-appb-000012
进行归一化操作，其表达公式如下： S3.4. Select the movement data of the poultry group in the time neighborhood with a radius of δ at the current moment j 0
Figure PCTCN2022126504-appb-000011
Then the poultry group exercise data in the time neighborhood
Figure PCTCN2022126504-appb-000012
Perform a normalization operation, and its expression formula is as follows:
Figure PCTCN2022126504-appb-000013
Figure PCTCN2022126504-appb-000013
式中，q j表示家禽所属分组群体的运动量数据的归一化结果； In the formula, q j represents the normalized result of the exercise data of the group to which the poultry belongs;
S3.5、结合KL散度计算群体同期评分d i，其表达公式如下： S3.5. Combined with KL divergence to calculate group contemporaneous score d i , its expression formula is as follows:
Figure PCTCN2022126504-appb-000014
Figure PCTCN2022126504-appb-000014
S3.6、对家禽个体i的个体同期评分c i和群体同期评分d i进行线性融合，得到个体评分a i，其表达公式如下： S3.6. Perform linear fusion on the individual contemporaneous score c i and the group contemporaneous score d i of poultry individual i to obtain the individual score a i , and its expression formula is as follows:
b i＝λc i+(1-λ)d i b i =λc i +(1-λ)d i
Figure PCTCN2022126504-appb-000015
Figure PCTCN2022126504-appb-000015
式中，λ为调节参数，取值范围为[0,1]；
Figure PCTCN2022126504-appb-000016
表示向上取整函数。 In the formula, λ is an adjustment parameter, and the value range is [0,1];
Figure PCTCN2022126504-appb-000016
Represents the round-up function.
在一实施例中，调节参数λ设置初始值为0.3。In an embodiment, the initial value of the adjustment parameter λ is set to 0.3.
本实施例中，为符合正常理解，将评分非线性压缩后放大至[100,0]，由上述个体评分a i的计算所得的结果，100分为最健康，0分为最不健康。将计算后的家禽评分结果储存，用于后续判断。 In this embodiment, in order to conform to the normal understanding, the scores are compressed nonlinearly and enlarged to [100,0]. According to the calculation results of the individual scores a i above, 100 is the healthiest and 0 is the unhealthiest. The calculated poultry scoring results are stored for subsequent judgment.
如图3所示，为本实施例中对家禽进行个体评分的流程图。As shown in FIG. 3 , it is a flowchart of individual scoring of poultry in this embodiment.
本实施例中主要采用KL散度的估计方法，并结合个体同期评分和群体同期评分进行个体评分计算。其中个体同期评分主要通过比较单个个体在不同日期同一时刻的数据，鉴于单个个体的运动量变化存在一定的局部性，本实施例采用高斯分布对不同日期的比较结果卷积得到个体分数。而群体同期评分直接比较个体数据与短期群体数据，最后将个体同期评分和群体同期评分进行线性融合，得到最终的个体评分。In this embodiment, the KL divergence estimation method is mainly used, and the individual score is calculated in combination with the individual contemporaneous score and the group contemporaneous score. Among them, the individual score at the same time is mainly by comparing the data of a single individual at the same time on different days. In view of the locality of the change in the amount of exercise of a single individual, this embodiment uses the Gaussian distribution to convolve the comparison results of different days to obtain the individual score. The group contemporaneous score directly compares individual data with short-term group data, and finally linearly fuses the individual contemporaneous score and the group contemporaneous score to obtain the final individual score.
进一步的，对家禽进行群体评分可采用两种方法：Further, two methods can be used for group scoring of poultry:
方法1：计算家禽对应的群体分组中N个家禽个体的个体评分a i，取均值得到家禽个体i对应的群体分组的家禽个体评分均值o i后，经过非线性压缩得到群体评分t i： Method 1: Calculate the individual score a i of N poultry individuals in the group group corresponding to the poultry, and take the mean value to obtain the average score o i of the group group corresponding to the poultry individual i, and then obtain the group score t i through nonlinear compression:
Figure PCTCN2022126504-appb-000017
Figure PCTCN2022126504-appb-000017
Figure PCTCN2022126504-appb-000018
Figure PCTCN2022126504-appb-000018
方法2：Method 2:
1)采用线性估计方法计算当前时刻j 0的理论运动量
Figure PCTCN2022126504-appb-000019
1) Use the linear estimation method to calculate the theoretical motion of j 0 at the current moment
Figure PCTCN2022126504-appb-000019
2)将当前时刻j 0的理论运动量
Figure PCTCN2022126504-appb-000020
与家禽个体评分均值
Figure PCTCN2022126504-appb-000021
进行比较，得到短期评分
Figure PCTCN2022126504-appb-000022
2) The theoretical movement amount at the current moment j 0
Figure PCTCN2022126504-appb-000020
individual poultry score mean
Figure PCTCN2022126504-appb-000021
Compare and get short-term scores
Figure PCTCN2022126504-appb-000022
Figure PCTCN2022126504-appb-000023
Figure PCTCN2022126504-appb-000023
3)选取当前时刻j 0的一个时间邻域的家禽群体运动量数据
Figure PCTCN2022126504-appb-000024
其中 j∈[j 0-δ,j 0]；对同一时间邻域内的家禽群体运动量数据进行归一化操作： 3) Select the movement data of the poultry group in a time neighborhood of the current moment j 0
Figure PCTCN2022126504-appb-000024
Where j∈[j 0 -δ,j 0 ]; normalize the movement data of poultry groups in the same time neighborhood:
Figure PCTCN2022126504-appb-000025
Figure PCTCN2022126504-appb-000025
式中，l ,j,k表示同一时间邻域内的家禽群体的历史运动量数据的归一化结果，
Figure PCTCN2022126504-appb-000026
表示第k天前同一时间邻域内的家禽群体的历史运动量数据。 In the formula, l , j, k represent the normalization results of the historical exercise data of poultry groups in the same time neighborhood,
Figure PCTCN2022126504-appb-000026
Represents the historical exercise data of the poultry group in the same time range before the kth day.
4)计算第k天前比较结果
Figure PCTCN2022126504-appb-000027
4) Calculate the comparison result before the kth day
Figure PCTCN2022126504-appb-000027
5)利用高斯卷积融合得到初步长期估计评分
Figure PCTCN2022126504-appb-000028
5) Use Gaussian convolution fusion to get preliminary long-term estimation score
Figure PCTCN2022126504-appb-000028
6)将短期评分
Figure PCTCN2022126504-appb-000029
和初步长期估计评分
Figure PCTCN2022126504-appb-000030
进行线性融合后，经过非线性压缩得到群体评分t i： 6) will short-term scoring
Figure PCTCN2022126504-appb-000029
and the initial long-term estimate score
Figure PCTCN2022126504-appb-000030
After linear fusion, the group score t i is obtained through nonlinear compression:
Figure PCTCN2022126504-appb-000031
Figure PCTCN2022126504-appb-000031
Figure PCTCN2022126504-appb-000032
Figure PCTCN2022126504-appb-000032
式中，ξ为调节参数。In the formula, ξ is an adjustment parameter.
如图4所示，为本实施例采用方法2对家禽进行群体评分的流程图。As shown in FIG. 4 , it is a flow chart of group scoring of poultry by method 2 in this embodiment.
本实施例中，方法1直接采用群体分组中N个家禽个体的个体评分a i的均值并进行非线性压缩得到，适用于群体分组中家禽个体数量较多、计算条件有限等应用场景。方法2采用小时间邻域的比较方式，将群体评分分为长期群体评估和短期群体评估，其中长期群体评估比较不同日期中同一时间邻域的分布，而后针对不同日期采用高斯卷积得到评分；短期群体评估利用变换的连续性，用短期数据估计出理论上的线性预估值，将预估值与实际值比较得到评分。经过两种评分的线性融合，得到最终评分。方法2能够更直观地反映群体分组中家禽的活跃情况。 In this embodiment, Method 1 is obtained by directly using the mean value of the individual scores a i of N poultry individuals in the group group and performing nonlinear compression, which is suitable for application scenarios such as a large number of poultry individuals in the group group and limited calculation conditions. Method 2 uses the comparison method of small temporal neighborhoods to divide group scoring into long-term group evaluation and short-term group evaluation. The long-term group evaluation compares the distribution of the same time neighborhood on different dates, and then uses Gaussian convolution to obtain scores for different dates; Short-term group assessment utilizes the continuity of transformation, uses short-term data to estimate a theoretical linear forecast value, and compares the forecasted value with the actual value to obtain a score. After a linear fusion of the two scores, the final score is obtained. Method 2 can more intuitively reflect the activity of poultry in group groups.
进一步的，结合家禽运动量数据、个体评分结果和群体评分结果对家禽健康状态进行评估，其具体步骤包括：Further, the poultry health status is evaluated in combination with poultry exercise data, individual scoring results and group scoring results, and the specific steps include:
S4.1、根据家禽i的个体评分a i及预设的第一阈值η 1、第二阈值η 2进行判断： S4.1. Judging according to the individual score a i of poultry i and the preset first threshold η 1 and second threshold η 2 :
当家禽i的个体评分a i小于预设的第一阈值η 1，则将家禽i评估为死亡； When the individual score a i of poultry i is less than the preset first threshold η 1 , the poultry i is evaluated as dead;
当家禽i的个体评分a i大于或等于预设的第一阈值η 1，且小于预设的第二阈值η 2，则将家禽i评估为患病； When the individual score a i of poultry i is greater than or equal to the preset first threshold η 1 and smaller than the preset second threshold η 2 , the poultry i is evaluated as sick;
当家禽i的个体评分a i大于或等于预设的第二阈值η 2，则将家禽i评估为正 常。 When the individual score a i of the poultry i is greater than or equal to the preset second threshold η 2 , the poultry i is evaluated as normal.
其中第一阈值η 1小于第二阈值η 2，具体根据实际应用情况设置。在一具体实施例中，第一阈值η 1设置为25，第二阈值η 2设置为100。 Wherein the first threshold η 1 is smaller than the second threshold η 2 , which is specifically set according to actual application conditions. In a specific embodiment, the first threshold η1 is set to 25, and the second threshold η2 is set to 100.
S4.2、根据家禽运动量数据计算家禽i所属群体分组z的群体运动量方差
Figure PCTCN2022126504-appb-000033
S4.2. Calculating the group exercise variance of group z to which poultry i belongs according to the poultry exercise data
Figure PCTCN2022126504-appb-000033
Figure PCTCN2022126504-appb-000034
Figure PCTCN2022126504-appb-000034
式中，m i,z为家禽i的运动量数据。 In the formula, m i, z is the exercise data of poultry i.
S4.3、根据群体运动量方差
Figure PCTCN2022126504-appb-000035
进行判断： S4.3, according to group exercise variance
Figure PCTCN2022126504-appb-000035
Make a judgment:
若家禽i的家禽运动量数据
Figure PCTCN2022126504-appb-000036
则将家禽i评估为健壮家禽； If the poultry exercise data of poultry i
Figure PCTCN2022126504-appb-000036
Then the poultry i is evaluated as a healthy poultry;
若家禽i的家禽运动量数据
Figure PCTCN2022126504-appb-000037
则将家禽i评估为残次家禽； If the poultry exercise data of poultry i
Figure PCTCN2022126504-appb-000037
Then evaluate poultry i as defective poultry;
其他情况则将家禽i评估为普通家禽；In other cases, poultry i is assessed as ordinary poultry;
其中，v a、v b为调整参数，取值范围为[0,5]。在一实施例中，v a＝v b＝3。 Wherein, v a and v b are adjustment parameters, and the value range is [0,5]. In one embodiment, v a =v b =3.
如图5所示，为本实施例对家禽健康状态进行评估的流程图。本实施例中，对家禽残次、患病、死亡、健壮的分类分为两个分类问题。As shown in FIG. 5 , it is a flow chart of evaluating the health status of poultry in this embodiment. In this embodiment, the classification of defective, diseased, dead and healthy poultry is divided into two classification problems.
第一类为患病、死亡和正常的分类问题。S4.1步骤用于评估家禽的患病、死亡和正常等健康状态，通过所得的个体健康评分基于阈值计算，其中患病、死亡可以通过短期内的评分变化表示，评分变为极低则为死亡，评分较低为患病。The first category is the classification problem of illness, death and normal. Step S4.1 is used to evaluate the health status of poultry such as sickness, death and normal, and the obtained individual health score is calculated based on the threshold value, wherein the sickness and death can be expressed by the change of the score in a short period of time, and the score becomes extremely low when it is Death, lower score is disease.
第二类为家禽的残次、健壮和普通，经过S4.2、S4.3步骤实现评估，其中残次、健壮使用群体运动量分布基于阈值取得，运动量最高的一批作为健壮家禽，运动量最小的一批作为残次家禽。The second category is the defective, robust and ordinary poultry, which are evaluated through steps S4.2 and S4.3. Among them, the group exercise distribution of defective and robust is obtained based on the threshold value. One batch as defective poultry.
实施例4Example 4
本实施例提出一种基于群体运动量统计特征的家禽健康评估系统，应用于实施例1～3任一提出的家禽健康评估方法。如图6所示，为本实施例的家禽健康评估系统的架构图。This embodiment proposes a poultry health assessment system based on statistical characteristics of group exercise, which is applied to any of the poultry health assessment methods proposed in Examples 1-3. As shown in FIG. 6, it is a structure diagram of the poultry health assessment system of this embodiment.
本实施例提出的基于群体运动量统计特征的家禽健康评估系统中，包括：In the poultry health assessment system based on group exercise statistical characteristics that the present embodiment proposes, include:
脚环1，所述脚环1配置有三轴传感器101和通信模块102，且所述脚环1佩戴设置在家禽的脚部；A foot ring 1, the foot ring 1 is configured with a three-axis sensor 101 and a communication module 102, and the foot ring 1 is worn on the foot of a poultry;
数据接收模块2，用于接收由脚环1以一定时间间隔回传的家禽数据；The data receiving module 2 is used to receive the poultry data returned by the foot ring 1 at a certain time interval;
数据预处理模块3，用于对接收的家禽数据进行预处理，得到家禽运动量数据；The data preprocessing module 3 is used to preprocess the received poultry data to obtain poultry exercise data;
异常检测和修正模块4，用于对所述家禽运动量数据进行异常数据检测和修正，得到经过修正的家禽运动量数据；Abnormality detection and correction module 4, used for abnormal data detection and correction of the poultry exercise data, to obtain corrected poultry exercise data;
个体评分模块5，用于根据所述经过修正的家禽运动量数据，对同一个体不同时间段的家禽运动量数据进行统计及数据分布，并进行个体评分；The individual scoring module 5 is used to perform statistics and data distribution on the poultry exercise data of the same individual in different time periods according to the corrected poultry exercise data, and perform individual scoring;
群体评分模块6，用于根据所述经过修正的家禽运动量数据，对同一个体所属群体分组在不同时间段的家禽运动量数据进行统计及数据分布，并进行群体评分；The group scoring module 6 is used to perform statistics and data distribution on the poultry exercise data of groups belonging to the same individual in different time periods according to the corrected poultry exercise data, and perform group scoring;
评估模块7，用于结合家禽运动量数据、个体评分结果和群体评分结果对家禽健康状态进行评估，生成家禽健康评估结果。The evaluation module 7 is used to evaluate the poultry health status by combining the poultry exercise data, individual scoring results and group scoring results, and generate poultry health evaluation results.
在具体实施过程中，将配置有三轴传感器101和通信模块102的脚环1佩戴在待评估的家禽个体，其中三轴传感器101实时采集家禽个体在三轴方向上的位移量，然后以预设的数据回传时间间隔Δt，通过通信模块102将当前时间间隔Δt时段内采集的若干时刻的家禽数据发送到数据接收模块2。In the specific implementation process, the foot ring 1 configured with the three-axis sensor 101 and the communication module 102 is worn on the poultry individual to be evaluated, wherein the three-axis sensor 101 collects the displacement of the poultry individual in the three-axis direction in real time, and then uses the preset The data return time interval Δt, and the poultry data collected at several moments in the current time interval Δt period are sent to the data receiving module 2 through the communication module 102 .
数据接收模块2接收到脚环1发出的家禽数据后传输至数据预处理模块3中进行处理。其中，数据预处理模块3计算所述脚环1当前回传的数据中检测的家禽的位移累加量、转动角分量和转动角累加量，进一步计算得到当前回传的数据中的家禽运动量，最后将计算得到的家禽运动量传输至异常检测和修正模块4。The data receiving module 2 transmits the poultry data sent by the foot ring 1 to the data preprocessing module 3 for processing. Wherein, the data preprocessing module 3 calculates the accumulated displacement, the rotation angle component and the accumulated amount of rotation angle of the poultry detected in the data currently returned by the foot ring 1, and further calculates the movement amount of the poultry in the currently returned data, and finally The calculated poultry movement is transmitted to the anomaly detection and correction module 4 .
异常检测和修正模块4对家禽运动量数据进行异常数据检测和修正时，具体的，采用二阶差分方法将第t时刻家禽运动量在小时间邻域内进行线性拟合，然后利用预设的阈值th对第t时刻家禽运动量二阶差分值进行异常数据检测，并对检测为异常的数据分别采用针对不同时间序列的个体线性修正和针对群体运动量的众数修正方法，再经过线性结合，得到一个异常数据的最终修正结果。When the abnormality detection and correction module 4 detects and corrects the abnormal data of the poultry movement data, specifically, the second order difference method is used to linearly fit the poultry movement quantity at the tth time in the small time neighborhood, and then use the preset threshold th to Abnormal data detection is performed on the second-order difference value of poultry exercise at time t, and the individual linear correction for different time series and the mode correction method for group exercise are respectively used for the data detected as abnormal, and then an abnormal data is obtained through linear combination The final correction result of .
异常检测和修正模块4将检测为正常数据以及经过修正的数据，按照对应家禽个体的日龄进行群体分组。The abnormality detection and correction module 4 will detect the normal data and the corrected data, and perform group grouping according to the age of the corresponding poultry individual.
个体评分模块5读取家禽个体的正常数据以及经过修正的数据，对同一个体不同时间段的家禽运动量数据进行统计及数据分布，并进行个体评分。The individual scoring module 5 reads the normal data and corrected data of poultry individuals, performs statistics and data distribution on the poultry exercise data of the same individual in different time periods, and performs individual scoring.
具体的，个体评分模块5采用KL散度的估计方法，并结合个体同期评分和群体同期评分进行个体评分计算。其中个体同期评分主要通过比较单个个体在不同日期同一时刻的数据，采用高斯分布对不同日期的比较结果卷积得到个体分数；群体同期评分直接比较个体数据与短期群体数据，最后将个体同期评分和群 体同期评分进行线性融合，得到最终的个体评分，并发送至评估模块7。Specifically, the individual scoring module 5 uses an estimation method of KL divergence, and combines individual contemporaneity scores and group contemporaneity scores to calculate individual scores. Among them, the individual score at the same time is mainly obtained by comparing the data of a single individual at the same time on different dates, and the Gaussian distribution is used to convolve the comparison results on different days to obtain the individual score; the group score at the same time directly compares the individual data with the short-term group data, and finally the individual score at the same time and the short-term group data The group contemporaneous scores are linearly fused to obtain the final individual scores, which are sent to the evaluation module 7 .
同时，群体评分模块6采用群体分组中N个家禽个体的个体评分a i的均值并进行非线性压缩得到群体评分结果；或采用小时间邻域的比较方式，将群体评分分为长期群体评估和短期群体评估，其中长期群体评估比较不同日期中同一时间邻域的分布，而后针对不同日期采用高斯卷积得到评分；短期群体评估利用变换的连续性，用短期数据估计出理论上的线性预估值，将预估值与实际值比较得到评分。经过两种评分的线性融合，得到最终群体评分，并发送至评估模块7。 At the same time, the group scoring module 6 adopts the mean value of the individual scores a i of N poultry individuals in the group grouping and performs nonlinear compression to obtain the group scoring result; or adopts a small time neighborhood comparison method to divide the group scoring into long-term group evaluation and Short-term group evaluation, in which long-term group evaluation compares the distribution of the same temporal neighborhood on different dates, and then uses Gaussian convolution to obtain scores for different dates; short-term group evaluation uses the continuity of the transformation to estimate the theoretical linear prediction with short-term data Value, compare the estimated value with the actual value to get a score. After a linear fusion of the two scores, the final group score is obtained and sent to the evaluation module 7 .
评估模块7接收到最终的个体评分和群体评分后，结合家禽运动量数据、个体评分结果和群体评分结果对家禽健康状态进行评估。After receiving the final individual scores and group scores, the evaluation module 7 evaluates the poultry health status in combination with the poultry exercise data, individual score results and group score results.
具体的，评估模块7首先对家禽的患病、死亡和正常状态进行评估，通过所得的个体健康评分基于阈值计算，其中患病、死亡可以通过短期内的评分变化表示，评分变为极低则为死亡，评分较低为患病。Specifically, the evaluation module 7 first evaluates the sickness, death and normal state of the poultry, and the obtained individual health score is calculated based on the threshold value, wherein the sickness and death can be represented by the score change in a short period of time, and the score becomes extremely low. for death, lower score for disease.
评估模块7进一步对家禽的家禽的残次、健壮和普通状态进行评估，其中残次、健壮使用群体运动量分布基于阈值取得，运动量最高的一批作为健壮家禽，运动量最小的一批作为残次家禽，其他情况则评估为普通家禽。The evaluation module 7 further evaluates the defective, robust and normal states of the poultry, wherein the defective and robust use group exercise distribution is obtained based on a threshold, the batch with the highest exercise amount is regarded as healthy poultry, and the batch with the smallest exercise amount is regarded as defective poultry , otherwise it is assessed as ordinary poultry.
显然，本发明的上述实施例仅仅是为清楚地说明本发明所作的举例，而并非是对本发明的实施方式的限定。对于所属领域的普通技术人员来说，在上述说明的基础上还可以做出其它不同形式的变化或变动。这里无需也无法对所有的实施方式予以穷举。凡在本发明的精神和原则之内所作的任何修改、等同替换和改进等，均应包含在本发明权利要求的保护范围之内。Apparently, the above-mentioned embodiments of the present invention are only examples for clearly illustrating the present invention, rather than limiting the implementation of the present invention. For those of ordinary skill in the art, other changes or changes in different forms can be made on the basis of the above description. It is not necessary and impossible to exhaustively list all the implementation manners here. All modifications, equivalent replacements and improvements made within the spirit and principles of the present invention shall be included within the protection scope of the claims of the present invention.

Claims (9)

  1. 一种基于群体运动量统计特征的家禽健康评估方法，其特征在于，包括以下步骤：A poultry health assessment method based on statistical characteristics of group exercise, characterized in that it comprises the following steps:
    S1、从家禽所佩戴的带三轴传感器的脚环获取家禽数据，并对获取的数据进行预处理，得到家禽运动量数据；其中，获取的家禽的数据包括家禽的位移量v，其中位移量v包括在三轴方向上的位移分量x、y、z；则对获取的数据进行预处理的步骤包括：S1. Obtain poultry data from foot rings with three-axis sensors worn by poultry, and preprocess the acquired data to obtain poultry exercise data; wherein, the acquired poultry data includes poultry displacement v, where displacement v Including displacement components x, y, z in the three-axis direction; then the steps of preprocessing the acquired data include:
    S1.1、计算所述脚环当前回传的数据中检测的家禽的位移累加量V，其表达公式如下：S1.1. Calculate the cumulative displacement V of the poultry detected in the data currently returned by the foot ring, and its expression formula is as follows:
    V＝∑ nv n V＝∑ n v n
    式中，v n表示脚环的数据回传时间间隔Δt内第n次检测的家禽的位移量； In the formula, v n represents the displacement of the poultry detected for the nth time within the data return time interval Δt of the foot ring;
    S1.2、计算数据回传时间间隔Δt内每一次检测的家禽的转动角分量w，再采用指数缩放方式计算数据回传时间间隔Δt内家禽的转动角累加量W，其表达公式如下：S1.2. Calculate the rotation angle component w of each detected poultry within the data return time interval Δt, and then use the exponential scaling method to calculate the cumulative amount W of the poultry’s rotation angle within the data return time interval Δt. The expression formula is as follows:
    Figure PCTCN2022126504-appb-100001
    Figure PCTCN2022126504-appb-100001
    Figure PCTCN2022126504-appb-100002
    Figure PCTCN2022126504-appb-100002
    式中，x n、y n、z n表示第n次检测的家禽的三轴方向上的位移分量，τ为常数；α为用于调整指数缩放比例的参数； In the formula, x n , y n , z n represent the displacement components in the three-axis direction of the poultry detected for the nth time, τ is a constant; α is a parameter used to adjust the exponential scaling;
    S1.3、根据家禽的位移累加量V和转动角累加量W，计算得到当前回传的数据中的家禽运动量m＝V+W；S1.3. According to the cumulative displacement V and the cumulative rotation angle W of the poultry, calculate the poultry movement m=V+W in the currently returned data;
    S1.4、对家禽运动量数据进行群体分组，将日龄相同的家禽对应的运动量作为一组群体数据；S1.4. Carry out group grouping to poultry exercise data, and use the corresponding exercise amount of poultry with the same day age as a group of group data;
    S2、对所述家禽运动量数据进行异常数据检测和修正，得到经过修正的家禽运动量数据；S2. Perform abnormal data detection and correction on the poultry exercise data to obtain corrected poultry exercise data;
    S3、根据所述经过修正的家禽运动量数据，对同一个体以及群体在不同时间段的家禽运动量数据进行统计及数据分布，并根据家禽运动量数据的统计及数据分布结果对家禽进行个体评分及群体评分；S3. According to the corrected poultry exercise data, perform statistics and data distribution on the poultry exercise data of the same individual and group in different time periods, and perform individual scoring and group scoring on the poultry according to the statistics and data distribution results of the poultry exercise data ;
    S4、结合家禽运动量数据、个体评分结果和群体评分结果对家禽健康状态进行评估，得到家禽健康评估结果。S4. Combining the poultry exercise data, the individual score results and the group score results to evaluate the poultry health status, and obtain the poultry health assessment results.
  2. 根据权利要求1所述的基于群体运动量统计特征的家禽健康评估方法，其特征在于，所述S2步骤中，采用二阶差分方法将第t时刻家禽运动量m t在小时间邻域内进行线性拟合，其表达公式如下： The poultry health assessment method based on the statistical characteristics of group exercise according to claim 1, wherein in the step S2, the second-order difference method is used to linearly fit the poultry exercise m at the tth moment in a small time neighborhood , its expression formula is as follows:
    Δm t＝m t+1-2m t+m t-1 Δm t ＝m t+1 -2m t +m t-1
    利用预设的阈值th对第t时刻家禽运动量二阶差分值进行异常数据检测：Use the preset threshold th to detect the abnormal data of the second-order difference value of the poultry exercise amount at the tth moment:
    若第t时刻家禽运动量二阶差分值Δm t大于或等于预设的阈值th，则判断为存在异常数据，并对第t时刻家禽运动量m t进行修正； If the second-order difference value Δm t of poultry movement at the t-th moment is greater than or equal to the preset threshold th, it is judged that there is abnormal data, and the poultry movement m t at the t-th moment is corrected;
    若第t时刻家禽运动量二阶差分值Δm t小于预设的阈值th，则判断为正常数据。 If the second-order difference value Δm t of poultry movement amount at the tth moment is smaller than the preset threshold th, it is judged as normal data.
  3. 根据权利要求2所述的基于群体运动量统计特征的家禽健康评估方法，其特征在于，所述阈值th由以下任一方法得到：The poultry health assessment method based on group exercise statistics according to claim 2, wherein said threshold th is obtained by any of the following methods:
    (1)统计一定时间段内的家禽运动量二阶差分值，取该时间段内家禽运动量二阶差分值的最大值作为阈值th，其表达公式如下：(1) Count the second-order difference of poultry exercise in a certain period of time, and take the maximum value of the second-order difference of poultry exercise in this period of time as the threshold th, and its expression formula is as follows:
    th＝βmaxΔmth=βmaxΔm
    式中，β为取值大于1的调节参数，Δm表示一定时间段内各时刻的家禽运动量二阶差分值；In the formula, β is an adjustment parameter with a value greater than 1, and Δm represents the second-order difference value of poultry exercise at each moment within a certain period of time;
    (2)统计一定时间段内的家禽运动量二阶差分值，根据该时间段内家禽运动量二阶差分值的均值
    Figure PCTCN2022126504-appb-100003
    和方差Δm 2计算阈值th，其表达公式如下：     (2) Statistics of the second-order difference value of poultry exercise amount in a certain period of time, according to the mean value of the second-order difference value of poultry exercise amount in this time period
    Figure PCTCN2022126504-appb-100003
    and the variance Δm 2 to calculate the threshold th, the expression formula is as follows:
    Figure PCTCN2022126504-appb-100004
    Figure PCTCN2022126504-appb-100004
    式中，θ为调节参数。In the formula, θ is an adjustment parameter.
  4. 根据权利要求2所述的基于群体运动量统计特征的家禽健康评估方法，其特征在于，所述S2步骤中，对异常数据进行修正的步骤包括：The poultry health assessment method based on the statistical characteristics of group exercise according to claim 2, wherein, in the S2 step, the step of correcting abnormal data includes:
    对于第t时刻家禽运动量m t，采用基于可线性拟合的线性假设进行修正，其表达公式如下： For the poultry movement m t at the tth moment, the linear assumption based on linear fitting is used for correction, and the expression formula is as follows:
    m t,1＝2m t-1-m t-2 m t,1 = 2m t-1 -m t-2
    式中，m t,1表示第t时刻家禽运动量的拟合值； In the formula, m t,1 represents the fitting value of poultry exercise at the tth moment;
    获取同一组家禽运动量数据，计算该分组的家禽群体在第t时刻的众数m t,2，根据第t时刻的家禽运动量的拟合值m t,1和众数m t,2进行修正，其表达公式如下： Obtain the same group of poultry exercise data, calculate the mode m t,2 of the poultry population in this group at time t, and make corrections according to the fitting value m t,1 and mode m t,2 of poultry exercise at time t, Its expression formula is as follows:
    m t＝γ·m t,1+(1-γ)m t,2 m t ＝γ·m t,1 +(1-γ)m t,2
    式中，γ为调节参数。In the formula, γ is an adjustment parameter.
  5. 根据权利要求1～4任一项所述的基于群体运动量统计特征的家禽健康评估方法，其特征在于，所述S2步骤中，还包括以下步骤：The poultry health assessment method based on statistical characteristics of group exercise according to any one of claims 1 to 4, wherein the step S2 further includes the following steps:
    完成异常数据检测和修正后，将判断为正常数据和/或完成修正的家禽运动量数据进行数据结构化，并将其存储在相应分组的缓存队列结构中；所述缓存队列结构存储有属于同一分组的N个不同家禽个体近d天的家禽个体运动量数据m i,j，以及该分组的群体运动量数据
    Figure PCTCN2022126504-appb-100005
    其中群体运动量数据
    Figure PCTCN2022126504-appb-100006
    为该分组内近d天的家禽运动量的均值，i∈[0,N]，j∈[0,d×24/Δt]，Δt为所述脚环的数据回传时间间隔。     After the abnormal data detection and correction are completed, the data is structured as normal data and/or the corrected poultry exercise data, and stored in the cache queue structure of the corresponding group; the cache queue structure stores data belonging to the same group The poultry individual exercise data m i,j of N different poultry individuals in the past d days, and the group exercise data of this group
    Figure PCTCN2022126504-appb-100005
    Among them, group exercise data
    Figure PCTCN2022126504-appb-100006
    is the average value of poultry exercise in the group in the last d days, i∈[0,N], j∈[0,d×24/Δt], and Δt is the data return time interval of the ankle ring.
  6. 根据权利要求5所述的基于群体运动量统计特征的家禽健康评估方法，其特征在于，所述S3步骤中，对家禽进行个体评分的步骤包括：The poultry health assessment method based on group exercise statistical characteristics according to claim 5, wherein, in the S3 step, the step of carrying out individual scoring to poultry comprises:
    S3.1、选取当前时刻j 0的一个时间邻域的家禽个体运动量数据m i,j，其中j∈[j 0-δ,j 0]，δ为邻域参数；对同一时间邻域内的家禽个体历史运动量数据进行归一化操作，其表达公式如下： S3.1. Select poultry individual exercise data m i,j in a time neighborhood of the current moment j 0 , where j∈[j 0 -δ,j 0 ], δ is the neighborhood parameter; for poultry in the same time neighborhood The individual historical exercise volume data is normalized, and the expression formula is as follows:
    Figure PCTCN2022126504-appb-100007
    Figure PCTCN2022126504-appb-100007
    式中，p i,j,k表示同一时间邻域内的家禽个体i的历史运动量数据的归一化结果，m i,j-24*k表示第k天前同一时间邻域内的家禽个体i的历史运动量数据； In the formula, p i,j,k represents the normalization result of the historical exercise data of poultry individual i in the same time neighborhood, m i,j-24*k represents the physical activity of poultry individual i in the same time neighborhood before the kth day Historical exercise data;
    S3.2、根据家禽运动量数据的归一化结果p i,j,k计算第k天前的个体同期比较结果s i,k，其表达公式如下： S3.2. According to the normalized results p i,j,k of poultry exercise data, calculate the individual comparison results s i,k before the kth day. The expression formula is as follows:
    Figure PCTCN2022126504-appb-100008
    Figure PCTCN2022126504-appb-100008
    式中，ω为调节参数；In the formula, ω is the adjustment parameter;
    S3.3、采用高斯卷积对k个个体同期比较结果进行融合，得到家禽个体i的个体同期评分c i，其表达公式如下： S3.3. Gaussian convolution is used to fuse the comparison results of k individuals over the same period to obtain the individual period score c i of poultry individual i, and its expression formula is as follows:
    Figure PCTCN2022126504-appb-100009
    Figure PCTCN2022126504-appb-100009
    式中，
    Figure PCTCN2022126504-appb-100010
    为均值为0标准差为1的正态分布概率密度函数；σ为调节参数，且σ∈(0,5]；     In the formula,
    Figure PCTCN2022126504-appb-100010
    is a normal distribution probability density function with mean 0 and standard deviation 1; σ is an adjustment parameter, and σ∈(0,5];
    S3.4、选取当前时刻j 0的半径为δ的时间邻域上的家禽群体运动量数据
    Figure PCTCN2022126504-appb-100011
    然后对该时间邻域内的家禽群体运动量数据
    Figure PCTCN2022126504-appb-100012
    进行归一化操作，其表达公式如 下：     S3.4. Select the movement data of the poultry group in the time neighborhood with a radius of δ at the current moment j 0
    Figure PCTCN2022126504-appb-100011
    Then the poultry group exercise data in the time neighborhood
    Figure PCTCN2022126504-appb-100012
    Perform a normalization operation, and its expression formula is as follows:
    Figure PCTCN2022126504-appb-100013
    Figure PCTCN2022126504-appb-100013
    式中，q j表示家禽所属分组群体的运动量数据的归一化结果； In the formula, q j represents the normalized result of the exercise data of the group to which the poultry belongs;
    S3.5、结合KL散度计算群体同期评分d i，其表达公式如下： S3.5. Combined with KL divergence to calculate group contemporaneous score d i , its expression formula is as follows:
    Figure PCTCN2022126504-appb-100014
    Figure PCTCN2022126504-appb-100014
    S3.6、对家禽个体i的个体同期评分c i和群体同期评分d i进行线性融合，得到个体评分a i，其表达公式如下： S3.6. Perform linear fusion on the individual contemporaneous score c i and the group contemporaneous score d i of poultry individual i to obtain the individual score a i , and its expression formula is as follows:
    b i＝λc i+(1-λ)d i b i =λc i +(1-λ)d i
    Figure PCTCN2022126504-appb-100015
    Figure PCTCN2022126504-appb-100015
    式中，λ为调节参数，
    Figure PCTCN2022126504-appb-100016
    表示向上取整函数。     In the formula, λ is the adjustment parameter,
    Figure PCTCN2022126504-appb-100016
    Represents the round-up function.
  7. 根据权利要求6所述的基于群体运动量统计特征的家禽健康评估方法，其特征在于，所述S3步骤中，对家禽进行群体评分的步骤包括以下任一步骤：The poultry health assessment method based on group exercise statistics according to claim 6, characterized in that, in the S3 step, the step of carrying out group scores to poultry includes any of the following steps:
    (1)计算家禽对应的群体分组中N个家禽个体的个体评分a i，取均值后经过非线性压缩得到群体评分t i： (1) Calculate the individual score a i of the N poultry individuals in the group group corresponding to the poultry, take the mean value and perform nonlinear compression to obtain the group score t i :
    Figure PCTCN2022126504-appb-100017
    Figure PCTCN2022126504-appb-100017
    Figure PCTCN2022126504-appb-100018
    Figure PCTCN2022126504-appb-100018
    式中，o i表示家禽个体i对应的群体分组中的家禽个体评分均值； In the formula, o i represents the mean value of poultry individual scores in the group group corresponding to poultry individual i;
    (2)采用线性估计方法计算当前时刻j 0的理论运动量
    Figure PCTCN2022126504-appb-100019
    (2) Use the linear estimation method to calculate the theoretical motion of j 0 at the current moment
    Figure PCTCN2022126504-appb-100019
    Figure PCTCN2022126504-appb-100020
    Figure PCTCN2022126504-appb-100020
    将当前时刻j 0的理论运动量
    Figure PCTCN2022126504-appb-100021
    与家禽个体评分均值
    Figure PCTCN2022126504-appb-100022
    进行比较，得到短期评分
    Figure PCTCN2022126504-appb-100023
    The theoretical movement amount of the current moment j 0
    Figure PCTCN2022126504-appb-100021
    individual poultry score mean
    Figure PCTCN2022126504-appb-100022
    Compare and get short-term scores
    Figure PCTCN2022126504-appb-100023
    Figure PCTCN2022126504-appb-100024
    Figure PCTCN2022126504-appb-100024
    选取当前时刻j 0的一个时间邻域的家禽群体运动量数据
    Figure PCTCN2022126504-appb-100025
    其中j∈[j 0-δ,j 0]；对同一时间邻域内的家禽群体运动量数据进行归一化操作：     Select the movement data of the poultry group in a time neighborhood at the current moment j 0
    Figure PCTCN2022126504-appb-100025
    Where j∈[j 0 -δ,j 0 ]; normalize the movement data of poultry groups in the same time neighborhood:
    Figure PCTCN2022126504-appb-100026
    Figure PCTCN2022126504-appb-100026
    式中，l ,j,k表示同一时间邻域内的家禽群体的历史运动量数据的归一化结果，
    Figure PCTCN2022126504-appb-100027
    表示第k天前同一时间邻域内的家禽群体的历史运动量数据；     In the formula, l , j, k represent the normalization results of the historical exercise data of poultry groups in the same time neighborhood,
    Figure PCTCN2022126504-appb-100027
    Represents the historical exercise data of the poultry group in the same time neighborhood before the kth day;
    计算第k天前比较结果
    Figure PCTCN2022126504-appb-100028
    Calculate the comparison result before the kth day
    Figure PCTCN2022126504-appb-100028
    Figure PCTCN2022126504-appb-100029
    Figure PCTCN2022126504-appb-100029
    利用高斯卷积融合得到初步长期估计评分
    Figure PCTCN2022126504-appb-100030
    Gaussian Convolution Fusion to Get Initial Long-Term Estimation Score
    Figure PCTCN2022126504-appb-100030
    Figure PCTCN2022126504-appb-100031
    Figure PCTCN2022126504-appb-100031
    将短期评分
    Figure PCTCN2022126504-appb-100032
    和初步长期估计评分
    Figure PCTCN2022126504-appb-100033
    进行线性融合后，经过非线性压缩得到群体评分t i：     Will short-term score
    Figure PCTCN2022126504-appb-100032
    and the initial long-term estimate score
    Figure PCTCN2022126504-appb-100033
    After linear fusion, the group score t i is obtained through nonlinear compression:
    Figure PCTCN2022126504-appb-100034
    Figure PCTCN2022126504-appb-100034
    Figure PCTCN2022126504-appb-100035
    Figure PCTCN2022126504-appb-100035
    式中，ξ为调节参数。In the formula, ξ is an adjustment parameter.
  8. 根据权利要求7所述的基于群体运动量统计特征的家禽健康评估方法，其特征在于，所述S4步骤中，结合家禽运动量数据、个体评分结果和群体评分结果对家禽健康状态进行评估的步骤包括：The poultry health assessment method based on group exercise statistical characteristics according to claim 7, wherein in the S4 step, the step of evaluating poultry health status in combination with poultry exercise data, individual scoring results and group scoring results comprises:
    S4.1、根据家禽i的个体评分a i及预设的第一阈值η 1、第二阈值η 2进行判断： S4.1. Judging according to the individual score a i of poultry i and the preset first threshold η 1 and second threshold η 2 :
    当家禽i的个体评分a i小于预设的第一阈值η 1，则将家禽i评估为死亡； When the individual score a i of poultry i is less than the preset first threshold η 1 , the poultry i is evaluated as dead;
    当家禽i的个体评分a i大于或等于预设的第一阈值η 1，且小于预设的第二阈值η 2，则将家禽i评估为患病； When the individual score a i of poultry i is greater than or equal to the preset first threshold η 1 and smaller than the preset second threshold η 2 , the poultry i is evaluated as sick;
    当家禽i的个体评分a i大于或等于预设的第二阈值η 2，则将家禽i评估为正常； When the individual score a i of poultry i is greater than or equal to the preset second threshold η 2 , the poultry i is evaluated as normal;
    S4.2、根据家禽运动量数据计算家禽i所属群体分组z的群体运动量方差
    Figure PCTCN2022126504-appb-100036
    S4.2. Calculating the group exercise variance of group z to which poultry i belongs according to the poultry exercise data
    Figure PCTCN2022126504-appb-100036
    Figure PCTCN2022126504-appb-100037
    Figure PCTCN2022126504-appb-100037
    式中，m i,z为家禽i的运动量数据； In the formula, m i, z are the exercise data of poultry i;
    S4.3、根据群体运动量方差
    Figure PCTCN2022126504-appb-100038
    进行判断：     S4.3, according to group exercise variance
    Figure PCTCN2022126504-appb-100038
    Make a judgment:
    若家禽i的家禽运动量数据
    Figure PCTCN2022126504-appb-100039
    则将家禽i评估为健壮家禽；     If the poultry exercise data of poultry i
    Figure PCTCN2022126504-appb-100039
    Then the poultry i is evaluated as a healthy poultry;
    若家禽i的家禽运动量数据
    Figure PCTCN2022126504-appb-100040
    则将家禽i评估为残次家禽；     If the poultry exercise data of poultry i
    Figure PCTCN2022126504-appb-100040
    Then evaluate poultry i as defective poultry;
    其他情况则将家禽i评估为普通家禽；In other cases, poultry i is assessed as ordinary poultry;
    其中，v a、v b为调整参数。 Among them, v a and v b are adjustment parameters.
  9. 一种基于群体运动量统计特征的家禽健康评估系统，应用于权利要求1～8任一项所述的家禽健康评估方法，其特征在于，包括：A poultry health assessment system based on statistical characteristics of group exercise, applied to the poultry health assessment method according to any one of claims 1 to 8, characterized in that it includes:
    脚环，所述脚环配置有三轴传感器和通信模块，且所述脚环佩戴设置在家禽的脚部；A foot ring, the foot ring is configured with a three-axis sensor and a communication module, and the foot ring is worn on the foot of the poultry;
    数据接收模块，用于接收由脚环以一定时间间隔回传的家禽数据；The data receiving module is used to receive the poultry data returned by the foot ring at a certain time interval;
    数据预处理模块，用于对接收的家禽数据进行预处理，得到家禽运动量数据；The data preprocessing module is used to preprocess the received poultry data to obtain poultry exercise data;
    异常检测和修正模块，用于对所述家禽运动量数据进行异常数据检测和修正，得到经过修正的家禽运动量数据；An abnormality detection and correction module is used for abnormal data detection and correction of the poultry exercise data to obtain corrected poultry exercise data;
    个体评分模块，用于根据所述经过修正的家禽运动量数据，对同一个体不同时间段的家禽运动量数据进行统计及数据分布，并进行个体评分；The individual scoring module is used to perform statistics and data distribution on the poultry exercise data of the same individual in different time periods according to the corrected poultry exercise data, and perform individual scoring;
    群体评分模块，用于根据所述经过修正的家禽运动量数据，对同一个体所属群体分组在不同时间段的家禽运动量数据进行统计及数据分布，并进行群体评分；The group scoring module is used to perform statistics and data distribution on the poultry exercise data of groups belonging to the same individual in different time periods according to the corrected poultry exercise data, and perform group scoring;
    评估模块，用于结合家禽运动量数据、个体评分结果和群体评分结果对家禽健康状态进行评估，生成家禽健康评估结果。The evaluation module is used to evaluate the poultry health status by combining the poultry exercise data, individual scoring results and group scoring results, and generate poultry health evaluation results.
PCT/CN2022/126504 2022-03-01 2022-10-20 Poultry health assessment method and system based on group motion amount statistical characteristics WO2023165141A1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210190861.5A CN114258877B (en) 2022-03-01 2022-03-01 Poultry health assessment method and system based on group motion quantity statistical characteristics
CN202210190861.5 2022-03-01

Publications (1)

Publication Number Publication Date
WO2023165141A1 true WO2023165141A1 (en) 2023-09-07

Family

ID=80833779

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/126504 WO2023165141A1 (en) 2022-03-01 2022-10-20 Poultry health assessment method and system based on group motion amount statistical characteristics

Country Status (2)

Country Link
CN (1) CN114258877B (en)
WO (1) WO2023165141A1 (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114258877B (en) * 2022-03-01 2022-05-31 华南农业大学 Poultry health assessment method and system based on group motion quantity statistical characteristics
CN115985502B (en) * 2023-03-16 2023-06-13 华南农业大学 Lion head goose health decision method and device based on multi-element perception
CN116701886B (en) * 2023-08-07 2023-10-27 知心健(南京)科技有限公司 Intelligent analysis management system for exercise lung test data

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1543721A1 (en) * 2003-12-04 2005-06-22 Gantner Pigeon Systems GmbH Method and device for monitoring the health of animals
CN107135983A (en) * 2017-05-05 2017-09-08 北京农业信息技术研究中心 Poultry health monitoring method, motion pin ring, server and system
CN108040918A (en) * 2017-12-07 2018-05-18 上海海事大学 A kind of new chicken feet ring
CN113575459A (en) * 2021-07-23 2021-11-02 华南农业大学 Cow nose ring and cow health monitoring device and method thereof
CN114258877A (en) * 2022-03-01 2022-04-01 华南农业大学 Poultry health assessment method and system based on group motion quantity statistical characteristics

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN209000052U (en) * 2018-12-28 2019-06-18 仲恺农业工程学院 Poultry behavior analysis system based on big data analysis

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1543721A1 (en) * 2003-12-04 2005-06-22 Gantner Pigeon Systems GmbH Method and device for monitoring the health of animals
CN107135983A (en) * 2017-05-05 2017-09-08 北京农业信息技术研究中心 Poultry health monitoring method, motion pin ring, server and system
CN108040918A (en) * 2017-12-07 2018-05-18 上海海事大学 A kind of new chicken feet ring
CN113575459A (en) * 2021-07-23 2021-11-02 华南农业大学 Cow nose ring and cow health monitoring device and method thereof
CN114258877A (en) * 2022-03-01 2022-04-01 华南农业大学 Poultry health assessment method and system based on group motion quantity statistical characteristics

Also Published As

Publication number Publication date
CN114258877A (en) 2022-04-01
CN114258877B (en) 2022-05-31

Similar Documents

Publication Publication Date Title
WO2023165141A1 (en) Poultry health assessment method and system based on group motion amount statistical characteristics
Jun et al. Estimating pig weights from images without constraint on posture and illumination
MX2015006775A (en) Systems and methods for predicting the outcome of a state of a subject.
Zhang et al. Design of sick chicken automatic detection system based on improved residual network
CN111681209B (en) Intelligent detection system for cleavage state of blastomere
WO2019126971A1 (en) Pest disaster monitoring method based on machine vision
CN112257608A (en) Yak breeding health state monitoring method
KR102425523B1 (en) Method and appartus for checking symptom of pig and managing cloud service
CN116033142B (en) Ambient light measurement method and system based on camera device
CN111680587B (en) Multi-target tracking-based chicken flock activity real-time estimation method and system
KR102047428B1 (en) Health care system for cow using detecting rumination
CN114980011B (en) Livestock and poultry body temperature monitoring method with cooperation of wearable sensor and infrared camera
CN116824626A (en) Artificial intelligent identification method for abnormal state of animal
CN113378638B (en) Method for identifying abnormal behavior of turbine operator based on human body joint point detection and D-GRU network
CN113456063B (en) Artificial intelligence-based dead chicken disease detection system and detection method
Pacheco et al. Deep learning-based model classifies thermal conditions in dairy cows using infrared thermography
CN117115688A (en) Dead fish identification and counting system and method based on deep learning under low-brightness environment
CN116563758A (en) Lion head goose monitoring method, device, equipment and storage medium
CN117029904A (en) Intelligent cage-rearing poultry inspection system
CN111652084B (en) Abnormal layer identification method and device
CN112115888B (en) Plant disease diagnosis system based on disease spot correlation
US20220383652A1 (en) Monitoring Animal Pose Dynamics from Monocular Images
Xie et al. Behavior Recognition of a Broiler Chicken using Long Short-Term Memory with Convolution Neural Networks
Tao et al. Implementation of kitchen food safety regulations detection system based on deep learning
CN114793948B (en) Sow oestrus intelligent management system

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22929570

Country of ref document: EP

Kind code of ref document: A1