Method for evaluating cold stress of livestock in cold and arid region

Abstract

A cold stress evaluation method for livestock in cold and arid regions, characterized in that the evaluation method comprises the following steps: preprocessing related data, establishing an evaluation index system; constructing a cold stress fuzzy comprehensive evaluation model; applying the cold stress fuzzy comprehensive evaluation model, analyzing the cold stress comprehensive evaluation results, and obtaining accurate evaluation results of the cold stress of livestock; the specific cold stress conditions of the livestock group and different individuals in the livestock group can be accurately judged, thereby providing reasonable guidance and decision-making for the cold region farm in terms of warmth, ventilation, feed adjustment and medication.

Description

Technical Field

[0001] This invention relates to a method for evaluating cold stress in livestock in cold and arid regions. Background Technology

[0002] Cold stress refers to the various physiological responses that animals produce in order to maintain their core body temperature when the ambient temperature drops to the lower limit of their thermoneutrality zone. After experiencing cold stress, most of the energy from the feed that livestock ingest is used for heat production rather than muscle growth, resulting in reduced feed conversion rate, slower weight gain, and impaired immune function.

[0003] Currently, the assessment of cold stress mainly relies on the temperature and humidity index and wind chill, which can be calculated by measuring only a few environmental parameters. However, it only considers factors such as temperature, humidity, and wind speed, without taking into account the impact of corresponding air pollutants. Furthermore, in order to increase the temperature inside the livestock shed, doors and windows need to be kept closed, which causes these air pollutants to accumulate more and more. Since they are not included in the air quality index, the overall stress load of livestock is greater than that indicated by a single temperature index.

[0004] Furthermore, traditional indices provide uniform values ​​at the group level, which cannot reflect the differences in stress response between individuals with different physical conditions within the same group, and therefore cannot accurately assess cold stress in livestock in cold and arid regions. Summary of the Invention

[0005] This invention provides a method for evaluating cold stress in livestock in cold and arid regions. The method is rationally designed, with the primary goal of accurate multi-indicator evaluation of cold stress. It is implemented through three levels: on-site monitoring, data modeling, and process algorithms. Thermal environment parameters, air quality indicators, and livestock ear tag body temperatures are collected and aggregated. An entropy weight algorithm is used to determine the specific weights of each indicator, and a five-level cold stress classification model based on fuzzy comprehensive evaluation is constructed. This model accurately determines the specific cold stress status of livestock herds and different individuals within herds, thus providing reasonable guidance and decisions for livestock farms in high-altitude and cold regions regarding insulation, ventilation, feed adjustment, and medication. This addresses the problems existing in current technologies.

[0006] The technical solution adopted by the present invention to solve the above-mentioned technical problems is as follows: A method for evaluating cold stress in livestock in arid and cold regions, the method comprising the following steps: S1, preprocess the relevant data and establish an evaluation index system; S2, Construct a fuzzy comprehensive evaluation model for cold stress; S3 applies the fuzzy comprehensive assessment model for cold stress, analyzes the comprehensive assessment results of cold stress, and obtains accurate evaluation results of cold stress in livestock.

[0007] The preprocessing of relevant data and the establishment of an evaluation index system include the following steps: S1.1, Obtain the characteristics of the raw monitoring data and clean up outliers; S1.2, Data expansion based on Monte Carlo simulation; S1.3, Construct a comprehensive evaluation index system for cold stress; Obtaining the characteristics of raw monitoring data and cleaning outliers includes the following steps: S1.1.1, adopt an indoor environmental monitoring system, an outdoor weather station and a livestock ear tag system, and set corresponding multi-category collection time intervals to obtain multi-source raw monitoring data; S1.1.2, three criteria are set to identify and remove outliers: physical law outliers, statistical law outliers, and time-related outliers. Physical law outliers are values ​​that violate basic physical or physiological common sense. Statistical law outliers are data points with extremely extreme values ​​that do not trigger physical law criteria. Box plots are used for judgment. Specifically, a statistical law outlier is marked when the value of a certain indicator at a certain moment is greater than or less than three times the difference between the upper and lower quartiles of all valid data for that indicator.

[0008] A three-times-wide spacing between fans is used instead to prevent environmental changes caused by normal operating conditions such as sudden weather changes, fan start-up and shutdown from being mistaken for anomalies. After outliers are marked, they are not immediately removed, but the original data and field records are retrieved to find a reasonable explanation. If the reasonableness can be determined, it is saved; if the reasonableness cannot be determined, it is represented by missing value (NaN). The time-related outliers are applicable to data points that show consecutive sharp jumps and do not recover afterward. The trend of data before and after the time period is used to make a comprehensive judgment. If all data in the entire time period shows abnormal fluctuations, the entire time period is marked as suspicious. S1.1.3, For some random missing values ​​that are not part of the above-mentioned segmental anomalies, linear interpolation is used to fill them in. The linear interpolation method is as follows:

[0009] Where t1 is the most recent complete time before the missing point, t2 is the most recent complete time after the missing point, and x is the corresponding missing value; After the missing values ​​are filled, the three sets of data should be aligned according to a unified timeline.

[0010] Data augmentation based on Monte Carlo simulations includes the following steps: S1.2.1, Simulation parameter settings and statistical characteristic constraints: For temperature-related indicators, simulation constraints are obtained by superimposing a daily variation baseline with random perturbations. The superimposed random perturbation term follows a normal distribution N(0, σ²), with σ taking 0.8 to 1.2 times the standard deviation of the original data. For concentration-related indicators, which follow a log-normal distribution, random numbers satisfying the log-normal distribution are generated during simulation, and then truncated according to the extreme value range of the original data. For ear temperature indicators, an additional covariance constraint with Tin is applied. The Pearson correlation coefficient r between Tin and EarTemp is calculated based on the original data. This correlation structure is introduced when generating simulated Tin and EarTemp, so that the generated data statistically reproduces the true correlation direction between ambient temperature and ear temperature in the original data. S1.2.2 compares the distribution characteristics of the main indicators in the simulated dataset with the original data, expands the validity verification of the dataset, removes non-physiological abnormal fluctuation factors, and makes the simulated ear temperature closer to the stable variation range under physiological conditions.

[0011] The steps to construct a comprehensive evaluation index system for cold stress are as follows: S1.3.1, perform indicator homogenization processing to ensure that all indicators are in the same direction, and analyze the substantive relationship between each indicator and cold stress risk; S1.3.2, the range standardization method is used to uniformly map each indicator to the interval [0, 1]. The standardization formula for the positive indicator is:

[0012] Where, x i Let X be the original value, min(X) and max(X) be the minimum and maximum values ​​of the index in all samples, respectively, and Z(x) be the index. i The value of )∈[0,1] indicates that the relative level of the indicator in the dataset is higher, that is, the risk of cold stress is higher; S1.3.3 Descriptive statistical analysis was performed on the six standardized indicators to construct a comprehensive evaluation index system for cold stress. The indicators include indoor temperature (Tin), relative humidity (RH), CO2 concentration, NH3 concentration, PM1 concentration, and outdoor wind speed (Wind).

[0013] The steps involved in constructing a fuzzy comprehensive assessment model for cold stress are as follows: S2.1, Calculate objective weights based on the entropy weight method; S2.2, Construct the cold stress membership function and evaluation set; S2.3 uses the core computing module to analyze and build a cold stress fuzzy comprehensive evaluation model, and provides visualization output and result storage; Calculating objective weights based on the entropy weight method includes the following steps: S2.1.1 constructs an evaluation matrix using the standardized data of the six indicators and performs automated calculations using the entropyWeights function; S2.1.2, after standardization, the information entropy and specific entropy weights corresponding to the six indicator data are obtained; Section S2.1.3 discusses and analyzes the weighting results.

[0014] Constructing the cold stress membership function and evaluation set includes the following steps: S2.2.1, classifying cold stress levels, the five-level cold stress system is qualitatively described as follows: Unstressed animals are in a thermally comfortable environment and do not require additional physiological compensation; Mildly stressed animals begin to show peripheral vasoconstriction, slightly reduced activity, and production performance is not significantly affected; Moderately stressed animals exhibit shivering-induced thermogenesis, increased feed intake, and decreased feed conversion rate; Severely stressed animals continue to shiver, huddle together, and their immune function begins to decline; Extremely stressed animals experience a drop in body temperature, significantly reduced activity, and an immediate risk of health damage. S2.2.2, the comprehensive score S_new is used to determine the level threshold and comment set for cold stress, wherein the comment set is: V = {v1, v2, v3, v4, v5} = no stress, mild stress, moderate stress, severe stress, extreme stress; The formula for defining the comprehensive score S_new is:

[0015] Among them, W j Z is the entropy weight of the j-th index. j Let S_new be the standardized value of the j-th indicator, where S_new ∈ [0, 1]. The larger S_new is, the higher the risk of cold stress. The grade threshold is set as a four-segment threshold:

[0016] The specific division rules are as follows: S_new < 0.25 → no stress, 0.25 ≤ S_new < 0.40 → mild stress, 0.40 ≤ S_new < 0.55 → moderate stress, 0.55 ≤ S_new < 0.70 → severe stress, S_new ≥ 0.70 → extreme stress; S2.2.3, a combination of trapezoidal and semi-trapezoidal membership functions is used to evaluate the membership degree of each level threshold. The membership function for each level threshold is constructed using the comprehensive score S_new as the independent variable. The membership function μ1(x) for the no-stress level is a descending semi-trapezoidal function. The formula is as follows:

[0017] The membership function μ2(x) for mild stress levels is an intermediate trapezoidal function with the following formula:

[0018] For each evaluation time point i, its membership vector μ to the five levels is calculated. i = [μ i1 μ i2 μ i3 μ i4 μ i5 The final cold stress level at that time point is determined based on the principle of maximum membership.

[0019] The following steps are taken to apply the fuzzy comprehensive assessment model for cold stress and analyze the comprehensive assessment results to obtain accurate evaluation results of cold stress in livestock: S3.1, Obtain the environmental parameters and spatiotemporal distribution characteristics of livestock sheds; S3.2, Analyze the comprehensive assessment results of cold stress; S3.3, perform model performance verification and comparative analysis to obtain accurate evaluation results of livestock cold stress.

[0020] Obtaining livestock barn environmental parameters and spatiotemporal distribution characteristics includes the following steps: S3.1.1, Determine the temporal variation patterns of temperature and humidity; S3.1.2, determine the dynamic change law of harmful gas and particulate matter concentration; S3.1.3, Analyze the impact of outdoor meteorological conditions; The analysis of the comprehensive assessment results of cold stress includes the following steps: S3.2.1, Construct the time-series distribution of the comprehensive cold stress score; S3.2.2, Count the frequency of relevant time points according to the cold stress level; S3.2.3, the mean values ​​of each core indicator under the five cold stress levels are calculated to obtain the mean value characteristics of environmental and physiological indicators under different levels; S3.2.4 After standardizing the mean values ​​of each indicator by range, a radar chart is drawn to visually compare the characteristics of environmental parameter combinations at each level.

[0021] To conduct model performance validation and comparative analysis and obtain accurate evaluation results of livestock cold stress, the following steps are included: S3.3.1 Correlation analysis was performed between the three indicators—temperature and humidity index, wind chill temperature, and comprehensive score—and the standardized inverse values ​​of ear tag temperature to verify the model performance. S3.3.2, Analyze the response of ear tag temperature to cold stress score and obtain ear temperature distribution box plot; S3.3.3, the contribution rate C of each environmental factor to the comprehensive score, which is composed of entropy weight and standardized value. j for:

[0022] S3.3.4 Based on the above analysis and combined with the specific environmental conditions of the livestock shed in winter, an accurate evaluation result of cold stress in livestock is obtained.

[0023] This invention employs the aforementioned method to obtain a standard evaluation index matrix through preprocessing of raw data. It uses multi-source monitoring data and combines physical and statistical methods to identify outliers. Logical judgment and box plot methods are used to remove abnormal records such as sensor reset errors and ear tag detachment distortions. Random missing values ​​are filled in using linear interpolation and multi-source time alignment is performed. An applicability analysis of cold stress evaluation is conducted using the entropy weight method and a five-step calculation process. The rationality of the weighting results is analyzed from both information content and actual production perspectives. An entropy weight-fuzzy comprehensive evaluation model is established to provide a complete tool for result analysis. By applying the cold stress fuzzy comprehensive evaluation model and analyzing the cold stress comprehensive evaluation results, an accurate evaluation result of the overall cold stress in livestock herds is obtained, which has the advantages of accuracy, practicality, safety, and reliability. Attached Figure Description

[0024] Figure 1 This is a schematic diagram of the process of the present invention.

[0025] Figure 2 This is a statistical characteristic table of core indicators of the original data after cleaning, as presented in this invention.

[0026] Figure 3 This is a graph showing the weight calculation results of the entropy weight method of the present invention.

[0027] Figure 4 This is a bar chart showing the entropy weight distribution of the various indicators in this invention.

[0028] Figure 5 This is a schematic diagram of the membership function of the cold stress level in this invention.

[0029] Figure 6 This is a standardized comparison radar chart of environmental parameters under different cold stress levels according to the present invention.

[0030] Figure 7 Box plots showing the ear tag temperature distribution under different cold stress levels according to the present invention. Detailed Implementation

[0031] To clearly illustrate the technical features of this solution, the invention will be described in detail below through specific implementation methods and in conjunction with the accompanying drawings.

[0032] like Figures 1-7 As shown, a method for evaluating cold stress in livestock in arid and cold regions includes the following steps: S1, preprocess the relevant data and establish an evaluation index system; S2, Construct a fuzzy comprehensive evaluation model for cold stress; S3 applies the fuzzy comprehensive assessment model for cold stress, analyzes the comprehensive assessment results of cold stress, and obtains accurate evaluation results of cold stress in livestock.

[0033] The preprocessing of relevant data and the establishment of an evaluation index system include the following steps: S1.1, Obtain the characteristics of the raw monitoring data and clean up outliers; S1.2, Data expansion based on Monte Carlo simulation; S1.3, Construct a comprehensive evaluation index system for cold stress; Obtaining the characteristics of raw monitoring data and cleaning outliers includes the following steps: S1.1.1, adopt an indoor environmental monitoring system, an outdoor weather station and a livestock ear tag system, and set corresponding multi-category collection time intervals to obtain multi-source raw monitoring data; S1.1.2, three criteria are set to identify and remove outliers: physical law outliers, statistical law outliers, and time-related outliers. Physical law outliers are values ​​that violate basic physical or physiological common sense. Statistical law outliers are data points with extremely extreme values ​​that do not trigger physical law criteria. Box plots are used for judgment. Specifically, a statistical law outlier is marked when the value of a certain indicator at a certain moment is greater than or less than three times the difference between the upper and lower quartiles of all valid data for that indicator.

[0034] A three-times-wide spacing between fans is used instead to prevent environmental changes caused by normal operating conditions such as sudden weather changes, fan start-up and shutdown from being mistaken for anomalies. After outliers are marked, they are not immediately removed, but the original data and field records are retrieved to find a reasonable explanation. If the reasonableness can be determined, it is saved; if the reasonableness cannot be determined, it is represented by missing value (NaN). The time-related outliers are applicable to data points that show consecutive sharp jumps and do not recover afterward. The trend of data before and after the time period is used to make a comprehensive judgment. If all data in the entire time period shows abnormal fluctuations, the entire time period is marked as suspicious. S1.1.3, For some random missing values ​​that are not part of the above-mentioned segmental anomalies, linear interpolation is used to fill them in. The linear interpolation method is as follows:

[0035] Where t1 is the most recent complete time before the missing point, t2 is the most recent complete time after the missing point, and x is the corresponding missing value; After the missing values ​​are filled, the three sets of data should be aligned according to a unified timeline.

[0036] Data augmentation based on Monte Carlo simulations includes the following steps: S1.2.1, Simulation parameter settings and statistical characteristic constraints: For temperature-related indicators, simulation constraints are obtained by superimposing a daily variation baseline with random perturbations. The superimposed random perturbation term follows a normal distribution N(0, σ²), with σ taking 0.8 to 1.2 times the standard deviation of the original data. For concentration-related indicators, which follow a log-normal distribution, random numbers satisfying the log-normal distribution are generated during simulation, and then truncated according to the extreme value range of the original data. For ear temperature indicators, an additional covariance constraint with Tin is applied. The Pearson correlation coefficient r between Tin and EarTemp is calculated based on the original data. This correlation structure is introduced when generating simulated Tin and EarTemp, so that the generated data statistically reproduces the true correlation direction between ambient temperature and ear temperature in the original data. S1.2.2 compares the distribution characteristics of the main indicators in the simulated dataset with the original data, expands the validity verification of the dataset, removes non-physiological abnormal fluctuation factors, and makes the simulated ear temperature closer to the stable variation range under physiological conditions.

[0037] The steps to construct a comprehensive evaluation index system for cold stress are as follows: S1.3.1, perform indicator homogenization processing to ensure that all indicators are in the same direction, and analyze the substantive relationship between each indicator and cold stress risk; S1.3.2, the range standardization method is used to uniformly map each indicator to the interval [0, 1]. The standardization formula for the positive indicator is:

[0038] Where, x i Let X be the original value, min(X) and max(X) be the minimum and maximum values ​​of the index in all samples, respectively, and Z(x) be the index. i The value of )∈[0,1] indicates that the relative level of the indicator in the dataset is higher, that is, the risk of cold stress is higher; S1.3.3 Descriptive statistical analysis was performed on the six standardized indicators to construct a comprehensive evaluation index system for cold stress. The indicators include indoor temperature (Tin), relative humidity (RH), CO2 concentration, NH3 concentration, PM1 concentration, and outdoor wind speed (Wind).

[0039] The steps involved in constructing a fuzzy comprehensive assessment model for cold stress are as follows: S2.1, Calculate objective weights based on the entropy weight method; S2.2, Construct the cold stress membership function and evaluation set; S2.3 uses the core computing module to analyze and build a cold stress fuzzy comprehensive evaluation model, and provides visualization output and result storage; Calculating objective weights based on the entropy weight method includes the following steps: S2.1.1 constructs an evaluation matrix using the standardized data of the six indicators and performs automated calculations using the entropyWeights function; S2.1.2, after standardization, the information entropy and specific entropy weights corresponding to the six indicator data are obtained; Section S2.1.3 discusses and analyzes the weighting results.

[0040] Constructing the cold stress membership function and evaluation set includes the following steps: S2.2.1, classifying cold stress levels, the five-level cold stress system is qualitatively described as follows: Unstressed animals are in a thermally comfortable environment and do not require additional physiological compensation; Mildly stressed animals begin to show peripheral vasoconstriction, slightly reduced activity, and production performance is not significantly affected; Moderately stressed animals exhibit shivering-induced thermogenesis, increased feed intake, and decreased feed conversion rate; Severely stressed animals continue to shiver, huddle together, and their immune function begins to decline; Extremely stressed animals experience a drop in body temperature, significantly reduced activity, and an immediate risk of health damage. S2.2.2, the comprehensive score S_new is used to determine the level threshold and comment set for cold stress, wherein the comment set is: V = {v1, v2, v3, v4, v5} = no stress, mild stress, moderate stress, severe stress, extreme stress; The formula for defining the comprehensive score S_new is:

[0041] Among them, W j Z is the entropy weight of the j-th index. j Let S_new be the standardized value of the j-th indicator, where S_new ∈ [0, 1]. The larger S_new is, the higher the risk of cold stress. The grade threshold is set as a four-segment threshold:

[0042] The specific division rules are as follows: S_new < 0.25 → no stress, 0.25 ≤ S_new < 0.40 → mild stress, 0.40 ≤ S_new < 0.55 → moderate stress, 0.55 ≤ S_new < 0.70 → severe stress, S_new ≥ 0.70 → extreme stress; S2.2.3, a combination of trapezoidal and semi-trapezoidal membership functions is used to evaluate the membership degree of each level threshold. The membership function for each level threshold is constructed using the comprehensive score S_new as the independent variable. The membership function μ1(x) for the no-stress level is a descending semi-trapezoidal function. The formula is as follows:

[0043] The membership function μ2(x) for mild stress levels is an intermediate trapezoidal function with the following formula:

[0044] For each evaluation time point i, its membership vector μ to the five levels is calculated. i = [μ i1 μ i2 μ i3 μ i4 μ i5 The final cold stress level at that time point is determined based on the principle of maximum membership.

[0045] The following steps are taken to apply the fuzzy comprehensive assessment model for cold stress and analyze the comprehensive assessment results to obtain accurate evaluation results of cold stress in livestock: S3.1, Obtain the environmental parameters and spatiotemporal distribution characteristics of livestock sheds; S3.2, Analyze the comprehensive assessment results of cold stress; S3.3, perform model performance verification and comparative analysis to obtain accurate evaluation results of livestock cold stress.

[0046] Obtaining livestock barn environmental parameters and spatiotemporal distribution characteristics includes the following steps: S3.1.1, Determine the temporal variation patterns of temperature and humidity; S3.1.2, determine the dynamic change law of harmful gas and particulate matter concentration; S3.1.3, Analyze the impact of outdoor meteorological conditions; The analysis of the comprehensive assessment results of cold stress includes the following steps: S3.2.1, Construct the time-series distribution of the comprehensive cold stress score; S3.2.2, Count the frequency of relevant time points according to the cold stress level; S3.2.3, the mean values ​​of each core indicator under the five cold stress levels are calculated to obtain the mean value characteristics of environmental and physiological indicators under different levels; S3.2.4 After standardizing the mean values ​​of each indicator by range, a radar chart is drawn to visually compare the characteristics of environmental parameter combinations at each level.

[0047] To conduct model performance validation and comparative analysis and obtain accurate evaluation results of livestock cold stress, the following steps are included: S3.3.1 Correlation analysis was performed between the three indicators—temperature and humidity index, wind chill temperature, and comprehensive score—and the standardized inverse values ​​of ear tag temperature to verify the model performance. S3.3.2, Analyze the response of ear tag temperature to cold stress score and obtain ear temperature distribution box plot; S3.3.3, the contribution rate C of each environmental factor to the comprehensive score, which is composed of entropy weight and standardized value. j for:

[0048] S3.3.4 Based on the above analysis and combined with the specific environmental conditions of the livestock shed in winter, an accurate evaluation result of cold stress in livestock is obtained.

[0049] The working principle of a cold stress assessment method for livestock in cold and arid regions according to an embodiment of the present invention is as follows: With the main goal of accurate multi-indicator evaluation of cold stress, the method is set up and operated from three levels: on-site monitoring, data modeling, and process algorithm. Thermal environment parameters, air quality indicators, and livestock ear tag body temperature are collected and combined. The specific weight of each indicator is determined by combining the entropy weight algorithm. A five-level classification model of cold stress based on fuzzy comprehensive evaluation is constructed to accurately determine the specific cold stress status of livestock herds and different individuals within herds. This provides reasonable guidance and decision-making for farms in high-altitude and cold regions in terms of heat preservation, ventilation, feed adjustment, and medication.

[0050] Currently, the three main cold stress evaluation indicators commonly used in livestock production at home and abroad are the Temperature and Humidity Index (THI), Wind-Cold Temperature (WCT), and Comprehensive Climate Index (CCI). Specifically, THI (Heat and Humidity Index) is used to evaluate the effects of temperature and humidity in hot environments on heat stress in livestock. Later, with modifications, it was also applied to cold stress assessment. The advantage of THI is its simple formula, which only requires two common meteorological parameters: air temperature and relative humidity. Therefore, it is widely used in livestock weather warning systems around the world. However, its formula does not include wind speed and air quality parameters. In winter, when livestock sheds are closed and ventilation is insufficient, THI has no effect on the accumulation of CO2 and NH3 inside the sheds. Compared to THI, WCT incorporates wind speed parameters, which can more accurately reflect the impact of strong winds on the perceived temperature of livestock in open pastures or semi-open sheds. However, in northern winters, livestock sheds are mostly fully enclosed or slightly open, resulting in very low wind speeds inside. In this case, the calculated value of WCT is close to the actual air temperature, and its design advantage of reflecting the "wind chill effect" cannot be fully realized. The Compressed Criterion Index (CCI) is currently the most comprehensive livestock stress assessment index considering environmental parameters, incorporating information from four dimensions—temperature, humidity, wind speed, and solar radiation—into an evaluation system. However, there are two problems with the practical application of CCI. First, it requires inputting solar radiation data, which small and medium-sized farms generally lack total radiation meters. Second, CCI needs to be adjusted for "effective ambient temperature" based on animal factors such as breed and coat thickness before it can be used for production guidance; directly applying general thresholds will introduce errors.

[0051] The above analysis reveals common problems with these three traditional indices. First, they are all based solely on environmental parameters, neglecting air quality indicators. In winter, CO2 concentrations in enclosed livestock sheds can reach 2000–5000 ppm, NH3 concentrations can reach 10–50 ppm, and PM2.5 and PM10 levels rise suddenly during feeding and cleaning periods. These factors, combined with low temperatures, cause the overall stress load on animals to exceed what a single temperature indicator could represent. Second, traditional indices provide uniform values ​​at the group level, failing to reflect differences in stress responses among individuals with different body conditions within the same group. Therefore, creating a comprehensive evaluation system that integrates thermal environment, air quality, and individual animal physiological indicators is crucial for improving the scientific rigor and accuracy of cold stress assessment in high-altitude and cold regions.

[0052] In this invention, the main purpose is to evaluate cold stress using multiple indicators. The process involves three aspects: on-site monitoring, data modeling, and process algorithms. An Internet of Things (IoT) monitoring system is built in the livestock shed to collect thermal environment parameters, air quality indicators, and livestock ear tag body temperature. The entropy weight method is used to objectively determine the weight of each indicator to obtain accurate evaluation results, which are then visualized and output.

[0053] In the overall scheme, the evaluation method includes the following steps: S1, preprocess the relevant data and establish an evaluation index system; S2, Construct a fuzzy comprehensive evaluation model for cold stress; S3 applies the fuzzy comprehensive assessment model for cold stress, analyzes the comprehensive assessment results of cold stress, and obtains accurate evaluation results of cold stress in livestock.

[0054] Specifically, the preprocessing of relevant data and the establishment of an evaluation indicator system include the following steps: Obtain the characteristics of the raw monitoring data and clean up outliers; Data augmentation based on Monte Carlo simulation; Construct a comprehensive evaluation index system for cold stress; Obtaining the characteristics of raw monitoring data and cleaning outliers includes the following steps: An indoor environmental monitoring system, an outdoor weather station, and a livestock ear tag system were used, with corresponding multi-category data collection time intervals set to acquire multi-source raw monitoring data. Three criteria are established to identify and remove outliers: physical law outliers, statistical law outliers, and time-related outliers. Physical law outliers are values ​​that violate basic physical or physiological principles. Statistical law outliers are data points with extremely extreme values ​​that do not trigger physical law criteria. Box plots are used for identification. Specifically, a statistical law outlier is defined as a value of a certain indicator at a given moment that is greater than or less than three times the difference between the upper and lower quartiles of all valid data for that indicator.

[0055] A three-times-wide spacing between fans is used instead to prevent environmental changes caused by normal operating conditions such as sudden weather changes, fan start-up and shutdown from being mistaken for anomalies. After outliers are marked, they are not immediately removed, but the original data and field records are retrieved to find a reasonable explanation. If the reasonableness can be determined, it is saved; if the reasonableness cannot be determined, it is represented by missing value (NaN). The time-related outliers are applicable to data points that show consecutive sharp jumps and do not recover afterward. The trend of data before and after the time period is used to make a comprehensive judgment. If all data in the entire time period shows abnormal fluctuations, the entire time period is marked as suspicious. For some random, non-segmented missing values, linear interpolation is used to fill them in. The linear interpolation method is as follows:

[0056] Where t1 is the most recent complete time before the missing point, t2 is the most recent complete time after the missing point, and x is the corresponding missing value; After the missing values ​​are filled, the three sets of data should be aligned according to a unified timeline.

[0057] Furthermore, data augmentation based on Monte Carlo simulations includes the following steps: The simulation parameters and statistical constraints were set as follows: For temperature-related indicators, the simulation constraints were obtained by superimposing a random perturbation on a daily variation baseline. The random perturbation term followed a normal distribution N(0, σ²), with σ ranging from 0.8 to 1.2 times the standard deviation of the original data. For concentration-related indicators, which followed a log-normal distribution, random numbers satisfying the log-normal distribution were generated during the simulation, and then truncated based on the extreme value range of the original data. For ear temperature indicators, an additional covariance constraint with Tin was applied. The Pearson correlation coefficient r between Tin and EarTemp was calculated based on the original data and introduced when generating the simulated Tin and EarTemp. This correlation structure ensured that the generated data statistically reproduced the true correlation between ambient temperature and ear temperature in the original data. The distribution characteristics of the main indicators in the simulated dataset were compared with those of the original data to expand the validity of the dataset and remove non-physiological abnormal fluctuation factors, so that the simulated ear temperature is closer to the stable variation range under physiological conditions.

[0058] The steps to construct a comprehensive evaluation index system for cold stress are as follows: Indicator homogenization was performed to ensure that all indicators were in the same direction, and the substantive relationship between each indicator and the risk of cold stress was analyzed. The lower the indoor temperature, the more the animal’s body temperature regulation relies on physiological compensation to maintain, and the higher the risk of cold stress. Therefore, indoor temperature is a negative indicator and needs to be positively converted.

[0059] Relative humidity (RH), CO2 concentration, NH3 concentration, PM 10 Concentration and outdoor wind speed (Wind): The higher the index value, the colder it feels or the worse the air quality, and the higher the risk of cold stress. Therefore, all five indicators are positive and do not require directional conversion.

[0060] The negative indicator, indoor temperature Tin, was subjected to a negative value conversion method, i.e., Tin' = -Tin. After conversion, the larger the value of Tin', the higher the risk of cold stress, and the same direction as the other five positive indicators.

[0061] Because the dimensions and magnitudes of the various indicators differ greatly, they cannot be directly weighted and synthesized. Therefore, in this invention, a range standardization method is used to uniformly map each indicator to the interval [0, 1]. The standardization formula for the positive indicator is:

[0062] Where, x i Let X be the original value, min(X) and max(X) be the minimum and maximum values ​​of the index in all samples, respectively, and Z(x) be the index. iThe value of )∈[0,1] indicates that the relative level of the indicator in the dataset is higher, that is, the risk of cold stress is higher; The reason for using range standardization instead of Z-score standardization is that range standardization maps the data to the [0, 1] interval, and the probability matrix P can be directly calculated from the Z value when calculating the entropy value, without having to consider the problem of negative values; at the same time, range standardization preserves the relative distance information of the original data, which is suitable for the construction of membership functions in fuzzy comprehensive evaluation.

[0063] Descriptive statistical analysis was performed on the six standardized indicators to construct a comprehensive evaluation index system for cold stress. The indicators include indoor temperature (Tin), relative humidity (RH), CO2 concentration, NH3 concentration, PM1 concentration, and outdoor wind speed (Wind).

[0064] After complete preprocessing of the original data, a standard evaluation index matrix was obtained. The statistical consistency between the simulated dataset and the original data was verified. Finally, based on the index direction, Tin was determined to be a negative index, while the others were positive indices. All indices were homogenized by taking the negative sign, and then the range standardization method was used to map all indices to the interval [0, 1]. This laid a standardized data foundation for the subsequent calculation of weights using the entropy weight method and fuzzy comprehensive evaluation.

[0065] The specific steps for constructing a fuzzy comprehensive evaluation model for cold stress are as follows: Calculate objective weights based on the entropy weight method; Constructing a membership function and evaluation set for cold stress; The core computing module is used to analyze and build a fuzzy comprehensive evaluation model for cold stress, and to provide visualization output and result storage. pass Figure 3 and Figure 4 The calculation results shown can be summarized into the following characteristics: (1) PM 10 The weight is significantly the highest (0.4119), which is 3.5 times that of RH (0.1192), which ranks second. Consistent with the information entropy calculation logic, PM 10 The distribution of the 1000 simulated data points is very uneven, with PM being the most frequent (over 85%) point. 10 The concentration was 0 μg / m³ (Z=0), but PM levels were higher during feeding periods or periods of increased animal activity. 10 High. This "long-tail distribution" results in extremely low information entropy (E). 10 =-0.1601), coefficient of difference D 10 Extremely high (1.1601), thus occupying the largest weight. From a statistical data perspective, PM 10PM2.5 is the best indicator for distinguishing environmental conditions at different times, but from a practical production perspective, PM2.5 is not the best indicator for distinguishing environmental conditions at different times. 10 The fluctuations in standard values ​​across different time periods may be due to monitoring accuracy, and the actual production significance of its weights requires further research.

[0066] (2) The weights of the four indicators RH, CO2, NH3, and Wind are close (0.1167-0.1192), all around 0.118, indicating that these four indicators have similar information content in the current dataset. The small difference in weights indicates that the ten indicators have the same degree of dispersion in multiple samples.

[0067] (3) Tin' has the lowest weight (0.1153), but the difference among the six indicators is not significant. The reason why Tin''s weight is not the highest may be because although temperature fluctuates in the multi-group simulation data, the temperature changes are relatively gradual and the dispersion is small, resulting in high information entropy and small difference coefficient. This is not entirely consistent with the intuitive understanding that temperature is the main influencing factor of cold stress in traditional cognition, and it must be understood in combination with the model application scenario and the correlation between indicators.

[0068] It should be noted that the weights derived by the entropy weight method are based on the objective weights of the data's information content, reflecting "which indicator contributes more to the differences between samples," rather than "which indicator is more important in terms of physiological mechanisms." Even if a certain indicator has a low weight, its role as a basic discriminant factor for cold stress cannot be ignored.

[0069] Calculating objective weights based on the entropy weight method includes the following steps: An evaluation matrix is ​​constructed using the standardized data of the six indicators, and the entropyWeights function is used for automated calculation. After standardization, the information entropy and specific entropy weights corresponding to the six indicator data are obtained; Discuss and analyze the weighting results.

[0070] In this invention, the construction of the membership function and the evaluation set is a core innovation, which specifically includes the following steps: The cold stress system is classified into five levels, which can be qualitatively described as follows: Unstressed animals are in a thermally comfortable environment and do not require additional physiological compensation; Mildly stressed animals begin to show peripheral vasoconstriction, slightly reduced activity, and production performance is not significantly affected; Moderately stressed animals exhibit shivering-induced thermogenesis, increased feed intake, and decreased feed conversion rate; Severely stressed animals continue to shiver, huddle together, and their immune function begins to decline; Extremely stressed animals experience a drop in body temperature, significantly reduced activity, and an immediate risk of health damage. A comprehensive score S_new is used to determine the level threshold and the set of comments for cold stress. The set of comments is as follows: V = {v1, v2, v3, v4, v5} = no stress, mild stress, moderate stress, severe stress, extreme stress; The formula for defining the comprehensive score S_new is:

[0071] Among them, W j Z is the entropy weight of the j-th index. j Let S_new be the standardized value of the j-th indicator, where S_new ∈ [0, 1]. The larger S_new is, the higher the risk of cold stress. The grade threshold is set as a four-segment threshold:

[0072] The specific division rules are as follows: S_new < 0.25 → no stress, 0.25 ≤ S_new < 0.40 → mild stress, 0.40 ≤ S_new < 0.55 → moderate stress, 0.55 ≤ S_new < 0.70 → severe stress, S_new ≥ 0.70 → extreme stress; The membership degree of each level threshold is evaluated by combining trapezoidal and semi-trapezoidal membership functions. The membership function for each level threshold is constructed using the comprehensive score S_new as the independent variable. The membership function μ1(x) for the no-stress level is a descending semi-trapezoidal function. The formula is as follows:

[0073] The membership function μ2(x) for mild stress levels is an intermediate trapezoidal function with the following formula:

[0074] For each evaluation time point i, its membership vector μ to the five levels is calculated. i = [μ i1 μ i2 μ i3 μ i4 μ i5 The final cold stress level at that time point is determined based on the principle of maximum membership.

[0075] Typically, after membership function calculations, multiple insets are output, including: temperature time series plot, air quality time series plot, correlation heatmap, weight distribution plot, comprehensive assessment scatter plot, model comparison plot, cold stress risk time series plot, ear thermometer line graph, radar chart, and level percentage pie chart. All images are exported as PNG format at 300 dpi resolution. Simultaneously, three data tables are output: Monitoring Data Summary (including cold stress level table), Mean Table of Key Indicators for Each Level, and Monitoring Data Summary (cleaned and generated by MATLAB).

[0076] The constructed fuzzy comprehensive evaluation model for cold stress can provide a complete and effective tool for subsequent applications and results analysis.

[0077] Furthermore, the fuzzy comprehensive assessment model for cold stress is applied, and the comprehensive assessment results are analyzed to obtain accurate evaluation results of cold stress in livestock, including the following steps: Obtain environmental parameters and spatiotemporal distribution characteristics of livestock sheds; Analyze the results of the comprehensive assessment of cold stress; Model performance was validated and compared to obtain accurate evaluation results of cold stress in livestock.

[0078] Obtaining livestock barn environmental parameters and spatiotemporal distribution characteristics includes the following steps: Determine the temporal variation patterns of temperature and humidity; Determine the dynamic changes in the concentrations of harmful gases and particulate matter; Analyze the impact of outdoor meteorological conditions; The analysis of the comprehensive assessment results of cold stress includes the following steps: Constructing the time-series distribution of the comprehensive cold stress score; The frequency of relevant time points was statistically analyzed according to the level of cold stress. The mean values ​​of each core indicator under five levels of cold stress were calculated to obtain the mean characteristics of environmental and physiological indicators under different levels. After standardizing the mean values ​​of each indicator by range, a radar chart is drawn to visually compare the characteristics of environmental parameter combinations at each level.

[0079] The radar chart clearly shows that the standardized wind speed value is the highest among all levels for extreme stress, the standardized PM10 value is high, and the standardized temperature value is in the middle range. This further illustrates that the cold stress risk assessment in this model is multidimensional. Extreme stress is not defined by the lowest temperature, but rather by the combined effects of low temperature, strong winds, and high particulate matter concentration.

[0080] To conduct model performance validation and comparative analysis and obtain accurate evaluation results of livestock cold stress, the following steps are included: The three indicators—temperature and humidity index, wind chill temperature, and comprehensive score—were correlated with the standardized inverse values ​​of ear tag temperature to verify the model performance. The response of ear tag temperature to cold stress scores was analyzed, and a box plot of ear temperature distribution was obtained. The contribution rate C of each environmental factor to the overall score is calculated by combining entropy weight and standardized value. j for:

[0081] Based on the above analysis and the specific environmental conditions of the livestock shed in winter, an accurate assessment of cold stress in livestock was obtained.

[0082] The ear temperature distribution box plot shows the ear temperature distribution under five levels. The median ear temperature is slightly higher in the moderate stress level, while the median ear temperature is not significantly different between the no stress and extreme stress levels, which is consistent with the results of the previous mean analysis. It should be noted again that ear temperature is the only physiological feedback indicator, and its limitations are evident here. Incorporating multidimensional physiological indicators such as respiratory rate and surface infrared temperature in later studies can greatly strengthen the physiological validation basis of the model.

[0083] It should be noted that in subsequent practice, more sensitive physiological indicators such as respiratory rate and body surface infrared temperature can be added to further ensure the accuracy and authenticity of the evaluation results.

[0084] In summary, the cold stress assessment method for livestock in cold and arid regions proposed in this invention aims to accurately assess cold stress using multiple indicators. It involves setting up operations at three levels: on-site monitoring, data modeling, and process algorithms. Thermal environment parameters, air quality indicators, and livestock ear tag body temperatures are collected and aggregated. An entropy weight algorithm is used to determine the specific weights of each indicator, and a five-level cold stress classification model based on fuzzy comprehensive evaluation is constructed. This model accurately determines the specific cold stress status of livestock herds and different individuals within herds, thereby providing reasonable guidance and decisions for farms in high-altitude and cold regions regarding insulation, ventilation, feed adjustment, and medication.

[0085] The above specific embodiments should not be construed as limiting the scope of protection of the present invention. For those skilled in the art, any alternative improvements or modifications made to the embodiments of the present invention shall fall within the scope of protection of the present invention.

[0086] Any aspects of this invention not described in detail are well-known to those skilled in the art.

Claims

1. A method for evaluating cold stress in livestock in arid and cold regions, characterized in that, The evaluation method includes the following steps: S1, preprocess the relevant data and establish an evaluation index system; S2, Construct a fuzzy comprehensive evaluation model for cold stress; S3 applies the fuzzy comprehensive assessment model for cold stress, analyzes the comprehensive assessment results of cold stress, and obtains accurate evaluation results of cold stress in livestock.

2. The method for evaluating cold stress in livestock in arid and cold regions according to claim 1, characterized in that, The preprocessing of relevant data and the establishment of an evaluation index system include the following steps: S1.1, Obtain the characteristics of the raw monitoring data and clean up outliers; S1.2, Data expansion based on Monte Carlo simulation; S1.3, Construct a comprehensive evaluation index system for cold stress; Obtaining the characteristics of raw monitoring data and cleaning outliers includes the following steps: S1.1.1, adopt an indoor environmental monitoring system, an outdoor weather station and a livestock ear tag system, and set corresponding multi-category collection time intervals to obtain multi-source raw monitoring data; S1.1.2, three criteria are set to identify and remove outliers: physical law outliers, statistical law outliers, and time-related outliers. Physical law outliers are values ​​that violate basic physical or physiological common sense. Statistical law outliers are data points with extremely extreme values ​​that do not trigger physical law criteria. Box plots are used for judgment. Specifically, a statistical law outlier is marked when the value of a certain indicator at a certain moment is greater than or less than three times the difference between the upper and lower quartiles of all valid data for that indicator. A three-times-wide spacing between fans is used instead to prevent environmental changes caused by normal operating conditions such as sudden weather changes, fan start-up and shutdown from being mistaken for anomalies. After outliers are marked, they are not immediately removed, but the original data and field records are retrieved to find a reasonable explanation. If the reasonableness can be determined, it is saved; if the reasonableness cannot be determined, it is represented by missing value (NaN). The time-related outliers are applicable to data points that show consecutive sharp jumps and do not recover afterward. The trend of data before and after the time period is used to make a comprehensive judgment. If all data in the entire time period shows abnormal fluctuations, the entire time period is marked as suspicious. S1.1.3, For some random missing values ​​that are not part of the above-mentioned segmental anomalies, linear interpolation is used to fill them in. The linear interpolation method is as follows: Where t1 is the most recent complete time before the missing point, t2 is the most recent complete time after the missing point, and x is the corresponding missing value; After the missing values ​​are filled, the three sets of data should be aligned according to a unified timeline.

3. The method for evaluating cold stress in livestock in arid and cold regions according to claim 2, characterized in that, Data augmentation based on Monte Carlo simulations includes the following steps: S1.2.1, Simulation parameter settings and statistical characteristic constraints: For temperature-related indicators, simulation constraints are obtained by superimposing a daily variation baseline with random perturbations. The superimposed random perturbation term follows a normal distribution N(0, σ²), with σ taking 0.8 to 1.2 times the standard deviation of the original data. For concentration-related indicators, which follow a log-normal distribution, random numbers satisfying the log-normal distribution are generated during simulation, and then truncated according to the extreme value range of the original data. For ear temperature indicators, an additional covariance constraint with Tin is applied. The Pearson correlation coefficient r between Tin and EarTemp is calculated based on the original data. This correlation structure is introduced when generating simulated Tin and EarTemp, so that the generated data statistically reproduces the true correlation direction between ambient temperature and ear temperature in the original data. S1.2.2 compares the distribution characteristics of the main indicators in the simulated dataset with the original data, expands the validity verification of the dataset, removes non-physiological abnormal fluctuation factors, and makes the simulated ear temperature closer to the stable variation range under physiological conditions.

4. The method for evaluating cold stress in livestock in cold and arid regions according to claim 2, characterized in that, The steps to construct a comprehensive evaluation index system for cold stress are as follows: S1.3.1, perform indicator homogenization processing to ensure that all indicators are in the same direction, and analyze the substantive relationship between each indicator and cold stress risk; S1.3.2, the range standardization method is used to uniformly map each indicator to the interval [0, 1]. The standardization formula for the positive indicator is: Where, x i Let X be the original value, min(X) and max(X) be the minimum and maximum values ​​of the index in all samples, respectively, and Z(x) be the index. i The value of )∈[0,1] indicates that the relative level of the indicator in the dataset is higher, that is, the risk of cold stress is higher; S1.3.3 Descriptive statistical analysis was performed on the six standardized indicators to construct a comprehensive evaluation index system for cold stress. The indicators include indoor temperature (Tin), relative humidity (RH), CO2 concentration, NH3 concentration, PM1 concentration, and outdoor wind speed (Wind).

5. The method for evaluating cold stress in livestock in arid and cold regions according to claim 1, characterized in that, The steps involved in constructing a fuzzy comprehensive assessment model for cold stress are as follows: S2.1, Calculate objective weights based on the entropy weight method; S2.2, Construct the cold stress membership function and evaluation set; S2.3 uses the core computing module to analyze and build a cold stress fuzzy comprehensive evaluation model, and provides visualization output and result storage; Calculating objective weights based on the entropy weight method includes the following steps: S2.1.1 constructs an evaluation matrix using the standardized data of the six indicators and performs automated calculations using the entropyWeights function; S2.1.2, after standardization, the information entropy and specific entropy weights corresponding to the six indicator data are obtained; Section S2.1.3 discusses and analyzes the weighting results.

6. The method for evaluating cold stress in livestock in cold and arid regions according to claim 5, characterized in that, Constructing the cold stress membership function and evaluation set includes the following steps: S2.2.1, classifying cold stress levels, the five-level cold stress system is qualitatively described as follows: Unstressed animals are in a thermally comfortable environment and do not require additional physiological compensation; Mildly stressed animals begin to show peripheral vasoconstriction, slightly reduced activity, and production performance is not significantly affected; Moderately stressed animals exhibit shivering-induced thermogenesis, increased feed intake, and decreased feed conversion rate; Severely stressed animals continue to shiver, huddle together, and their immune function begins to decline; Extremely stressed animals experience a drop in body temperature, significantly reduced activity, and an immediate risk of health damage. S2.2.2, the comprehensive score S_new is used to determine the level threshold and comment set for cold stress, wherein the comment set is: V = {v1, v2, v3, v4, v5} = no stress, mild stress, moderate stress, severe stress, extreme stress; The formula for defining the comprehensive score S_new is: Among them, W j Z is the entropy weight of the j-th index. j Let S_new be the standardized value of the j-th indicator, where S_new ∈ [0, 1]. The larger S_new is, the higher the risk of cold stress. The grade threshold is set as a four-segment threshold: The specific division rules are as follows: S_new < 0.25 → no stress, 0.25 ≤ S_new < 0.40 → mild stress, 0.40 ≤ S_new < 0.55 → moderate stress, 0.55 ≤ S_new < 0.70 → severe stress, S_new ≥ 0.70 → extreme stress; S2.2.3, a combination of trapezoidal and semi-trapezoidal membership functions is used to evaluate the membership degree of each level threshold. The membership function for each level threshold is constructed using the comprehensive score S_new as the independent variable. The membership function μ1(x) for the no-stress level is a descending semi-trapezoidal function. The formula is as follows: The membership function μ2(x) for mild stress levels is an intermediate trapezoidal function with the following formula: For each evaluation time point i, its membership vector μ to the five levels is calculated. i = [μ i1 μ i2 μ i3 μ i4 μ i5 The final cold stress level at that time point is determined based on the principle of maximum membership.

7. The method for evaluating cold stress in livestock in cold and arid regions according to claim 1, characterized in that, The following steps are taken to apply the fuzzy comprehensive assessment model for cold stress and analyze the comprehensive assessment results to obtain accurate evaluation results of cold stress in livestock: S3.1, Obtain the environmental parameters and spatiotemporal distribution characteristics of livestock sheds; S3.2, Analyze the comprehensive assessment results of cold stress; S3.3, perform model performance verification and comparative analysis to obtain accurate evaluation results of livestock cold stress.

8. The method for evaluating cold stress in livestock in arid and cold regions according to claim 7, characterized in that, Obtaining livestock barn environmental parameters and spatiotemporal distribution characteristics includes the following steps: S3.1.1, Determine the temporal variation patterns of temperature and humidity; S3.1.2, determine the dynamic change law of harmful gas and particulate matter concentration; S3.1.3, Analyze the impact of outdoor meteorological conditions; The analysis of the comprehensive assessment results of cold stress includes the following steps: S3.2.1, Construct the time-series distribution of the comprehensive cold stress score; S3.2.2, Count the frequency of relevant time points according to the cold stress level; S3.2.3, the mean values ​​of each core indicator under the five cold stress levels are calculated to obtain the mean value characteristics of environmental and physiological indicators under different levels; S3.2.4 After standardizing the mean values ​​of each indicator by range, a radar chart is drawn to visually compare the characteristics of environmental parameter combinations at each level.

9. A method for evaluating cold stress in livestock in arid and cold regions according to claim 7, characterized in that, To conduct model performance validation and comparative analysis and obtain accurate evaluation results of livestock cold stress, the following steps are included: S3.3.1 Correlation analysis was performed between the three indicators—temperature and humidity index, wind chill temperature, and comprehensive score—and the standardized inverse values ​​of ear tag temperature to verify the model performance. S3.3.2, Analyze the response of ear tag temperature to cold stress score and obtain ear temperature distribution box plot; S3.3.3, the contribution rate C of each environmental factor to the comprehensive score, which is composed of entropy weight and standardized value. j for: S3.3.4 Based on the above analysis and combined with the specific environmental conditions of the livestock shed in winter, an accurate evaluation result of cold stress in livestock is obtained.

Images