Livestock and poultry respiration calorimetry data acquisition and accuracy verification processing system

Abstract

The present application relates to the technical field of data preprocessing, and particularly relates to a livestock and poultry respiration calorimetry data acquisition and accuracy verification processing system, the system comprises: an identification module, which is used for determining the comprehensive abnormal score of each data point according to the time sequence variation characteristics of the data points in the concentration time sequence, and determining the abnormal points based on the score; a stress analysis module, which is used for obtaining the precursor abnormal index and the successor abnormal index according to the comprehensive abnormal score and the distribution trend of the abnormal points in the precursor and successor time windows of each abnormal point; determining the relative time index of the abnormal points according to the gas concentration of the abnormal points and the gas concentration of the first and last data points of the concentration time sequence; determining the correlation degree by combining the three indexes; a correction module, which is used for determining the correction weight according to the correlation degree and the comprehensive abnormal score; correcting the gas concentration of the abnormal points by using the correction weight to obtain the corrected gas concentration, so that the corrected data is closer to the real metabolic state of the livestock and poultry.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data preprocessing, in particular to a livestock and poultry respiratory calorimetry data acquisition and accuracy verification processing system. BACKGROUND

[0002] Livestock and poultry energy metabolism is an important research basis for animal nutrition and feed science, and accurate energy metabolism evaluation is crucial for formulating scientific feeding standards and optimizing feed allocation. Respiratory calorimetry is one of the current mainstream non-destructive detection methods, among which, respiratory mask method is widely used due to its relatively simple operation. However, respiratory mask method has a significant inherent defect: at the beginning of the test, livestock and poultry produce a strong stress response due to the sudden wearing of a sealed mask, showing rapid and unstable breathing. This physiological stress response can seriously interfere with the authenticity of the collected gas concentration data, resulting in deviations in the calculated energy metabolism indicators.

[0003] Currently, existing technologies often directly use clustering algorithms to screen and correct abnormal respiratory gas concentration data. This method usually assumes that abnormal data caused by stress has consistency in features, and is based on this to screen and fit for correction. However, this method has obvious limitations: first, the randomness of livestock and poultry stress response is strong, resulting in significant differences in the amplitude and frequency of abnormal data exhibited by livestock and poultry, making it difficult to establish the "consistency" assumption on which clustering algorithms rely, resulting in inaccurate screening results. Second, directly correcting based on clustering results and normal data fitting values completely ignores the time sequence correlation information and physiological context contained in abnormal data, making the correction process one-sided and leading to low reliability of the final respiratory heat analysis results. SUMMARY

[0004] To solve the technical problem of directly using clustering algorithms to screen and correct abnormal respiratory gas concentration data without considering the randomness of livestock and poultry stress response, resulting in inaccurate screening results and affecting the accuracy of the final respiratory heat analysis results, the present application provides a livestock and poultry respiratory calorimetry data acquisition and accuracy verification processing system, the technical solution adopted is as follows:

[0005] The present application provides a livestock and poultry respiratory calorimetry data acquisition and accuracy verification processing system, the system comprises:

[0006] The acquisition module is used to acquire the concentration time series of livestock and poultry respiratory gas;

[0007] The identification module is used to determine the comprehensive abnormal score of each data point according to the time sequence variation characteristics of the data points in the concentration time series, and mark the data points whose scores meet the local abnormality condition as abnormal points;

[0008] a stress analysis module, configured to obtain a precursor abnormality index and a successor abnormality index of each abnormal point according to a comprehensive abnormality score and a distribution trend of the abnormal point in a precursor time window and a successor time window of the abnormal point respectively; determine a relative time index of the abnormal point according to a gas concentration of the abnormal point and a gas concentration of a first data point and a last data point in the concentration time sequence; and determine a correlation degree between each abnormal point and an initial stress state of the livestock and poultry by combining the precursor abnormality index, the successor abnormality index and the relative time index;

[0009] a correction module, configured to determine a correction weight according to the correlation degree and the comprehensive abnormality score; estimate an ideal gas concentration of each abnormal point based on the concentration time sequence after all abnormal points are removed; and obtain a corrected gas concentration by weighted fusion of the collected gas concentration of the abnormal point and the ideal gas concentration by using the correction weight, and generate a corrected concentration time sequence of the gas concentration;

[0010] an evaluation module, configured to analyze an energy metabolism state of the livestock and poultry based on the corrected concentration time sequence of the gas concentration, and output a respiratory calorimetry result that has been verified for accuracy.

[0011] Further, the comprehensive abnormality score determination process comprises:

[0012] based on the concentration time sequence, calculating a first-order difference of each data point and forming a first-order difference sequence;

[0013] based on the first-order difference sequence, determining a local mutation index of each data point;

[0014] based on the overall trend of the concentration time sequence, determining a global deviation index of each data point;

[0015] calculating a product of the local mutation index and the global deviation index as a comprehensive abnormality score of each data point.

[0016] Further, the determination of the local mutation index of each data point based on the first-order difference sequence comprises:

[0017] calculating a quartile range of the first-order difference sequence;

[0018] for each data point, calculating a sum of absolute values of a difference between a first-order difference value of the data point and first-order difference values of a previous data point and a next data point of the data point as a local mutation reference value;

[0019] calculating a ratio of the local mutation reference value to the quartile range as the local mutation index.

[0020] Further, the determination of the global deviation index of each data point based on the overall trend of the concentration time sequence comprises:

[0021] for each data point, calculating an instantaneous slope of each data point in the concentration time sequence by a central difference method.

[0022] Linear fitting of the concentration time series to obtain the overall slope;

[0023] Calculate the absolute difference between the instantaneous slope and the overall slope as a global deviation index.

[0024] Further, the data points with scores satisfying the local anomaly condition are marked as anomaly points, including:

[0025] Based on the comprehensive anomaly score of each data point, a comprehensive anomaly score sequence is constructed in chronological order;

[0026] Traverse each data point in the comprehensive anomaly score sequence except the first and last data points, and determine whether the comprehensive anomaly score of the current traversed data point is greater than the score of the previous data point and the score of the next data point; if so, the current traversed data point is determined as a local maximum point;

[0027] From the local maximum points, filter out data points with a comprehensive anomaly score greater than a preset threshold, and mark them as anomaly points.

[0028] Further, the method further comprises:

[0029] The time period before the occurrence time of the anomaly point in the concentration time sequence is defined as the precursor time window of the anomaly point, wherein the precursor time window starts at the beginning of the concentration time sequence and ends at the occurrence time of the anomaly point;

[0030] The time period after the occurrence time of the anomaly point in the concentration time sequence is defined as the successor time window of the anomaly point, wherein the successor time window starts at the occurrence time of the current anomaly point and ends at the end of the concentration time sequence.

[0031] Further, the method further comprises:

[0032] Obtain the time interval from the occurrence time of each anomaly point to the occurrence time of the next adjacent anomaly point;

[0033] Calculate the arithmetic mean of the comprehensive anomaly scores of all anomaly points in the time window as the anomaly intensity baseline value of the time window;

[0034] Calculate the difference between the comprehensive anomaly score of each anomaly point in the time window and the anomaly intensity baseline as the anomaly metric value of each anomaly point;

[0035] Take the sum of the absolute values of the anomaly metric values as a reference to normalize each anomaly metric value to obtain the relative prominence weight of each anomaly point;

[0036] The relative prominence weight of each abnormal point is multiplied by the reciprocal of the time interval corresponding to the abnormal point to obtain a product result; and the sum of all product results in the time window is calculated as an abnormal index of the time window, wherein the abnormal index is a precursor abnormal index or a subsequent abnormal index.

[0037] Further, the relative time index determination process comprises:

[0038] The initial gas concentration value and the ending gas concentration value of the concentration time sequence are obtained, and the gas concentration value at the moment when each abnormal point occurs is obtained;

[0039] The absolute difference between the ending gas concentration value and the gas concentration value of each abnormal point is calculated as a first difference value;

[0040] The absolute difference between the initial gas concentration value and the ending gas concentration value is calculated as a second difference value;

[0041] The ratio of the first difference value to the second difference value is calculated as the relative time index of each abnormal point.

[0042] Further, the correlation degree determination process comprises:

[0043] The difference between the precursor abnormal index and the subsequent abnormal index is calculated as an abnormal index difference value;

[0044] The product of the abnormal index difference value and the relative time index is calculated as the correlation degree of each abnormal point and the initial stress state of livestock and poultry.

[0045] Further, the correction weight determination process comprises:

[0046] The absolute value of the correlation degree of all abnormal points is calculated, and the maximum value thereof is determined as a maximum correlation degree; the ratio of the correlation degree of each abnormal point to the maximum correlation degree is calculated as a first ratio value;

[0047] For each abnormal point, the number of abnormal points with continuously strictly increasing comprehensive abnormal scores is counted from the immediately preceding abnormal point of the abnormal point to the front;

[0048] The product of the first ratio value and the comprehensive abnormal score is calculated as an intermediate value;

[0049] The intermediate value is divided by the sum of the number of continuously increasing abnormal points and 1 to obtain a quotient result, and the quotient result is normalized to obtain a correction weight.

[0050] The present application has the following beneficial effects:

[0051] By comprehensively scoring anomalies, traditional clustering algorithms that rely on the consistency of static data features are abandoned, making it easier to accurately capture real abnormal data points caused by stress responses such as shortness of breath, effectively reducing the probability of misjudgment and missed judgment. Considering that when livestock and poultry are first put on masks, the initial stress response usually shows that the stress level gradually decreases during the process of soothing and self-adaptation, and the respiratory instability of livestock and poultry also gradually decreases. Therefore, this attribution analysis mechanism based on the physiological model of stress attenuation is introduced. By quantitatively analyzing the intensity of abnormal patterns in the preceding and following time windows of each abnormal point, the preceding abnormality indicators are determined. By combining subsequent abnormal indicators with the relative time indicators of the occurrence of abnormal points, the system intelligently judges the possibility (i.e., correlation) that the abnormal points are caused by the initial stress, thus realizing the identification of the root cause of abnormal data. Based on the calculated correlation and comprehensive abnormal score, a correction weight is dynamically generated for each abnormal point. Through weighted fusion, the collected gas concentration is combined with the ideal gas concentration established based on normal data. This gradient correction strategy not only eliminates stress interference to the greatest extent, but also retains the reference for the real metabolic trend, making the corrected data more scientific and closer to the real metabolic state of livestock and poultry. Attached Figure Description

[0052] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0053] Figure 1 This is a schematic diagram of a livestock and poultry respiratory calorimetry data acquisition and accuracy verification processing system according to an embodiment of the present invention;

[0054] Figure 2 This is a schematic diagram of the comprehensive anomaly score determination process provided in one embodiment of the present invention;

[0055] Figure 3 This is a schematic diagram illustrating the correlation determination process provided in one embodiment of the present invention. Detailed Implementation

[0056] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a livestock and poultry respiratory thermometry data acquisition and accuracy verification processing system proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0057] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0058] The following description, in conjunction with the accompanying drawings, details the specific scheme of the livestock and poultry respiratory thermometry data acquisition and accuracy verification processing system provided by the present invention.

[0059] Please see Figure 1 The diagram illustrates a structural schematic of a livestock and poultry respiratory calorimetry data acquisition and accuracy verification processing system according to an embodiment of the present invention. The system includes:

[0060] The acquisition module 101 is used to acquire the concentration time series of respiratory gases of livestock and poultry.

[0061] To ensure that the physiological condition of livestock and poultry meets the testing requirements, standardized pretreatment is required before data collection begins. Standardized pretreatment typically includes health screening and following specific fasting or feeding procedures to eliminate interference from digestive activity on basal metabolic measurements. Subsequently, a sealed breathing mask is securely placed on the head of the livestock and poultry.

[0062] It should be noted that the sealed breathing mask must have good airtightness to minimize gas exchange with the outside environment. At the same time, it is necessary to ensure that there is sufficient oxygen supply inside the mask, or to prevent the livestock from experiencing hypoxia during the test by connecting a breathing circuit with safeguards.

[0063] It should be noted that a high-precision gas concentration sensor array is deployed inside the sealed breathing mask. The gas concentration sensor array includes at least sensors for monitoring oxygen and carbon dioxide concentrations. These sensors are configured to collect the concentration of the corresponding gases inside the mask in real time and continuously at a preset fixed sampling frequency.

[0064] The specific value of the preset fixed sampling frequency is determined according to the actual situation. This embodiment does not make a specific limitation. For example, it may be 1 time or 10 times per second.

[0065] It should be noted that each data point contains two core pieces of information: a precise timestamp (and the gas concentration value corresponding to that moment). From the start to the end of the test, all gas concentration data points arranged in chronological order are stored, thus forming a gas concentration time series.

[0066] The identification module 102 is used to determine the comprehensive anomaly score of each data point based on the temporal change characteristics of the data points in the concentration time series, and to record the data points whose scores meet the local anomaly conditions as anomaly points.

[0067] It's important to understand that livestock and poultry generally maintain relatively slow breathing most of the time, with a relatively stable rate of exhalation (or absorption) of relevant gases and a high degree of consistency in gas concentration fluctuations. In contrast, when livestock and poultry are under stress or in a pathological state, the most direct manifestation of the stress response is a disruption of the respiratory rhythm, resulting in brief, rapid exhalation or inhalation. Moreover, some data may show localized changes, but these are consistent with the animal's overall metabolic trend (such as slow, deep breathing). True stress abnormalities, however, will cause the respiratory behavior at this moment to deviate significantly from the livestock's normal data. Therefore, when analyzing whether a data point is abnormal, it must show a high degree of abnormality both locally and globally to ensure that the data point corresponds to a real physiological stress event with a high probability.

[0068] The process of determining the comprehensive anomaly score is as follows: Figure 2 As shown, it includes:

[0069] S101-1, based on the concentration time series, calculates the first difference of each data point and constructs a first difference sequence.

[0070] For example, the original gas concentration time series is ,in Let represent the concentration value at the i-th sampling time, and n represent the total number of data points in the gas concentration time series. Then, the first-order difference sequence C is calculated as follows: Each element in the sequence This represents the change in gas concentration from data point i to data point i+1.

[0071] It's important to understand that an animal's stress response (such as rapid breathing) is directly reflected in unstable changes in respiratory rate and depth, rather than directly analyzing concentration values. This change information will be ignored, while the first-order difference sequence Essentially, it is an approximation of the rate of concentration change. By using a first-order difference sequence, the focus of the analysis can be shifted from the absolute level of concentration to the rate of concentration change, allowing the instability of respiratory rhythm to be directly observed and analyzed, laying the foundation for subsequent detection of "abrupt changes" in concentration.

[0072] It should be noted that the method for calculating the first-order difference is a common computational technique, and will not be described in detail in this embodiment.

[0073] S101-2, based on the first-order difference sequence, determines the local mutation index of each data point.

[0074] Local mutation index is used to quantify the degree of drastic fluctuation in the rate of change near each point.

[0075] It's important to understand that a rapid breath will appear in the data as follows: the rate of change initially rises sharply (or falls), then quickly returns to normal. This process will produce a distinct "peak" or "trough" in the first-order difference sequence C. Therefore, by using local mutation indicators, those brief, intense, and unstable parts of respiratory behavior can be effectively captured.

[0076] In this embodiment, the interquartile range of the first-order difference sequence is calculated; for each data point, the sum of the absolute values ​​of the difference between the first-order difference value of the data point and the first-order difference value of the preceding and following data points is calculated as the local mutation benchmark value; the ratio of the local mutation benchmark value to the interquartile range is calculated as the local mutation index.

[0077] It should be noted that the method for calculating the interquartile range is a common technique, which will not be elaborated in this embodiment. For example, first calculate the first quartile (represented by Q1): it is the value in the first 25% position after the first difference sequence is arranged in ascending order, that is, 25% of the data are less than or equal to Q1; then calculate the third quartile (represented by Q3): it is the value in the first 75% position after the sequence is arranged in ascending order; finally calculate the interquartile range (represented by IQR): IQR = Q3 - Q1.

[0078] Local mutation baseline values ​​are used to quantify the magnitude of a sharp shift in a concentration time series near a certain data point.

[0079] It should be noted that the first-order difference sequence obtained from the known calculations... Where m = n-1, to avoid boundary issues, the effective range of the index i when calculating the local mutation baseline value is from 2 to m-1. Therefore, the calculation of its local mutation baseline value can be expressed as: .

[0080] in, This represents the local mutation baseline value of the i-th data point. This represents the change in gas concentration from data point i-1 to data point i; || represents the change in gas concentration from data point i+2 to data point i+1; || represents taking the absolute value.

[0081] It should be noted that respiration is a continuous physiological process driven by the autonomic nervous system and metabolic activities. As long as livestock and poultry are alive, their respiration will not be as absolutely precise as that of a machine. Therefore, in any test with a certain length of time (e.g., a few minutes), the first-order difference sequence C generated by respiration will inevitably fluctuate and it is impossible for all values ​​to be completely equal. Thus, the interquartile range cannot be zero.

[0082] It's important to understand that the local mutation baseline measures the severity of the instantaneous change in the respiratory rate at the moment data point i occurs. A larger local mutation baseline for a data point indicates a more abrupt change in the respiratory rate. The interquartile range measures the inherent, normal fluctuation range of the respiratory rate throughout the test period. A larger interquartile range indicates more unstable respiration in livestock. Therefore, if a data point has a large local mutation baseline and a large interquartile range, but the local mutation index is moderate or small, it suggests that this may not be a typical case of sudden stress; the livestock's respiration may already be unstable (a high IQR may indicate agitation or illness). In this case, the mutation (…) The large value is only part of its unstable performance; the abnormality of this data point is not particularly prominent compared to its "normal" state.

[0083] S101-3, based on the overall trend of the concentration time series, determines the global deviation index of each data point.

[0084] It is important to understand that relying solely on local mutation indicators may misjudge some large-amplitude normal respiratory fluctuations as abnormal. Therefore, a macroscopic perspective is also needed to determine whether data points have abnormal fluctuations. For example, even if a data point has local fluctuations, if such fluctuations are consistent with the overall respiratory and metabolic trends of livestock and poultry, then the data point is not considered "abnormal".

[0085] In this embodiment, for each data point, the instantaneous slope of each data point in the concentration time series is calculated by the central difference method; the overall slope is obtained by linear fitting of the concentration time series; and the absolute difference between the instantaneous slope and the overall slope is calculated as a global deviation index.

[0086] It should be noted that the central difference method is a common technique, which will not be described in detail in this embodiment. For example, for concentration time series... For internal data points within the range, considering boundary conditions, the effective calculation range for the index i when calculating the instantaneous slope is from i=2 to i=n−1. Therefore, the instantaneous slope... Expressed using the central difference method:

[0087]

[0088] in, The instantaneous slope of the i-th data point is represented by t; t represents the fixed sampling time interval. This represents the gas concentration value at the (i-1)th data point; This represents the gas concentration value at the (i+1)th data point.

[0089] It should be noted that the least squares linear regression algorithm can be used to fit a straight line to the entire concentration time series Q, and the slope of the fitted line is the overall slope.

[0090] It should be noted that the specific method of linear fitting is a well-known technique in the art, and will not be described in detail in this embodiment.

[0091] It is important to understand that the greater the instantaneous slope of a data point, the further the livestock's respiratory behavior deviates from its normal state, and the higher the likelihood that the data point is a significant anomaly. The overall slope is learned from the data of the livestock being tested, while the global deviation index focuses on relative deviation. If the absolute difference between the instantaneous slope and the overall slope is greater, the value of the global deviation index will also be greater. This is a strong abnormal signal, indicating that the livestock suddenly exhaled rapidly (for carbon dioxide gas), causing the gas concentration change rate at that instant to be much higher than its normal average level.

[0092] S101-4 Calculate the product of the local mutation index and the global deviation index, which serves as the comprehensive anomaly score for each data point.

[0093] It is important to understand that if the local mutation index of a certain data point is larger, it indicates that the livestock and poultry's breathing experienced a violent instantaneous tremor at the time the data point appeared. Furthermore, if the global deviation index of the data point is also larger, it indicates that the livestock and poultry's breathing deviated significantly from its normal state at the time the data point appeared. This is consistent with the physiological manifestations of acute stress (such as being startled or experiencing rapid panting).

[0094] In this embodiment, a comprehensive anomaly score sequence is constructed based on the comprehensive anomaly score of each data point in chronological order. Each data point in the comprehensive anomaly score sequence, excluding the first and last data points, is traversed. It is determined whether the comprehensive anomaly score of the currently traversed data point is simultaneously greater than the scores of the preceding and following data points. If so, the currently traversed data point is determined to be a local maximum. From the local maximum points, data points with comprehensive anomaly scores greater than a preset threshold are selected and marked as anomalies.

[0095] It should be noted that, for the first data point, it is determined whether its comprehensive anomaly score is greater than the score of the next data point; for the last data point, it is determined whether its comprehensive anomaly score is greater than the score of the previous data point; and for other data points besides the first and last data points...

[0096] It's important to understand that a stress-induced respiratory event (such as a shortness of breath) will produce a continuous abnormal fluctuation in the data, forming a "peak," rather than an isolated spike. The apex of this peak is a local maximum. Although a local maximum is a peak, it may contain some low-absolute peaks caused by measurement noise or minor physiological fluctuations. Therefore, to filter out these insignificant fluctuations and ensure that the marked outliers are sufficiently significant and severe, a threshold can be set.

[0097] It should be noted that the specific value of the preset threshold is dynamically determined based on the statistical characteristics of the comprehensive anomaly scoring sequence. For example, the threshold can be set as the sum of the average value (denoted by μ) of all data points in the comprehensive anomaly scoring sequence and three times the standard deviation (denoted by 3σ), i.e., threshold = μ + 3σ.

[0098] The stress analysis module 103 is used to obtain precursor and successor abnormal indicators for each abnormal point based on the comprehensive abnormal score and distribution trend of the abnormal points within the preceding and succeeding time windows; determine the relative time indicators of the abnormal points based on the gas concentration of the abnormal points and the gas concentration of the first and last data points of the concentration time series; and determine the correlation between each abnormal point and the initial stress state of livestock and poultry by combining the precursor abnormal indicators, successor abnormal indicators and relative time indicators.

[0099] It is important to understand that livestock and poultry experience the strongest stress when they first start wearing masks. As they gradually adapt, the stress response gradually diminishes. This stress response can be called initial stress. Therefore, a true anomaly caused by initial stress should exhibit the following data pattern: before this data point, anomalous events should be more frequent and intense, i.e., high precursor anomalous indicators; after this data point, anomalous events should become sparser and weaker, i.e., low subsequent anomalous indicators.

[0100] In this embodiment, the time period before the occurrence of an anomaly in the concentration time series is defined as the preceding time window of the anomaly, wherein the preceding time window begins at the start of the concentration time series and ends at the occurrence of the anomaly; the time period after the occurrence of the anomaly in the concentration time series is defined as the following time window of the anomaly, wherein the following time window begins at the occurrence of the current anomaly and ends at the end of the concentration time series.

[0101] The precursor time window contains all historical information from the initial stress caused by livestock wearing masks until the current anomalous point occurs. The precursor time window can be used to analyze how the stress state of livestock evolved before the occurrence of the anomalous point.

[0102] The subsequent time window contains all subsequent information from the occurrence of the current anomaly until the end of the test. The subsequent time window can be used to analyze whether the stress state of livestock and poultry continues, intensifies, or is alleviated after the occurrence of the anomaly.

[0103] To accurately obtain precursor and successor anomaly indicators, as an example, the time interval from the occurrence of each anomaly point to the occurrence of its next adjacent anomaly point is obtained; the arithmetic mean of the comprehensive anomaly scores of all anomalies within the time window is calculated as the anomaly intensity baseline value of the time window; the difference between the comprehensive anomaly score of each anomaly point within the time window and the anomaly intensity baseline is calculated as the anomaly metric value of each anomaly point; based on the sum of the absolute values ​​of the anomaly metrics, each anomaly metric value is normalized to obtain the relative prominence weight of each anomaly point; the relative prominence weight of each anomaly point is multiplied by the reciprocal of the time interval corresponding to each anomaly point to obtain the product result; and the sum of all product results within the time window is calculated as the anomaly indicator of the time window, where the anomaly indicator is either a precursor or successor anomaly indicator.

[0104] For example, suppose the outliers within the window are arranged in chronological order as [point P, point Q, point R]. Then calculate: the time interval for point P. ,in, This represents the time interval of outlier point P. Indicates the moment when outlier Q occurs; This indicates the time when outlier point P occurs; similarly, the time interval for point Q can be obtained. Since point R is the last point in the window, its time interval can be defined as the time until the end of the window, or it can be ignored in the calculation.

[0105] Anomaly intensity baseline is a benchmark used to quantify the average severity of all anomalous events within a specific time window (precursor or successor window). The anomaly intensity baseline represents the "average level" of anomaly intensity within that specific time window.

[0106] Anomaly metric measures the prominence of a single outlier relative to the average level within a window. If the anomaly metric of an outlier is greater than zero, it means that the anomaly intensity of that outlier is higher than the average level of its window.

[0107] It should be noted that normalizing each anomaly metric means dividing the anomaly metric of each anomaly point by the sum of the absolute values ​​of the anomaly metrics to obtain the relative prominence weight of each anomaly point.

[0108] It should be noted that the comprehensive abnormality score of the abnormal points comes from the respiratory gas concentration data of livestock and poultry. As long as the livestock and poultry are alive, there will always be differences in the intensity of abnormal events caused by stress. Therefore, under the background of respiratory pyrography, the intensity of all stress-induced abnormal respiratory events that occur within a certain period of time must be absolutely consistent and without any difference. This is physiologically impossible, so the sum of the absolute values ​​of the abnormality measurement values ​​cannot be zero.

[0109] It's important to understand that anomaly metrics measure the prominence and direction of a particular outlier relative to the average level within a given time window. The sign of an anomaly metric indicates whether it's "above average" (positive) or "below average" (negative). The sum of the absolute values ​​of anomaly metrics measures the overall prominence of all outliers within that time window. Therefore, if an anomaly metric for a particular outlier is positive and the larger its absolute value, it indicates that the outlier is significantly higher than the window's average level, making it a major contributor to defining that window as a "high-intensity anomaly period."

[0110] The time interval corresponding to an anomaly point refers to the time difference between the occurrence of an anomaly point itself and the occurrence of the next anomaly point immediately following it within the time window. The time difference between the two does not reflect how long the anomaly point lasts, but rather how long it takes from the occurrence of this anomaly event (i.e., the anomaly point) to the occurrence of the next anomaly event. It indicates the time interval between anomaly events. Therefore, the smaller the time interval corresponding to an anomaly point, the shorter the waiting time for the next anomaly, the more frequently the anomaly events occur, and the more likely it is to be a "high-intensity anomaly period".

[0111] It is important to understand that if, within a certain time window, the relative prominence weight of a certain anomaly is a larger positive number and the time interval corresponding to that anomaly is shorter, this indicates that after a very prominent anomaly, a new anomaly quickly occurs. This is a typical characteristic of "dense and high-intensity outbreaks," strongly indicating an active stress period.

[0112] It is important to understand that since the initial stress is strongest at the beginning of the test, the earlier an anomaly appears, the greater the likelihood that the anomaly is caused by the initial stress. Therefore, the characteristic that gas concentration is monotonically cumulative (e.g., CO2) or consumed (e.g., O2) can be used to cleverly use the changing trend of the gas concentration data itself to reflect the time process, so that the relative stage of an anomaly in the test can be determined without relying on absolute timestamps.

[0113] In this embodiment, the initial and final gas concentration values ​​of the concentration time series are obtained, as well as the gas concentration values ​​at the time of occurrence of each anomaly point; the absolute difference between the final gas concentration value and the gas concentration values ​​of each anomaly point is calculated as the first difference; the absolute difference between the initial and final gas concentration values ​​is calculated as the second difference; and the ratio of the first difference to the second difference is calculated as the relative time index of each anomaly point.

[0114] The initial gas concentration value represents the gas concentration value of the first data point in the concentration time series.

[0115] The final gas concentration value represents the gas concentration value of the last data point in the concentration time series.

[0116] The first difference refers to the absolute value of the difference between the gas concentration value at the end and the gas concentration value at the time of an anomaly. It reflects how much the gas concentration can change from the moment the anomaly occurs until the end of the entire data collection.

[0117] The second difference reflects the total change in gas concentration throughout the entire collection period, from start to finish.

[0118] It should be noted that the aforementioned calculation process (including the determination of the first difference, the second difference, and the relative time index) is based on an important premise: all calculations are performed independently for the concentration time series of the same gas.

[0119] It should be noted that if the second difference is zero, it means that the gas concentration at the end is exactly the same as the gas concentration at the beginning of the collection. However, in a successful and effective respiratory calorimetry test, due to the continuous metabolism and respiration of livestock and poultry, the gas concentration will inevitably undergo observable changes from the beginning to the end, so the second difference cannot be zero.

[0120] It should be noted that in a sealed mask, the concentration of carbon dioxide increases monotonically due to the continuous exhalation of livestock and poultry, while the concentration of oxygen decreases monotonically due to the continuous inhalation and consumption of livestock and poultry. Therefore, by using absolute values, the formulas for calculating the first and second differences can be applied to both CO2 (rising) and O2 (falling) gases.

[0121] It is important to understand that if the first difference of a certain anomaly is larger and closer to the second difference, it reflects that the remaining concentration change potential from that anomaly to the end of the collection is greater, thus indicating that the anomaly appeared in the very early stage; if the first difference of a certain anomaly is smaller, then the relative time index will be closer to zero, reflecting that the remaining concentration change potential from that anomaly to the end of the collection is smaller, thus indicating that the anomaly appeared near the end of the collection.

[0122] The process of determining the degree of correlation is as follows Figure 3 As shown, it includes:

[0123] S102-1, Calculate the difference between the precursor abnormality index and the successor abnormality index, and use it as the abnormality index difference value.

[0124] The difference in anomalous indicators was used to directly test the decay model of the initial stress response being "more severe at the beginning and milder at the end." If the difference in anomalous indicators at a certain anomaly point is greater than zero, it indicates that the preceding anomalous indicator is greater than the subsequent anomalous indicator, meaning that the intensity of the anomalous pattern before the anomaly point is greater than that after it. This is consistent with the initial stress decay model and is strong evidence that the anomaly point was caused by the initial stress. If the difference in anomalous indicators at a certain anomaly point is less than zero, it indicates that the preceding anomalous indicator is less than the subsequent anomalous indicator, meaning that the intensity of the anomalous pattern before the anomaly point is less than that after it. This violates the decay model and indicates that the anomaly point is likely caused by other factors (such as sudden interference during the test).

[0125] S102-2, calculate the product of the difference in abnormal indicators and the relative time indicators, as the correlation between each abnormal point and the initial stress state of livestock and poultry.

[0126] It's important to understand that a larger relative time index for an anomaly indicates that the anomaly occurred earlier. Since the initial stress is strongest at the start of data collection, the anomaly is more likely to be caused by this initial stress. Furthermore, a larger positive anomaly index difference for an anomaly indicates that the anomaly occurred early and the preceding and following patterns conform to typical stress decay characteristics. This is the primary target for subsequent corrections.

[0127] The correction module 104 is used to determine the correction weight based on the correlation degree and the comprehensive anomaly score; estimate the ideal gas concentration of each anomaly point based on the concentration time series after removing all anomalies; use the correction weight to perform weighted fusion of the gas concentrations collected at the anomalies and the ideal gas concentrations to obtain the corrected gas concentrations, and generate the corrected gas concentration time series.

[0128] In this embodiment, the absolute value of the correlation degree of all anomalies is calculated, and the maximum value is determined as the maximum correlation degree; the ratio of the correlation degree of each anomaly to the maximum correlation degree is calculated as the first ratio; for each anomaly, starting from the anomaly closest to the previous anomaly, the number of anomalies with continuously strictly increasing comprehensive anomaly scores is counted; the product of the first ratio and the comprehensive anomaly score is calculated as the median value; the median value is divided by the sum of the continuously increasing number of anomalies and 1 to obtain the quotient result, and the quotient result is normalized to obtain the corrected weight.

[0129] The first ratio is the ratio of the correlation of a certain outlier to the maximum correlation among all outliers. The value ranges from -1 to 1. If the value of an outlier is larger (i.e., closer to 1), it means that the correlation of the outlier is closer to the highest level in this collection (i.e., the outlier reflecting the most typical initial stress state). If the value of an outlier is larger (i.e., closer to -1), it means that the correlation of the outlier is closer to the lowest level in this collection (i.e., the outlier reflecting the most typical non-initial stress state).

[0130] It should be noted that since the "correction weight calculation" step has been entered, it means that the previous "anomaly identification" step has successfully screened out at least one anomaly (that is, there is at least one local maximum point in the comprehensive anomaly score sequence and its score exceeds the preset threshold), so the maximum correlation degree cannot be zero.

[0131] It should be noted that a continuously increasing comprehensive anomaly score means that several anomalies appearing consecutively in chronological order have increasingly higher comprehensive anomaly scores than the previous ones, and this increase must be uninterrupted and sequential. For example, suppose there are 5 anomalies in chronological order with the following comprehensive anomaly scores: E(2.0)→D(4.0)→C(5.5)→B(4.8)→A(6.0). For anomaly A, we trace back to anomaly B. Since 6.0 is greater than 4.8, the score increases, so the count is incremented by 1. Then we look at the anomaly C(5.5) preceding anomaly B: Is 5.5 less than 4.8? If yes, we stop tracing. The final conclusion is that for anomaly A, the number of continuously increasing anomalies is 1.

[0132] It's important to understand that a continuously increasing number of outliers measures which stage of a stress sequence a particular outlier is in. A larger number of continuously increasing outliers indicates that the outlier is at the end of a longer, increasingly stressful sequence, suggesting that the high outlier value is predictable and the data distortion at that outlier may be relatively low.

[0133] It is important to understand that if the first ratio of an outlier is a large positive number and the overall outlier score is also large, it means that the outlier is an object that needs to be corrected. If the first ratio of an outlier is negative (and is judged to be a non-initial stress point), no matter how large the overall outlier score is, the median value will be negative, which will lead to the final value being smaller, thus realizing the logic of not correcting it.

[0134] It should be noted that if the sum of the continuously increasing number of outliers and 1 for a certain outlier is very small (e.g., 1), it indicates that the outlier erupted suddenly (without prior context). Furthermore, if the median value of the outlier is a large positive number, the outlier is identified as a stressed and severe point, reflecting that the data for this outlier is extremely unreliable. Therefore, it needs to be assigned a very high correction weight and undergo significant correction. Conversely, if the sum of the continuously increasing number of outliers and 1 for a certain outlier is very large, even if the median value of the outlier is a large positive number, the quotient result will be lowered. This indicates that the outlier is also identified as a stressed and severe point, but it is at the end of a long-term, escalating sequence, meaning that the high outlier value of this outlier is "expected." The data for this outlier still contains some real trend information, therefore its correction weight should be suppressed, and its own value should be considered more during correction.

[0135] It should be noted that the min-max normalization method can be used to normalize the quotient result.

[0136] It should be noted that the ideal gas concentration can be estimated using a linear regression algorithm. For example, from the original concentration time series, all identified outliers are removed. Then, using these remaining "normal" data points, a curve is fitted using a linear regression algorithm (such as the least squares method) to obtain an optimal fitted line. This fitted line represents the overall trend of gas concentration change after excluding sudden stress interference. Finally, the time corresponding to each outlier is substituted into this fitted line, and the calculated gas concentration value is the ideal gas concentration of that outlier.

[0137] It is important to understand that if the correction weight of an outlier is larger, it means that the system has determined that the outlier is a highly unreliable outlier (such as a sudden stress point), and the correction process will mainly rely on the baseline model's estimate. If the correction weight of an outlier is smaller, it means that the system has determined that the outlier is basically reliable or not caused by stress, and the correction process will basically retain its original observation value, without correction or with only very minor adjustments. Therefore, the corrected gas concentration = correction weight × ideal gas concentration + (1 - correction weight) × collected gas concentration.

[0138] The evaluation module 105 is used to analyze the energy metabolism status of livestock and poultry based on the corrected gas concentration time series and output respiratory calorimetry results with accuracy verified.

[0139] It is important to understand that the corrected gas concentration time series is no longer the original, coarse data containing stress interference, but rather a corrected gas concentration time series that has undergone the "anomaly identification-attribution analysis-data reconstruction" process. The corrected gas concentration time series reflects the true gas exchange of livestock and poultry under stable conditions to the greatest extent. Therefore, the analysis results obtained through the corrected gas concentration time series are no longer the original gas concentration data that may mislead decision-making or the unverified metabolic results, but a reliable respiratory calorimetry analysis report with an "accuracy verified" statement.

[0140] It should be noted that analyzing the corrected gas concentration time series to output respiratory calorimetry results is a common data analysis method for those skilled in the art, and will not be elaborated upon in this embodiment.

[0141] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0142] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A system for acquiring and verifying the accuracy of respiratory thermometry data in livestock and poultry, characterized in that, The system includes: The data acquisition module is used to collect the concentration time series of respiratory gases from livestock and poultry. The identification module is used to determine the comprehensive anomaly score of each data point based on the temporal change characteristics of the data points in the concentration time series, and to record the data points whose scores meet the local anomaly conditions as anomaly points; The stress analysis module is used to obtain precursor and successor abnormality indicators for each abnormal point based on the comprehensive abnormality score and distribution trend of the abnormal points within the preceding and succeeding time windows; determine the relative time indicators of the abnormal points based on the gas concentration of the abnormal points and the gas concentration of the first and last data points of the concentration time series; and determine the correlation between each abnormal point and the initial stress state of livestock and poultry by combining the precursor abnormality indicators, successor abnormality indicators, and relative time indicators. The correction module is used to determine the correction weight based on the correlation degree and the comprehensive anomaly score; estimate the ideal gas concentration of each anomaly point based on the concentration time series after removing all anomalies; use the correction weight to perform weighted fusion of the gas concentrations collected at the anomalies and the ideal gas concentrations to obtain the corrected gas concentrations, and generate the corrected gas concentration time series. The assessment module is used to analyze the energy metabolism status of livestock and poultry based on the corrected gas concentration time series and output respiratory calorimetry results with accuracy verified.

2. The livestock and poultry respiratory calorimetry data acquisition and accuracy verification system according to claim 1, characterized in that, The comprehensive anomaly scoring determination process includes: Based on the concentration time series, the first difference of each data point is calculated and a first difference sequence is constructed. Based on the first-order difference sequence, determine the local mutation index of each data point; Based on the overall trend of the concentration time series, determine the global deviation index for each data point; The product of the local mutation index and the global deviation index is calculated as the comprehensive anomaly score for each data point.

3. The livestock and poultry respiratory calorimetry data acquisition and accuracy verification system according to claim 2, characterized in that, The determination of local mutation indices for each data point based on the first-order difference sequence includes: Calculate the interquartile range of a first-order difference sequence; For each data point, the absolute values ​​of the differences between the first-order difference of the data point and the first-order difference of the data point are calculated and used as the local mutation baseline value. The ratio of the local mutation baseline value to the interquartile range is calculated as an indicator of local mutation.

4. The livestock and poultry respiratory calorimetry data acquisition and accuracy verification system according to claim 2, characterized in that, The global deviation index for each data point, determined based on the overall trend of the concentration time series, includes: For each data point, the instantaneous slope of each data point in the concentration time series is calculated using the central difference method; The overall slope is obtained by linear fitting of the concentration time series. The absolute difference between the instantaneous slope and the overall slope is calculated and used as a global deviation index.

5. The livestock and poultry respiratory calorimetry data acquisition and accuracy verification system according to claim 1, characterized in that, The process of defining data points whose scores meet the local anomaly criteria as anomalies includes: Based on the comprehensive anomaly score of each data point, a comprehensive anomaly score sequence is constructed in chronological order. Traverse each data point in the comprehensive anomaly score sequence except for the first and last data points, and determine whether the comprehensive anomaly score of the currently traversed data point is simultaneously greater than the scores of the previous data point and the next data point; if so, determine that the currently traversed data point is a local maximum point. From the local maxima, data points with a comprehensive anomaly score greater than a preset threshold are selected and marked as anomalies.

6. The livestock and poultry respiratory calorimetry data acquisition and accuracy verification system according to claim 1, characterized in that, The system also includes: The time period preceding the occurrence of an outlier in the concentration time series is defined as the precursor time window of the outlier, which begins at the start of the concentration time series and ends at the occurrence of the outlier. The time period following the occurrence of an outlier in the concentration time series is defined as the subsequent time window of the outlier. The subsequent time window begins at the occurrence time of the current outlier and ends at the end of the concentration time series.

7. The livestock and poultry respiratory calorimetry data acquisition and accuracy verification system according to claim 6, characterized in that, For each anomaly, based on the comprehensive anomaly score and distribution trend of anomalies within its predecessor and successor time windows, precursor and successor anomaly indices are obtained, including: Obtain the time interval between the occurrence time of each anomaly and the occurrence time of its next adjacent anomaly; Calculate the arithmetic mean of the comprehensive anomaly scores of all anomalies within the time window, and use it as the baseline value of the anomaly intensity for the time window; The difference between the comprehensive anomaly score of each anomaly point within the calculation time window and the anomaly intensity baseline is used as the anomaly metric of each anomaly point. Based on the sum of the absolute values ​​of the anomaly metrics, each anomaly metric is normalized to obtain the relative prominence weight of each anomaly. The relative prominence weight of each outlier is multiplied by the reciprocal of the time interval corresponding to each outlier to obtain the product result; and the sum of all product results within the time window is calculated as the outlier index of the time window, where the outlier index is either a preceding outlier index or a successor outlier index.

8. The livestock and poultry respiratory calorimetry data acquisition and accuracy verification system according to claim 1, characterized in that, The process of determining the relative time index includes: Obtain the initial and final gas concentration values ​​of the concentration time series, as well as the gas concentration values ​​at the time of occurrence of each anomaly point; Calculate the absolute difference between the final gas concentration value and the gas concentration values ​​at each anomaly point, and use this as the first difference value; Calculate the absolute difference between the initial gas concentration value and the final gas concentration value, and use it as the second difference value; Calculate the ratio of the first difference to the second difference, and use it as a relative time indicator for each outlier.

9. The livestock and poultry respiratory calorimetry data acquisition and accuracy verification system according to claim 1, characterized in that, The correlation determination process includes: Calculate the difference between the precursor abnormality index and the successor abnormality index, and use it as the abnormality index difference. The product of the difference in abnormal indicators and the relative time indicators is calculated as the correlation between each abnormal point and the initial stress state of livestock and poultry.

10. The livestock and poultry respiratory calorimetry data acquisition and accuracy verification system according to claim 1, characterized in that, The process of determining the corrected weights includes: Calculate the absolute value of the correlation degree of all outliers and determine the maximum value as the maximum correlation degree; calculate the ratio of the correlation degree of each outlier to the maximum correlation degree, and use it as the first ratio. For each outlier, starting from the nearest preceding outlier, count the number of outliers whose comprehensive outlier scores increase continuously and strictly. Calculate the product of the first ratio and the comprehensive anomaly score, and use it as the median value; Divide the median value by the sum of the number of continuously increasing outliers and 1 to obtain the quotient, and then normalize the quotient to obtain the corrected weight.

Images