Animal husbandry abnormality early warning system based on the Internet of Things

Abstract

The present invention relates to the field of data processing technology, and in particular to an abnormal early warning system for animal husbandry based on the Internet of Things, including: obtaining the degree of deviation of an animal health data sequence at each acquisition moment in each local neighborhood sequence subset according to the dimensional data correlation between a stable average value and a local neighborhood sequence subset; obtaining the degree of noise of an animal health data sequence at each acquisition moment in each local neighborhood sequence subset; obtaining a corrected filtering value of an animal health data sequence at each acquisition moment according to the degree of noise; denoising a health data sequence set of each animal according to the corrected filtering value to obtain a denoised health data sequence set of each animal; obtaining abnormal data and performing abnormal early warning through the denoised health data sequence set of each animal. The present invention can improve the reliability of an abnormal early warning system for animal husbandry.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to an abnormal early warning system for animal husbandry based on the Internet of Things. Background Art

[0002] As an important part of agriculture, animal husbandry is directly related to food safety and economic stability. IoT technology obtains various information data in the animal husbandry process through sensors, and monitors and analyzes these data in real time. For example, temperature and humidity sensors can monitor changes in the breeding environment in real time, and biosensors can monitor the physiological parameters of animals. The real-time processing and analysis of these data is the core of the abnormal warning system, so it is necessary to ensure the accuracy of these data.

[0003] In the process of animal husbandry, the use of sensors to collect data may cause some noise in the collected data due to the accuracy of the sensors themselves and environmental changes. The traditional denoising method is non-local mean filtering, which uses Gaussian weighted Euclidean distance to measure the distance between neighbors, thereby further expressing the similarity between neighbors. Based on this, each data is given a weight, which can ensure that the influence of adjacent and very different data on each other is reduced when averaging within the window. However, in Gaussian weighting, the size of the weight depends on the distance between the data and other data, that is, the closer the data is to the data, the higher the weight will be, and the more attention should be paid to the calculation of the average, while the data farther away from the data will have a lower weight. This may make the weight of some noise data too high, resulting in excessive emphasis on the influence of noise points in weighted averaging, resulting in excessive smoothing of the data, which in turn affects the accuracy of analyzing abnormal data. Summary of the invention

[0004] The present invention provides an abnormal early warning system for animal husbandry based on the Internet of Things to solve the existing problems.

[0005] The livestock breeding abnormality early warning system based on the Internet of Things of the present invention adopts the following technical solutions:

[0006] Includes the following modules:

[0007] A data acquisition module is used to obtain a set of health data sequences of several kinds of animals in the livestock breeding process and a sequence of position coordinates of each animal;

[0008] The offset degree acquisition module is used to evenly divide the health data sequence set of each animal into several local neighborhood sequence subsets; obtain the stable average value of each local neighborhood sequence subset according to the difference of the animal health data sequence at different collection times in each local neighborhood sequence subset and the animal position coordinate sequence; obtain the dimensional data correlation of each local neighborhood sequence subset; obtain the offset degree of the animal health data sequence at each collection time in each local neighborhood sequence subset according to the dimensional data correlation of the stable average value and the local neighborhood sequence subset;

[0009] A data denoising module is used to obtain the noise level of the animal health data sequence at each collection moment in each local neighborhood sequence subset according to the correlation between the animal health data sequence in each local neighborhood sequence subset and its collection time, as well as the degree of offset; obtain the corrected filtering value of the animal health data sequence at each collection moment according to the noise level; denoise the health data sequence set of each animal according to the corrected filtering value to obtain the denoised health data sequence set of each animal;

[0010] The abnormal warning module is used to obtain abnormal data and issue abnormal warnings through the denoised health data sequence set of each animal.

[0011] Preferably, the method of evenly dividing the health data sequence set of each animal into a plurality of local neighborhood sequence subsets includes:

[0012] Preset a partition parameter For any animal health data sequence set in the livestock breeding process, the animal health data sequence set is evenly divided into a number of local neighborhood sequence subsets, each of which includes A sequence of health data of an animal at each collection moment.

[0013] Preferably, the method of obtaining the stable average value of each local neighborhood sequence subset according to the difference of the animal health data sequence at different collection times in each local neighborhood sequence subset and the animal position coordinate sequence includes the following specific methods:

[0014] For The first Animal health data series at the time of collection, according to The difference in the collection time between the animal health data series at the collection time and the animal health data series at other collection times is used to obtain the first Stable difference factor for animal health data series at each collection moment;

[0015] For any animal in the livestock breeding process, the said animal The position coordinates of the first collection time and the The Euclidean distance between the position coordinates of the collected moments is recorded as The motion displacement at the acquisition moment; The variance of the mean motion displacement of all animals in the local neighborhood sequence subset at all acquisition times is denoted as Animal behavior fluctuation values ​​of a subset of local neighborhood sequences;

[0016] The first The mean of the stable difference factors of the animal health data series at all acquisition moments in the local neighborhood sequence subset is The inverse of the animal behavior fluctuation value of the local neighborhood sequence subset is taken as the The stable average of a subset of local neighborhood sequences.

[0017] Preferably, the obtaining The specific method for calculating the stable difference factor of the animal health data series at each collection moment is:

[0018] The first The first The cumulative sum of the data values ​​of all dimensions in the animal health data sequence at the collection time is recorded as the first cumulative sum; The first The collection time and The first The reciprocal of the absolute value of the difference between the first and second acquisition moments is recorded as The time factor of the collection moment; The product of the cumulative sum of the time factors of all acquisition moments in the local neighborhood sequence subset and the first cumulative sum is taken as the first Stable difference factor for animal health data series at different collection moments.

[0019] Preferably, the obtaining of the dimensional data correlation of each local neighborhood sequence subset includes the following specific methods:

[0020] The first The first animal health data sequence of all collected moments in the local neighborhood sequence subset The sequence of dimensional data is the same as the The first animal health data sequence of all collected moments in the local neighborhood sequence subset The Pearson correlation coefficient between the sequences composed of dimensional data is recorded as The relevant factors of the dimension data; The normalized value of the cumulative sum of the correlation factors of all dimensional data in the animal health data sequence in the local neighborhood sequence subset is taken as the first Dimensional data correlation of a subset of local neighborhood sequences.

[0021] Preferably, the specific method for obtaining the degree of deviation of the animal health data sequence at each collection moment in each local neighborhood sequence subset according to the correlation between the stable average value and the dimensional data of the local neighborhood sequence subset is:

[0022] The first The first The cumulative sum of the data values ​​of all dimensions in the animal health data sequence at the collection time is equal to the The absolute value of the difference between the stable average values ​​of the local neighborhood sequence subsets is denoted as The first The offset factor of the animal health data series at the collection time; The inversely proportional normalized value of the dimensional data correlation of the local neighborhood sequence subset is recorded as the dimensional data correlation factor; The first The product of the offset factor of the animal health data series at the collection time and the dimensional data correlation factor is used as the The first The degree of deviation of the animal health data series at each collection moment.

[0023] Preferably, the method of obtaining the noise level of the animal health data sequence at each collection moment in each local neighborhood sequence subset according to the correlation between the animal health data sequence in each local neighborhood sequence subset and its collection time, and the degree of offset, includes the following specific methods:

[0024] Get the The acquisition time series of the local neighborhood sequence subset is the same as the The first subset of the local neighborhood sequence The correlation between the data series of each dimension;

[0025] The first The acquisition time series of the local neighborhood sequence subset is the same as the The mean of the correlations between all dimensional data sequences of the local neighborhood sequence subsets is taken as the The temporal correlation of a subset of local neighborhood sequences;

[0026] The first The first The degree of deviation of the animal health data series at the first collection moment is different from that at the The normalized value of the ratio of the temporal correlations between the subsets of local neighborhood sequences is taken as the The first The degree of noise in the animal health data series at each collection moment.

[0027] Preferably, the obtaining The acquisition time series of the local neighborhood sequence subset is the same as the The first subset of the local neighborhood sequence The correlation between the data series of each dimension includes the following specific methods:

[0028] The first The sequence of all acquisition moments in the local neighborhood sequence subset is taken as the first The acquisition time series of a subset of local neighborhood sequences;

[0029] The first The first animal health data sequence in all the collected moments in the local neighborhood sequence subset The sequence of data values ​​of dimensions is used as the The first subset of the local neighborhood sequence Dimensional data sequence;

[0030] The first The acquisition time series of the local neighborhood sequence subset is the same as the The first subset of the local neighborhood sequence The Pearson correlation coefficient between the data sequences of the dimensions is used as the The acquisition time series of the local neighborhood sequence subset is the same as the The first subset of the local neighborhood sequence The correlation between the data series of the dimensions.

[0031] Preferably, the method of obtaining the corrected filter value of the animal health data sequence at each acquisition moment according to the noise level includes:

[0032] A non-mean filtering algorithm is used to obtain a filtering value of the animal health data sequence at each acquisition moment in each local neighborhood sequence subset;

[0033] The first The first The filtered value of the animal health data series at the first collection moment is The first The ratio of the noise levels of the animal health data series at the first collection moment is taken as the The first Corrected filtered values ​​of the animal health data series at each collection moment.

[0034] Preferably, the method of obtaining abnormal data and performing abnormal warning through the denoised health data sequence set of each animal includes the following specific methods:

[0035] A three-dimensional space is constructed through body temperature data, heart rate data and respiratory rate data; for any animal in the livestock breeding process, the denoised health data sequence set of any animal is input into the three-dimensional space to obtain a number of data points, and a density clustering algorithm is performed on all data points, and outlier data points are recorded as abnormal data; if there is abnormal data, the animal corresponding to the abnormal data is marked, and an early warning prompt is issued.

[0036] The beneficial effects of the technical solution of the present invention are as follows: the present invention obtains the noise level of the animal health data sequence at each collection moment in each local neighborhood sequence subset according to the correlation between the animal health data sequence and its collection time in each local neighborhood sequence subset, as well as the degree of offset; obtains the corrected filter value of the animal health data sequence at each collection moment according to the noise level; denoises the health data sequence set of each animal according to the corrected filter value to obtain the denoised health data sequence set of each animal; obtains abnormal data and performs abnormal warning through the denoised health data sequence set of each animal; thereby denoises various types of animal health data obtained by the sensor in the animal husbandry process, obtains accurate animal health data in the animal husbandry process, thereby reducing the incidence of false alarms and improving the reliability of the animal husbandry abnormal warning system. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0038] Figure 1 This is a structural block diagram of the livestock breeding abnormality early warning system based on the Internet of Things of the present invention;

[0039] Figure 2 The present invention is a characteristic relationship flow chart of the livestock breeding abnormality early warning system based on the Internet of Things. DETAILED DESCRIPTION

[0040] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation method, structure, characteristics and effects of the livestock breeding abnormal warning system based on the Internet of Things proposed by the present invention are described in detail below in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0041] Unless defined otherwise, 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 belongs.

[0042] The specific scheme of the livestock breeding abnormality early warning system based on the Internet of Things provided by the present invention is described in detail below with reference to the accompanying drawings.

[0043] See also Figure 1 , which shows a structural block diagram of an abnormal early warning system for animal husbandry based on the Internet of Things provided by an embodiment of the present invention, the system includes the following modules:

[0044] The data acquisition module is used to obtain a set of health data sequences of several kinds of animals in the livestock breeding process and a sequence of position coordinates of each animal.

[0045] Specifically, the health data series of several kinds of animals in the livestock breeding process are collected. The specific process is as follows:

[0046] Take any animal in the livestock breeding process as an example:

[0047] Every 1 minute is a sampling moment, and the three dimensional data types of body temperature data, heart rate data, and respiratory rate data of this type of animal are collected in turn each time, for a total of 120 minutes; the three dimensional data of the body temperature data, heart rate data, and respiratory rate data of this type of animal at each sampling moment are used as the animal health data sequence at each sampling moment; the animal health data sequences at all sampling moments are constituted into a sequence set, which is used as the health data sequence set of this type of animal; wherein the body temperature data of this type of animal is measured by a body temperature sensor; the heart rate data of this type of animal is monitored by a heart rate sensor; the respiratory rate data of this type of animal is measured by a respiratory rate sensor, and the respiratory rate data, heart rate data, and body temperature data of the animal corresponding to the abnormal data are standardized by z-score.

[0048] A GPS positioning device is installed on each animal in the livestock breeding process to obtain the position coordinates of each animal at each collection time; the sequence consisting of the position coordinates of the animals at all collection times is recorded as the position coordinate sequence of each animal.

[0049] So far, the above method has been used to obtain a set of health data sequences of several kinds of animals in the livestock breeding process and a sequence of position coordinates of each animal.

[0050] The offset degree acquisition module is used to evenly divide the health data sequence set of each animal into several local neighborhood sequence subsets; obtain the stable average value of each local neighborhood sequence subset according to the difference of the animal health data sequence at different collection times in each local neighborhood sequence subset and the animal's position coordinate sequence; obtain the dimensional data correlation of each local neighborhood sequence subset; obtain the offset degree of the animal health data sequence at each collection time in each local neighborhood sequence subset according to the stable average value and the dimensional data correlation of the local neighborhood sequence subset.

[0051] It should be noted that the use of sensors to collect data during the livestock breeding process may cause some noise in the collected data due to the accuracy of the sensors themselves and environmental changes. Various types of animal health data often fluctuate and change over time. Since collection is a continuous time process, the data values ​​of each dimension in the animal health data sequence tend to be stable.

[0052] The sensor collects data such as body temperature, respiratory rate, heart rate, etc. of each animal in real time. Due to changes in ambient temperature and humidity, some data may have certain deviations. For example, on a hot afternoon, the body temperature readings of some animals may be high, but this does not mean that they have health problems, but because the ambient temperature affects the reading of the sensor; at the same time, when animals are engaged in intense activities, such as running or struggling, their physiological indicators such as body temperature and heart rate will fluctuate sharply; at this time, it is difficult to accurately reflect the health status of the animal based on the health data at a single moment; therefore, the "local neighborhood sequence subset" analysis method can effectively solve this noise problem; for example, if the health data of multiple adjacent animals show an increase in body temperature and an increase in respiratory rate within a certain period of time, this may mean data fluctuations caused by changes in environmental conditions, rather than health problems or noise problems of a single animal; therefore, by analyzing the local neighborhood sequence subset, if there is a large fluctuation in the local neighborhood sequence subset, there will be a large degree of deviation between the animal health data sequence at any collection time and the stable average value of the local neighborhood sequence subset to which it belongs.

[0053] Preferably, in some implementations of the embodiments of the present invention, the specific method of evenly dividing the health data sequence set of each animal into a plurality of local neighborhood sequence subsets is:

[0054] Preset a partition parameter , wherein this embodiment is based on This example is described as an example, and this embodiment is not specifically limited. Depends on the specific implementation situation;

[0055] For any animal health data sequence set in the livestock breeding process, the animal health data sequence set is evenly divided into a number of local neighborhood sequence subsets, each of which includes A sequence of health data of an animal at each collection moment.

[0056] Preferably, in some implementations of the embodiments of the present invention, the specific method for obtaining the stable average value of each local neighborhood sequence subset is as follows:

[0057] For The first Animal health data series at the time of collection, according to The difference in the collection time between the animal health data series at the collection time and the animal health data series at other collection times is used to obtain the first Stable difference factor for animal health data series at each collection moment;

[0058] For any animal in the livestock breeding process, the said animal The position coordinates of the first collection time and the The Euclidean distance between the position coordinates of the collected moments is recorded as The motion displacement at the acquisition moment; The variance of the mean motion displacement of all animals in the local neighborhood sequence subset at all acquisition times is denoted as Animal behavior fluctuation values ​​of a subset of local neighborhood sequences;

[0059] The first The mean of the stable difference factors of the animal health data series at all acquisition moments in the local neighborhood sequence subset is The inverse of the animal behavior fluctuation value of the local neighborhood sequence subset is taken as the The stable average of a subset of local neighborhood sequences;

[0060] Preferably, in some implementations of the embodiments of the present invention, obtaining the The specific method for calculating the stable difference factor of the animal health data series at each collection moment is:

[0061] The first The first The cumulative sum of the data values ​​of all dimensions in the animal health data sequence at the collection time is recorded as the first cumulative sum; The first The collection time and The first The reciprocal of the absolute value of the difference between the first and second acquisition moments is recorded as The time factor of the collection moment; The product of the cumulative sum of the time factors of all acquisition moments in the local neighborhood sequence subset and the first cumulative sum is taken as the first Stable difference factor for animal health data series at each collection moment;

[0062] The specific formula is:

[0063]

[0064] In the formula, Indicates Stable difference factor for animal health data series at each collection moment; Indicates The number of all acquisition moments in a subset of local neighborhood sequences; Indicates The first A collection moment; Indicates The first A collection moment; Indicates The first The number of all dimensional data types in the animal health data series at each collection moment; Indicates The first The first animal health data in the series of collection time The data value of each dimension; Indicates taking the absolute value.

[0065] It should be noted that in order to better show the impact of animal health data and make the stable average value of the local neighborhood sequence subset closer to the actual situation; by assigning a larger weight to the data with similar collection time, because of the data with smaller weight, even if the animal health data sequence deviates from the stable average value of the local neighborhood sequence subset, the impact on the final calculation result will be relatively small; the above operation can more accurately reflect the actual situation of the data values ​​in the animal health data sequence in the local neighborhood sequence subset.

[0066] Preferably, in some implementations of the present invention, when an animal has sudden physiological movements, its body temperature rises, and its respiratory rate and heart rate will change. Therefore, the multi-dimensional data are correlated. Therefore, it is necessary to analyze the correlation between the changes in the multi-dimensional data; then obtain the first The specific method for calculating the dimensional data correlation of a subset of local neighborhood sequences is:

[0067] The first The first animal health data sequence of all collected moments in the local neighborhood sequence subset The sequence of dimensional data is the same as the The first animal health data sequence of all collected moments in the local neighborhood sequence subset The Pearson correlation coefficient between the sequences composed of dimensional data is recorded as The relevant factors of the dimension data; The normalized value of the cumulative sum of the correlation factors of all dimensional data in the animal health data sequence in the local neighborhood sequence subset is taken as the first Dimensional data correlation of a subset of local neighborhood sequences;

[0068] The specific formula is:

[0069]

[0070] In the formula, Indicates Dimensional data correlation of a subset of local neighborhood sequences; Indicates The first animal health data sequence of all collected moments in the local neighborhood sequence subset A sequence of dimensional data; Indicates The first animal health data sequence of all collected moments in the local neighborhood sequence subset A sequence of dimensional data; Indicates The number of all dimensional data types in the animal health data sequence in the local neighborhood sequence subset; Indicates obtaining the Pearson correlation coefficient between two series.

[0071] It should be noted that when analyzing the health data of animals, we found that the body temperature of the animals continued to fluctuate in the past 10 minutes, with the highest temperature reaching 39.5°C and the respiratory rate reaching 50 times per minute. At this time, by comparing with the local neighborhood sequence subset, we found that other dimensional data of the animals during this period also had similar fluctuations. It is necessary to combine the correlation between other dimensional data to ultimately determine that this is just a normal physiological reaction caused by environmental fluctuations, rather than a disease warning. This method can effectively eliminate the influence of noise data, making the analysis results of animal health data more accurate, and can also detect and respond to sudden physiological abnormalities in a timely manner, without being misled by environmental fluctuations or deviations from short-term data.

[0072] Preferably, in some implementations of the embodiments of the present invention, according to the correlation between the stable average value and the dimensional data of the local neighborhood sequence subset, the specific method for obtaining the degree of deviation of the animal health data sequence at each collection moment in each local neighborhood sequence subset is:

[0073] The first The first The cumulative sum of the data values ​​of all dimensions in the animal health data sequence at the collection time is equal to the The absolute value of the difference between the stable average values ​​of the local neighborhood sequence subsets is denoted as The first The offset factor of the animal health data series at the collection time; The inversely proportional normalized value of the dimensional data correlation of the local neighborhood sequence subset is recorded as the dimensional data correlation factor; The first The product of the offset factor of the animal health data series at the collection time and the dimensional data correlation factor is used as the The first The degree of deviation of the animal health data series at each collection moment;

[0074] The specific formula is:

[0075]

[0076] In the formula, Indicates The first The degree of deviation of the animal health data series at each collection moment; Indicates The stable average of a subset of local neighborhood sequences; Indicates The first The number of all dimensional data types in the animal health data series at each collection moment; Indicates The first The first animal health data in the series of collection time The data value of each dimension; Indicates taking the absolute value; Indicates Dimensional data correlation of a subset of local neighborhood sequences; represents an exponential function with a natural constant as the base, and the embodiment adopts Model to present inverse proportional relationship and normalization processing, As the input of the model, the implementer can choose the inverse proportional function and the normalization function according to the actual situation.

[0077] So far, the deviation degree of the animal health data sequence at each collection moment in each local neighborhood sequence subset is obtained through the above method.

[0078] The data denoising module is used to obtain the noise level of the animal health data sequence at each collection moment in each local neighborhood sequence subset according to the correlation between the animal health data sequence in each local neighborhood sequence subset and its collection time, as well as the degree of offset; obtain the corrected filtering value of the animal health data sequence at each collection moment according to the noise level; denoise the health data sequence set of each animal according to the corrected filtering value to obtain the denoised health data sequence set of each animal.

[0079] It should be noted that if the animal health data values ​​in the local neighborhood sequence subset suddenly fluctuate, it may be caused by problems with the animal health, which will cause a large degree of deviation in the animal health data sequence. However, due to the influence of noise data, there will also be a large degree of deviation in the animal health data sequence. Therefore, if only the degree of deviation of the animal health data sequence is considered to make an abnormal judgment on the animal health data, misjudgment may occur, resulting in inaccurate detection results; in actual scenarios, due to external factors, the sudden fluctuation of animal health data values ​​​​will not be a change in the animal health data sequence at a single collection moment, but the change trend of the animal health data sequence at all collection moments in the local neighborhood sequence subset is consistent; if any animal health data sequence The greater the correlation between the local neighborhood sequence subset and its acquisition time series, the stronger the correlation is, the stronger the animal health data in the local neighborhood sequence subset is over time, which is consistent with the actual scenario, indicating that the animal health data sequence cannot be noise data; if the correlation between the local neighborhood sequence subset and its acquisition time series of any animal health data sequence is smaller, it indicates that there is an oscillation phenomenon in the change of animal health data in the local neighborhood sequence subset, indicating that the animal health data sequence may be noise data. Therefore, the noise degree of the animal health data sequence at each acquisition moment in each local neighborhood sequence subset can be obtained based on the correlation between the animal health data sequence at all acquisition moments in each local neighborhood sequence subset and its acquisition time.

[0080] Preferably, in some implementations of the embodiments of the present invention, the noise level of the animal health data sequence at each acquisition moment in each local neighborhood sequence subset is obtained according to the correlation between the animal health data sequence in each local neighborhood sequence subset and its acquisition time, as well as the offset degree, including the specific method of:

[0081] The first The sequence of all acquisition moments in the local neighborhood sequence subset is taken as the first The acquisition time series of a subset of local neighborhood sequences;

[0082] The first The animal health data sequence of all the collected moments in the local neighborhood sequence subset The sequence of data values ​​of dimensions is used as the The first subset of local neighborhood sequences Dimensional data sequence;

[0083] The first The acquisition time series of the local neighborhood sequence subset is the same as the The first subset of local neighborhood sequences The Pearson correlation coefficient between the data sequences of the dimensions is used as the The acquisition time series of the local neighborhood sequence subset is the same as the The first subset of the local neighborhood sequence The correlation between the data series of each dimension;

[0084] The first The acquisition time series of the local neighborhood sequence subset is the same as the The mean of the correlations between all dimensional data sequences of the local neighborhood sequence subsets is taken as the The temporal correlation of a subset of local neighborhood sequences;

[0085] The specific formula is:

[0086]

[0087] In the formula, Indicates The temporal correlation of a subset of local neighborhood sequences; Indicates The number of types of all dimensional data in the animal health data sequence at each collection moment in the local neighborhood sequence subset; Indicates The acquisition time series of the local neighborhood sequence subset is the same as the The first subset of the local neighborhood sequence Pearson correlation coefficient between the data series of each dimension; Indicates taking the absolute value.

[0088] The first The first The degree of deviation of the animal health data series at the first collection moment is different from that at the The normalized value of the ratio of the temporal correlations between the subsets of local neighborhood sequences is taken as the The first The noise level of the animal health data series at each collection moment;

[0089] The specific formula is:

[0090]

[0091] In the formula, Indicates The first The noise level of the animal health data series at each collection moment; Indicates The first The degree of deviation of the animal health data series at each collection moment; Indicates The temporal correlation of a subset of local neighborhood sequences; represents the linear normalization function.

[0092] Wherein, obtaining the Pearson correlation coefficient between two sequences is a prior art, and will not be described in detail in this embodiment.

[0093] Preferably, in some implementations of the embodiments of the present invention, the modified filtering value of the animal health data sequence at each acquisition moment is obtained according to the noise level, including the specific method of:

[0094] A non-mean filtering algorithm is used to obtain a filtering value of the animal health data sequence at each acquisition moment in each local neighborhood sequence subset;

[0095] The first The first The filtered value of the animal health data series at the first collection moment is The first The ratio of the noise levels of the animal health data series at the first collection moment is taken as the The first Corrected filtered values ​​of the animal health data series at each collection moment;

[0096] The specific formula is:

[0097]

[0098] In the formula, Indicates The first Corrected filtered values ​​of the animal health data series at each collection moment; Indicates The first The filtered value of the animal health data series at each collection moment; Indicates The first The degree of noise in the animal health data series at each collection moment.

[0099] The corrected filtering value of the animal health data sequence at each acquisition moment in each local neighborhood sequence subset is input into the non-mean filtering algorithm, and the health data sequence set of the animal is filtered and denoised to obtain the denoised health data sequence set of the animal.

[0100] Among them, the non-mean filtering algorithm is a prior art and will not be described in detail in this embodiment.

[0101] At this point, the denoised health data sequence set of each animal is obtained through the above method.

[0102] The abnormal warning module is used to obtain abnormal data and issue abnormal warnings through the denoised health data sequence set of each animal.

[0103] A three-dimensional space is constructed through body temperature data, heart rate data and respiratory rate data. For any animal in the livestock breeding process, the denoised healthy data sequence set of the animal corresponding to the abnormal data is input into the three-dimensional space to obtain several data points. A density clustering algorithm is performed on all data points, and outlier data points are recorded as abnormal data. If abnormal data exists, it means that there is an unhealthy existence in the animal, and an early warning is issued, and the animal is given timely medical treatment.

[0104] Among them, the density clustering algorithm is an existing technology, and this embodiment will not be described in detail here.

[0105] See also Figure 2 , which shows a characteristic relationship flow chart of an abnormal early warning system for animal husbandry based on the Internet of Things;

[0106] At this point, this embodiment is completed.

[0107] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. The abnormal early warning system for animal husbandry based on the Internet of Things is characterized by: The system includes the following modules: A data acquisition module is used to obtain a set of health data sequences of several kinds of animals in the livestock breeding process and a sequence of position coordinates of each animal; The offset degree acquisition module is used to evenly divide the health data sequence set of each animal into several local neighborhood sequence subsets; obtain the stable average value of each local neighborhood sequence subset according to the difference of the animal health data sequence at different collection times in each local neighborhood sequence subset and the animal position coordinate sequence; obtain the dimensional data correlation of each local neighborhood sequence subset; obtain the offset degree of the animal health data sequence at each collection time in each local neighborhood sequence subset according to the dimensional data correlation of the stable average value and the local neighborhood sequence subset; A data denoising module, for obtaining the noise level of the animal health data sequence at each collection moment in each local neighborhood sequence subset according to the correlation between the animal health data sequence in each local neighborhood sequence subset and its collection time, as well as the degree of offset; Obtaining a corrected filter value of the animal health data sequence at each acquisition moment according to the noise level; denoising the health data sequence set of each animal according to the corrected filter value to obtain a denoised health data sequence set of each animal; The abnormal warning module is used to obtain abnormal data and issue abnormal warnings through the denoised health data sequence set of each animal; The specific method for obtaining the degree of deviation of the animal health data sequence at each acquisition moment in each local neighborhood sequence subset is: The first The first The cumulative sum of the data values ​​of all dimensions in the animal health data sequence at the collection time is equal to the The absolute value of the difference between the stable average values ​​of the local neighborhood sequence subsets is denoted as The first The offset factor of the animal health data series at the collection time; The inversely proportional normalized value of the dimensional data correlation of the local neighborhood sequence subset is recorded as the dimensional data correlation factor; The first The product of the offset factor of the animal health data series at the collection time and the dimensional data correlation factor is used as the The first The degree of deviation of the animal health data series at each collection moment.

2. The abnormal early warning system for animal husbandry based on the Internet of Things according to claim 1 is characterized in that: The specific method of evenly dividing the health data sequence set of each animal into a plurality of local neighborhood sequence subsets includes: Preset a partition parameter For any animal health data sequence set in the livestock breeding process, the animal health data sequence set is evenly divided into a number of local neighborhood sequence subsets, each of which includes A sequence of health data of an animal at each collection moment.

3. The abnormal early warning system for animal husbandry based on the Internet of Things according to claim 1 is characterized in that: The method of obtaining the stable average value of each local neighborhood sequence subset according to the difference of the animal health data sequence at different collection times in each local neighborhood sequence subset and the animal position coordinate sequence includes the following specific methods: For The first Animal health data series at the time of collection, according to The difference in the collection time between the animal health data series at the collection time and the animal health data series at other collection times is used to obtain the first Stable difference factor for animal health data series at each collection moment; For any animal in the livestock breeding process, the said animal The position coordinates of the first collection time and the The Euclidean distance between the position coordinates of the collected moments is recorded as The motion displacement at the acquisition moment; The variance of the mean motion displacement of all animals in the local neighborhood sequence subset at all acquisition times is denoted as Animal behavior fluctuation values ​​of a subset of local neighborhood sequences; The first The mean of the stable difference factors of the animal health data series at all acquisition moments in the local neighborhood sequence subset is The inverse of the animal behavior fluctuation value of the local neighborhood sequence subset is taken as the The stable average of a subset of local neighborhood sequences.

4. The livestock breeding abnormality early warning system based on the Internet of Things according to claim 3 is characterized in that: The acquisition The specific method for calculating the stable difference factor of the animal health data series at each collection moment is: The first The first The cumulative sum of the data values ​​of all dimensions in the animal health data sequence at the collection time is recorded as the first cumulative sum; The first The collection time and The first The reciprocal of the absolute value of the difference between the first and second acquisition moments is recorded as The time factor of the collection moment; The product of the cumulative sum of the time factors of all acquisition moments in the local neighborhood sequence subset and the first cumulative sum is taken as the first Stable difference factor for animal health data series at different collection moments.

5. The livestock breeding abnormality early warning system based on the Internet of Things according to claim 1 is characterized in that: The specific method of obtaining the dimensional data correlation of each local neighborhood sequence subset includes: The first The first animal health data sequence of all collected moments in the local neighborhood sequence subset The sequence of dimensional data is the same as the The first animal health data sequence of all collected moments in the local neighborhood sequence subset The Pearson correlation coefficient between the sequences composed of dimensional data is recorded as The relevant factors of the dimension data; The normalized value of the cumulative sum of the correlation factors of all dimensional data in the animal health data sequence in the local neighborhood sequence subset is taken as the first Dimensional data correlation of a subset of local neighborhood sequences.

6. The abnormal early warning system for animal husbandry based on the Internet of Things according to claim 1 is characterized in that: The method of obtaining the noise level of the animal health data sequence at each collection moment in each local neighborhood sequence subset according to the correlation between the animal health data sequence in each local neighborhood sequence subset and its collection time, and the degree of offset, includes the following specific methods: Get the The acquisition time series of the local neighborhood sequence subset is The first subset of the local neighborhood sequence The correlation between the data series of each dimension; The first The acquisition time series of the local neighborhood sequence subset is the same as the The mean of the correlations between all dimensional data sequences of the local neighborhood sequence subsets is taken as the The temporal correlation of a subset of local neighborhood sequences; The first The first The degree of deviation of the animal health data series at the first collection moment is different from that at the The normalized value of the ratio of the temporal correlations between the subsets of local neighborhood sequences is taken as the The first The degree of noise in the animal health data series at each collection moment.

7. The abnormal early warning system for animal husbandry based on the Internet of Things according to claim 6 is characterized in that: The acquisition The acquisition time series of the local neighborhood sequence subset is The first subset of the local neighborhood sequence The correlation between the data series of each dimension includes the following specific methods: The first The sequence of all acquisition moments in the local neighborhood sequence subset is taken as the first The acquisition time series of a subset of local neighborhood sequences; The first The animal health data sequence of all the collected moments in the local neighborhood sequence subset The sequence of data values ​​of dimensions is used as the The first subset of the local neighborhood sequence Dimensional data sequence; The first The acquisition time series of the local neighborhood sequence subset is The first subset of the local neighborhood sequence The Pearson correlation coefficient between the data sequences of the dimensions is used as the The acquisition time series of the local neighborhood sequence subset is The first subset of the local neighborhood sequence The correlation between the data series of the dimensions.

8. The livestock breeding abnormality early warning system based on the Internet of Things according to claim 1 is characterized in that: The specific method of obtaining the corrected filter value of the animal health data sequence at each acquisition moment according to the noise level is as follows: A non-mean filtering algorithm is used to obtain a filtering value of the animal health data sequence at each acquisition moment in each local neighborhood sequence subset; The first The first The filtered value of the animal health data series at the first collection moment is The first The ratio of the noise levels of the animal health data series at the first collection moment is taken as the The first Corrected filtered values ​​of the animal health data series at each collection moment.

9. The livestock breeding abnormality early warning system based on the Internet of Things according to claim 1 is characterized in that: The specific method of obtaining abnormal data and performing abnormal warning through the denoised healthy data sequence set of each animal is as follows: A three-dimensional space is constructed through body temperature data, heart rate data and respiratory rate data; for any animal in the livestock breeding process, a denoised health data sequence set of any animal is input into the three-dimensional space to obtain a number of data points, a density clustering algorithm is performed on all the data points, and outlier data points are recorded as abnormal data; If there is abnormal data, the animal corresponding to the abnormal data will be marked and an early warning will be issued.

Images